diff --git a/.gitattributes b/.gitattributes index a6344aac8c09253b3b630fb776ae94478aa0275b..15908e8d6230d90dec2f561bbacce783eb625ed9 100644 --- a/.gitattributes +++ b/.gitattributes @@ -33,3 +33,8 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text *.zip filter=lfs diff=lfs merge=lfs -text *.zst filter=lfs diff=lfs merge=lfs -text *tfevents* filter=lfs diff=lfs merge=lfs -text +checkpoint-4711/tokenizer.json filter=lfs diff=lfs merge=lfs -text +checkpoint-5384/tokenizer.json filter=lfs diff=lfs merge=lfs -text +checkpoint-6057/tokenizer.json filter=lfs diff=lfs merge=lfs -text +checkpoint-6730/tokenizer.json filter=lfs diff=lfs merge=lfs -text +tokenizer.json filter=lfs diff=lfs merge=lfs -text diff --git a/README.md b/README.md new file mode 100644 index 0000000000000000000000000000000000000000..47184fb1491e794474b937328cce0d0158e3876b --- /dev/null +++ b/README.md @@ -0,0 +1,181 @@ +--- +license: apache-2.0 +library_name: peft +tags: +- axolotl +- generated_from_trainer +base_model: unsloth/gemma-7b +model-index: +- name: gemma_odia_7b_unsloth + results: [] +--- + + + +[Built with Axolotl](https://github.com/OpenAccess-AI-Collective/axolotl) +
See axolotl config + +axolotl version: `0.4.0` +```yaml +# use google/gemma-7b if you have access +base_model: unsloth/gemma-7b +model_type: AutoModelForCausalLM +tokenizer_type: AutoTokenizer + +load_in_8bit: false +load_in_4bit: true +strict: false + +# huggingface repo +datasets: + - path: OdiaGenAIdata/culturax-gemma-data + type: completion +val_set_size: 0.1 +output_dir: ./gemma-odia-7b-pretrain-unsloth +hub_model_id: sam2ai/gemma_odia_7b_unsloth + +adapter: qlora +lora_r: 32 +lora_alpha: 16 +lora_dropout: 0.05 +lora_target_linear: true + +sequence_len: 4096 +sample_packing: true +pad_to_sequence_len: true + +wandb_project: gemma-completion-7b-odia-unsloth +wandb_entity: +wandb_watch: +wandb_name: +wandb_log_model: + + +gradient_accumulation_steps: 8 +micro_batch_size: 2 +num_epochs: 10 +optimizer: adamw_bnb_8bit +lr_scheduler: cosine +learning_rate: 0.0002 + +train_on_inputs: false +group_by_length: false +bf16: auto +fp16: +tf32: false + +gradient_checkpointing: true +early_stopping_patience: +resume_from_checkpoint: +local_rank: +logging_steps: 1 +xformers_attention: +flash_attention: false + +warmup_ratio: 0.1 +evals_per_epoch: 4 +eval_table_size: +eval_max_new_tokens: 128 +saves_per_epoch: 1 +debug: +deepspeed: +weight_decay: 0.0 +fsdp: +fsdp_config: +special_tokens: + +``` + +

+ +# gemma_odia_7b_unsloth + +This model is a fine-tuned version of [unsloth/gemma-7b](https://huggingface.co/unsloth/gemma-7b) on the None dataset. +It achieves the following results on the evaluation set: +- Loss: 2.9914 + +## Model description + +More information needed + +## Intended uses & limitations + +More information needed + +## Training and evaluation data + +More information needed + +## Training procedure + +### Training hyperparameters + +The following hyperparameters were used during training: +- learning_rate: 0.0002 +- train_batch_size: 2 +- eval_batch_size: 2 +- seed: 42 +- distributed_type: multi-GPU +- num_devices: 8 +- gradient_accumulation_steps: 8 +- total_train_batch_size: 128 +- total_eval_batch_size: 16 +- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 +- lr_scheduler_type: cosine +- lr_scheduler_warmup_steps: 32 +- num_epochs: 10 + +### Training results + +| Training Loss | Epoch | Step | Validation Loss | +|:-------------:|:-----:|:----:|:---------------:| +| 39.5782 | 0.0 | 1 | 39.2579 | +| 7.2511 | 0.25 | 169 | 7.0771 | +| 4.2519 | 0.5 | 338 | 4.0654 | +| 3.7348 | 0.75 | 507 | 3.5937 | +| 3.4573 | 1.0 | 676 | 3.3126 | +| 3.4299 | 1.24 | 845 | 3.2429 | +| 3.4908 | 1.49 | 1014 | 3.2063 | +| 3.3588 | 1.74 | 1183 | 3.1614 | +| 3.3646 | 1.99 | 1352 | 3.1313 | +| 3.2672 | 2.23 | 1521 | 3.0885 | +| 3.2706 | 2.48 | 1690 | 3.0678 | +| 3.173 | 2.73 | 1859 | 3.0410 | +| 3.7319 | 2.98 | 2028 | 3.5392 | +| 3.3142 | 3.22 | 2197 | 3.1610 | +| 3.2931 | 3.47 | 2366 | 3.1339 | +| 3.3045 | 3.72 | 2535 | 3.0710 | +| 3.2423 | 3.97 | 2704 | 3.0920 | +| 3.2565 | 4.2 | 2873 | 3.0311 | +| 3.1167 | 4.45 | 3042 | 3.0039 | +| 3.1624 | 4.71 | 3211 | 3.0108 | +| 3.1697 | 4.96 | 3380 | 3.1008 | +| 3.1434 | 5.19 | 3549 | 2.9915 | +| 3.2301 | 5.44 | 3718 | 3.0033 | +| 3.1686 | 5.69 | 3887 | 2.9893 | +| 3.9959 | 5.95 | 4056 | 3.7561 | +| 3.3066 | 6.18 | 4225 | 3.1076 | +| 3.2567 | 6.43 | 4394 | 3.0679 | +| 3.1764 | 6.68 | 4563 | 3.0459 | +| 3.1848 | 6.93 | 4732 | 3.0342 | +| 3.181 | 7.17 | 4901 | 3.0279 | +| 3.1688 | 7.42 | 5070 | 3.0203 | +| 3.1474 | 7.67 | 5239 | 3.0131 | +| 3.1672 | 7.92 | 5408 | 3.0080 | +| 3.1202 | 8.16 | 5577 | 3.0036 | +| 3.1368 | 8.41 | 5746 | 2.9999 | +| 3.1104 | 8.66 | 5915 | 2.9968 | +| 3.1236 | 8.91 | 6084 | 2.9939 | +| 3.1055 | 9.15 | 6253 | 2.9924 | +| 3.1563 | 9.4 | 6422 | 2.9918 | +| 3.1373 | 9.65 | 6591 | 2.9914 | + + +### Framework versions + +- PEFT 0.9.0 +- Transformers 4.40.0.dev0 +- Pytorch 2.4.0.dev20240326+rocm6.0 +- Datasets 2.18.0 +- Tokenizers 0.15.0 \ No newline at end of file diff --git a/adapter_config.json b/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..35971161ab187cbdc8af6c44079edb2d2b677921 --- /dev/null +++ b/adapter_config.json @@ -0,0 +1,33 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "unsloth/gemma-7b", + "bias": "none", + "fan_in_fan_out": null, + "inference_mode": true, + "init_lora_weights": true, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 16, + "lora_dropout": 0.05, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 32, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "v_proj", + "o_proj", + "down_proj", + "k_proj", + "up_proj", + "gate_proj", + "q_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": false +} \ No newline at end of file diff --git a/adapter_model.bin b/adapter_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..3b590699cb71e09f1fc72dc85fc7303691f00539 --- /dev/null +++ b/adapter_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b1bb5f9724c8cf5fb21556cffa7ce4a52c38412c0ff2d816fe758a4d741efed +size 400173482 diff --git a/adapter_model.safetensors b/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1b9951526b63b0d03b0872d48fa5ba3f8fb9cecb --- /dev/null +++ b/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58ce96e81ec854cab49155c3e49332b00651e7bde0a5073948f61d7cbab7ffa4 +size 400084608 diff --git a/checkpoint-4711/README.md b/checkpoint-4711/README.md new file mode 100644 index 0000000000000000000000000000000000000000..dbeb1c7761226042e1d72de6378a4fa69fdb3f06 --- /dev/null +++ b/checkpoint-4711/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: unsloth/gemma-7b +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.9.0 \ No newline at end of file diff --git a/checkpoint-4711/adapter_config.json b/checkpoint-4711/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..35971161ab187cbdc8af6c44079edb2d2b677921 --- /dev/null +++ b/checkpoint-4711/adapter_config.json @@ -0,0 +1,33 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "unsloth/gemma-7b", + "bias": "none", + "fan_in_fan_out": null, + "inference_mode": true, + "init_lora_weights": true, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 16, + "lora_dropout": 0.05, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 32, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "v_proj", + "o_proj", + "down_proj", + "k_proj", + "up_proj", + "gate_proj", + "q_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": false +} \ No newline at end of file diff --git a/checkpoint-4711/adapter_model.safetensors b/checkpoint-4711/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d0c3014c45c4cf9fb7334f1a033cb938ae4789ce --- /dev/null +++ b/checkpoint-4711/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be64216487ca9032ad606e1c0f50243658668eae5e1fdb6c7fcdab9505ad37ea +size 400084608 diff --git a/checkpoint-4711/optimizer.pt b/checkpoint-4711/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..8c73914e14f363c79e9a47cf095e0bc24526f3b0 --- /dev/null +++ b/checkpoint-4711/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96c7632e04d7266fd49835f02e6b54c7ec26e5083b00abb6ab9b4449dfc45838 +size 200866804 diff --git a/checkpoint-4711/rng_state_0.pth b/checkpoint-4711/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..2ad6dc733bc509591ec1159cc78b8738c19dd0af --- /dev/null +++ b/checkpoint-4711/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ec2c09b704b8a15ac1eda5e2ce142d72111e33e7e0dd50ce78fc20b161a81b4b +size 15984 diff --git a/checkpoint-4711/rng_state_1.pth b/checkpoint-4711/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..d501c2333c829904d51a84e9305b915899fbe163 --- /dev/null +++ b/checkpoint-4711/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d3640867e9b9068557f1edbe5bd9b2e5aea92a422ee3b4b6eb24846b1e48c0f +size 15984 diff --git a/checkpoint-4711/rng_state_2.pth b/checkpoint-4711/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..4eb4fd53c636243c096c36e63dda9bdcfdf6955f --- /dev/null +++ b/checkpoint-4711/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:26ef31764269b32b57e6fde62c6001b1ece1d8d8182b91df6b3a1f73754bc068 +size 15984 diff --git a/checkpoint-4711/rng_state_3.pth b/checkpoint-4711/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..16fdd18b47e7e8237a5c803320926aa6a8345800 --- /dev/null +++ b/checkpoint-4711/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4557907a66006cf7a9737a737ea949a6af7c7a4833ac6a463b3ed37657b1807b +size 15984 diff --git a/checkpoint-4711/rng_state_4.pth b/checkpoint-4711/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..01b9d4367c5312e412bfe74768ddf9201f6e6620 --- /dev/null +++ b/checkpoint-4711/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1003136506a46e6e78d82ca2966c8c919e83162cbd4e1d56d56ee73fbd4d2315 +size 15984 diff --git a/checkpoint-4711/rng_state_5.pth b/checkpoint-4711/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..957189b2944b07a5ae32f96c9e54da4f62242cdc --- /dev/null +++ b/checkpoint-4711/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:215e47b5585f71ac12dcf21ae220e3825aefc1c49694eaff619726f9598ae935 +size 15984 diff --git a/checkpoint-4711/rng_state_6.pth b/checkpoint-4711/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..81049272fed854b386fbc650cd05ea02b84bd48d --- /dev/null +++ b/checkpoint-4711/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ffc9c082f173e0203c942c9abc65e52c433ab82cf7d6dc73c985145809aad45a +size 15984 diff --git a/checkpoint-4711/rng_state_7.pth b/checkpoint-4711/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..c3fc71f1fb38db6b0d349aa311cea77a4cd5faa6 --- /dev/null +++ b/checkpoint-4711/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b86154e1fa40a706e75ee4e87346c6433b2c2d96e59e155e4d20cb9153d78a4 +size 15984 diff --git a/checkpoint-4711/scheduler.pt b/checkpoint-4711/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..6f07d4082eb9a8e21d46910c47c5ce7de9195f5b --- /dev/null +++ b/checkpoint-4711/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b8939de54fdf80f47b7dbaa478b322d84f14e3317b1e5380c1f9c872f28f71a1 +size 1064 diff --git a/checkpoint-4711/special_tokens_map.json b/checkpoint-4711/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80 --- /dev/null +++ b/checkpoint-4711/special_tokens_map.json @@ -0,0 +1,34 @@ +{ + "additional_special_tokens": [ + "", + "" + ], + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-4711/tokenizer.json b/checkpoint-4711/tokenizer.json new file mode 100644 index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044 --- /dev/null +++ b/checkpoint-4711/tokenizer.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee +size 17477929 diff --git a/checkpoint-4711/tokenizer.model b/checkpoint-4711/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8 --- /dev/null +++ b/checkpoint-4711/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2 +size 4241003 diff --git a/checkpoint-4711/tokenizer_config.json b/checkpoint-4711/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..1451ca400a880c5056ab40cb960930cf827de2de --- /dev/null +++ b/checkpoint-4711/tokenizer_config.json @@ -0,0 +1,70 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "3": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "106": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "107": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "additional_special_tokens": [ + "", + "" + ], + "bos_token": "", + "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '' + role + '\n' + message['content'] | trim + '\n' }}{% endfor %}{% if add_generation_prompt %}{{'model\n'}}{% endif %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": null, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "tokenizer_class": "GemmaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-4711/trainer_state.json b/checkpoint-4711/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..c2d16728596a0c84a608b1d8263285cd063e1030 --- /dev/null +++ b/checkpoint-4711/trainer_state.json @@ -0,0 +1,33222 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 6.9029144236124, + "eval_steps": 169, + "global_step": 4711, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.10426733642816544, + "learning_rate": 6.25e-06, + "loss": 39.5782, + "step": 1 + }, + { + "epoch": 0.0, + "eval_loss": 39.257938385009766, + "eval_runtime": 1143.223, + "eval_samples_per_second": 14.994, + "eval_steps_per_second": 0.938, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 0.10457848757505417, + "learning_rate": 1.25e-05, + "loss": 39.5213, + "step": 2 + }, + { + "epoch": 0.0, + "grad_norm": 0.10474710911512375, + "learning_rate": 1.8750000000000002e-05, + "loss": 39.4605, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.11379425972700119, + "learning_rate": 2.5e-05, + "loss": 39.4968, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.12108173966407776, + "learning_rate": 3.125e-05, + "loss": 39.4831, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.13807319104671478, + "learning_rate": 3.7500000000000003e-05, + "loss": 39.5751, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.1619432121515274, + "learning_rate": 4.375e-05, + "loss": 39.4458, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.1936173141002655, + "learning_rate": 5e-05, + "loss": 39.4696, + "step": 8 + }, + { + "epoch": 0.01, + "grad_norm": 0.23368136584758759, + "learning_rate": 5.6250000000000005e-05, + "loss": 39.5398, + "step": 9 + }, + { + "epoch": 0.01, + "grad_norm": 0.28147873282432556, + "learning_rate": 6.25e-05, + "loss": 39.5004, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.3371683359146118, + "learning_rate": 6.875e-05, + "loss": 39.4975, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.4020785391330719, + "learning_rate": 7.500000000000001e-05, + "loss": 39.5237, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.4757504463195801, + "learning_rate": 8.125000000000001e-05, + "loss": 39.5889, + "step": 13 + }, + { + "epoch": 0.02, + "grad_norm": 0.5553041100502014, + "learning_rate": 8.75e-05, + "loss": 39.5117, + "step": 14 + }, + { + "epoch": 0.02, + "grad_norm": 0.6490126848220825, + "learning_rate": 9.375e-05, + "loss": 39.5925, + "step": 15 + }, + { + "epoch": 0.02, + "grad_norm": 0.7478038668632507, + "learning_rate": 0.0001, + "loss": 39.5638, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.8497499227523804, + "learning_rate": 0.00010625000000000001, + "loss": 39.513, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.9426423907279968, + "learning_rate": 0.00011250000000000001, + "loss": 39.1585, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.9930742979049683, + "learning_rate": 0.00011875, + "loss": 38.4983, + "step": 19 + }, + { + "epoch": 0.03, + "grad_norm": 0.9656581878662109, + "learning_rate": 0.000125, + "loss": 37.7409, + "step": 20 + }, + { + "epoch": 0.03, + "grad_norm": 1.0122301578521729, + "learning_rate": 0.00013125000000000002, + "loss": 37.248, + "step": 21 + }, + { + "epoch": 0.03, + "grad_norm": 1.0719304084777832, + "learning_rate": 0.0001375, + "loss": 37.0875, + "step": 22 + }, + { + "epoch": 0.03, + "grad_norm": 1.1606364250183105, + "learning_rate": 0.00014375, + "loss": 36.8509, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 1.2129034996032715, + "learning_rate": 0.00015000000000000001, + "loss": 36.5097, + "step": 24 + }, + { + "epoch": 0.04, + "grad_norm": 1.3106859922409058, + "learning_rate": 0.00015625, + "loss": 36.101, + "step": 25 + }, + { + "epoch": 0.04, + "grad_norm": 1.395408272743225, + "learning_rate": 0.00016250000000000002, + "loss": 35.5219, + "step": 26 + }, + { + "epoch": 0.04, + "grad_norm": 1.5144782066345215, + "learning_rate": 0.00016875, + "loss": 35.1062, + "step": 27 + }, + { + "epoch": 0.04, + "grad_norm": 1.6357661485671997, + "learning_rate": 0.000175, + "loss": 34.562, + "step": 28 + }, + { + "epoch": 0.04, + "grad_norm": 1.7252875566482544, + "learning_rate": 0.00018125000000000001, + "loss": 33.7745, + "step": 29 + }, + { + "epoch": 0.04, + "grad_norm": 1.8169721364974976, + "learning_rate": 0.0001875, + "loss": 33.1849, + "step": 30 + }, + { + "epoch": 0.05, + "grad_norm": 1.8421262502670288, + "learning_rate": 0.00019375000000000002, + "loss": 32.4654, + "step": 31 + }, + { + "epoch": 0.05, + "grad_norm": 1.9798814058303833, + "learning_rate": 0.0002, + "loss": 31.5573, + "step": 32 + }, + { + "epoch": 0.05, + "grad_norm": 2.1344172954559326, + "learning_rate": 0.0001999999890003351, + "loss": 30.9011, + "step": 33 + }, + { + "epoch": 0.05, + "grad_norm": 2.1522722244262695, + "learning_rate": 0.00019999995600134275, + "loss": 30.0969, + "step": 34 + }, + { + "epoch": 0.05, + "grad_norm": 1.9861739873886108, + "learning_rate": 0.0001999999010030303, + "loss": 29.2185, + "step": 35 + }, + { + "epoch": 0.05, + "grad_norm": 1.967444896697998, + "learning_rate": 0.00019999982400540977, + "loss": 28.5804, + "step": 36 + }, + { + "epoch": 0.05, + "grad_norm": 1.9654927253723145, + "learning_rate": 0.00019999972500849814, + "loss": 27.8538, + "step": 37 + }, + { + "epoch": 0.06, + "grad_norm": 1.9123767614364624, + "learning_rate": 0.00019999960401231715, + "loss": 27.1902, + "step": 38 + }, + { + "epoch": 0.06, + "grad_norm": 1.7705096006393433, + "learning_rate": 0.00019999946101689346, + "loss": 26.4983, + "step": 39 + }, + { + "epoch": 0.06, + "grad_norm": 1.6448909044265747, + "learning_rate": 0.00019999929602225854, + "loss": 25.8354, + "step": 40 + }, + { + "epoch": 0.06, + "grad_norm": 1.5614784955978394, + "learning_rate": 0.00019999910902844862, + "loss": 25.2961, + "step": 41 + }, + { + "epoch": 0.06, + "grad_norm": 1.5540868043899536, + "learning_rate": 0.0001999989000355049, + "loss": 24.8097, + "step": 42 + }, + { + "epoch": 0.06, + "grad_norm": 1.4986692667007446, + "learning_rate": 0.00019999866904347332, + "loss": 24.3735, + "step": 43 + }, + { + "epoch": 0.07, + "grad_norm": 1.4384242296218872, + "learning_rate": 0.00019999841605240475, + "loss": 23.8929, + "step": 44 + }, + { + "epoch": 0.07, + "grad_norm": 1.290089726448059, + "learning_rate": 0.00019999814106235478, + "loss": 23.4627, + "step": 45 + }, + { + "epoch": 0.07, + "grad_norm": 1.3345509767532349, + "learning_rate": 0.0001999978440733839, + "loss": 23.1114, + "step": 46 + }, + { + "epoch": 0.07, + "grad_norm": 1.5627790689468384, + "learning_rate": 0.00019999752508555752, + "loss": 22.8306, + "step": 47 + }, + { + "epoch": 0.07, + "grad_norm": 1.67068612575531, + "learning_rate": 0.00019999718409894577, + "loss": 22.5982, + "step": 48 + }, + { + "epoch": 0.07, + "grad_norm": 1.6429119110107422, + "learning_rate": 0.00019999682111362368, + "loss": 22.2669, + "step": 49 + }, + { + "epoch": 0.07, + "grad_norm": 1.5682146549224854, + "learning_rate": 0.00019999643612967106, + "loss": 21.9562, + "step": 50 + }, + { + "epoch": 0.08, + "grad_norm": 1.3983449935913086, + "learning_rate": 0.00019999602914717266, + "loss": 21.6858, + "step": 51 + }, + { + "epoch": 0.08, + "grad_norm": 1.3296594619750977, + "learning_rate": 0.000199995600166218, + "loss": 21.4453, + "step": 52 + }, + { + "epoch": 0.08, + "grad_norm": 1.2462741136550903, + "learning_rate": 0.00019999514918690142, + "loss": 21.1379, + "step": 53 + }, + { + "epoch": 0.08, + "grad_norm": 1.2661601305007935, + "learning_rate": 0.0001999946762093222, + "loss": 20.8178, + "step": 54 + }, + { + "epoch": 0.08, + "grad_norm": 1.5054320096969604, + "learning_rate": 0.00019999418123358432, + "loss": 20.5357, + "step": 55 + }, + { + "epoch": 0.08, + "grad_norm": 1.8500558137893677, + "learning_rate": 0.0001999936642597967, + "loss": 20.2271, + "step": 56 + }, + { + "epoch": 0.08, + "grad_norm": 2.102200508117676, + "learning_rate": 0.00019999312528807308, + "loss": 19.9845, + "step": 57 + }, + { + "epoch": 0.09, + "grad_norm": 2.3885788917541504, + "learning_rate": 0.000199992564318532, + "loss": 19.6821, + "step": 58 + }, + { + "epoch": 0.09, + "grad_norm": 1.3269203901290894, + "learning_rate": 0.00019999198135129689, + "loss": 19.4197, + "step": 59 + }, + { + "epoch": 0.09, + "grad_norm": 1.0758070945739746, + "learning_rate": 0.000199991376386496, + "loss": 19.0985, + "step": 60 + }, + { + "epoch": 0.09, + "grad_norm": 1.1309177875518799, + "learning_rate": 0.00019999074942426246, + "loss": 18.8138, + "step": 61 + }, + { + "epoch": 0.09, + "grad_norm": 1.230851173400879, + "learning_rate": 0.00019999010046473412, + "loss": 18.5005, + "step": 62 + }, + { + "epoch": 0.09, + "grad_norm": 1.2155430316925049, + "learning_rate": 0.00019998942950805377, + "loss": 18.1152, + "step": 63 + }, + { + "epoch": 0.1, + "grad_norm": 1.2195416688919067, + "learning_rate": 0.00019998873655436905, + "loss": 17.7053, + "step": 64 + }, + { + "epoch": 0.1, + "grad_norm": 1.2259167432785034, + "learning_rate": 0.0001999880216038324, + "loss": 17.3123, + "step": 65 + }, + { + "epoch": 0.1, + "grad_norm": 1.2353639602661133, + "learning_rate": 0.00019998728465660105, + "loss": 16.9048, + "step": 66 + }, + { + "epoch": 0.1, + "grad_norm": 1.3028600215911865, + "learning_rate": 0.00019998652571283715, + "loss": 16.477, + "step": 67 + }, + { + "epoch": 0.1, + "grad_norm": 1.3070034980773926, + "learning_rate": 0.00019998574477270772, + "loss": 16.0892, + "step": 68 + }, + { + "epoch": 0.1, + "grad_norm": 1.231437087059021, + "learning_rate": 0.00019998494183638446, + "loss": 15.6727, + "step": 69 + }, + { + "epoch": 0.1, + "grad_norm": 1.143466591835022, + "learning_rate": 0.00019998411690404412, + "loss": 15.3314, + "step": 70 + }, + { + "epoch": 0.11, + "grad_norm": 1.4420065879821777, + "learning_rate": 0.0001999832699758681, + "loss": 15.1217, + "step": 71 + }, + { + "epoch": 0.11, + "grad_norm": 1.9184247255325317, + "learning_rate": 0.00019998240105204276, + "loss": 15.0641, + "step": 72 + }, + { + "epoch": 0.11, + "grad_norm": 2.11586332321167, + "learning_rate": 0.0001999815101327592, + "loss": 14.9107, + "step": 73 + }, + { + "epoch": 0.11, + "grad_norm": 2.1635212898254395, + "learning_rate": 0.0001999805972182135, + "loss": 14.6949, + "step": 74 + }, + { + "epoch": 0.11, + "grad_norm": 1.9647817611694336, + "learning_rate": 0.00019997966230860643, + "loss": 14.3436, + "step": 75 + }, + { + "epoch": 0.11, + "grad_norm": 1.6679894924163818, + "learning_rate": 0.00019997870540414367, + "loss": 13.917, + "step": 76 + }, + { + "epoch": 0.11, + "grad_norm": 1.407972812652588, + "learning_rate": 0.0001999777265050358, + "loss": 13.6019, + "step": 77 + }, + { + "epoch": 0.12, + "grad_norm": 1.442781686782837, + "learning_rate": 0.00019997672561149806, + "loss": 13.3002, + "step": 78 + }, + { + "epoch": 0.12, + "grad_norm": 1.4343981742858887, + "learning_rate": 0.00019997570272375074, + "loss": 13.111, + "step": 79 + }, + { + "epoch": 0.12, + "grad_norm": 1.4623301029205322, + "learning_rate": 0.00019997465784201882, + "loss": 12.9296, + "step": 80 + }, + { + "epoch": 0.12, + "grad_norm": 1.2028108835220337, + "learning_rate": 0.00019997359096653216, + "loss": 12.7946, + "step": 81 + }, + { + "epoch": 0.12, + "grad_norm": 0.7906871438026428, + "learning_rate": 0.0001999725020975255, + "loss": 12.6467, + "step": 82 + }, + { + "epoch": 0.12, + "grad_norm": 0.8495603799819946, + "learning_rate": 0.0001999713912352384, + "loss": 12.5962, + "step": 83 + }, + { + "epoch": 0.12, + "grad_norm": 1.952924370765686, + "learning_rate": 0.00019997025837991514, + "loss": 12.4475, + "step": 84 + }, + { + "epoch": 0.13, + "grad_norm": 2.176668405532837, + "learning_rate": 0.00019996910353180502, + "loss": 12.3597, + "step": 85 + }, + { + "epoch": 0.13, + "grad_norm": 1.4191577434539795, + "learning_rate": 0.0001999679266911621, + "loss": 12.2764, + "step": 86 + }, + { + "epoch": 0.13, + "grad_norm": 1.0766239166259766, + "learning_rate": 0.00019996672785824526, + "loss": 12.2163, + "step": 87 + }, + { + "epoch": 0.13, + "grad_norm": 0.8922137022018433, + "learning_rate": 0.00019996550703331825, + "loss": 12.004, + "step": 88 + }, + { + "epoch": 0.13, + "grad_norm": 0.8033492565155029, + "learning_rate": 0.00019996426421664963, + "loss": 11.8298, + "step": 89 + }, + { + "epoch": 0.13, + "grad_norm": 0.8409807682037354, + "learning_rate": 0.00019996299940851277, + "loss": 11.7005, + "step": 90 + }, + { + "epoch": 0.14, + "grad_norm": 1.0452065467834473, + "learning_rate": 0.000199961712609186, + "loss": 11.5059, + "step": 91 + }, + { + "epoch": 0.14, + "grad_norm": 1.1938868761062622, + "learning_rate": 0.00019996040381895237, + "loss": 11.3725, + "step": 92 + }, + { + "epoch": 0.14, + "grad_norm": 1.1088273525238037, + "learning_rate": 0.0001999590730380998, + "loss": 11.1496, + "step": 93 + }, + { + "epoch": 0.14, + "grad_norm": 0.9627100229263306, + "learning_rate": 0.00019995772026692107, + "loss": 10.9487, + "step": 94 + }, + { + "epoch": 0.14, + "grad_norm": 0.9412399530410767, + "learning_rate": 0.00019995634550571377, + "loss": 10.7797, + "step": 95 + }, + { + "epoch": 0.14, + "grad_norm": 0.962394654750824, + "learning_rate": 0.00019995494875478032, + "loss": 10.5443, + "step": 96 + }, + { + "epoch": 0.14, + "grad_norm": 0.7746237516403198, + "learning_rate": 0.000199953530014428, + "loss": 10.3263, + "step": 97 + }, + { + "epoch": 0.15, + "grad_norm": 0.821438729763031, + "learning_rate": 0.00019995208928496897, + "loss": 10.1812, + "step": 98 + }, + { + "epoch": 0.15, + "grad_norm": 1.4739282131195068, + "learning_rate": 0.00019995062656672012, + "loss": 9.9212, + "step": 99 + }, + { + "epoch": 0.15, + "grad_norm": 0.9201889038085938, + "learning_rate": 0.00019994914186000328, + "loss": 9.8514, + "step": 100 + }, + { + "epoch": 0.15, + "grad_norm": 0.909538745880127, + "learning_rate": 0.00019994763516514505, + "loss": 9.7503, + "step": 101 + }, + { + "epoch": 0.15, + "grad_norm": 0.8758969902992249, + "learning_rate": 0.0001999461064824769, + "loss": 9.6219, + "step": 102 + }, + { + "epoch": 0.15, + "grad_norm": 0.9529918432235718, + "learning_rate": 0.00019994455581233514, + "loss": 9.5321, + "step": 103 + }, + { + "epoch": 0.15, + "grad_norm": 1.0058927536010742, + "learning_rate": 0.0001999429831550609, + "loss": 9.3729, + "step": 104 + }, + { + "epoch": 0.16, + "grad_norm": 0.9739552140235901, + "learning_rate": 0.00019994138851100015, + "loss": 9.3018, + "step": 105 + }, + { + "epoch": 0.16, + "grad_norm": 0.8659080266952515, + "learning_rate": 0.0001999397718805037, + "loss": 9.1481, + "step": 106 + }, + { + "epoch": 0.16, + "grad_norm": 0.9334455132484436, + "learning_rate": 0.00019993813326392722, + "loss": 9.0994, + "step": 107 + }, + { + "epoch": 0.16, + "grad_norm": 0.9508799910545349, + "learning_rate": 0.00019993647266163117, + "loss": 9.0184, + "step": 108 + }, + { + "epoch": 0.16, + "grad_norm": 0.8480530381202698, + "learning_rate": 0.00019993479007398088, + "loss": 8.8884, + "step": 109 + }, + { + "epoch": 0.16, + "grad_norm": 0.7846106290817261, + "learning_rate": 0.0001999330855013465, + "loss": 8.7288, + "step": 110 + }, + { + "epoch": 0.16, + "grad_norm": 0.7611198425292969, + "learning_rate": 0.00019993135894410304, + "loss": 8.7147, + "step": 111 + }, + { + "epoch": 0.17, + "grad_norm": 0.7758539915084839, + "learning_rate": 0.00019992961040263033, + "loss": 8.7001, + "step": 112 + }, + { + "epoch": 0.17, + "grad_norm": 0.6993403434753418, + "learning_rate": 0.000199927839877313, + "loss": 8.5369, + "step": 113 + }, + { + "epoch": 0.17, + "grad_norm": 0.6209515929222107, + "learning_rate": 0.0001999260473685406, + "loss": 8.4196, + "step": 114 + }, + { + "epoch": 0.17, + "grad_norm": 0.6134201884269714, + "learning_rate": 0.00019992423287670748, + "loss": 8.3379, + "step": 115 + }, + { + "epoch": 0.17, + "grad_norm": 0.6407685875892639, + "learning_rate": 0.00019992239640221276, + "loss": 8.2981, + "step": 116 + }, + { + "epoch": 0.17, + "grad_norm": 0.5949248671531677, + "learning_rate": 0.00019992053794546047, + "loss": 8.266, + "step": 117 + }, + { + "epoch": 0.18, + "grad_norm": 0.5390576720237732, + "learning_rate": 0.0001999186575068595, + "loss": 8.2384, + "step": 118 + }, + { + "epoch": 0.18, + "grad_norm": 0.5707695484161377, + "learning_rate": 0.00019991675508682349, + "loss": 8.1337, + "step": 119 + }, + { + "epoch": 0.18, + "grad_norm": 0.6072248816490173, + "learning_rate": 0.00019991483068577097, + "loss": 8.1228, + "step": 120 + }, + { + "epoch": 0.18, + "grad_norm": 0.5436020493507385, + "learning_rate": 0.00019991288430412532, + "loss": 8.0083, + "step": 121 + }, + { + "epoch": 0.18, + "grad_norm": 0.43990185856819153, + "learning_rate": 0.00019991091594231464, + "loss": 7.943, + "step": 122 + }, + { + "epoch": 0.18, + "grad_norm": 0.7439638376235962, + "learning_rate": 0.0001999089256007721, + "loss": 7.9075, + "step": 123 + }, + { + "epoch": 0.18, + "grad_norm": 0.5559173822402954, + "learning_rate": 0.00019990691327993548, + "loss": 7.8031, + "step": 124 + }, + { + "epoch": 0.19, + "grad_norm": 0.5491620302200317, + "learning_rate": 0.0001999048789802475, + "loss": 7.7261, + "step": 125 + }, + { + "epoch": 0.19, + "grad_norm": 0.7270621657371521, + "learning_rate": 0.00019990282270215563, + "loss": 7.8013, + "step": 126 + }, + { + "epoch": 0.19, + "grad_norm": 0.5974888205528259, + "learning_rate": 0.0001999007444461123, + "loss": 7.6982, + "step": 127 + }, + { + "epoch": 0.19, + "grad_norm": 0.4842529892921448, + "learning_rate": 0.00019989864421257467, + "loss": 7.6482, + "step": 128 + }, + { + "epoch": 0.19, + "grad_norm": 0.6567961573600769, + "learning_rate": 0.00019989652200200485, + "loss": 7.6721, + "step": 129 + }, + { + "epoch": 0.19, + "grad_norm": 0.5840482711791992, + "learning_rate": 0.00019989437781486963, + "loss": 7.6075, + "step": 130 + }, + { + "epoch": 0.19, + "grad_norm": 0.41609910130500793, + "learning_rate": 0.0001998922116516408, + "loss": 7.5145, + "step": 131 + }, + { + "epoch": 0.2, + "grad_norm": 0.4465920329093933, + "learning_rate": 0.00019989002351279483, + "loss": 7.5013, + "step": 132 + }, + { + "epoch": 0.2, + "grad_norm": 0.6557760238647461, + "learning_rate": 0.0001998878133988131, + "loss": 7.5361, + "step": 133 + }, + { + "epoch": 0.2, + "grad_norm": 0.4710114598274231, + "learning_rate": 0.00019988558131018186, + "loss": 7.4989, + "step": 134 + }, + { + "epoch": 0.2, + "grad_norm": 0.67153000831604, + "learning_rate": 0.00019988332724739215, + "loss": 7.6135, + "step": 135 + }, + { + "epoch": 0.2, + "grad_norm": 1.6115596294403076, + "learning_rate": 0.00019988105121093982, + "loss": 7.8275, + "step": 136 + }, + { + "epoch": 0.2, + "grad_norm": 0.7458434700965881, + "learning_rate": 0.0001998787532013256, + "loss": 7.9861, + "step": 137 + }, + { + "epoch": 0.2, + "grad_norm": 0.6269705891609192, + "learning_rate": 0.000199876433219055, + "loss": 8.0249, + "step": 138 + }, + { + "epoch": 0.21, + "grad_norm": 0.7430456876754761, + "learning_rate": 0.00019987409126463846, + "loss": 8.0029, + "step": 139 + }, + { + "epoch": 0.21, + "grad_norm": 1.2338321208953857, + "learning_rate": 0.0001998717273385912, + "loss": 7.9936, + "step": 140 + }, + { + "epoch": 0.21, + "grad_norm": 0.7371585965156555, + "learning_rate": 0.00019986934144143316, + "loss": 7.8833, + "step": 141 + }, + { + "epoch": 0.21, + "grad_norm": 0.6107997894287109, + "learning_rate": 0.00019986693357368934, + "loss": 7.7912, + "step": 142 + }, + { + "epoch": 0.21, + "grad_norm": 1.231048583984375, + "learning_rate": 0.00019986450373588946, + "loss": 7.7342, + "step": 143 + }, + { + "epoch": 0.21, + "grad_norm": 0.6497113704681396, + "learning_rate": 0.00019986205192856795, + "loss": 7.7583, + "step": 144 + }, + { + "epoch": 0.22, + "grad_norm": 1.1496895551681519, + "learning_rate": 0.0001998595781522643, + "loss": 7.7372, + "step": 145 + }, + { + "epoch": 0.22, + "grad_norm": 0.949022114276886, + "learning_rate": 0.00019985708240752268, + "loss": 7.7063, + "step": 146 + }, + { + "epoch": 0.22, + "grad_norm": 3.4438350200653076, + "learning_rate": 0.00019985456469489213, + "loss": 7.6564, + "step": 147 + }, + { + "epoch": 0.22, + "grad_norm": 1.3692785501480103, + "learning_rate": 0.00019985202501492656, + "loss": 7.7306, + "step": 148 + }, + { + "epoch": 0.22, + "grad_norm": 1.341970443725586, + "learning_rate": 0.00019984946336818463, + "loss": 7.6558, + "step": 149 + }, + { + "epoch": 0.22, + "grad_norm": 0.8111562728881836, + "learning_rate": 0.00019984687975522993, + "loss": 7.5931, + "step": 150 + }, + { + "epoch": 0.22, + "grad_norm": 1.028735876083374, + "learning_rate": 0.00019984427417663085, + "loss": 7.5077, + "step": 151 + }, + { + "epoch": 0.23, + "grad_norm": 4.875123023986816, + "learning_rate": 0.00019984164663296058, + "loss": 7.4798, + "step": 152 + }, + { + "epoch": 0.23, + "grad_norm": 1.2443698644638062, + "learning_rate": 0.00019983899712479715, + "loss": 7.5151, + "step": 153 + }, + { + "epoch": 0.23, + "grad_norm": 38.03964614868164, + "learning_rate": 0.00019983632565272346, + "loss": 7.3952, + "step": 154 + }, + { + "epoch": 0.23, + "grad_norm": 2.3863110542297363, + "learning_rate": 0.00019983363221732716, + "loss": 7.37, + "step": 155 + }, + { + "epoch": 0.23, + "grad_norm": 1.0871200561523438, + "learning_rate": 0.00019983091681920084, + "loss": 7.4005, + "step": 156 + }, + { + "epoch": 0.23, + "grad_norm": 1.156063437461853, + "learning_rate": 0.0001998281794589419, + "loss": 7.3377, + "step": 157 + }, + { + "epoch": 0.23, + "grad_norm": 3.350292682647705, + "learning_rate": 0.00019982542013715244, + "loss": 7.0855, + "step": 158 + }, + { + "epoch": 0.24, + "grad_norm": 4.875136375427246, + "learning_rate": 0.00019982263885443956, + "loss": 7.0308, + "step": 159 + }, + { + "epoch": 0.24, + "grad_norm": 4.6525068283081055, + "learning_rate": 0.00019981983561141516, + "loss": 7.0757, + "step": 160 + }, + { + "epoch": 0.24, + "grad_norm": 2.9209208488464355, + "learning_rate": 0.00019981701040869585, + "loss": 7.1687, + "step": 161 + }, + { + "epoch": 0.24, + "grad_norm": 2.691115379333496, + "learning_rate": 0.00019981416324690317, + "loss": 7.2762, + "step": 162 + }, + { + "epoch": 0.24, + "grad_norm": 240.15269470214844, + "learning_rate": 0.00019981129412666358, + "loss": 7.3205, + "step": 163 + }, + { + "epoch": 0.24, + "grad_norm": 5.614213466644287, + "learning_rate": 0.00019980840304860813, + "loss": 7.2591, + "step": 164 + }, + { + "epoch": 0.25, + "grad_norm": 6.848481178283691, + "learning_rate": 0.0001998054900133729, + "loss": 7.3047, + "step": 165 + }, + { + "epoch": 0.25, + "grad_norm": 1.3473995923995972, + "learning_rate": 0.00019980255502159873, + "loss": 7.314, + "step": 166 + }, + { + "epoch": 0.25, + "grad_norm": 1.2133899927139282, + "learning_rate": 0.00019979959807393132, + "loss": 7.2487, + "step": 167 + }, + { + "epoch": 0.25, + "grad_norm": 46.43593215942383, + "learning_rate": 0.00019979661917102115, + "loss": 7.2762, + "step": 168 + }, + { + "epoch": 0.25, + "grad_norm": 1.915299892425537, + "learning_rate": 0.00019979361831352357, + "loss": 7.2511, + "step": 169 + }, + { + "epoch": 0.25, + "eval_loss": 7.077121257781982, + "eval_runtime": 1142.7781, + "eval_samples_per_second": 15.0, + "eval_steps_per_second": 0.938, + "step": 169 + }, + { + "epoch": 0.25, + "grad_norm": 2.1094720363616943, + "learning_rate": 0.00019979059550209876, + "loss": 7.2082, + "step": 170 + }, + { + "epoch": 0.25, + "grad_norm": 1.6243547201156616, + "learning_rate": 0.0001997875507374117, + "loss": 7.2193, + "step": 171 + }, + { + "epoch": 0.26, + "grad_norm": 1.493200659751892, + "learning_rate": 0.0001997844840201322, + "loss": 7.2206, + "step": 172 + }, + { + "epoch": 0.26, + "grad_norm": 5.889941215515137, + "learning_rate": 0.00019978139535093496, + "loss": 7.2216, + "step": 173 + }, + { + "epoch": 0.26, + "grad_norm": 16.500181198120117, + "learning_rate": 0.00019977828473049947, + "loss": 7.2762, + "step": 174 + }, + { + "epoch": 0.26, + "grad_norm": 1.3627996444702148, + "learning_rate": 0.00019977515215951, + "loss": 7.2056, + "step": 175 + }, + { + "epoch": 0.26, + "grad_norm": 3.392564535140991, + "learning_rate": 0.0001997719976386557, + "loss": 7.2832, + "step": 176 + }, + { + "epoch": 0.26, + "grad_norm": 4.66141939163208, + "learning_rate": 0.00019976882116863061, + "loss": 7.2547, + "step": 177 + }, + { + "epoch": 0.26, + "grad_norm": 1.6790283918380737, + "learning_rate": 0.00019976562275013344, + "loss": 7.1852, + "step": 178 + }, + { + "epoch": 0.27, + "grad_norm": 3.375530242919922, + "learning_rate": 0.00019976240238386787, + "loss": 7.2132, + "step": 179 + }, + { + "epoch": 0.27, + "grad_norm": 2.9105639457702637, + "learning_rate": 0.0001997591600705424, + "loss": 7.2096, + "step": 180 + }, + { + "epoch": 0.27, + "grad_norm": 3.507265329360962, + "learning_rate": 0.00019975589581087023, + "loss": 7.1184, + "step": 181 + }, + { + "epoch": 0.27, + "grad_norm": 3.2614657878875732, + "learning_rate": 0.0001997526096055695, + "loss": 7.174, + "step": 182 + }, + { + "epoch": 0.27, + "grad_norm": 20.52471923828125, + "learning_rate": 0.0001997493014553632, + "loss": 7.1458, + "step": 183 + }, + { + "epoch": 0.27, + "grad_norm": 7.462634563446045, + "learning_rate": 0.00019974597136097904, + "loss": 7.2018, + "step": 184 + }, + { + "epoch": 0.27, + "grad_norm": 3.244502544403076, + "learning_rate": 0.00019974261932314972, + "loss": 7.2068, + "step": 185 + }, + { + "epoch": 0.28, + "grad_norm": 3.8728437423706055, + "learning_rate": 0.0001997392453426125, + "loss": 7.2269, + "step": 186 + }, + { + "epoch": 0.28, + "grad_norm": 1.4866065979003906, + "learning_rate": 0.00019973584942010978, + "loss": 7.3578, + "step": 187 + }, + { + "epoch": 0.28, + "grad_norm": 11.506189346313477, + "learning_rate": 0.0001997324315563886, + "loss": 7.3122, + "step": 188 + }, + { + "epoch": 0.28, + "grad_norm": 30.93811798095703, + "learning_rate": 0.0001997289917522008, + "loss": 7.3178, + "step": 189 + }, + { + "epoch": 0.28, + "grad_norm": 27.9945125579834, + "learning_rate": 0.00019972553000830324, + "loss": 7.3128, + "step": 190 + }, + { + "epoch": 0.28, + "grad_norm": 144.04150390625, + "learning_rate": 0.00019972204632545736, + "loss": 7.2427, + "step": 191 + }, + { + "epoch": 0.29, + "grad_norm": 77.6709213256836, + "learning_rate": 0.00019971854070442962, + "loss": 7.1204, + "step": 192 + }, + { + "epoch": 0.29, + "grad_norm": 56.658504486083984, + "learning_rate": 0.00019971501314599118, + "loss": 7.1243, + "step": 193 + }, + { + "epoch": 0.29, + "grad_norm": 12.49243450164795, + "learning_rate": 0.00019971146365091813, + "loss": 7.0943, + "step": 194 + }, + { + "epoch": 0.29, + "grad_norm": 12.344218254089355, + "learning_rate": 0.0001997078922199913, + "loss": 7.0644, + "step": 195 + }, + { + "epoch": 0.29, + "grad_norm": 542.6148071289062, + "learning_rate": 0.0001997042988539964, + "loss": 7.1231, + "step": 196 + }, + { + "epoch": 0.29, + "grad_norm": 13.391535758972168, + "learning_rate": 0.00019970068355372393, + "loss": 7.0059, + "step": 197 + }, + { + "epoch": 0.29, + "grad_norm": 438.4042663574219, + "learning_rate": 0.00019969704631996927, + "loss": 7.0568, + "step": 198 + }, + { + "epoch": 0.3, + "grad_norm": 918.09326171875, + "learning_rate": 0.00019969338715353257, + "loss": 7.0436, + "step": 199 + }, + { + "epoch": 0.3, + "grad_norm": 8.38797664642334, + "learning_rate": 0.00019968970605521876, + "loss": 7.0259, + "step": 200 + }, + { + "epoch": 0.3, + "grad_norm": 99.3614501953125, + "learning_rate": 0.00019968600302583774, + "loss": 7.016, + "step": 201 + }, + { + "epoch": 0.3, + "grad_norm": 9.193567276000977, + "learning_rate": 0.0001996822780662041, + "loss": 7.0366, + "step": 202 + }, + { + "epoch": 0.3, + "grad_norm": 20.37701988220215, + "learning_rate": 0.00019967853117713739, + "loss": 7.0319, + "step": 203 + }, + { + "epoch": 0.3, + "grad_norm": 13.956972122192383, + "learning_rate": 0.00019967476235946175, + "loss": 7.1477, + "step": 204 + }, + { + "epoch": 0.3, + "grad_norm": 18.673934936523438, + "learning_rate": 0.00019967097161400642, + "loss": 7.0995, + "step": 205 + }, + { + "epoch": 0.31, + "grad_norm": 52.66343688964844, + "learning_rate": 0.0001996671589416053, + "loss": 7.1156, + "step": 206 + }, + { + "epoch": 0.31, + "grad_norm": 54.673892974853516, + "learning_rate": 0.00019966332434309716, + "loss": 7.0896, + "step": 207 + }, + { + "epoch": 0.31, + "grad_norm": 120.72140502929688, + "learning_rate": 0.00019965946781932558, + "loss": 7.0867, + "step": 208 + }, + { + "epoch": 0.31, + "grad_norm": 22.450502395629883, + "learning_rate": 0.00019965558937113894, + "loss": 7.1196, + "step": 209 + }, + { + "epoch": 0.31, + "grad_norm": 94.62016296386719, + "learning_rate": 0.00019965168899939052, + "loss": 7.1317, + "step": 210 + }, + { + "epoch": 0.31, + "grad_norm": 47.635772705078125, + "learning_rate": 0.00019964776670493837, + "loss": 7.1411, + "step": 211 + }, + { + "epoch": 0.31, + "grad_norm": 5.324208736419678, + "learning_rate": 0.00019964382248864536, + "loss": 7.1979, + "step": 212 + }, + { + "epoch": 0.32, + "grad_norm": 11.261038780212402, + "learning_rate": 0.00019963985635137915, + "loss": 7.1543, + "step": 213 + }, + { + "epoch": 0.32, + "grad_norm": 463.3558044433594, + "learning_rate": 0.0001996358682940123, + "loss": 7.1743, + "step": 214 + }, + { + "epoch": 0.32, + "grad_norm": 3.5599043369293213, + "learning_rate": 0.00019963185831742218, + "loss": 7.1325, + "step": 215 + }, + { + "epoch": 0.32, + "grad_norm": 3.592991828918457, + "learning_rate": 0.00019962782642249092, + "loss": 7.1094, + "step": 216 + }, + { + "epoch": 0.32, + "grad_norm": 59.75080108642578, + "learning_rate": 0.00019962377261010551, + "loss": 7.1571, + "step": 217 + }, + { + "epoch": 0.32, + "grad_norm": 3.108335494995117, + "learning_rate": 0.00019961969688115782, + "loss": 7.061, + "step": 218 + }, + { + "epoch": 0.33, + "grad_norm": 87.59527587890625, + "learning_rate": 0.0001996155992365444, + "loss": 7.0367, + "step": 219 + }, + { + "epoch": 0.33, + "grad_norm": 234.99960327148438, + "learning_rate": 0.00019961147967716673, + "loss": 7.1225, + "step": 220 + }, + { + "epoch": 0.33, + "grad_norm": 2.509389877319336, + "learning_rate": 0.00019960733820393114, + "loss": 7.0294, + "step": 221 + }, + { + "epoch": 0.33, + "grad_norm": 4.749375343322754, + "learning_rate": 0.00019960317481774865, + "loss": 6.9989, + "step": 222 + }, + { + "epoch": 0.33, + "grad_norm": 6.5035319328308105, + "learning_rate": 0.00019959898951953524, + "loss": 6.9879, + "step": 223 + }, + { + "epoch": 0.33, + "grad_norm": 15.936602592468262, + "learning_rate": 0.0001995947823102116, + "loss": 6.8966, + "step": 224 + }, + { + "epoch": 0.33, + "grad_norm": 27.570510864257812, + "learning_rate": 0.00019959055319070332, + "loss": 6.9171, + "step": 225 + }, + { + "epoch": 0.34, + "grad_norm": 31.393901824951172, + "learning_rate": 0.00019958630216194073, + "loss": 6.993, + "step": 226 + }, + { + "epoch": 0.34, + "grad_norm": 2.980260133743286, + "learning_rate": 0.0001995820292248591, + "loss": 7.0092, + "step": 227 + }, + { + "epoch": 0.34, + "grad_norm": 2.1385061740875244, + "learning_rate": 0.0001995777343803984, + "loss": 6.9541, + "step": 228 + }, + { + "epoch": 0.34, + "grad_norm": 4.7213053703308105, + "learning_rate": 0.00019957341762950344, + "loss": 6.8875, + "step": 229 + }, + { + "epoch": 0.34, + "grad_norm": 2.533954381942749, + "learning_rate": 0.000199569078973124, + "loss": 6.934, + "step": 230 + }, + { + "epoch": 0.34, + "grad_norm": 3.432656764984131, + "learning_rate": 0.00019956471841221438, + "loss": 6.8725, + "step": 231 + }, + { + "epoch": 0.34, + "grad_norm": 9.07863998413086, + "learning_rate": 0.000199560335947734, + "loss": 6.8456, + "step": 232 + }, + { + "epoch": 0.35, + "grad_norm": 2.4700138568878174, + "learning_rate": 0.0001995559315806469, + "loss": 6.8803, + "step": 233 + }, + { + "epoch": 0.35, + "grad_norm": 15.215128898620605, + "learning_rate": 0.0001995515053119221, + "loss": 6.7947, + "step": 234 + }, + { + "epoch": 0.35, + "grad_norm": 10.798171997070312, + "learning_rate": 0.00019954705714253328, + "loss": 6.8543, + "step": 235 + }, + { + "epoch": 0.35, + "grad_norm": 2.1929962635040283, + "learning_rate": 0.000199542587073459, + "loss": 6.7983, + "step": 236 + }, + { + "epoch": 0.35, + "grad_norm": 18.292057037353516, + "learning_rate": 0.00019953809510568271, + "loss": 6.704, + "step": 237 + }, + { + "epoch": 0.35, + "grad_norm": 0.7752164602279663, + "learning_rate": 0.00019953358124019253, + "loss": 6.7268, + "step": 238 + }, + { + "epoch": 0.35, + "grad_norm": 1.0236197710037231, + "learning_rate": 0.00019952904547798154, + "loss": 6.6618, + "step": 239 + }, + { + "epoch": 0.36, + "grad_norm": 1.3812897205352783, + "learning_rate": 0.00019952448782004756, + "loss": 6.6367, + "step": 240 + }, + { + "epoch": 0.36, + "grad_norm": 1.219442367553711, + "learning_rate": 0.00019951990826739323, + "loss": 6.6027, + "step": 241 + }, + { + "epoch": 0.36, + "grad_norm": 1.485183596611023, + "learning_rate": 0.00019951530682102604, + "loss": 6.5468, + "step": 242 + }, + { + "epoch": 0.36, + "grad_norm": 20.273746490478516, + "learning_rate": 0.00019951068348195828, + "loss": 6.5281, + "step": 243 + }, + { + "epoch": 0.36, + "grad_norm": 3.194899082183838, + "learning_rate": 0.00019950603825120706, + "loss": 6.5781, + "step": 244 + }, + { + "epoch": 0.36, + "grad_norm": 0.9288484454154968, + "learning_rate": 0.00019950137112979425, + "loss": 6.5043, + "step": 245 + }, + { + "epoch": 0.37, + "grad_norm": 0.9781804084777832, + "learning_rate": 0.00019949668211874665, + "loss": 6.4531, + "step": 246 + }, + { + "epoch": 0.37, + "grad_norm": 5.633649826049805, + "learning_rate": 0.00019949197121909575, + "loss": 6.4262, + "step": 247 + }, + { + "epoch": 0.37, + "grad_norm": 4.805322170257568, + "learning_rate": 0.00019948723843187796, + "loss": 6.3919, + "step": 248 + }, + { + "epoch": 0.37, + "grad_norm": 1.2492502927780151, + "learning_rate": 0.00019948248375813445, + "loss": 6.3459, + "step": 249 + }, + { + "epoch": 0.37, + "grad_norm": 1.5589463710784912, + "learning_rate": 0.0001994777071989112, + "loss": 6.2473, + "step": 250 + }, + { + "epoch": 0.37, + "grad_norm": 1.4179363250732422, + "learning_rate": 0.00019947290875525908, + "loss": 6.2753, + "step": 251 + }, + { + "epoch": 0.37, + "grad_norm": 1.064170479774475, + "learning_rate": 0.00019946808842823366, + "loss": 6.2785, + "step": 252 + }, + { + "epoch": 0.38, + "grad_norm": 1.806328296661377, + "learning_rate": 0.00019946324621889535, + "loss": 6.2752, + "step": 253 + }, + { + "epoch": 0.38, + "grad_norm": 1.2437814474105835, + "learning_rate": 0.0001994583821283095, + "loss": 6.1444, + "step": 254 + }, + { + "epoch": 0.38, + "grad_norm": 17.912715911865234, + "learning_rate": 0.00019945349615754608, + "loss": 6.1001, + "step": 255 + }, + { + "epoch": 0.38, + "grad_norm": 0.8929802775382996, + "learning_rate": 0.00019944858830768002, + "loss": 6.0709, + "step": 256 + }, + { + "epoch": 0.38, + "grad_norm": 2.8683338165283203, + "learning_rate": 0.000199443658579791, + "loss": 5.9837, + "step": 257 + }, + { + "epoch": 0.38, + "grad_norm": 1.3912129402160645, + "learning_rate": 0.00019943870697496357, + "loss": 5.9728, + "step": 258 + }, + { + "epoch": 0.38, + "grad_norm": 1.9068537950515747, + "learning_rate": 0.00019943373349428698, + "loss": 5.9657, + "step": 259 + }, + { + "epoch": 0.39, + "grad_norm": 2.0884199142456055, + "learning_rate": 0.0001994287381388554, + "loss": 5.963, + "step": 260 + }, + { + "epoch": 0.39, + "grad_norm": 0.8210564851760864, + "learning_rate": 0.00019942372090976779, + "loss": 5.8253, + "step": 261 + }, + { + "epoch": 0.39, + "grad_norm": 3.3612587451934814, + "learning_rate": 0.0001994186818081279, + "loss": 5.7743, + "step": 262 + }, + { + "epoch": 0.39, + "grad_norm": 2.1916985511779785, + "learning_rate": 0.00019941362083504425, + "loss": 5.7697, + "step": 263 + }, + { + "epoch": 0.39, + "grad_norm": 1.3054827451705933, + "learning_rate": 0.00019940853799163026, + "loss": 5.7867, + "step": 264 + }, + { + "epoch": 0.39, + "grad_norm": 12.717972755432129, + "learning_rate": 0.00019940343327900416, + "loss": 5.6708, + "step": 265 + }, + { + "epoch": 0.4, + "grad_norm": 3.4190762042999268, + "learning_rate": 0.00019939830669828888, + "loss": 5.6869, + "step": 266 + }, + { + "epoch": 0.4, + "grad_norm": 2.753885507583618, + "learning_rate": 0.00019939315825061223, + "loss": 5.5825, + "step": 267 + }, + { + "epoch": 0.4, + "grad_norm": 1.1797378063201904, + "learning_rate": 0.0001993879879371069, + "loss": 5.6165, + "step": 268 + }, + { + "epoch": 0.4, + "grad_norm": 2.9885246753692627, + "learning_rate": 0.00019938279575891033, + "loss": 5.5817, + "step": 269 + }, + { + "epoch": 0.4, + "grad_norm": 1.8179508447647095, + "learning_rate": 0.00019937758171716468, + "loss": 5.4549, + "step": 270 + }, + { + "epoch": 0.4, + "grad_norm": 7.133963584899902, + "learning_rate": 0.00019937234581301707, + "loss": 5.5333, + "step": 271 + }, + { + "epoch": 0.4, + "grad_norm": 7.718517780303955, + "learning_rate": 0.00019936708804761932, + "loss": 5.4891, + "step": 272 + }, + { + "epoch": 0.41, + "grad_norm": 1.2075421810150146, + "learning_rate": 0.00019936180842212818, + "loss": 5.3542, + "step": 273 + }, + { + "epoch": 0.41, + "grad_norm": 6.181577205657959, + "learning_rate": 0.000199356506937705, + "loss": 5.4501, + "step": 274 + }, + { + "epoch": 0.41, + "grad_norm": 6.531129837036133, + "learning_rate": 0.00019935118359551624, + "loss": 5.5179, + "step": 275 + }, + { + "epoch": 0.41, + "grad_norm": 4.549463272094727, + "learning_rate": 0.00019934583839673287, + "loss": 5.4167, + "step": 276 + }, + { + "epoch": 0.41, + "grad_norm": 1.8859062194824219, + "learning_rate": 0.00019934047134253081, + "loss": 5.2628, + "step": 277 + }, + { + "epoch": 0.41, + "grad_norm": 7.954891681671143, + "learning_rate": 0.00019933508243409086, + "loss": 5.3723, + "step": 278 + }, + { + "epoch": 0.41, + "grad_norm": 6.0001444816589355, + "learning_rate": 0.00019932967167259847, + "loss": 5.2979, + "step": 279 + }, + { + "epoch": 0.42, + "grad_norm": 6.450177192687988, + "learning_rate": 0.000199324239059244, + "loss": 5.2947, + "step": 280 + }, + { + "epoch": 0.42, + "grad_norm": 5.8015642166137695, + "learning_rate": 0.00019931878459522254, + "loss": 5.2849, + "step": 281 + }, + { + "epoch": 0.42, + "grad_norm": 8.846211433410645, + "learning_rate": 0.00019931330828173408, + "loss": 5.1841, + "step": 282 + }, + { + "epoch": 0.42, + "grad_norm": 4.6721086502075195, + "learning_rate": 0.00019930781011998338, + "loss": 5.2056, + "step": 283 + }, + { + "epoch": 0.42, + "grad_norm": 3.9085495471954346, + "learning_rate": 0.00019930229011118, + "loss": 5.1857, + "step": 284 + }, + { + "epoch": 0.42, + "grad_norm": 1.6466888189315796, + "learning_rate": 0.00019929674825653826, + "loss": 5.0631, + "step": 285 + }, + { + "epoch": 0.42, + "grad_norm": 2.1051926612854004, + "learning_rate": 0.0001992911845572774, + "loss": 5.033, + "step": 286 + }, + { + "epoch": 0.43, + "grad_norm": 3.247199058532715, + "learning_rate": 0.00019928559901462133, + "loss": 5.098, + "step": 287 + }, + { + "epoch": 0.43, + "grad_norm": 4.04020881652832, + "learning_rate": 0.00019927999162979885, + "loss": 5.0759, + "step": 288 + }, + { + "epoch": 0.43, + "grad_norm": 1.5520354509353638, + "learning_rate": 0.00019927436240404359, + "loss": 5.0395, + "step": 289 + }, + { + "epoch": 0.43, + "grad_norm": 1.2532438039779663, + "learning_rate": 0.00019926871133859388, + "loss": 4.9942, + "step": 290 + }, + { + "epoch": 0.43, + "grad_norm": 3.3151402473449707, + "learning_rate": 0.00019926303843469296, + "loss": 4.9396, + "step": 291 + }, + { + "epoch": 0.43, + "grad_norm": 1.311003565788269, + "learning_rate": 0.00019925734369358883, + "loss": 4.9452, + "step": 292 + }, + { + "epoch": 0.44, + "grad_norm": 3.930495500564575, + "learning_rate": 0.00019925162711653425, + "loss": 4.9488, + "step": 293 + }, + { + "epoch": 0.44, + "grad_norm": 1.0052605867385864, + "learning_rate": 0.00019924588870478687, + "loss": 4.8627, + "step": 294 + }, + { + "epoch": 0.44, + "grad_norm": 7.544490814208984, + "learning_rate": 0.00019924012845960908, + "loss": 4.9602, + "step": 295 + }, + { + "epoch": 0.44, + "grad_norm": 6.155104637145996, + "learning_rate": 0.00019923434638226812, + "loss": 4.9091, + "step": 296 + }, + { + "epoch": 0.44, + "grad_norm": 3.3449931144714355, + "learning_rate": 0.00019922854247403599, + "loss": 5.002, + "step": 297 + }, + { + "epoch": 0.44, + "grad_norm": 3.9289393424987793, + "learning_rate": 0.0001992227167361895, + "loss": 4.8124, + "step": 298 + }, + { + "epoch": 0.44, + "grad_norm": 6.716512680053711, + "learning_rate": 0.00019921686917001032, + "loss": 4.8421, + "step": 299 + }, + { + "epoch": 0.45, + "grad_norm": 4.2615132331848145, + "learning_rate": 0.00019921099977678484, + "loss": 4.8303, + "step": 300 + }, + { + "epoch": 0.45, + "grad_norm": 4.952175617218018, + "learning_rate": 0.00019920510855780428, + "loss": 4.8718, + "step": 301 + }, + { + "epoch": 0.45, + "grad_norm": 1.830321192741394, + "learning_rate": 0.00019919919551436468, + "loss": 4.8395, + "step": 302 + }, + { + "epoch": 0.45, + "grad_norm": 2.7058656215667725, + "learning_rate": 0.0001991932606477669, + "loss": 4.721, + "step": 303 + }, + { + "epoch": 0.45, + "grad_norm": 2.4603841304779053, + "learning_rate": 0.00019918730395931649, + "loss": 4.7365, + "step": 304 + }, + { + "epoch": 0.45, + "grad_norm": 1.5384434461593628, + "learning_rate": 0.00019918132545032396, + "loss": 4.6979, + "step": 305 + }, + { + "epoch": 0.45, + "grad_norm": 5.274871349334717, + "learning_rate": 0.0001991753251221045, + "loss": 4.7287, + "step": 306 + }, + { + "epoch": 0.46, + "grad_norm": 3.8402764797210693, + "learning_rate": 0.00019916930297597817, + "loss": 4.744, + "step": 307 + }, + { + "epoch": 0.46, + "grad_norm": 3.417302131652832, + "learning_rate": 0.0001991632590132698, + "loss": 4.6439, + "step": 308 + }, + { + "epoch": 0.46, + "grad_norm": 4.113287925720215, + "learning_rate": 0.00019915719323530898, + "loss": 4.6228, + "step": 309 + }, + { + "epoch": 0.46, + "grad_norm": 3.57785964012146, + "learning_rate": 0.0001991511056434302, + "loss": 4.6077, + "step": 310 + }, + { + "epoch": 0.46, + "grad_norm": 1.568590521812439, + "learning_rate": 0.00019914499623897264, + "loss": 4.6056, + "step": 311 + }, + { + "epoch": 0.46, + "grad_norm": 3.293492317199707, + "learning_rate": 0.0001991388650232804, + "loss": 4.5294, + "step": 312 + }, + { + "epoch": 0.46, + "grad_norm": 3.416661262512207, + "learning_rate": 0.00019913271199770217, + "loss": 4.6239, + "step": 313 + }, + { + "epoch": 0.47, + "grad_norm": 2.6851394176483154, + "learning_rate": 0.0001991265371635917, + "loss": 4.6183, + "step": 314 + }, + { + "epoch": 0.47, + "grad_norm": 3.812194585800171, + "learning_rate": 0.0001991203405223074, + "loss": 4.5595, + "step": 315 + }, + { + "epoch": 0.47, + "grad_norm": 1.8244884014129639, + "learning_rate": 0.0001991141220752124, + "loss": 4.5329, + "step": 316 + }, + { + "epoch": 0.47, + "grad_norm": 2.763340950012207, + "learning_rate": 0.0001991078818236748, + "loss": 4.5336, + "step": 317 + }, + { + "epoch": 0.47, + "grad_norm": 0.6997628808021545, + "learning_rate": 0.0001991016197690674, + "loss": 4.5623, + "step": 318 + }, + { + "epoch": 0.47, + "grad_norm": 6.273808479309082, + "learning_rate": 0.0001990953359127678, + "loss": 4.549, + "step": 319 + }, + { + "epoch": 0.48, + "grad_norm": 4.5749359130859375, + "learning_rate": 0.00019908903025615842, + "loss": 4.4894, + "step": 320 + }, + { + "epoch": 0.48, + "grad_norm": 4.788200378417969, + "learning_rate": 0.00019908270280062643, + "loss": 4.514, + "step": 321 + }, + { + "epoch": 0.48, + "grad_norm": 5.6347551345825195, + "learning_rate": 0.00019907635354756385, + "loss": 4.477, + "step": 322 + }, + { + "epoch": 0.48, + "grad_norm": 5.96523904800415, + "learning_rate": 0.00019906998249836745, + "loss": 4.4759, + "step": 323 + }, + { + "epoch": 0.48, + "grad_norm": 2.5423848628997803, + "learning_rate": 0.00019906358965443888, + "loss": 4.4077, + "step": 324 + }, + { + "epoch": 0.48, + "grad_norm": 1.0939940214157104, + "learning_rate": 0.00019905717501718444, + "loss": 4.372, + "step": 325 + }, + { + "epoch": 0.48, + "grad_norm": 2.4741616249084473, + "learning_rate": 0.00019905073858801536, + "loss": 4.4112, + "step": 326 + }, + { + "epoch": 0.49, + "grad_norm": 2.774264335632324, + "learning_rate": 0.00019904428036834763, + "loss": 4.4412, + "step": 327 + }, + { + "epoch": 0.49, + "grad_norm": 1.9465824365615845, + "learning_rate": 0.00019903780035960194, + "loss": 4.4442, + "step": 328 + }, + { + "epoch": 0.49, + "grad_norm": 1.6578731536865234, + "learning_rate": 0.0001990312985632039, + "loss": 4.4368, + "step": 329 + }, + { + "epoch": 0.49, + "grad_norm": 1.1936014890670776, + "learning_rate": 0.00019902477498058387, + "loss": 4.3445, + "step": 330 + }, + { + "epoch": 0.49, + "grad_norm": 1.1214628219604492, + "learning_rate": 0.00019901822961317697, + "loss": 4.4032, + "step": 331 + }, + { + "epoch": 0.49, + "grad_norm": 1.4763222932815552, + "learning_rate": 0.00019901166246242316, + "loss": 4.3713, + "step": 332 + }, + { + "epoch": 0.49, + "grad_norm": 3.219590425491333, + "learning_rate": 0.00019900507352976714, + "loss": 4.3248, + "step": 333 + }, + { + "epoch": 0.5, + "grad_norm": 1.5973373651504517, + "learning_rate": 0.00019899846281665843, + "loss": 4.3571, + "step": 334 + }, + { + "epoch": 0.5, + "grad_norm": 2.3610401153564453, + "learning_rate": 0.0001989918303245514, + "loss": 4.2991, + "step": 335 + }, + { + "epoch": 0.5, + "grad_norm": 2.8560168743133545, + "learning_rate": 0.0001989851760549051, + "loss": 4.3024, + "step": 336 + }, + { + "epoch": 0.5, + "grad_norm": 0.9701719880104065, + "learning_rate": 0.0001989785000091834, + "loss": 4.2602, + "step": 337 + }, + { + "epoch": 0.5, + "grad_norm": 1.6666287183761597, + "learning_rate": 0.00019897180218885507, + "loss": 4.2519, + "step": 338 + }, + { + "epoch": 0.5, + "eval_loss": 4.065431594848633, + "eval_runtime": 1149.6322, + "eval_samples_per_second": 14.911, + "eval_steps_per_second": 0.932, + "step": 338 + }, + { + "epoch": 0.5, + "grad_norm": 2.7144992351531982, + "learning_rate": 0.00019896508259539353, + "loss": 4.2504, + "step": 339 + }, + { + "epoch": 0.5, + "grad_norm": 1.088392734527588, + "learning_rate": 0.00019895834123027703, + "loss": 4.2467, + "step": 340 + }, + { + "epoch": 0.51, + "grad_norm": 1.7031919956207275, + "learning_rate": 0.00019895157809498868, + "loss": 4.2498, + "step": 341 + }, + { + "epoch": 0.51, + "grad_norm": 3.883507013320923, + "learning_rate": 0.00019894479319101626, + "loss": 4.289, + "step": 342 + }, + { + "epoch": 0.51, + "grad_norm": 1.1266521215438843, + "learning_rate": 0.00019893798651985246, + "loss": 4.2371, + "step": 343 + }, + { + "epoch": 0.51, + "grad_norm": 2.436180591583252, + "learning_rate": 0.00019893115808299468, + "loss": 4.213, + "step": 344 + }, + { + "epoch": 0.51, + "grad_norm": 6.882200241088867, + "learning_rate": 0.00019892430788194513, + "loss": 4.2289, + "step": 345 + }, + { + "epoch": 0.51, + "grad_norm": 4.444315433502197, + "learning_rate": 0.00019891743591821077, + "loss": 4.1858, + "step": 346 + }, + { + "epoch": 0.52, + "grad_norm": 2.7933340072631836, + "learning_rate": 0.00019891054219330344, + "loss": 4.1988, + "step": 347 + }, + { + "epoch": 0.52, + "grad_norm": 1.5705324411392212, + "learning_rate": 0.00019890362670873973, + "loss": 4.1639, + "step": 348 + }, + { + "epoch": 0.52, + "grad_norm": 4.15967321395874, + "learning_rate": 0.00019889668946604093, + "loss": 4.1391, + "step": 349 + }, + { + "epoch": 0.52, + "grad_norm": 3.6706178188323975, + "learning_rate": 0.00019888973046673324, + "loss": 4.1706, + "step": 350 + }, + { + "epoch": 0.52, + "grad_norm": 4.234200477600098, + "learning_rate": 0.00019888274971234757, + "loss": 4.3108, + "step": 351 + }, + { + "epoch": 0.52, + "grad_norm": 2.6971170902252197, + "learning_rate": 0.00019887574720441967, + "loss": 4.1871, + "step": 352 + }, + { + "epoch": 0.52, + "grad_norm": 1.4427449703216553, + "learning_rate": 0.00019886872294448997, + "loss": 4.1327, + "step": 353 + }, + { + "epoch": 0.53, + "grad_norm": 4.426568984985352, + "learning_rate": 0.00019886167693410384, + "loss": 4.195, + "step": 354 + }, + { + "epoch": 0.53, + "grad_norm": 4.152656555175781, + "learning_rate": 0.00019885460917481136, + "loss": 4.1858, + "step": 355 + }, + { + "epoch": 0.53, + "grad_norm": 5.731381893157959, + "learning_rate": 0.0001988475196681673, + "loss": 4.179, + "step": 356 + }, + { + "epoch": 0.53, + "grad_norm": 17.160917282104492, + "learning_rate": 0.0001988404084157314, + "loss": 4.0972, + "step": 357 + }, + { + "epoch": 0.53, + "grad_norm": 1.9398918151855469, + "learning_rate": 0.00019883327541906802, + "loss": 4.1238, + "step": 358 + }, + { + "epoch": 0.53, + "grad_norm": 2.0568935871124268, + "learning_rate": 0.00019882612067974643, + "loss": 4.0953, + "step": 359 + }, + { + "epoch": 0.53, + "grad_norm": 3.45890212059021, + "learning_rate": 0.00019881894419934056, + "loss": 4.1171, + "step": 360 + }, + { + "epoch": 0.54, + "grad_norm": 1.6760789155960083, + "learning_rate": 0.00019881174597942923, + "loss": 4.1128, + "step": 361 + }, + { + "epoch": 0.54, + "grad_norm": 307.13031005859375, + "learning_rate": 0.000198804526021596, + "loss": 4.0658, + "step": 362 + }, + { + "epoch": 0.54, + "grad_norm": 15.57320785522461, + "learning_rate": 0.00019879728432742918, + "loss": 4.2352, + "step": 363 + }, + { + "epoch": 0.54, + "grad_norm": 8.536359786987305, + "learning_rate": 0.00019879002089852197, + "loss": 4.2513, + "step": 364 + }, + { + "epoch": 0.54, + "grad_norm": 9.536478996276855, + "learning_rate": 0.00019878273573647218, + "loss": 4.2273, + "step": 365 + }, + { + "epoch": 0.54, + "grad_norm": 11.019887924194336, + "learning_rate": 0.00019877542884288256, + "loss": 4.2037, + "step": 366 + }, + { + "epoch": 0.55, + "grad_norm": 7.40153169631958, + "learning_rate": 0.00019876810021936053, + "loss": 4.1092, + "step": 367 + }, + { + "epoch": 0.55, + "grad_norm": 5.204662322998047, + "learning_rate": 0.0001987607498675184, + "loss": 4.0828, + "step": 368 + }, + { + "epoch": 0.55, + "grad_norm": 3.2306785583496094, + "learning_rate": 0.00019875337778897315, + "loss": 4.0828, + "step": 369 + }, + { + "epoch": 0.55, + "grad_norm": 9.242484092712402, + "learning_rate": 0.00019874598398534658, + "loss": 4.1478, + "step": 370 + }, + { + "epoch": 0.55, + "grad_norm": 10.423043251037598, + "learning_rate": 0.00019873856845826536, + "loss": 4.1709, + "step": 371 + }, + { + "epoch": 0.55, + "grad_norm": 2.9181222915649414, + "learning_rate": 0.00019873113120936074, + "loss": 4.0819, + "step": 372 + }, + { + "epoch": 0.55, + "grad_norm": 5.6168718338012695, + "learning_rate": 0.00019872367224026895, + "loss": 4.09, + "step": 373 + }, + { + "epoch": 0.56, + "grad_norm": 2.738548994064331, + "learning_rate": 0.00019871619155263087, + "loss": 4.1129, + "step": 374 + }, + { + "epoch": 0.56, + "grad_norm": 8.364383697509766, + "learning_rate": 0.0001987086891480922, + "loss": 4.0793, + "step": 375 + }, + { + "epoch": 0.56, + "grad_norm": 8.516968727111816, + "learning_rate": 0.00019870116502830346, + "loss": 4.1287, + "step": 376 + }, + { + "epoch": 0.56, + "grad_norm": 2.941425323486328, + "learning_rate": 0.00019869361919491987, + "loss": 4.1145, + "step": 377 + }, + { + "epoch": 0.56, + "grad_norm": 4.925361633300781, + "learning_rate": 0.00019868605164960145, + "loss": 4.1046, + "step": 378 + }, + { + "epoch": 0.56, + "grad_norm": 1.9746365547180176, + "learning_rate": 0.0001986784623940131, + "loss": 3.9944, + "step": 379 + }, + { + "epoch": 0.56, + "grad_norm": 8.789060592651367, + "learning_rate": 0.00019867085142982426, + "loss": 4.0482, + "step": 380 + }, + { + "epoch": 0.57, + "grad_norm": 7.4179301261901855, + "learning_rate": 0.0001986632187587094, + "loss": 4.0702, + "step": 381 + }, + { + "epoch": 0.57, + "grad_norm": 3.4922187328338623, + "learning_rate": 0.00019865556438234763, + "loss": 4.0579, + "step": 382 + }, + { + "epoch": 0.57, + "grad_norm": 5.763650417327881, + "learning_rate": 0.00019864788830242284, + "loss": 4.0776, + "step": 383 + }, + { + "epoch": 0.57, + "grad_norm": 3.3361189365386963, + "learning_rate": 0.00019864019052062375, + "loss": 4.0426, + "step": 384 + }, + { + "epoch": 0.57, + "grad_norm": 8.98303508758545, + "learning_rate": 0.0001986324710386438, + "loss": 4.0599, + "step": 385 + }, + { + "epoch": 0.57, + "grad_norm": 8.836548805236816, + "learning_rate": 0.00019862472985818124, + "loss": 4.0534, + "step": 386 + }, + { + "epoch": 0.57, + "grad_norm": 2.0397250652313232, + "learning_rate": 0.00019861696698093902, + "loss": 4.0105, + "step": 387 + }, + { + "epoch": 0.58, + "grad_norm": 3.091047763824463, + "learning_rate": 0.000198609182408625, + "loss": 3.9637, + "step": 388 + }, + { + "epoch": 0.58, + "grad_norm": 3.192955255508423, + "learning_rate": 0.00019860137614295168, + "loss": 3.9963, + "step": 389 + }, + { + "epoch": 0.58, + "grad_norm": 1.5237922668457031, + "learning_rate": 0.00019859354818563643, + "loss": 3.964, + "step": 390 + }, + { + "epoch": 0.58, + "grad_norm": 4.192271709442139, + "learning_rate": 0.00019858569853840131, + "loss": 3.9638, + "step": 391 + }, + { + "epoch": 0.58, + "grad_norm": 2.3870139122009277, + "learning_rate": 0.00019857782720297319, + "loss": 4.0272, + "step": 392 + }, + { + "epoch": 0.58, + "grad_norm": 6.654936790466309, + "learning_rate": 0.00019856993418108376, + "loss": 3.9616, + "step": 393 + }, + { + "epoch": 0.59, + "grad_norm": 4.117982387542725, + "learning_rate": 0.00019856201947446937, + "loss": 3.9983, + "step": 394 + }, + { + "epoch": 0.59, + "grad_norm": 5.394196033477783, + "learning_rate": 0.00019855408308487124, + "loss": 4.0334, + "step": 395 + }, + { + "epoch": 0.59, + "grad_norm": 5.410153865814209, + "learning_rate": 0.0001985461250140353, + "loss": 4.0342, + "step": 396 + }, + { + "epoch": 0.59, + "grad_norm": 1.220145583152771, + "learning_rate": 0.00019853814526371228, + "loss": 3.9964, + "step": 397 + }, + { + "epoch": 0.59, + "grad_norm": 1.3831374645233154, + "learning_rate": 0.00019853014383565772, + "loss": 3.8909, + "step": 398 + }, + { + "epoch": 0.59, + "grad_norm": 3.4708709716796875, + "learning_rate": 0.0001985221207316318, + "loss": 3.9548, + "step": 399 + }, + { + "epoch": 0.59, + "grad_norm": 1.6609090566635132, + "learning_rate": 0.00019851407595339958, + "loss": 3.9481, + "step": 400 + }, + { + "epoch": 0.6, + "grad_norm": 1.5527194738388062, + "learning_rate": 0.00019850600950273086, + "loss": 3.9567, + "step": 401 + }, + { + "epoch": 0.6, + "grad_norm": 6.021573066711426, + "learning_rate": 0.00019849792138140022, + "loss": 4.0133, + "step": 402 + }, + { + "epoch": 0.6, + "grad_norm": 4.150272846221924, + "learning_rate": 0.00019848981159118699, + "loss": 3.9614, + "step": 403 + }, + { + "epoch": 0.6, + "grad_norm": 7.311338901519775, + "learning_rate": 0.00019848168013387522, + "loss": 3.9789, + "step": 404 + }, + { + "epoch": 0.6, + "grad_norm": 7.797035217285156, + "learning_rate": 0.00019847352701125385, + "loss": 4.0036, + "step": 405 + }, + { + "epoch": 0.6, + "grad_norm": 2.067911386489868, + "learning_rate": 0.0001984653522251165, + "loss": 3.9421, + "step": 406 + }, + { + "epoch": 0.6, + "grad_norm": 5.364695072174072, + "learning_rate": 0.00019845715577726148, + "loss": 3.9846, + "step": 407 + }, + { + "epoch": 0.61, + "grad_norm": 3.663616895675659, + "learning_rate": 0.00019844893766949207, + "loss": 3.9255, + "step": 408 + }, + { + "epoch": 0.61, + "grad_norm": 8.939522743225098, + "learning_rate": 0.00019844069790361611, + "loss": 3.9433, + "step": 409 + }, + { + "epoch": 0.61, + "grad_norm": 9.791443824768066, + "learning_rate": 0.00019843243648144635, + "loss": 3.9879, + "step": 410 + }, + { + "epoch": 0.61, + "grad_norm": 1.5382161140441895, + "learning_rate": 0.00019842415340480025, + "loss": 3.9755, + "step": 411 + }, + { + "epoch": 0.61, + "grad_norm": 7.2470879554748535, + "learning_rate": 0.00019841584867549998, + "loss": 3.9603, + "step": 412 + }, + { + "epoch": 0.61, + "grad_norm": 7.930785655975342, + "learning_rate": 0.0001984075222953726, + "loss": 3.9416, + "step": 413 + }, + { + "epoch": 0.61, + "grad_norm": 3.421830654144287, + "learning_rate": 0.00019839917426624977, + "loss": 3.9023, + "step": 414 + }, + { + "epoch": 0.62, + "grad_norm": 13.186482429504395, + "learning_rate": 0.00019839080458996806, + "loss": 4.0143, + "step": 415 + }, + { + "epoch": 0.62, + "grad_norm": 18.97665023803711, + "learning_rate": 0.0001983824132683687, + "loss": 4.0361, + "step": 416 + }, + { + "epoch": 0.62, + "grad_norm": 12.884719848632812, + "learning_rate": 0.0001983740003032978, + "loss": 4.0086, + "step": 417 + }, + { + "epoch": 0.62, + "grad_norm": 2.3282957077026367, + "learning_rate": 0.00019836556569660608, + "loss": 3.8853, + "step": 418 + }, + { + "epoch": 0.62, + "grad_norm": 5.4361186027526855, + "learning_rate": 0.0001983571094501491, + "loss": 3.8995, + "step": 419 + }, + { + "epoch": 0.62, + "grad_norm": 2.480142116546631, + "learning_rate": 0.00019834863156578724, + "loss": 3.9204, + "step": 420 + }, + { + "epoch": 0.63, + "grad_norm": 8.683335304260254, + "learning_rate": 0.0001983401320453855, + "loss": 4.04, + "step": 421 + }, + { + "epoch": 0.63, + "grad_norm": 10.6290864944458, + "learning_rate": 0.00019833161089081381, + "loss": 3.9174, + "step": 422 + }, + { + "epoch": 0.63, + "grad_norm": 1.7693558931350708, + "learning_rate": 0.00019832306810394664, + "loss": 3.937, + "step": 423 + }, + { + "epoch": 0.63, + "grad_norm": 6.862966537475586, + "learning_rate": 0.00019831450368666348, + "loss": 3.9314, + "step": 424 + }, + { + "epoch": 0.63, + "grad_norm": 6.495865345001221, + "learning_rate": 0.0001983059176408484, + "loss": 3.8863, + "step": 425 + }, + { + "epoch": 0.63, + "grad_norm": 2.8406107425689697, + "learning_rate": 0.0001982973099683902, + "loss": 3.9525, + "step": 426 + }, + { + "epoch": 0.63, + "grad_norm": 10.725767135620117, + "learning_rate": 0.00019828868067118255, + "loss": 3.894, + "step": 427 + }, + { + "epoch": 0.64, + "grad_norm": 12.51372241973877, + "learning_rate": 0.0001982800297511239, + "loss": 4.0417, + "step": 428 + }, + { + "epoch": 0.64, + "grad_norm": 5.210300445556641, + "learning_rate": 0.00019827135721011736, + "loss": 3.9486, + "step": 429 + }, + { + "epoch": 0.64, + "grad_norm": 8.358055114746094, + "learning_rate": 0.0001982626630500708, + "loss": 4.0152, + "step": 430 + }, + { + "epoch": 0.64, + "grad_norm": 6.729689598083496, + "learning_rate": 0.0001982539472728969, + "loss": 3.8904, + "step": 431 + }, + { + "epoch": 0.64, + "grad_norm": 4.138473033905029, + "learning_rate": 0.00019824520988051306, + "loss": 3.8004, + "step": 432 + }, + { + "epoch": 0.64, + "grad_norm": 6.071408271789551, + "learning_rate": 0.00019823645087484147, + "loss": 4.0332, + "step": 433 + }, + { + "epoch": 0.64, + "grad_norm": 4.820131301879883, + "learning_rate": 0.000198227670257809, + "loss": 4.0076, + "step": 434 + }, + { + "epoch": 0.65, + "grad_norm": 3.6242270469665527, + "learning_rate": 0.00019821886803134738, + "loss": 3.9052, + "step": 435 + }, + { + "epoch": 0.65, + "grad_norm": 4.531324863433838, + "learning_rate": 0.00019821004419739307, + "loss": 3.8389, + "step": 436 + }, + { + "epoch": 0.65, + "grad_norm": 3.6086678504943848, + "learning_rate": 0.00019820119875788716, + "loss": 3.8785, + "step": 437 + }, + { + "epoch": 0.65, + "grad_norm": 3.600398063659668, + "learning_rate": 0.00019819233171477566, + "loss": 3.8425, + "step": 438 + }, + { + "epoch": 0.65, + "grad_norm": 3.8471834659576416, + "learning_rate": 0.00019818344307000925, + "loss": 3.8667, + "step": 439 + }, + { + "epoch": 0.65, + "grad_norm": 3.2653310298919678, + "learning_rate": 0.00019817453282554333, + "loss": 3.8744, + "step": 440 + }, + { + "epoch": 0.65, + "grad_norm": 4.489375591278076, + "learning_rate": 0.00019816560098333817, + "loss": 3.8705, + "step": 441 + }, + { + "epoch": 0.66, + "grad_norm": 3.816256284713745, + "learning_rate": 0.00019815664754535862, + "loss": 3.7886, + "step": 442 + }, + { + "epoch": 0.66, + "grad_norm": 6.624678134918213, + "learning_rate": 0.00019814767251357448, + "loss": 3.865, + "step": 443 + }, + { + "epoch": 0.66, + "grad_norm": 6.498514652252197, + "learning_rate": 0.00019813867588996012, + "loss": 3.8771, + "step": 444 + }, + { + "epoch": 0.66, + "grad_norm": 3.340604305267334, + "learning_rate": 0.00019812965767649474, + "loss": 3.8211, + "step": 445 + }, + { + "epoch": 0.66, + "grad_norm": 5.6404805183410645, + "learning_rate": 0.0001981206178751623, + "loss": 3.8503, + "step": 446 + }, + { + "epoch": 0.66, + "grad_norm": 2.282567024230957, + "learning_rate": 0.00019811155648795153, + "loss": 3.865, + "step": 447 + }, + { + "epoch": 0.67, + "grad_norm": 4.082645893096924, + "learning_rate": 0.00019810247351685587, + "loss": 3.7833, + "step": 448 + }, + { + "epoch": 0.67, + "grad_norm": 2.7963573932647705, + "learning_rate": 0.00019809336896387344, + "loss": 3.7991, + "step": 449 + }, + { + "epoch": 0.67, + "grad_norm": 8.172618865966797, + "learning_rate": 0.00019808424283100726, + "loss": 3.8369, + "step": 450 + }, + { + "epoch": 0.67, + "grad_norm": 8.013967514038086, + "learning_rate": 0.000198075095120265, + "loss": 3.8526, + "step": 451 + }, + { + "epoch": 0.67, + "grad_norm": 0.9244366884231567, + "learning_rate": 0.00019806592583365907, + "loss": 3.7882, + "step": 452 + }, + { + "epoch": 0.67, + "grad_norm": 4.72100305557251, + "learning_rate": 0.00019805673497320662, + "loss": 3.7314, + "step": 453 + }, + { + "epoch": 0.67, + "grad_norm": 2.556580066680908, + "learning_rate": 0.00019804752254092968, + "loss": 3.8743, + "step": 454 + }, + { + "epoch": 0.68, + "grad_norm": 10.014892578125, + "learning_rate": 0.00019803828853885485, + "loss": 3.8263, + "step": 455 + }, + { + "epoch": 0.68, + "grad_norm": 12.621180534362793, + "learning_rate": 0.00019802903296901357, + "loss": 3.8423, + "step": 456 + }, + { + "epoch": 0.68, + "grad_norm": 6.725605010986328, + "learning_rate": 0.000198019755833442, + "loss": 3.7592, + "step": 457 + }, + { + "epoch": 0.68, + "grad_norm": 5.9739603996276855, + "learning_rate": 0.00019801045713418102, + "loss": 3.8252, + "step": 458 + }, + { + "epoch": 0.68, + "grad_norm": 6.546073913574219, + "learning_rate": 0.00019800113687327633, + "loss": 3.8135, + "step": 459 + }, + { + "epoch": 0.68, + "grad_norm": 1.279313325881958, + "learning_rate": 0.0001979917950527783, + "loss": 3.8242, + "step": 460 + }, + { + "epoch": 0.68, + "grad_norm": 14.773170471191406, + "learning_rate": 0.00019798243167474207, + "loss": 3.9442, + "step": 461 + }, + { + "epoch": 0.69, + "grad_norm": 20.262290954589844, + "learning_rate": 0.00019797304674122753, + "loss": 3.9303, + "step": 462 + }, + { + "epoch": 0.69, + "grad_norm": 13.257652282714844, + "learning_rate": 0.00019796364025429926, + "loss": 3.8547, + "step": 463 + }, + { + "epoch": 0.69, + "grad_norm": 1.5865988731384277, + "learning_rate": 0.00019795421221602668, + "loss": 3.781, + "step": 464 + }, + { + "epoch": 0.69, + "grad_norm": 4.687791347503662, + "learning_rate": 0.00019794476262848388, + "loss": 3.7924, + "step": 465 + }, + { + "epoch": 0.69, + "grad_norm": 1.8210766315460205, + "learning_rate": 0.0001979352914937497, + "loss": 3.8706, + "step": 466 + }, + { + "epoch": 0.69, + "grad_norm": 10.269219398498535, + "learning_rate": 0.00019792579881390768, + "loss": 3.8314, + "step": 467 + }, + { + "epoch": 0.7, + "grad_norm": 11.758221626281738, + "learning_rate": 0.00019791628459104625, + "loss": 3.8405, + "step": 468 + }, + { + "epoch": 0.7, + "grad_norm": 3.8874833583831787, + "learning_rate": 0.0001979067488272584, + "loss": 3.7889, + "step": 469 + }, + { + "epoch": 0.7, + "grad_norm": 8.348686218261719, + "learning_rate": 0.00019789719152464192, + "loss": 3.815, + "step": 470 + }, + { + "epoch": 0.7, + "grad_norm": 11.831920623779297, + "learning_rate": 0.0001978876126852994, + "loss": 3.8381, + "step": 471 + }, + { + "epoch": 0.7, + "grad_norm": 7.59721040725708, + "learning_rate": 0.00019787801231133814, + "loss": 3.8394, + "step": 472 + }, + { + "epoch": 0.7, + "grad_norm": 4.788586616516113, + "learning_rate": 0.00019786839040487005, + "loss": 3.8047, + "step": 473 + }, + { + "epoch": 0.7, + "grad_norm": 9.041956901550293, + "learning_rate": 0.00019785874696801202, + "loss": 3.7394, + "step": 474 + }, + { + "epoch": 0.71, + "grad_norm": 3.841315746307373, + "learning_rate": 0.00019784908200288546, + "loss": 3.7839, + "step": 475 + }, + { + "epoch": 0.71, + "grad_norm": 6.382961750030518, + "learning_rate": 0.0001978393955116166, + "loss": 3.7502, + "step": 476 + }, + { + "epoch": 0.71, + "grad_norm": 8.265791893005371, + "learning_rate": 0.00019782968749633642, + "loss": 3.7888, + "step": 477 + }, + { + "epoch": 0.71, + "grad_norm": 4.210597038269043, + "learning_rate": 0.00019781995795918063, + "loss": 3.7096, + "step": 478 + }, + { + "epoch": 0.71, + "grad_norm": 10.053391456604004, + "learning_rate": 0.00019781020690228963, + "loss": 3.7276, + "step": 479 + }, + { + "epoch": 0.71, + "grad_norm": 14.174814224243164, + "learning_rate": 0.0001978004343278086, + "loss": 3.891, + "step": 480 + }, + { + "epoch": 0.71, + "grad_norm": 7.2850661277771, + "learning_rate": 0.00019779064023788746, + "loss": 3.8149, + "step": 481 + }, + { + "epoch": 0.72, + "grad_norm": 5.421473503112793, + "learning_rate": 0.00019778082463468085, + "loss": 3.8328, + "step": 482 + }, + { + "epoch": 0.72, + "grad_norm": 8.712848663330078, + "learning_rate": 0.00019777098752034807, + "loss": 3.826, + "step": 483 + }, + { + "epoch": 0.72, + "grad_norm": 4.683780670166016, + "learning_rate": 0.00019776112889705334, + "loss": 3.7366, + "step": 484 + }, + { + "epoch": 0.72, + "grad_norm": 5.190871715545654, + "learning_rate": 0.00019775124876696536, + "loss": 3.7527, + "step": 485 + }, + { + "epoch": 0.72, + "grad_norm": 8.225262641906738, + "learning_rate": 0.00019774134713225778, + "loss": 3.8112, + "step": 486 + }, + { + "epoch": 0.72, + "grad_norm": 2.200317859649658, + "learning_rate": 0.00019773142399510883, + "loss": 3.7121, + "step": 487 + }, + { + "epoch": 0.72, + "grad_norm": 9.15087604522705, + "learning_rate": 0.0001977214793577016, + "loss": 3.7827, + "step": 488 + }, + { + "epoch": 0.73, + "grad_norm": 11.307198524475098, + "learning_rate": 0.0001977115132222238, + "loss": 3.7558, + "step": 489 + }, + { + "epoch": 0.73, + "grad_norm": 8.0049409866333, + "learning_rate": 0.00019770152559086794, + "loss": 3.7133, + "step": 490 + }, + { + "epoch": 0.73, + "grad_norm": 4.207780361175537, + "learning_rate": 0.0001976915164658312, + "loss": 3.7517, + "step": 491 + }, + { + "epoch": 0.73, + "grad_norm": 9.044132232666016, + "learning_rate": 0.00019768148584931554, + "loss": 3.7111, + "step": 492 + }, + { + "epoch": 0.73, + "grad_norm": 6.856382846832275, + "learning_rate": 0.00019767143374352762, + "loss": 3.7271, + "step": 493 + }, + { + "epoch": 0.73, + "grad_norm": 3.427558183670044, + "learning_rate": 0.00019766136015067886, + "loss": 3.7357, + "step": 494 + }, + { + "epoch": 0.74, + "grad_norm": 4.354124069213867, + "learning_rate": 0.00019765126507298534, + "loss": 3.7107, + "step": 495 + }, + { + "epoch": 0.74, + "grad_norm": 3.53426194190979, + "learning_rate": 0.00019764114851266796, + "loss": 3.7585, + "step": 496 + }, + { + "epoch": 0.74, + "grad_norm": 2.1305716037750244, + "learning_rate": 0.00019763101047195225, + "loss": 3.688, + "step": 497 + }, + { + "epoch": 0.74, + "grad_norm": 108.82621002197266, + "learning_rate": 0.00019762085095306856, + "loss": 3.6435, + "step": 498 + }, + { + "epoch": 0.74, + "grad_norm": 8.900164604187012, + "learning_rate": 0.00019761066995825183, + "loss": 3.7853, + "step": 499 + }, + { + "epoch": 0.74, + "grad_norm": 8.347526550292969, + "learning_rate": 0.0001976004674897419, + "loss": 3.7845, + "step": 500 + }, + { + "epoch": 0.74, + "grad_norm": 3.418429374694824, + "learning_rate": 0.00019759024354978322, + "loss": 3.7529, + "step": 501 + }, + { + "epoch": 0.75, + "grad_norm": 14.40571403503418, + "learning_rate": 0.00019757999814062495, + "loss": 3.795, + "step": 502 + }, + { + "epoch": 0.75, + "grad_norm": 21.624300003051758, + "learning_rate": 0.00019756973126452108, + "loss": 4.0035, + "step": 503 + }, + { + "epoch": 0.75, + "grad_norm": 14.849161148071289, + "learning_rate": 0.00019755944292373017, + "loss": 3.9246, + "step": 504 + }, + { + "epoch": 0.75, + "grad_norm": 7.422677993774414, + "learning_rate": 0.00019754913312051568, + "loss": 3.9148, + "step": 505 + }, + { + "epoch": 0.75, + "grad_norm": 11.331315994262695, + "learning_rate": 0.00019753880185714563, + "loss": 3.7997, + "step": 506 + }, + { + "epoch": 0.75, + "grad_norm": 34.16909408569336, + "learning_rate": 0.00019752844913589285, + "loss": 3.7348, + "step": 507 + }, + { + "epoch": 0.75, + "eval_loss": 3.593658447265625, + "eval_runtime": 1156.954, + "eval_samples_per_second": 14.816, + "eval_steps_per_second": 0.927, + "step": 507 + }, + { + "epoch": 0.75, + "grad_norm": 1562.1485595703125, + "learning_rate": 0.00019751807495903484, + "loss": 3.8149, + "step": 508 + }, + { + "epoch": 0.76, + "grad_norm": 12.771462440490723, + "learning_rate": 0.0001975076793288539, + "loss": 3.9449, + "step": 509 + }, + { + "epoch": 0.76, + "grad_norm": 7.921811580657959, + "learning_rate": 0.00019749726224763697, + "loss": 3.9061, + "step": 510 + }, + { + "epoch": 0.76, + "grad_norm": 4.624162197113037, + "learning_rate": 0.00019748682371767575, + "loss": 3.8572, + "step": 511 + }, + { + "epoch": 0.76, + "grad_norm": 33.93694305419922, + "learning_rate": 0.00019747636374126663, + "loss": 3.8013, + "step": 512 + }, + { + "epoch": 0.76, + "grad_norm": 67.59894561767578, + "learning_rate": 0.00019746588232071076, + "loss": 3.9281, + "step": 513 + }, + { + "epoch": 0.76, + "grad_norm": 18.626140594482422, + "learning_rate": 0.00019745537945831397, + "loss": 4.0322, + "step": 514 + }, + { + "epoch": 0.76, + "grad_norm": 9.813076972961426, + "learning_rate": 0.00019744485515638677, + "loss": 4.0781, + "step": 515 + }, + { + "epoch": 0.77, + "grad_norm": 12.949821472167969, + "learning_rate": 0.00019743430941724453, + "loss": 4.0615, + "step": 516 + }, + { + "epoch": 0.77, + "grad_norm": 9.89267635345459, + "learning_rate": 0.00019742374224320717, + "loss": 3.9752, + "step": 517 + }, + { + "epoch": 0.77, + "grad_norm": 4.396897792816162, + "learning_rate": 0.00019741315363659942, + "loss": 3.768, + "step": 518 + }, + { + "epoch": 0.77, + "grad_norm": 3.6423070430755615, + "learning_rate": 0.00019740254359975072, + "loss": 3.7554, + "step": 519 + }, + { + "epoch": 0.77, + "grad_norm": 5.352052688598633, + "learning_rate": 0.0001973919121349952, + "loss": 3.8254, + "step": 520 + }, + { + "epoch": 0.77, + "grad_norm": 4.971579074859619, + "learning_rate": 0.00019738125924467168, + "loss": 3.9223, + "step": 521 + }, + { + "epoch": 0.78, + "grad_norm": 3.2794764041900635, + "learning_rate": 0.00019737058493112376, + "loss": 3.7504, + "step": 522 + }, + { + "epoch": 0.78, + "grad_norm": 3.714524269104004, + "learning_rate": 0.00019735988919669968, + "loss": 3.7884, + "step": 523 + }, + { + "epoch": 0.78, + "grad_norm": 5.291292190551758, + "learning_rate": 0.0001973491720437525, + "loss": 3.8398, + "step": 524 + }, + { + "epoch": 0.78, + "grad_norm": 4.59107780456543, + "learning_rate": 0.00019733843347463984, + "loss": 3.9075, + "step": 525 + }, + { + "epoch": 0.78, + "grad_norm": 6.6806793212890625, + "learning_rate": 0.00019732767349172413, + "loss": 3.7432, + "step": 526 + }, + { + "epoch": 0.78, + "grad_norm": 48.79610824584961, + "learning_rate": 0.00019731689209737257, + "loss": 3.7873, + "step": 527 + }, + { + "epoch": 0.78, + "grad_norm": 7.798208236694336, + "learning_rate": 0.0001973060892939569, + "loss": 3.8176, + "step": 528 + }, + { + "epoch": 0.79, + "grad_norm": 7.387636661529541, + "learning_rate": 0.00019729526508385369, + "loss": 3.892, + "step": 529 + }, + { + "epoch": 0.79, + "grad_norm": 3.599566698074341, + "learning_rate": 0.00019728441946944422, + "loss": 3.754, + "step": 530 + }, + { + "epoch": 0.79, + "grad_norm": 6.45220947265625, + "learning_rate": 0.00019727355245311446, + "loss": 3.7462, + "step": 531 + }, + { + "epoch": 0.79, + "grad_norm": 6.4421563148498535, + "learning_rate": 0.00019726266403725503, + "loss": 3.8014, + "step": 532 + }, + { + "epoch": 0.79, + "grad_norm": 9.638814926147461, + "learning_rate": 0.00019725175422426135, + "loss": 3.9211, + "step": 533 + }, + { + "epoch": 0.79, + "grad_norm": 4.907318592071533, + "learning_rate": 0.00019724082301653345, + "loss": 3.8431, + "step": 534 + }, + { + "epoch": 0.79, + "grad_norm": 3.3494908809661865, + "learning_rate": 0.00019722987041647622, + "loss": 3.8407, + "step": 535 + }, + { + "epoch": 0.8, + "grad_norm": 2.774167060852051, + "learning_rate": 0.00019721889642649912, + "loss": 3.7459, + "step": 536 + }, + { + "epoch": 0.8, + "grad_norm": 4.917426586151123, + "learning_rate": 0.0001972079010490163, + "loss": 3.8871, + "step": 537 + }, + { + "epoch": 0.8, + "grad_norm": 2.627586841583252, + "learning_rate": 0.0001971968842864467, + "loss": 3.7812, + "step": 538 + }, + { + "epoch": 0.8, + "grad_norm": 6.549022197723389, + "learning_rate": 0.00019718584614121396, + "loss": 3.7288, + "step": 539 + }, + { + "epoch": 0.8, + "grad_norm": 13.424997329711914, + "learning_rate": 0.00019717478661574638, + "loss": 3.7931, + "step": 540 + }, + { + "epoch": 0.8, + "grad_norm": 10.896014213562012, + "learning_rate": 0.00019716370571247696, + "loss": 3.8668, + "step": 541 + }, + { + "epoch": 0.8, + "grad_norm": 7.297813415527344, + "learning_rate": 0.00019715260343384347, + "loss": 3.8884, + "step": 542 + }, + { + "epoch": 0.81, + "grad_norm": 3.8043060302734375, + "learning_rate": 0.00019714147978228832, + "loss": 3.8604, + "step": 543 + }, + { + "epoch": 0.81, + "grad_norm": 25.301651000976562, + "learning_rate": 0.0001971303347602586, + "loss": 3.8794, + "step": 544 + }, + { + "epoch": 0.81, + "grad_norm": 215.01597595214844, + "learning_rate": 0.0001971191683702062, + "loss": 3.9304, + "step": 545 + }, + { + "epoch": 0.81, + "grad_norm": 253.02139282226562, + "learning_rate": 0.0001971079806145876, + "loss": 4.3177, + "step": 546 + }, + { + "epoch": 0.81, + "grad_norm": 41.9672966003418, + "learning_rate": 0.00019709677149586409, + "loss": 4.7431, + "step": 547 + }, + { + "epoch": 0.81, + "grad_norm": 34.384029388427734, + "learning_rate": 0.0001970855410165015, + "loss": 5.1589, + "step": 548 + }, + { + "epoch": 0.82, + "grad_norm": 18.074525833129883, + "learning_rate": 0.00019707428917897058, + "loss": 5.4518, + "step": 549 + }, + { + "epoch": 0.82, + "grad_norm": 10.624735832214355, + "learning_rate": 0.0001970630159857466, + "loss": 5.0911, + "step": 550 + }, + { + "epoch": 0.82, + "grad_norm": 15.860513687133789, + "learning_rate": 0.00019705172143930953, + "loss": 4.8221, + "step": 551 + }, + { + "epoch": 0.82, + "grad_norm": 9.580286979675293, + "learning_rate": 0.00019704040554214422, + "loss": 4.7721, + "step": 552 + }, + { + "epoch": 0.82, + "grad_norm": 16.33072280883789, + "learning_rate": 0.00019702906829674002, + "loss": 4.7985, + "step": 553 + }, + { + "epoch": 0.82, + "grad_norm": 15.135660171508789, + "learning_rate": 0.00019701770970559105, + "loss": 4.7531, + "step": 554 + }, + { + "epoch": 0.82, + "grad_norm": 15.646889686584473, + "learning_rate": 0.0001970063297711961, + "loss": 4.6453, + "step": 555 + }, + { + "epoch": 0.83, + "grad_norm": 13.652432441711426, + "learning_rate": 0.00019699492849605874, + "loss": 4.671, + "step": 556 + }, + { + "epoch": 0.83, + "grad_norm": 18.923864364624023, + "learning_rate": 0.00019698350588268713, + "loss": 4.7306, + "step": 557 + }, + { + "epoch": 0.83, + "grad_norm": 15.4611177444458, + "learning_rate": 0.0001969720619335942, + "loss": 4.6082, + "step": 558 + }, + { + "epoch": 0.83, + "grad_norm": 6.054503440856934, + "learning_rate": 0.0001969605966512975, + "loss": 4.4091, + "step": 559 + }, + { + "epoch": 0.83, + "grad_norm": 11.314775466918945, + "learning_rate": 0.00019694911003831933, + "loss": 4.1236, + "step": 560 + }, + { + "epoch": 0.83, + "grad_norm": 12.968482971191406, + "learning_rate": 0.0001969376020971867, + "loss": 4.023, + "step": 561 + }, + { + "epoch": 0.83, + "grad_norm": 8.118325233459473, + "learning_rate": 0.00019692607283043124, + "loss": 3.9331, + "step": 562 + }, + { + "epoch": 0.84, + "grad_norm": 4.455306529998779, + "learning_rate": 0.00019691452224058933, + "loss": 3.9195, + "step": 563 + }, + { + "epoch": 0.84, + "grad_norm": 5.133811950683594, + "learning_rate": 0.000196902950330202, + "loss": 3.8438, + "step": 564 + }, + { + "epoch": 0.84, + "grad_norm": 4.062127590179443, + "learning_rate": 0.00019689135710181503, + "loss": 3.7435, + "step": 565 + }, + { + "epoch": 0.84, + "grad_norm": 4.880388259887695, + "learning_rate": 0.00019687974255797885, + "loss": 3.7828, + "step": 566 + }, + { + "epoch": 0.84, + "grad_norm": 5.3774027824401855, + "learning_rate": 0.00019686810670124856, + "loss": 3.7977, + "step": 567 + }, + { + "epoch": 0.84, + "grad_norm": 5.431920528411865, + "learning_rate": 0.00019685644953418397, + "loss": 3.8113, + "step": 568 + }, + { + "epoch": 0.84, + "grad_norm": 3.2028257846832275, + "learning_rate": 0.00019684477105934956, + "loss": 3.7359, + "step": 569 + }, + { + "epoch": 0.85, + "grad_norm": 3.04168701171875, + "learning_rate": 0.00019683307127931458, + "loss": 3.6924, + "step": 570 + }, + { + "epoch": 0.85, + "grad_norm": 3.9302849769592285, + "learning_rate": 0.00019682135019665286, + "loss": 3.7543, + "step": 571 + }, + { + "epoch": 0.85, + "grad_norm": 3.5611188411712646, + "learning_rate": 0.00019680960781394297, + "loss": 3.751, + "step": 572 + }, + { + "epoch": 0.85, + "grad_norm": 2.1138036251068115, + "learning_rate": 0.00019679784413376811, + "loss": 3.7196, + "step": 573 + }, + { + "epoch": 0.85, + "grad_norm": 3.7343742847442627, + "learning_rate": 0.0001967860591587163, + "loss": 3.6888, + "step": 574 + }, + { + "epoch": 0.85, + "grad_norm": 3.8373682498931885, + "learning_rate": 0.00019677425289138007, + "loss": 3.6819, + "step": 575 + }, + { + "epoch": 0.86, + "grad_norm": 3.580359935760498, + "learning_rate": 0.00019676242533435678, + "loss": 3.662, + "step": 576 + }, + { + "epoch": 0.86, + "grad_norm": 3.4737274646759033, + "learning_rate": 0.00019675057649024837, + "loss": 3.6293, + "step": 577 + }, + { + "epoch": 0.86, + "grad_norm": 2.3330490589141846, + "learning_rate": 0.00019673870636166156, + "loss": 3.6554, + "step": 578 + }, + { + "epoch": 0.86, + "grad_norm": 4.6455512046813965, + "learning_rate": 0.00019672681495120763, + "loss": 3.6991, + "step": 579 + }, + { + "epoch": 0.86, + "grad_norm": 2.803642988204956, + "learning_rate": 0.00019671490226150264, + "loss": 3.6473, + "step": 580 + }, + { + "epoch": 0.86, + "grad_norm": 5.52310848236084, + "learning_rate": 0.00019670296829516732, + "loss": 3.6435, + "step": 581 + }, + { + "epoch": 0.86, + "grad_norm": 5.066474914550781, + "learning_rate": 0.00019669101305482705, + "loss": 3.6769, + "step": 582 + }, + { + "epoch": 0.87, + "grad_norm": 2.860041379928589, + "learning_rate": 0.00019667903654311194, + "loss": 3.6205, + "step": 583 + }, + { + "epoch": 0.87, + "grad_norm": 2.413363456726074, + "learning_rate": 0.00019666703876265666, + "loss": 3.6287, + "step": 584 + }, + { + "epoch": 0.87, + "grad_norm": 4.433047771453857, + "learning_rate": 0.0001966550197161007, + "loss": 3.6912, + "step": 585 + }, + { + "epoch": 0.87, + "grad_norm": 117.41354370117188, + "learning_rate": 0.00019664297940608812, + "loss": 3.6996, + "step": 586 + }, + { + "epoch": 0.87, + "grad_norm": 2.949551582336426, + "learning_rate": 0.0001966309178352678, + "loss": 3.6726, + "step": 587 + }, + { + "epoch": 0.87, + "grad_norm": 6.152908802032471, + "learning_rate": 0.0001966188350062931, + "loss": 3.6525, + "step": 588 + }, + { + "epoch": 0.87, + "grad_norm": 6.42512321472168, + "learning_rate": 0.00019660673092182225, + "loss": 3.6739, + "step": 589 + }, + { + "epoch": 0.88, + "grad_norm": 3.4023945331573486, + "learning_rate": 0.000196594605584518, + "loss": 3.6757, + "step": 590 + }, + { + "epoch": 0.88, + "grad_norm": 4.428458213806152, + "learning_rate": 0.00019658245899704789, + "loss": 3.6438, + "step": 591 + }, + { + "epoch": 0.88, + "grad_norm": 4.402431488037109, + "learning_rate": 0.00019657029116208405, + "loss": 3.6902, + "step": 592 + }, + { + "epoch": 0.88, + "grad_norm": 4.742648124694824, + "learning_rate": 0.00019655810208230333, + "loss": 3.5998, + "step": 593 + }, + { + "epoch": 0.88, + "grad_norm": 1.4105639457702637, + "learning_rate": 0.00019654589176038728, + "loss": 3.5836, + "step": 594 + }, + { + "epoch": 0.88, + "grad_norm": 5.09644079208374, + "learning_rate": 0.00019653366019902206, + "loss": 3.616, + "step": 595 + }, + { + "epoch": 0.89, + "grad_norm": 5.025322914123535, + "learning_rate": 0.00019652140740089854, + "loss": 3.6783, + "step": 596 + }, + { + "epoch": 0.89, + "grad_norm": 4.536464691162109, + "learning_rate": 0.00019650913336871222, + "loss": 3.6274, + "step": 597 + }, + { + "epoch": 0.89, + "grad_norm": 3.1868488788604736, + "learning_rate": 0.00019649683810516336, + "loss": 3.6113, + "step": 598 + }, + { + "epoch": 0.89, + "grad_norm": 4.735551834106445, + "learning_rate": 0.00019648452161295682, + "loss": 3.5986, + "step": 599 + }, + { + "epoch": 0.89, + "grad_norm": 4.411576747894287, + "learning_rate": 0.0001964721838948021, + "loss": 3.5865, + "step": 600 + }, + { + "epoch": 0.89, + "grad_norm": 2.665224075317383, + "learning_rate": 0.00019645982495341348, + "loss": 3.5546, + "step": 601 + }, + { + "epoch": 0.89, + "grad_norm": 4.546963691711426, + "learning_rate": 0.00019644744479150981, + "loss": 3.6332, + "step": 602 + }, + { + "epoch": 0.9, + "grad_norm": 1.664005160331726, + "learning_rate": 0.00019643504341181466, + "loss": 3.6299, + "step": 603 + }, + { + "epoch": 0.9, + "grad_norm": 2.6972901821136475, + "learning_rate": 0.0001964226208170562, + "loss": 3.6158, + "step": 604 + }, + { + "epoch": 0.9, + "grad_norm": 3.175882339477539, + "learning_rate": 0.00019641017700996736, + "loss": 3.5895, + "step": 605 + }, + { + "epoch": 0.9, + "grad_norm": 3.3170359134674072, + "learning_rate": 0.00019639771199328574, + "loss": 3.5989, + "step": 606 + }, + { + "epoch": 0.9, + "grad_norm": 3.245979070663452, + "learning_rate": 0.00019638522576975346, + "loss": 3.5763, + "step": 607 + }, + { + "epoch": 0.9, + "grad_norm": 1.0223109722137451, + "learning_rate": 0.00019637271834211748, + "loss": 3.5668, + "step": 608 + }, + { + "epoch": 0.9, + "grad_norm": 1.9072734117507935, + "learning_rate": 0.00019636018971312932, + "loss": 3.5738, + "step": 609 + }, + { + "epoch": 0.91, + "grad_norm": 2.925701141357422, + "learning_rate": 0.00019634763988554522, + "loss": 3.616, + "step": 610 + }, + { + "epoch": 0.91, + "grad_norm": 2.6939620971679688, + "learning_rate": 0.00019633506886212603, + "loss": 3.5764, + "step": 611 + }, + { + "epoch": 0.91, + "grad_norm": 2.020845413208008, + "learning_rate": 0.0001963224766456373, + "loss": 3.5868, + "step": 612 + }, + { + "epoch": 0.91, + "grad_norm": 3.805741548538208, + "learning_rate": 0.0001963098632388492, + "loss": 3.5979, + "step": 613 + }, + { + "epoch": 0.91, + "grad_norm": 1.6603795289993286, + "learning_rate": 0.00019629722864453663, + "loss": 3.5792, + "step": 614 + }, + { + "epoch": 0.91, + "grad_norm": 8.938972473144531, + "learning_rate": 0.00019628457286547914, + "loss": 3.5978, + "step": 615 + }, + { + "epoch": 0.91, + "grad_norm": 9.437355041503906, + "learning_rate": 0.0001962718959044609, + "loss": 3.6115, + "step": 616 + }, + { + "epoch": 0.92, + "grad_norm": 1.7519173622131348, + "learning_rate": 0.0001962591977642707, + "loss": 3.5745, + "step": 617 + }, + { + "epoch": 0.92, + "grad_norm": 8.598402976989746, + "learning_rate": 0.00019624647844770213, + "loss": 3.5939, + "step": 618 + }, + { + "epoch": 0.92, + "grad_norm": 11.709813117980957, + "learning_rate": 0.0001962337379575533, + "loss": 3.7499, + "step": 619 + }, + { + "epoch": 0.92, + "grad_norm": 8.366117477416992, + "learning_rate": 0.00019622097629662708, + "loss": 3.5727, + "step": 620 + }, + { + "epoch": 0.92, + "grad_norm": 2.4304378032684326, + "learning_rate": 0.0001962081934677309, + "loss": 3.6257, + "step": 621 + }, + { + "epoch": 0.92, + "grad_norm": 6.7245025634765625, + "learning_rate": 0.00019619538947367692, + "loss": 3.5817, + "step": 622 + }, + { + "epoch": 0.93, + "grad_norm": 2.964221239089966, + "learning_rate": 0.00019618256431728194, + "loss": 3.5903, + "step": 623 + }, + { + "epoch": 0.93, + "grad_norm": 5.846991539001465, + "learning_rate": 0.00019616971800136734, + "loss": 3.5198, + "step": 624 + }, + { + "epoch": 0.93, + "grad_norm": 6.497872829437256, + "learning_rate": 0.00019615685052875933, + "loss": 3.5556, + "step": 625 + }, + { + "epoch": 0.93, + "grad_norm": 1.711037039756775, + "learning_rate": 0.00019614396190228864, + "loss": 3.5108, + "step": 626 + }, + { + "epoch": 0.93, + "grad_norm": 9.816991806030273, + "learning_rate": 0.0001961310521247906, + "loss": 3.5854, + "step": 627 + }, + { + "epoch": 0.93, + "grad_norm": 12.794452667236328, + "learning_rate": 0.00019611812119910537, + "loss": 3.6549, + "step": 628 + }, + { + "epoch": 0.93, + "grad_norm": 6.7332916259765625, + "learning_rate": 0.0001961051691280776, + "loss": 3.6605, + "step": 629 + }, + { + "epoch": 0.94, + "grad_norm": 6.012147426605225, + "learning_rate": 0.00019609219591455675, + "loss": 3.6014, + "step": 630 + }, + { + "epoch": 0.94, + "grad_norm": 6.8346381187438965, + "learning_rate": 0.00019607920156139671, + "loss": 3.5713, + "step": 631 + }, + { + "epoch": 0.94, + "grad_norm": 3.088141679763794, + "learning_rate": 0.0001960661860714563, + "loss": 3.5767, + "step": 632 + }, + { + "epoch": 0.94, + "grad_norm": 8.472504615783691, + "learning_rate": 0.0001960531494475987, + "loss": 3.6236, + "step": 633 + }, + { + "epoch": 0.94, + "grad_norm": 9.718921661376953, + "learning_rate": 0.00019604009169269193, + "loss": 3.5673, + "step": 634 + }, + { + "epoch": 0.94, + "grad_norm": 2.083310127258301, + "learning_rate": 0.00019602701280960866, + "loss": 3.5786, + "step": 635 + }, + { + "epoch": 0.94, + "grad_norm": 8.658992767333984, + "learning_rate": 0.0001960139128012261, + "loss": 3.6109, + "step": 636 + }, + { + "epoch": 0.95, + "grad_norm": 10.503134727478027, + "learning_rate": 0.0001960007916704262, + "loss": 3.6035, + "step": 637 + }, + { + "epoch": 0.95, + "grad_norm": 7.250131130218506, + "learning_rate": 0.00019598764942009548, + "loss": 3.6211, + "step": 638 + }, + { + "epoch": 0.95, + "grad_norm": 4.671655178070068, + "learning_rate": 0.00019597448605312515, + "loss": 3.5175, + "step": 639 + }, + { + "epoch": 0.95, + "grad_norm": 9.631921768188477, + "learning_rate": 0.00019596130157241112, + "loss": 3.6144, + "step": 640 + }, + { + "epoch": 0.95, + "grad_norm": 5.835899353027344, + "learning_rate": 0.00019594809598085381, + "loss": 3.5662, + "step": 641 + }, + { + "epoch": 0.95, + "grad_norm": 5.0995893478393555, + "learning_rate": 0.00019593486928135841, + "loss": 3.5438, + "step": 642 + }, + { + "epoch": 0.95, + "grad_norm": 6.497475624084473, + "learning_rate": 0.00019592162147683473, + "loss": 3.536, + "step": 643 + }, + { + "epoch": 0.96, + "grad_norm": 2.704535722732544, + "learning_rate": 0.00019590835257019714, + "loss": 3.5268, + "step": 644 + }, + { + "epoch": 0.96, + "grad_norm": 8.847089767456055, + "learning_rate": 0.00019589506256436473, + "loss": 3.6095, + "step": 645 + }, + { + "epoch": 0.96, + "grad_norm": 10.879183769226074, + "learning_rate": 0.00019588175146226122, + "loss": 3.657, + "step": 646 + }, + { + "epoch": 0.96, + "grad_norm": 4.468491077423096, + "learning_rate": 0.00019586841926681495, + "loss": 3.5589, + "step": 647 + }, + { + "epoch": 0.96, + "grad_norm": 6.55201530456543, + "learning_rate": 0.00019585506598095897, + "loss": 3.5587, + "step": 648 + }, + { + "epoch": 0.96, + "grad_norm": 8.522590637207031, + "learning_rate": 0.0001958416916076308, + "loss": 3.5354, + "step": 649 + }, + { + "epoch": 0.97, + "grad_norm": 4.8020806312561035, + "learning_rate": 0.00019582829614977283, + "loss": 3.506, + "step": 650 + }, + { + "epoch": 0.97, + "grad_norm": 7.987406253814697, + "learning_rate": 0.00019581487961033188, + "loss": 3.5615, + "step": 651 + }, + { + "epoch": 0.97, + "grad_norm": 11.70871353149414, + "learning_rate": 0.00019580144199225953, + "loss": 3.5993, + "step": 652 + }, + { + "epoch": 0.97, + "grad_norm": 4.69012975692749, + "learning_rate": 0.00019578798329851203, + "loss": 3.542, + "step": 653 + }, + { + "epoch": 0.97, + "grad_norm": 6.301438331604004, + "learning_rate": 0.0001957745035320501, + "loss": 3.5313, + "step": 654 + }, + { + "epoch": 0.97, + "grad_norm": 8.368910789489746, + "learning_rate": 0.0001957610026958393, + "loss": 3.5855, + "step": 655 + }, + { + "epoch": 0.97, + "grad_norm": 5.793453693389893, + "learning_rate": 0.00019574748079284962, + "loss": 3.5403, + "step": 656 + }, + { + "epoch": 0.98, + "grad_norm": 4.990311622619629, + "learning_rate": 0.00019573393782605582, + "loss": 3.5196, + "step": 657 + }, + { + "epoch": 0.98, + "grad_norm": 9.033957481384277, + "learning_rate": 0.0001957203737984373, + "loss": 3.528, + "step": 658 + }, + { + "epoch": 0.98, + "grad_norm": 3.766148805618286, + "learning_rate": 0.00019570678871297804, + "loss": 3.5015, + "step": 659 + }, + { + "epoch": 0.98, + "grad_norm": 7.3363518714904785, + "learning_rate": 0.00019569318257266666, + "loss": 3.5311, + "step": 660 + }, + { + "epoch": 0.98, + "grad_norm": 9.119929313659668, + "learning_rate": 0.00019567955538049643, + "loss": 3.5767, + "step": 661 + }, + { + "epoch": 0.98, + "grad_norm": 5.856534957885742, + "learning_rate": 0.00019566590713946522, + "loss": 3.5836, + "step": 662 + }, + { + "epoch": 0.98, + "grad_norm": 4.941824436187744, + "learning_rate": 0.00019565223785257557, + "loss": 3.5366, + "step": 663 + }, + { + "epoch": 0.99, + "grad_norm": 8.7649507522583, + "learning_rate": 0.00019563854752283465, + "loss": 3.5562, + "step": 664 + }, + { + "epoch": 0.99, + "grad_norm": 3.0613012313842773, + "learning_rate": 0.00019562483615325418, + "loss": 3.5715, + "step": 665 + }, + { + "epoch": 0.99, + "grad_norm": 6.858389377593994, + "learning_rate": 0.00019561110374685062, + "loss": 3.4948, + "step": 666 + }, + { + "epoch": 0.99, + "grad_norm": 8.862404823303223, + "learning_rate": 0.00019559735030664496, + "loss": 3.5523, + "step": 667 + }, + { + "epoch": 0.99, + "grad_norm": 5.061213970184326, + "learning_rate": 0.00019558357583566297, + "loss": 3.4475, + "step": 668 + }, + { + "epoch": 0.99, + "grad_norm": 7.312166690826416, + "learning_rate": 0.0001955697803369348, + "loss": 3.5274, + "step": 669 + }, + { + "epoch": 0.99, + "grad_norm": 11.606597900390625, + "learning_rate": 0.00019555596381349547, + "loss": 3.5426, + "step": 670 + }, + { + "epoch": 1.0, + "grad_norm": 5.651201248168945, + "learning_rate": 0.0001955421262683845, + "loss": 3.533, + "step": 671 + }, + { + "epoch": 1.0, + "grad_norm": 5.392771244049072, + "learning_rate": 0.00019552826770464598, + "loss": 3.4873, + "step": 672 + }, + { + "epoch": 1.0, + "grad_norm": 7.194543361663818, + "learning_rate": 0.00019551438812532882, + "loss": 3.5313, + "step": 673 + }, + { + "epoch": 1.0, + "grad_norm": 3.8677308559417725, + "learning_rate": 0.00019550048753348638, + "loss": 3.5539, + "step": 674 + }, + { + "epoch": 1.0, + "grad_norm": 8.319040298461914, + "learning_rate": 0.0001954865659321767, + "loss": 3.5204, + "step": 675 + }, + { + "epoch": 1.0, + "grad_norm": 12.673453330993652, + "learning_rate": 0.00019547262332446241, + "loss": 3.4573, + "step": 676 + }, + { + "epoch": 1.0, + "eval_loss": 3.3125858306884766, + "eval_runtime": 1160.3441, + "eval_samples_per_second": 14.773, + "eval_steps_per_second": 0.924, + "step": 676 + }, + { + "epoch": 1.01, + "grad_norm": 6.508894920349121, + "learning_rate": 0.00019545865971341085, + "loss": 3.5108, + "step": 677 + }, + { + "epoch": 1.01, + "grad_norm": 5.700242519378662, + "learning_rate": 0.00019544467510209388, + "loss": 3.4928, + "step": 678 + }, + { + "epoch": 1.01, + "grad_norm": 7.889042854309082, + "learning_rate": 0.00019543066949358803, + "loss": 3.5309, + "step": 679 + }, + { + "epoch": 1.01, + "grad_norm": 4.5509724617004395, + "learning_rate": 0.0001954166428909744, + "loss": 3.5496, + "step": 680 + }, + { + "epoch": 1.01, + "grad_norm": 7.286065578460693, + "learning_rate": 0.00019540259529733883, + "loss": 3.5685, + "step": 681 + }, + { + "epoch": 1.01, + "grad_norm": 10.92095947265625, + "learning_rate": 0.00019538852671577165, + "loss": 3.5438, + "step": 682 + }, + { + "epoch": 1.01, + "grad_norm": 4.9438557624816895, + "learning_rate": 0.00019537443714936787, + "loss": 3.4908, + "step": 683 + }, + { + "epoch": 1.0, + "grad_norm": 6.148499011993408, + "learning_rate": 0.00019536032660122707, + "loss": 3.5438, + "step": 684 + }, + { + "epoch": 1.0, + "grad_norm": 8.24433708190918, + "learning_rate": 0.00019534619507445352, + "loss": 3.6016, + "step": 685 + }, + { + "epoch": 1.0, + "grad_norm": 5.07331657409668, + "learning_rate": 0.000195332042572156, + "loss": 3.6093, + "step": 686 + }, + { + "epoch": 1.01, + "grad_norm": 7.187658309936523, + "learning_rate": 0.00019531786909744803, + "loss": 3.5025, + "step": 687 + }, + { + "epoch": 1.01, + "grad_norm": 10.835110664367676, + "learning_rate": 0.0001953036746534476, + "loss": 3.5561, + "step": 688 + }, + { + "epoch": 1.01, + "grad_norm": 4.704497814178467, + "learning_rate": 0.00019528945924327751, + "loss": 3.5617, + "step": 689 + }, + { + "epoch": 1.01, + "grad_norm": 6.398228168487549, + "learning_rate": 0.00019527522287006495, + "loss": 3.5302, + "step": 690 + }, + { + "epoch": 1.01, + "grad_norm": 8.401050567626953, + "learning_rate": 0.00019526096553694187, + "loss": 3.5882, + "step": 691 + }, + { + "epoch": 1.01, + "grad_norm": 4.914839267730713, + "learning_rate": 0.00019524668724704478, + "loss": 3.4931, + "step": 692 + }, + { + "epoch": 1.01, + "grad_norm": 6.374270915985107, + "learning_rate": 0.00019523238800351483, + "loss": 3.4807, + "step": 693 + }, + { + "epoch": 1.02, + "grad_norm": 9.898938179016113, + "learning_rate": 0.0001952180678094977, + "loss": 3.5138, + "step": 694 + }, + { + "epoch": 1.02, + "grad_norm": 4.583561897277832, + "learning_rate": 0.00019520372666814378, + "loss": 3.4739, + "step": 695 + }, + { + "epoch": 1.02, + "grad_norm": 6.307168483734131, + "learning_rate": 0.00019518936458260807, + "loss": 3.5473, + "step": 696 + }, + { + "epoch": 1.02, + "grad_norm": 8.191679000854492, + "learning_rate": 0.00019517498155605003, + "loss": 3.5002, + "step": 697 + }, + { + "epoch": 1.02, + "grad_norm": 5.078609466552734, + "learning_rate": 0.0001951605775916339, + "loss": 3.4731, + "step": 698 + }, + { + "epoch": 1.02, + "grad_norm": 6.6793060302734375, + "learning_rate": 0.00019514615269252845, + "loss": 3.5543, + "step": 699 + }, + { + "epoch": 1.02, + "grad_norm": 10.510246276855469, + "learning_rate": 0.00019513170686190706, + "loss": 3.5589, + "step": 700 + }, + { + "epoch": 1.03, + "grad_norm": 4.769564151763916, + "learning_rate": 0.00019511724010294766, + "loss": 3.4587, + "step": 701 + }, + { + "epoch": 1.03, + "grad_norm": 6.157220363616943, + "learning_rate": 0.00019510275241883292, + "loss": 3.4658, + "step": 702 + }, + { + "epoch": 1.03, + "grad_norm": 8.317898750305176, + "learning_rate": 0.00019508824381274996, + "loss": 3.5147, + "step": 703 + }, + { + "epoch": 1.03, + "grad_norm": 4.906617164611816, + "learning_rate": 0.00019507371428789064, + "loss": 3.5319, + "step": 704 + }, + { + "epoch": 1.03, + "grad_norm": 6.44058084487915, + "learning_rate": 0.00019505916384745134, + "loss": 3.5268, + "step": 705 + }, + { + "epoch": 1.03, + "grad_norm": 10.473220825195312, + "learning_rate": 0.00019504459249463304, + "loss": 3.535, + "step": 706 + }, + { + "epoch": 1.04, + "grad_norm": 4.5666656494140625, + "learning_rate": 0.00019503000023264136, + "loss": 3.4976, + "step": 707 + }, + { + "epoch": 1.04, + "grad_norm": 6.2709221839904785, + "learning_rate": 0.0001950153870646865, + "loss": 3.5246, + "step": 708 + }, + { + "epoch": 1.04, + "grad_norm": 8.397104263305664, + "learning_rate": 0.00019500075299398324, + "loss": 3.5369, + "step": 709 + }, + { + "epoch": 1.04, + "grad_norm": 4.958198547363281, + "learning_rate": 0.00019498609802375097, + "loss": 3.4644, + "step": 710 + }, + { + "epoch": 1.04, + "grad_norm": 6.521581172943115, + "learning_rate": 0.00019497142215721373, + "loss": 3.4931, + "step": 711 + }, + { + "epoch": 1.04, + "grad_norm": 10.69038200378418, + "learning_rate": 0.00019495672539760007, + "loss": 3.4746, + "step": 712 + }, + { + "epoch": 1.04, + "grad_norm": 5.08820915222168, + "learning_rate": 0.0001949420077481432, + "loss": 3.4963, + "step": 713 + }, + { + "epoch": 1.05, + "grad_norm": 5.867018222808838, + "learning_rate": 0.00019492726921208092, + "loss": 3.4735, + "step": 714 + }, + { + "epoch": 1.05, + "grad_norm": 7.802560329437256, + "learning_rate": 0.00019491250979265554, + "loss": 3.514, + "step": 715 + }, + { + "epoch": 1.05, + "grad_norm": 4.548806190490723, + "learning_rate": 0.0001948977294931141, + "loss": 3.4695, + "step": 716 + }, + { + "epoch": 1.05, + "grad_norm": 7.293185710906982, + "learning_rate": 0.00019488292831670818, + "loss": 3.5301, + "step": 717 + }, + { + "epoch": 1.05, + "grad_norm": 10.711014747619629, + "learning_rate": 0.00019486810626669388, + "loss": 3.5166, + "step": 718 + }, + { + "epoch": 1.05, + "grad_norm": 5.207925796508789, + "learning_rate": 0.00019485326334633198, + "loss": 3.4741, + "step": 719 + }, + { + "epoch": 1.05, + "grad_norm": 5.915352821350098, + "learning_rate": 0.00019483839955888784, + "loss": 3.4837, + "step": 720 + }, + { + "epoch": 1.06, + "grad_norm": 8.018194198608398, + "learning_rate": 0.00019482351490763135, + "loss": 3.4991, + "step": 721 + }, + { + "epoch": 1.06, + "grad_norm": 4.433293342590332, + "learning_rate": 0.00019480860939583705, + "loss": 3.417, + "step": 722 + }, + { + "epoch": 1.06, + "grad_norm": 6.9495038986206055, + "learning_rate": 0.0001947936830267841, + "loss": 3.5061, + "step": 723 + }, + { + "epoch": 1.06, + "grad_norm": 10.787668228149414, + "learning_rate": 0.00019477873580375614, + "loss": 3.5036, + "step": 724 + }, + { + "epoch": 1.06, + "grad_norm": 5.106097221374512, + "learning_rate": 0.0001947637677300415, + "loss": 3.4732, + "step": 725 + }, + { + "epoch": 1.06, + "grad_norm": 5.890613555908203, + "learning_rate": 0.00019474877880893302, + "loss": 3.5222, + "step": 726 + }, + { + "epoch": 1.06, + "grad_norm": 7.86448335647583, + "learning_rate": 0.0001947337690437282, + "loss": 3.4886, + "step": 727 + }, + { + "epoch": 1.07, + "grad_norm": 4.345864772796631, + "learning_rate": 0.00019471873843772904, + "loss": 3.4902, + "step": 728 + }, + { + "epoch": 1.07, + "grad_norm": 7.271783351898193, + "learning_rate": 0.00019470368699424218, + "loss": 3.4832, + "step": 729 + }, + { + "epoch": 1.07, + "grad_norm": 11.227890968322754, + "learning_rate": 0.0001946886147165789, + "loss": 3.4815, + "step": 730 + }, + { + "epoch": 1.07, + "grad_norm": 5.0216264724731445, + "learning_rate": 0.00019467352160805498, + "loss": 3.4912, + "step": 731 + }, + { + "epoch": 1.07, + "grad_norm": 5.998202800750732, + "learning_rate": 0.00019465840767199073, + "loss": 3.4653, + "step": 732 + }, + { + "epoch": 1.07, + "grad_norm": 7.844666004180908, + "learning_rate": 0.00019464327291171117, + "loss": 3.5708, + "step": 733 + }, + { + "epoch": 1.08, + "grad_norm": 4.441153526306152, + "learning_rate": 0.00019462811733054588, + "loss": 3.479, + "step": 734 + }, + { + "epoch": 1.08, + "grad_norm": 7.58824348449707, + "learning_rate": 0.0001946129409318289, + "loss": 3.4787, + "step": 735 + }, + { + "epoch": 1.08, + "grad_norm": 11.293692588806152, + "learning_rate": 0.00019459774371889901, + "loss": 3.5108, + "step": 736 + }, + { + "epoch": 1.08, + "grad_norm": 6.113391399383545, + "learning_rate": 0.00019458252569509946, + "loss": 3.4456, + "step": 737 + }, + { + "epoch": 1.08, + "grad_norm": 4.897597312927246, + "learning_rate": 0.00019456728686377813, + "loss": 3.5001, + "step": 738 + }, + { + "epoch": 1.08, + "grad_norm": 6.808078765869141, + "learning_rate": 0.00019455202722828743, + "loss": 3.5318, + "step": 739 + }, + { + "epoch": 1.08, + "grad_norm": 2.7436025142669678, + "learning_rate": 0.0001945367467919844, + "loss": 3.452, + "step": 740 + }, + { + "epoch": 1.09, + "grad_norm": 10.119528770446777, + "learning_rate": 0.00019452144555823065, + "loss": 3.5012, + "step": 741 + }, + { + "epoch": 1.09, + "grad_norm": 15.063286781311035, + "learning_rate": 0.00019450612353039235, + "loss": 3.5615, + "step": 742 + }, + { + "epoch": 1.09, + "grad_norm": 9.542817115783691, + "learning_rate": 0.0001944907807118402, + "loss": 3.5215, + "step": 743 + }, + { + "epoch": 1.09, + "grad_norm": 3.414539098739624, + "learning_rate": 0.00019447541710594952, + "loss": 3.4854, + "step": 744 + }, + { + "epoch": 1.09, + "grad_norm": 5.853447437286377, + "learning_rate": 0.00019446003271610026, + "loss": 3.4727, + "step": 745 + }, + { + "epoch": 1.09, + "grad_norm": 2.5720930099487305, + "learning_rate": 0.00019444462754567682, + "loss": 3.4853, + "step": 746 + }, + { + "epoch": 1.09, + "grad_norm": 8.554189682006836, + "learning_rate": 0.0001944292015980683, + "loss": 3.4712, + "step": 747 + }, + { + "epoch": 1.1, + "grad_norm": 11.265769958496094, + "learning_rate": 0.00019441375487666822, + "loss": 3.5236, + "step": 748 + }, + { + "epoch": 1.1, + "grad_norm": 4.981095314025879, + "learning_rate": 0.00019439828738487483, + "loss": 3.5258, + "step": 749 + }, + { + "epoch": 1.1, + "grad_norm": 5.837092876434326, + "learning_rate": 0.00019438279912609082, + "loss": 3.5503, + "step": 750 + }, + { + "epoch": 1.1, + "grad_norm": 8.067007064819336, + "learning_rate": 0.00019436729010372355, + "loss": 3.5129, + "step": 751 + }, + { + "epoch": 1.1, + "grad_norm": 4.977865219116211, + "learning_rate": 0.00019435176032118486, + "loss": 3.4686, + "step": 752 + }, + { + "epoch": 1.1, + "grad_norm": 6.810112953186035, + "learning_rate": 0.00019433620978189122, + "loss": 3.4424, + "step": 753 + }, + { + "epoch": 1.1, + "grad_norm": 10.288273811340332, + "learning_rate": 0.00019432063848926365, + "loss": 3.4829, + "step": 754 + }, + { + "epoch": 1.11, + "grad_norm": 4.5541839599609375, + "learning_rate": 0.0001943050464467277, + "loss": 3.4592, + "step": 755 + }, + { + "epoch": 1.11, + "grad_norm": 5.883044719696045, + "learning_rate": 0.0001942894336577136, + "loss": 3.5044, + "step": 756 + }, + { + "epoch": 1.11, + "grad_norm": 7.786341667175293, + "learning_rate": 0.00019427380012565593, + "loss": 3.5129, + "step": 757 + }, + { + "epoch": 1.11, + "grad_norm": 4.628783226013184, + "learning_rate": 0.00019425814585399405, + "loss": 3.43, + "step": 758 + }, + { + "epoch": 1.11, + "grad_norm": 6.876189708709717, + "learning_rate": 0.00019424247084617177, + "loss": 3.4606, + "step": 759 + }, + { + "epoch": 1.11, + "grad_norm": 10.343745231628418, + "learning_rate": 0.00019422677510563748, + "loss": 3.5079, + "step": 760 + }, + { + "epoch": 1.12, + "grad_norm": 4.678600311279297, + "learning_rate": 0.00019421105863584413, + "loss": 3.5135, + "step": 761 + }, + { + "epoch": 1.12, + "grad_norm": 6.044957160949707, + "learning_rate": 0.0001941953214402493, + "loss": 3.5067, + "step": 762 + }, + { + "epoch": 1.12, + "grad_norm": 7.979968070983887, + "learning_rate": 0.00019417956352231502, + "loss": 3.4955, + "step": 763 + }, + { + "epoch": 1.12, + "grad_norm": 4.172381401062012, + "learning_rate": 0.00019416378488550787, + "loss": 3.4781, + "step": 764 + }, + { + "epoch": 1.12, + "grad_norm": 7.783565998077393, + "learning_rate": 0.00019414798553329916, + "loss": 3.4649, + "step": 765 + }, + { + "epoch": 1.12, + "grad_norm": 11.02284049987793, + "learning_rate": 0.00019413216546916457, + "loss": 3.5242, + "step": 766 + }, + { + "epoch": 1.12, + "grad_norm": 5.28484582901001, + "learning_rate": 0.00019411632469658443, + "loss": 3.4645, + "step": 767 + }, + { + "epoch": 1.13, + "grad_norm": 6.049994945526123, + "learning_rate": 0.00019410046321904356, + "loss": 3.4601, + "step": 768 + }, + { + "epoch": 1.13, + "grad_norm": 7.867702960968018, + "learning_rate": 0.00019408458104003143, + "loss": 3.4853, + "step": 769 + }, + { + "epoch": 1.13, + "grad_norm": 5.088557720184326, + "learning_rate": 0.00019406867816304204, + "loss": 3.5329, + "step": 770 + }, + { + "epoch": 1.13, + "grad_norm": 6.444199085235596, + "learning_rate": 0.00019405275459157383, + "loss": 3.4624, + "step": 771 + }, + { + "epoch": 1.13, + "grad_norm": 9.823208808898926, + "learning_rate": 0.00019403681032912994, + "loss": 3.4496, + "step": 772 + }, + { + "epoch": 1.13, + "grad_norm": 4.309892654418945, + "learning_rate": 0.00019402084537921798, + "loss": 3.4607, + "step": 773 + }, + { + "epoch": 1.13, + "grad_norm": 6.139223575592041, + "learning_rate": 0.00019400485974535014, + "loss": 3.4246, + "step": 774 + }, + { + "epoch": 1.14, + "grad_norm": 8.018882751464844, + "learning_rate": 0.00019398885343104315, + "loss": 3.4731, + "step": 775 + }, + { + "epoch": 1.14, + "grad_norm": 4.7042555809021, + "learning_rate": 0.00019397282643981826, + "loss": 3.4863, + "step": 776 + }, + { + "epoch": 1.14, + "grad_norm": 6.583994388580322, + "learning_rate": 0.00019395677877520137, + "loss": 3.4354, + "step": 777 + }, + { + "epoch": 1.14, + "grad_norm": 9.998130798339844, + "learning_rate": 0.00019394071044072283, + "loss": 3.4882, + "step": 778 + }, + { + "epoch": 1.14, + "grad_norm": 4.592165946960449, + "learning_rate": 0.00019392462143991753, + "loss": 3.4696, + "step": 779 + }, + { + "epoch": 1.14, + "grad_norm": 6.2591681480407715, + "learning_rate": 0.00019390851177632497, + "loss": 3.4393, + "step": 780 + }, + { + "epoch": 1.14, + "grad_norm": 8.170223236083984, + "learning_rate": 0.00019389238145348916, + "loss": 3.421, + "step": 781 + }, + { + "epoch": 1.15, + "grad_norm": 5.1918559074401855, + "learning_rate": 0.00019387623047495867, + "loss": 3.452, + "step": 782 + }, + { + "epoch": 1.15, + "grad_norm": 6.183322429656982, + "learning_rate": 0.0001938600588442866, + "loss": 3.5035, + "step": 783 + }, + { + "epoch": 1.15, + "grad_norm": 9.819025039672852, + "learning_rate": 0.00019384386656503064, + "loss": 3.4897, + "step": 784 + }, + { + "epoch": 1.15, + "grad_norm": 4.417552471160889, + "learning_rate": 0.00019382765364075292, + "loss": 3.5126, + "step": 785 + }, + { + "epoch": 1.15, + "grad_norm": 6.277004241943359, + "learning_rate": 0.0001938114200750202, + "loss": 3.4848, + "step": 786 + }, + { + "epoch": 1.15, + "grad_norm": 8.565668106079102, + "learning_rate": 0.0001937951658714038, + "loss": 3.5328, + "step": 787 + }, + { + "epoch": 1.16, + "grad_norm": 5.04703426361084, + "learning_rate": 0.00019377889103347945, + "loss": 3.4752, + "step": 788 + }, + { + "epoch": 1.16, + "grad_norm": 5.934195518493652, + "learning_rate": 0.0001937625955648276, + "loss": 3.5186, + "step": 789 + }, + { + "epoch": 1.16, + "grad_norm": 9.426629066467285, + "learning_rate": 0.00019374627946903305, + "loss": 3.4537, + "step": 790 + }, + { + "epoch": 1.16, + "grad_norm": 3.8387088775634766, + "learning_rate": 0.00019372994274968527, + "loss": 3.4172, + "step": 791 + }, + { + "epoch": 1.16, + "grad_norm": 6.488461971282959, + "learning_rate": 0.00019371358541037825, + "loss": 3.5177, + "step": 792 + }, + { + "epoch": 1.16, + "grad_norm": 8.451127052307129, + "learning_rate": 0.0001936972074547105, + "loss": 3.4701, + "step": 793 + }, + { + "epoch": 1.16, + "grad_norm": 5.298417568206787, + "learning_rate": 0.00019368080888628502, + "loss": 3.4579, + "step": 794 + }, + { + "epoch": 1.17, + "grad_norm": 6.457042694091797, + "learning_rate": 0.00019366438970870942, + "loss": 3.4666, + "step": 795 + }, + { + "epoch": 1.17, + "grad_norm": 9.55510139465332, + "learning_rate": 0.00019364794992559578, + "loss": 3.4449, + "step": 796 + }, + { + "epoch": 1.17, + "grad_norm": 3.862884998321533, + "learning_rate": 0.00019363148954056077, + "loss": 3.4185, + "step": 797 + }, + { + "epoch": 1.17, + "grad_norm": 6.591372966766357, + "learning_rate": 0.00019361500855722555, + "loss": 3.537, + "step": 798 + }, + { + "epoch": 1.17, + "grad_norm": 8.397138595581055, + "learning_rate": 0.0001935985069792158, + "loss": 3.4205, + "step": 799 + }, + { + "epoch": 1.17, + "grad_norm": 5.131923675537109, + "learning_rate": 0.00019358198481016182, + "loss": 3.4916, + "step": 800 + }, + { + "epoch": 1.17, + "grad_norm": 5.620177745819092, + "learning_rate": 0.00019356544205369834, + "loss": 3.4604, + "step": 801 + }, + { + "epoch": 1.18, + "grad_norm": 9.219866752624512, + "learning_rate": 0.00019354887871346463, + "loss": 3.467, + "step": 802 + }, + { + "epoch": 1.18, + "grad_norm": 4.237349033355713, + "learning_rate": 0.00019353229479310456, + "loss": 3.4847, + "step": 803 + }, + { + "epoch": 1.18, + "grad_norm": 6.673110485076904, + "learning_rate": 0.00019351569029626644, + "loss": 3.4412, + "step": 804 + }, + { + "epoch": 1.18, + "grad_norm": 8.31098747253418, + "learning_rate": 0.0001934990652266032, + "loss": 3.4838, + "step": 805 + }, + { + "epoch": 1.18, + "grad_norm": 5.009383201599121, + "learning_rate": 0.0001934824195877722, + "loss": 3.4408, + "step": 806 + }, + { + "epoch": 1.18, + "grad_norm": 5.483675956726074, + "learning_rate": 0.00019346575338343537, + "loss": 3.4663, + "step": 807 + }, + { + "epoch": 1.19, + "grad_norm": 9.371169090270996, + "learning_rate": 0.00019344906661725918, + "loss": 3.5116, + "step": 808 + }, + { + "epoch": 1.19, + "grad_norm": 3.6695902347564697, + "learning_rate": 0.00019343235929291462, + "loss": 3.4776, + "step": 809 + }, + { + "epoch": 1.19, + "grad_norm": 6.492842197418213, + "learning_rate": 0.00019341563141407714, + "loss": 3.4725, + "step": 810 + }, + { + "epoch": 1.19, + "grad_norm": 8.520849227905273, + "learning_rate": 0.00019339888298442682, + "loss": 3.4845, + "step": 811 + }, + { + "epoch": 1.19, + "grad_norm": 4.986138820648193, + "learning_rate": 0.00019338211400764816, + "loss": 3.4313, + "step": 812 + }, + { + "epoch": 1.19, + "grad_norm": 6.272872447967529, + "learning_rate": 0.00019336532448743023, + "loss": 3.4337, + "step": 813 + }, + { + "epoch": 1.19, + "grad_norm": 10.374552726745605, + "learning_rate": 0.00019334851442746664, + "loss": 3.4586, + "step": 814 + }, + { + "epoch": 1.2, + "grad_norm": 4.865425109863281, + "learning_rate": 0.00019333168383145547, + "loss": 3.5045, + "step": 815 + }, + { + "epoch": 1.2, + "grad_norm": 6.018640995025635, + "learning_rate": 0.0001933148327030993, + "loss": 3.4509, + "step": 816 + }, + { + "epoch": 1.2, + "grad_norm": 8.015512466430664, + "learning_rate": 0.00019329796104610537, + "loss": 3.4657, + "step": 817 + }, + { + "epoch": 1.2, + "grad_norm": 4.071451187133789, + "learning_rate": 0.00019328106886418522, + "loss": 3.4079, + "step": 818 + }, + { + "epoch": 1.2, + "grad_norm": 7.4237823486328125, + "learning_rate": 0.00019326415616105507, + "loss": 3.4906, + "step": 819 + }, + { + "epoch": 1.2, + "grad_norm": 10.776564598083496, + "learning_rate": 0.00019324722294043558, + "loss": 3.4639, + "step": 820 + }, + { + "epoch": 1.2, + "grad_norm": 5.143712997436523, + "learning_rate": 0.00019323026920605197, + "loss": 3.4356, + "step": 821 + }, + { + "epoch": 1.21, + "grad_norm": 5.7633891105651855, + "learning_rate": 0.00019321329496163397, + "loss": 3.428, + "step": 822 + }, + { + "epoch": 1.21, + "grad_norm": 7.827281475067139, + "learning_rate": 0.0001931963002109157, + "loss": 3.4913, + "step": 823 + }, + { + "epoch": 1.21, + "grad_norm": 4.470641136169434, + "learning_rate": 0.00019317928495763602, + "loss": 3.4441, + "step": 824 + }, + { + "epoch": 1.21, + "grad_norm": 7.5875091552734375, + "learning_rate": 0.00019316224920553807, + "loss": 3.468, + "step": 825 + }, + { + "epoch": 1.21, + "grad_norm": 10.94041919708252, + "learning_rate": 0.00019314519295836964, + "loss": 3.4932, + "step": 826 + }, + { + "epoch": 1.21, + "grad_norm": 5.624682426452637, + "learning_rate": 0.000193128116219883, + "loss": 3.4411, + "step": 827 + }, + { + "epoch": 1.21, + "grad_norm": 5.68918514251709, + "learning_rate": 0.00019311101899383492, + "loss": 3.4896, + "step": 828 + }, + { + "epoch": 1.22, + "grad_norm": 7.500881671905518, + "learning_rate": 0.00019309390128398665, + "loss": 3.4429, + "step": 829 + }, + { + "epoch": 1.22, + "grad_norm": 4.407763481140137, + "learning_rate": 0.000193076763094104, + "loss": 3.4694, + "step": 830 + }, + { + "epoch": 1.22, + "grad_norm": 7.296525955200195, + "learning_rate": 0.0001930596044279572, + "loss": 3.4065, + "step": 831 + }, + { + "epoch": 1.22, + "grad_norm": 11.3180513381958, + "learning_rate": 0.00019304242528932113, + "loss": 3.5025, + "step": 832 + }, + { + "epoch": 1.22, + "grad_norm": 5.404650688171387, + "learning_rate": 0.00019302522568197502, + "loss": 3.4129, + "step": 833 + }, + { + "epoch": 1.22, + "grad_norm": 5.595582485198975, + "learning_rate": 0.0001930080056097027, + "loss": 3.5212, + "step": 834 + }, + { + "epoch": 1.23, + "grad_norm": 7.934787273406982, + "learning_rate": 0.00019299076507629245, + "loss": 3.5551, + "step": 835 + }, + { + "epoch": 1.23, + "grad_norm": 4.486216068267822, + "learning_rate": 0.00019297350408553708, + "loss": 3.441, + "step": 836 + }, + { + "epoch": 1.23, + "grad_norm": 7.029922962188721, + "learning_rate": 0.0001929562226412339, + "loss": 3.4982, + "step": 837 + }, + { + "epoch": 1.23, + "grad_norm": 10.693680763244629, + "learning_rate": 0.0001929389207471847, + "loss": 3.4494, + "step": 838 + }, + { + "epoch": 1.23, + "grad_norm": 4.879317760467529, + "learning_rate": 0.00019292159840719576, + "loss": 3.4357, + "step": 839 + }, + { + "epoch": 1.23, + "grad_norm": 6.057565212249756, + "learning_rate": 0.00019290425562507792, + "loss": 3.4738, + "step": 840 + }, + { + "epoch": 1.23, + "grad_norm": 7.925189018249512, + "learning_rate": 0.0001928868924046465, + "loss": 3.4606, + "step": 841 + }, + { + "epoch": 1.24, + "grad_norm": 4.621508598327637, + "learning_rate": 0.0001928695087497212, + "loss": 3.4088, + "step": 842 + }, + { + "epoch": 1.24, + "grad_norm": 6.142604351043701, + "learning_rate": 0.00019285210466412635, + "loss": 3.4484, + "step": 843 + }, + { + "epoch": 1.24, + "grad_norm": 10.20028305053711, + "learning_rate": 0.00019283468015169076, + "loss": 3.5059, + "step": 844 + }, + { + "epoch": 1.24, + "grad_norm": 4.734007358551025, + "learning_rate": 0.0001928172352162477, + "loss": 3.4299, + "step": 845 + }, + { + "epoch": 1.24, + "eval_loss": 3.2429304122924805, + "eval_runtime": 1163.5436, + "eval_samples_per_second": 14.733, + "eval_steps_per_second": 0.921, + "step": 845 + }, + { + "epoch": 1.24, + "grad_norm": 6.036166191101074, + "learning_rate": 0.00019279976986163493, + "loss": 3.3913, + "step": 846 + }, + { + "epoch": 1.24, + "grad_norm": 8.014019012451172, + "learning_rate": 0.00019278228409169468, + "loss": 3.4981, + "step": 847 + }, + { + "epoch": 1.24, + "grad_norm": 4.856557369232178, + "learning_rate": 0.00019276477791027374, + "loss": 3.4562, + "step": 848 + }, + { + "epoch": 1.25, + "grad_norm": 6.650398254394531, + "learning_rate": 0.00019274725132122336, + "loss": 3.4407, + "step": 849 + }, + { + "epoch": 1.25, + "grad_norm": 10.177898406982422, + "learning_rate": 0.00019272970432839922, + "loss": 3.4437, + "step": 850 + }, + { + "epoch": 1.25, + "grad_norm": 4.559935569763184, + "learning_rate": 0.0001927121369356616, + "loss": 3.418, + "step": 851 + }, + { + "epoch": 1.25, + "grad_norm": 5.8679423332214355, + "learning_rate": 0.00019269454914687517, + "loss": 3.4338, + "step": 852 + }, + { + "epoch": 1.25, + "grad_norm": 8.044879913330078, + "learning_rate": 0.00019267694096590914, + "loss": 3.4146, + "step": 853 + }, + { + "epoch": 1.25, + "grad_norm": 4.654249668121338, + "learning_rate": 0.0001926593123966372, + "loss": 3.4353, + "step": 854 + }, + { + "epoch": 1.25, + "grad_norm": 6.326594352722168, + "learning_rate": 0.0001926416634429375, + "loss": 3.447, + "step": 855 + }, + { + "epoch": 1.26, + "grad_norm": 10.194587707519531, + "learning_rate": 0.0001926239941086927, + "loss": 3.4594, + "step": 856 + }, + { + "epoch": 1.26, + "grad_norm": 4.672416687011719, + "learning_rate": 0.00019260630439778994, + "loss": 3.4292, + "step": 857 + }, + { + "epoch": 1.26, + "grad_norm": 5.656294822692871, + "learning_rate": 0.00019258859431412083, + "loss": 3.4478, + "step": 858 + }, + { + "epoch": 1.26, + "grad_norm": 8.051904678344727, + "learning_rate": 0.00019257086386158148, + "loss": 3.4809, + "step": 859 + }, + { + "epoch": 1.26, + "grad_norm": 4.461113452911377, + "learning_rate": 0.00019255311304407243, + "loss": 3.4361, + "step": 860 + }, + { + "epoch": 1.26, + "grad_norm": 6.34873104095459, + "learning_rate": 0.00019253534186549883, + "loss": 3.424, + "step": 861 + }, + { + "epoch": 1.27, + "grad_norm": 10.255434036254883, + "learning_rate": 0.00019251755032977013, + "loss": 3.4578, + "step": 862 + }, + { + "epoch": 1.27, + "grad_norm": 4.895047187805176, + "learning_rate": 0.00019249973844080037, + "loss": 3.4606, + "step": 863 + }, + { + "epoch": 1.27, + "grad_norm": 6.042967319488525, + "learning_rate": 0.00019248190620250807, + "loss": 3.4398, + "step": 864 + }, + { + "epoch": 1.27, + "grad_norm": 7.916470050811768, + "learning_rate": 0.00019246405361881622, + "loss": 3.4921, + "step": 865 + }, + { + "epoch": 1.27, + "grad_norm": 4.985673904418945, + "learning_rate": 0.0001924461806936522, + "loss": 3.4425, + "step": 866 + }, + { + "epoch": 1.27, + "grad_norm": 6.128829002380371, + "learning_rate": 0.000192428287430948, + "loss": 3.4719, + "step": 867 + }, + { + "epoch": 1.27, + "grad_norm": 9.511234283447266, + "learning_rate": 0.00019241037383463998, + "loss": 3.4018, + "step": 868 + }, + { + "epoch": 1.28, + "grad_norm": 3.816689968109131, + "learning_rate": 0.000192392439908669, + "loss": 3.375, + "step": 869 + }, + { + "epoch": 1.28, + "grad_norm": 6.579503059387207, + "learning_rate": 0.00019237448565698046, + "loss": 3.4039, + "step": 870 + }, + { + "epoch": 1.28, + "grad_norm": 8.701712608337402, + "learning_rate": 0.00019235651108352416, + "loss": 3.3992, + "step": 871 + }, + { + "epoch": 1.28, + "grad_norm": 5.65900993347168, + "learning_rate": 0.00019233851619225432, + "loss": 3.4225, + "step": 872 + }, + { + "epoch": 1.28, + "grad_norm": 4.935523509979248, + "learning_rate": 0.00019232050098712974, + "loss": 3.4198, + "step": 873 + }, + { + "epoch": 1.28, + "grad_norm": 8.355915069580078, + "learning_rate": 0.00019230246547211367, + "loss": 3.4283, + "step": 874 + }, + { + "epoch": 1.28, + "grad_norm": 3.1533217430114746, + "learning_rate": 0.0001922844096511738, + "loss": 3.4058, + "step": 875 + }, + { + "epoch": 1.29, + "grad_norm": 7.010662078857422, + "learning_rate": 0.00019226633352828224, + "loss": 3.3914, + "step": 876 + }, + { + "epoch": 1.29, + "grad_norm": 8.548685073852539, + "learning_rate": 0.00019224823710741569, + "loss": 3.5008, + "step": 877 + }, + { + "epoch": 1.29, + "grad_norm": 5.439549446105957, + "learning_rate": 0.00019223012039255516, + "loss": 3.4421, + "step": 878 + }, + { + "epoch": 1.29, + "grad_norm": 5.446386814117432, + "learning_rate": 0.00019221198338768622, + "loss": 3.4071, + "step": 879 + }, + { + "epoch": 1.29, + "grad_norm": 9.378525733947754, + "learning_rate": 0.00019219382609679894, + "loss": 3.409, + "step": 880 + }, + { + "epoch": 1.29, + "grad_norm": 4.573349475860596, + "learning_rate": 0.0001921756485238878, + "loss": 3.4156, + "step": 881 + }, + { + "epoch": 1.29, + "grad_norm": 5.857920169830322, + "learning_rate": 0.00019215745067295169, + "loss": 3.433, + "step": 882 + }, + { + "epoch": 1.3, + "grad_norm": 7.654170036315918, + "learning_rate": 0.00019213923254799402, + "loss": 3.4011, + "step": 883 + }, + { + "epoch": 1.3, + "grad_norm": 4.262071132659912, + "learning_rate": 0.0001921209941530227, + "loss": 3.401, + "step": 884 + }, + { + "epoch": 1.3, + "grad_norm": 7.949652194976807, + "learning_rate": 0.00019210273549205003, + "loss": 3.5334, + "step": 885 + }, + { + "epoch": 1.3, + "grad_norm": 11.388256072998047, + "learning_rate": 0.0001920844565690928, + "loss": 3.4325, + "step": 886 + }, + { + "epoch": 1.3, + "grad_norm": 6.005974292755127, + "learning_rate": 0.00019206615738817225, + "loss": 3.4264, + "step": 887 + }, + { + "epoch": 1.3, + "grad_norm": 5.020950794219971, + "learning_rate": 0.0001920478379533141, + "loss": 3.426, + "step": 888 + }, + { + "epoch": 1.31, + "grad_norm": 6.8441386222839355, + "learning_rate": 0.00019202949826854844, + "loss": 3.4234, + "step": 889 + }, + { + "epoch": 1.31, + "grad_norm": 3.4975130558013916, + "learning_rate": 0.0001920111383379099, + "loss": 3.4667, + "step": 890 + }, + { + "epoch": 1.31, + "grad_norm": 7.94019889831543, + "learning_rate": 0.00019199275816543762, + "loss": 3.4208, + "step": 891 + }, + { + "epoch": 1.31, + "grad_norm": 11.041260719299316, + "learning_rate": 0.00019197435775517501, + "loss": 3.4392, + "step": 892 + }, + { + "epoch": 1.31, + "grad_norm": 5.074487209320068, + "learning_rate": 0.0001919559371111701, + "loss": 3.3906, + "step": 893 + }, + { + "epoch": 1.31, + "grad_norm": 5.909255027770996, + "learning_rate": 0.00019193749623747525, + "loss": 3.4508, + "step": 894 + }, + { + "epoch": 1.31, + "grad_norm": 7.794407367706299, + "learning_rate": 0.0001919190351381474, + "loss": 3.3957, + "step": 895 + }, + { + "epoch": 1.32, + "grad_norm": 4.833076000213623, + "learning_rate": 0.00019190055381724781, + "loss": 3.403, + "step": 896 + }, + { + "epoch": 1.32, + "grad_norm": 6.273806095123291, + "learning_rate": 0.0001918820522788423, + "loss": 3.4114, + "step": 897 + }, + { + "epoch": 1.32, + "grad_norm": 9.626479148864746, + "learning_rate": 0.00019186353052700104, + "loss": 3.4555, + "step": 898 + }, + { + "epoch": 1.32, + "grad_norm": 4.462099075317383, + "learning_rate": 0.00019184498856579868, + "loss": 3.4001, + "step": 899 + }, + { + "epoch": 1.32, + "grad_norm": 5.839742660522461, + "learning_rate": 0.00019182642639931438, + "loss": 3.4041, + "step": 900 + }, + { + "epoch": 1.32, + "grad_norm": 7.5265583992004395, + "learning_rate": 0.00019180784403163167, + "loss": 3.3898, + "step": 901 + }, + { + "epoch": 1.32, + "grad_norm": 4.089066028594971, + "learning_rate": 0.0001917892414668385, + "loss": 3.4162, + "step": 902 + }, + { + "epoch": 1.33, + "grad_norm": 7.329177379608154, + "learning_rate": 0.0001917706187090274, + "loss": 3.5439, + "step": 903 + }, + { + "epoch": 1.33, + "grad_norm": 10.022590637207031, + "learning_rate": 0.0001917519757622952, + "loss": 3.4009, + "step": 904 + }, + { + "epoch": 1.33, + "grad_norm": 5.030527591705322, + "learning_rate": 0.00019173331263074324, + "loss": 3.4314, + "step": 905 + }, + { + "epoch": 1.33, + "grad_norm": 5.504607677459717, + "learning_rate": 0.00019171462931847724, + "loss": 3.4463, + "step": 906 + }, + { + "epoch": 1.33, + "grad_norm": 7.219310283660889, + "learning_rate": 0.00019169592582960742, + "loss": 3.4878, + "step": 907 + }, + { + "epoch": 1.33, + "grad_norm": 4.046157360076904, + "learning_rate": 0.00019167720216824847, + "loss": 3.3643, + "step": 908 + }, + { + "epoch": 1.34, + "grad_norm": 7.463158130645752, + "learning_rate": 0.00019165845833851941, + "loss": 3.472, + "step": 909 + }, + { + "epoch": 1.34, + "grad_norm": 9.688533782958984, + "learning_rate": 0.00019163969434454382, + "loss": 3.4541, + "step": 910 + }, + { + "epoch": 1.34, + "grad_norm": 4.718085765838623, + "learning_rate": 0.00019162091019044958, + "loss": 3.4252, + "step": 911 + }, + { + "epoch": 1.34, + "grad_norm": 6.263822555541992, + "learning_rate": 0.00019160210588036911, + "loss": 3.4105, + "step": 912 + }, + { + "epoch": 1.34, + "grad_norm": 7.928195953369141, + "learning_rate": 0.00019158328141843928, + "loss": 3.4249, + "step": 913 + }, + { + "epoch": 1.34, + "grad_norm": 4.594407081604004, + "learning_rate": 0.00019156443680880126, + "loss": 3.453, + "step": 914 + }, + { + "epoch": 1.34, + "grad_norm": 6.126087188720703, + "learning_rate": 0.0001915455720556008, + "loss": 3.3944, + "step": 915 + }, + { + "epoch": 1.35, + "grad_norm": 9.558304786682129, + "learning_rate": 0.000191526687162988, + "loss": 3.3926, + "step": 916 + }, + { + "epoch": 1.35, + "grad_norm": 4.792506217956543, + "learning_rate": 0.0001915077821351174, + "loss": 3.4081, + "step": 917 + }, + { + "epoch": 1.35, + "grad_norm": 5.998870372772217, + "learning_rate": 0.00019148885697614797, + "loss": 3.4326, + "step": 918 + }, + { + "epoch": 1.35, + "grad_norm": 7.541780948638916, + "learning_rate": 0.00019146991169024315, + "loss": 3.4647, + "step": 919 + }, + { + "epoch": 1.35, + "grad_norm": 4.514527320861816, + "learning_rate": 0.00019145094628157078, + "loss": 3.3601, + "step": 920 + }, + { + "epoch": 1.35, + "grad_norm": 6.531849384307861, + "learning_rate": 0.0001914319607543031, + "loss": 3.4428, + "step": 921 + }, + { + "epoch": 1.35, + "grad_norm": 9.34991455078125, + "learning_rate": 0.00019141295511261678, + "loss": 3.393, + "step": 922 + }, + { + "epoch": 1.36, + "grad_norm": 4.035532474517822, + "learning_rate": 0.000191393929360693, + "loss": 3.3938, + "step": 923 + }, + { + "epoch": 1.36, + "grad_norm": 6.159792900085449, + "learning_rate": 0.0001913748835027172, + "loss": 3.4207, + "step": 924 + }, + { + "epoch": 1.36, + "grad_norm": 8.069183349609375, + "learning_rate": 0.00019135581754287944, + "loss": 3.4143, + "step": 925 + }, + { + "epoch": 1.36, + "grad_norm": 4.528735160827637, + "learning_rate": 0.00019133673148537405, + "loss": 3.4282, + "step": 926 + }, + { + "epoch": 1.36, + "grad_norm": 6.5554399490356445, + "learning_rate": 0.00019131762533439983, + "loss": 3.4149, + "step": 927 + }, + { + "epoch": 1.36, + "grad_norm": 9.77618408203125, + "learning_rate": 0.00019129849909416005, + "loss": 3.4509, + "step": 928 + }, + { + "epoch": 1.36, + "grad_norm": 4.406791687011719, + "learning_rate": 0.0001912793527688623, + "loss": 3.3955, + "step": 929 + }, + { + "epoch": 1.37, + "grad_norm": 6.1382622718811035, + "learning_rate": 0.00019126018636271867, + "loss": 3.4066, + "step": 930 + }, + { + "epoch": 1.37, + "grad_norm": 8.133618354797363, + "learning_rate": 0.00019124099987994565, + "loss": 3.4536, + "step": 931 + }, + { + "epoch": 1.37, + "grad_norm": 4.887189865112305, + "learning_rate": 0.0001912217933247641, + "loss": 3.4516, + "step": 932 + }, + { + "epoch": 1.37, + "grad_norm": 5.946772575378418, + "learning_rate": 0.0001912025667013994, + "loss": 3.3842, + "step": 933 + }, + { + "epoch": 1.37, + "grad_norm": 9.542755126953125, + "learning_rate": 0.00019118332001408122, + "loss": 3.3833, + "step": 934 + }, + { + "epoch": 1.37, + "grad_norm": 4.512613296508789, + "learning_rate": 0.00019116405326704373, + "loss": 3.3754, + "step": 935 + }, + { + "epoch": 1.38, + "grad_norm": 6.039595603942871, + "learning_rate": 0.00019114476646452547, + "loss": 3.4341, + "step": 936 + }, + { + "epoch": 1.38, + "grad_norm": 7.820281982421875, + "learning_rate": 0.0001911254596107694, + "loss": 3.3616, + "step": 937 + }, + { + "epoch": 1.38, + "grad_norm": 4.0570759773254395, + "learning_rate": 0.0001911061327100229, + "loss": 3.3999, + "step": 938 + }, + { + "epoch": 1.38, + "grad_norm": 6.659790992736816, + "learning_rate": 0.0001910867857665378, + "loss": 3.361, + "step": 939 + }, + { + "epoch": 1.38, + "grad_norm": 10.188064575195312, + "learning_rate": 0.00019106741878457026, + "loss": 3.3805, + "step": 940 + }, + { + "epoch": 1.38, + "grad_norm": 4.66610860824585, + "learning_rate": 0.0001910480317683809, + "loss": 3.3881, + "step": 941 + }, + { + "epoch": 1.38, + "grad_norm": 5.840310573577881, + "learning_rate": 0.0001910286247222347, + "loss": 3.4162, + "step": 942 + }, + { + "epoch": 1.39, + "grad_norm": 7.710467338562012, + "learning_rate": 0.00019100919765040113, + "loss": 3.396, + "step": 943 + }, + { + "epoch": 1.39, + "grad_norm": 4.53704833984375, + "learning_rate": 0.00019098975055715397, + "loss": 3.3605, + "step": 944 + }, + { + "epoch": 1.39, + "grad_norm": 6.515756607055664, + "learning_rate": 0.00019097028344677153, + "loss": 3.4119, + "step": 945 + }, + { + "epoch": 1.39, + "grad_norm": 10.256503105163574, + "learning_rate": 0.00019095079632353633, + "loss": 3.4173, + "step": 946 + }, + { + "epoch": 1.39, + "grad_norm": 4.8957672119140625, + "learning_rate": 0.0001909312891917355, + "loss": 3.3812, + "step": 947 + }, + { + "epoch": 1.39, + "grad_norm": 5.8118205070495605, + "learning_rate": 0.00019091176205566043, + "loss": 3.3734, + "step": 948 + }, + { + "epoch": 1.39, + "grad_norm": 7.7631025314331055, + "learning_rate": 0.00019089221491960695, + "loss": 3.4171, + "step": 949 + }, + { + "epoch": 1.4, + "grad_norm": 4.2992377281188965, + "learning_rate": 0.00019087264778787534, + "loss": 3.3724, + "step": 950 + }, + { + "epoch": 1.4, + "grad_norm": 6.633007049560547, + "learning_rate": 0.00019085306066477021, + "loss": 3.4198, + "step": 951 + }, + { + "epoch": 1.4, + "grad_norm": 9.867729187011719, + "learning_rate": 0.00019083345355460062, + "loss": 3.4247, + "step": 952 + }, + { + "epoch": 1.4, + "grad_norm": 4.476417541503906, + "learning_rate": 0.00019081382646167998, + "loss": 3.3617, + "step": 953 + }, + { + "epoch": 1.4, + "grad_norm": 5.963509559631348, + "learning_rate": 0.00019079417939032612, + "loss": 3.3945, + "step": 954 + }, + { + "epoch": 1.4, + "grad_norm": 7.921098232269287, + "learning_rate": 0.00019077451234486128, + "loss": 3.4305, + "step": 955 + }, + { + "epoch": 1.4, + "grad_norm": 4.772178649902344, + "learning_rate": 0.00019075482532961204, + "loss": 3.3836, + "step": 956 + }, + { + "epoch": 1.41, + "grad_norm": 6.6379265785217285, + "learning_rate": 0.00019073511834890949, + "loss": 3.3646, + "step": 957 + }, + { + "epoch": 1.41, + "grad_norm": 9.883600234985352, + "learning_rate": 0.00019071539140708894, + "loss": 3.398, + "step": 958 + }, + { + "epoch": 1.41, + "grad_norm": 4.0690178871154785, + "learning_rate": 0.00019069564450849025, + "loss": 3.3394, + "step": 959 + }, + { + "epoch": 1.41, + "grad_norm": 6.045716762542725, + "learning_rate": 0.00019067587765745757, + "loss": 3.4695, + "step": 960 + }, + { + "epoch": 1.41, + "grad_norm": 7.893038272857666, + "learning_rate": 0.00019065609085833947, + "loss": 3.4336, + "step": 961 + }, + { + "epoch": 1.41, + "grad_norm": 4.4083075523376465, + "learning_rate": 0.00019063628411548897, + "loss": 3.3471, + "step": 962 + }, + { + "epoch": 1.42, + "grad_norm": 6.27316951751709, + "learning_rate": 0.00019061645743326337, + "loss": 3.4311, + "step": 963 + }, + { + "epoch": 1.42, + "grad_norm": 9.92757797241211, + "learning_rate": 0.00019059661081602443, + "loss": 3.4114, + "step": 964 + }, + { + "epoch": 1.42, + "grad_norm": 5.099198341369629, + "learning_rate": 0.00019057674426813822, + "loss": 3.4386, + "step": 965 + }, + { + "epoch": 1.42, + "grad_norm": 5.424035549163818, + "learning_rate": 0.00019055685779397532, + "loss": 3.4275, + "step": 966 + }, + { + "epoch": 1.42, + "grad_norm": 7.513932704925537, + "learning_rate": 0.0001905369513979106, + "loss": 3.4135, + "step": 967 + }, + { + "epoch": 1.42, + "grad_norm": 4.388814449310303, + "learning_rate": 0.0001905170250843233, + "loss": 3.3889, + "step": 968 + }, + { + "epoch": 1.42, + "grad_norm": 6.554171562194824, + "learning_rate": 0.00019049707885759713, + "loss": 3.4184, + "step": 969 + }, + { + "epoch": 1.43, + "grad_norm": 9.21511173248291, + "learning_rate": 0.00019047711272212006, + "loss": 3.393, + "step": 970 + }, + { + "epoch": 1.43, + "grad_norm": 4.280778408050537, + "learning_rate": 0.00019045712668228456, + "loss": 3.3808, + "step": 971 + }, + { + "epoch": 1.43, + "grad_norm": 5.782205581665039, + "learning_rate": 0.0001904371207424874, + "loss": 3.3168, + "step": 972 + }, + { + "epoch": 1.43, + "grad_norm": 7.35162878036499, + "learning_rate": 0.00019041709490712978, + "loss": 3.4373, + "step": 973 + }, + { + "epoch": 1.43, + "grad_norm": 3.4769937992095947, + "learning_rate": 0.0001903970491806172, + "loss": 3.4039, + "step": 974 + }, + { + "epoch": 1.43, + "grad_norm": 8.196279525756836, + "learning_rate": 0.00019037698356735963, + "loss": 3.38, + "step": 975 + }, + { + "epoch": 1.43, + "grad_norm": 11.59374713897705, + "learning_rate": 0.00019035689807177136, + "loss": 3.4377, + "step": 976 + }, + { + "epoch": 1.44, + "grad_norm": 6.463580131530762, + "learning_rate": 0.00019033679269827105, + "loss": 3.3706, + "step": 977 + }, + { + "epoch": 1.44, + "grad_norm": 4.297844409942627, + "learning_rate": 0.00019031666745128177, + "loss": 3.4047, + "step": 978 + }, + { + "epoch": 1.44, + "grad_norm": 6.652123928070068, + "learning_rate": 0.00019029652233523093, + "loss": 3.3968, + "step": 979 + }, + { + "epoch": 1.44, + "grad_norm": 3.399991750717163, + "learning_rate": 0.00019027635735455027, + "loss": 3.3552, + "step": 980 + }, + { + "epoch": 1.44, + "grad_norm": 7.520792007446289, + "learning_rate": 0.00019025617251367604, + "loss": 3.3979, + "step": 981 + }, + { + "epoch": 1.44, + "grad_norm": 10.621402740478516, + "learning_rate": 0.00019023596781704872, + "loss": 3.4131, + "step": 982 + }, + { + "epoch": 1.44, + "grad_norm": 5.002892971038818, + "learning_rate": 0.0001902157432691132, + "loss": 3.3583, + "step": 983 + }, + { + "epoch": 1.45, + "grad_norm": 5.647467136383057, + "learning_rate": 0.00019019549887431877, + "loss": 3.384, + "step": 984 + }, + { + "epoch": 1.45, + "grad_norm": 7.81645393371582, + "learning_rate": 0.00019017523463711906, + "loss": 3.4198, + "step": 985 + }, + { + "epoch": 1.45, + "grad_norm": 4.508808135986328, + "learning_rate": 0.00019015495056197207, + "loss": 3.4205, + "step": 986 + }, + { + "epoch": 1.45, + "grad_norm": 6.004973411560059, + "learning_rate": 0.00019013464665334012, + "loss": 3.396, + "step": 987 + }, + { + "epoch": 1.45, + "grad_norm": 9.286575317382812, + "learning_rate": 0.00019011432291568995, + "loss": 3.4053, + "step": 988 + }, + { + "epoch": 1.45, + "grad_norm": 4.02089786529541, + "learning_rate": 0.00019009397935349268, + "loss": 3.3216, + "step": 989 + }, + { + "epoch": 1.46, + "grad_norm": 6.502218246459961, + "learning_rate": 0.00019007361597122373, + "loss": 3.4367, + "step": 990 + }, + { + "epoch": 1.46, + "grad_norm": 8.069449424743652, + "learning_rate": 0.0001900532327733629, + "loss": 3.3666, + "step": 991 + }, + { + "epoch": 1.46, + "grad_norm": 4.8657917976379395, + "learning_rate": 0.0001900328297643944, + "loss": 3.3401, + "step": 992 + }, + { + "epoch": 1.46, + "grad_norm": 5.57826042175293, + "learning_rate": 0.00019001240694880672, + "loss": 3.393, + "step": 993 + }, + { + "epoch": 1.46, + "grad_norm": 8.596120834350586, + "learning_rate": 0.00018999196433109274, + "loss": 3.416, + "step": 994 + }, + { + "epoch": 1.46, + "grad_norm": 4.3653130531311035, + "learning_rate": 0.0001899715019157497, + "loss": 3.3789, + "step": 995 + }, + { + "epoch": 1.46, + "grad_norm": 6.067572593688965, + "learning_rate": 0.0001899510197072792, + "loss": 3.3577, + "step": 996 + }, + { + "epoch": 1.47, + "grad_norm": 7.723760604858398, + "learning_rate": 0.00018993051771018719, + "loss": 3.4, + "step": 997 + }, + { + "epoch": 1.47, + "grad_norm": 4.619333744049072, + "learning_rate": 0.00018990999592898396, + "loss": 3.3979, + "step": 998 + }, + { + "epoch": 1.47, + "grad_norm": 6.598495006561279, + "learning_rate": 0.00018988945436818422, + "loss": 3.4331, + "step": 999 + }, + { + "epoch": 1.47, + "grad_norm": 9.450194358825684, + "learning_rate": 0.0001898688930323069, + "loss": 3.4156, + "step": 1000 + }, + { + "epoch": 1.47, + "grad_norm": 4.10836935043335, + "learning_rate": 0.00018984831192587538, + "loss": 3.3787, + "step": 1001 + }, + { + "epoch": 1.47, + "grad_norm": 6.087796211242676, + "learning_rate": 0.00018982771105341737, + "loss": 3.3969, + "step": 1002 + }, + { + "epoch": 1.47, + "grad_norm": 7.915310859680176, + "learning_rate": 0.00018980709041946497, + "loss": 3.3948, + "step": 1003 + }, + { + "epoch": 1.48, + "grad_norm": 4.227004528045654, + "learning_rate": 0.0001897864500285545, + "loss": 3.3238, + "step": 1004 + }, + { + "epoch": 1.48, + "grad_norm": 6.717720031738281, + "learning_rate": 0.00018976578988522675, + "loss": 3.3781, + "step": 1005 + }, + { + "epoch": 1.48, + "grad_norm": 10.050957679748535, + "learning_rate": 0.0001897451099940268, + "loss": 3.4204, + "step": 1006 + }, + { + "epoch": 1.48, + "grad_norm": 4.475175857543945, + "learning_rate": 0.0001897244103595041, + "loss": 3.4221, + "step": 1007 + }, + { + "epoch": 1.48, + "grad_norm": 5.719995975494385, + "learning_rate": 0.00018970369098621244, + "loss": 3.3555, + "step": 1008 + }, + { + "epoch": 1.48, + "grad_norm": 7.594791889190674, + "learning_rate": 0.00018968295187870996, + "loss": 3.4073, + "step": 1009 + }, + { + "epoch": 1.49, + "grad_norm": 4.333056926727295, + "learning_rate": 0.00018966219304155906, + "loss": 3.3556, + "step": 1010 + }, + { + "epoch": 1.49, + "grad_norm": 6.078528881072998, + "learning_rate": 0.00018964141447932656, + "loss": 3.3484, + "step": 1011 + }, + { + "epoch": 1.49, + "grad_norm": 8.950298309326172, + "learning_rate": 0.00018962061619658366, + "loss": 3.4055, + "step": 1012 + }, + { + "epoch": 1.49, + "grad_norm": 4.340825080871582, + "learning_rate": 0.0001895997981979058, + "loss": 3.334, + "step": 1013 + }, + { + "epoch": 1.49, + "grad_norm": 6.103879451751709, + "learning_rate": 0.00018957896048787278, + "loss": 3.4908, + "step": 1014 + }, + { + "epoch": 1.49, + "eval_loss": 3.206282138824463, + "eval_runtime": 1164.0205, + "eval_samples_per_second": 14.727, + "eval_steps_per_second": 0.921, + "step": 1014 + }, + { + "epoch": 1.49, + "grad_norm": 7.772497653961182, + "learning_rate": 0.00018955810307106881, + "loss": 3.3821, + "step": 1015 + }, + { + "epoch": 1.49, + "grad_norm": 4.639206886291504, + "learning_rate": 0.00018953722595208236, + "loss": 3.3776, + "step": 1016 + }, + { + "epoch": 1.5, + "grad_norm": 5.658569812774658, + "learning_rate": 0.00018951632913550626, + "loss": 3.339, + "step": 1017 + }, + { + "epoch": 1.5, + "grad_norm": 8.45776081085205, + "learning_rate": 0.00018949541262593762, + "loss": 3.4214, + "step": 1018 + }, + { + "epoch": 1.5, + "grad_norm": 3.2400307655334473, + "learning_rate": 0.000189474476427978, + "loss": 3.3666, + "step": 1019 + }, + { + "epoch": 1.5, + "grad_norm": 6.397870063781738, + "learning_rate": 0.0001894535205462332, + "loss": 3.4133, + "step": 1020 + }, + { + "epoch": 1.5, + "grad_norm": 8.270928382873535, + "learning_rate": 0.00018943254498531337, + "loss": 3.4133, + "step": 1021 + }, + { + "epoch": 1.5, + "grad_norm": 5.132183074951172, + "learning_rate": 0.00018941154974983298, + "loss": 3.3984, + "step": 1022 + }, + { + "epoch": 1.5, + "grad_norm": 5.842728137969971, + "learning_rate": 0.0001893905348444109, + "loss": 3.4019, + "step": 1023 + }, + { + "epoch": 1.51, + "grad_norm": 8.618907928466797, + "learning_rate": 0.00018936950027367017, + "loss": 3.4036, + "step": 1024 + }, + { + "epoch": 1.51, + "grad_norm": 3.155228853225708, + "learning_rate": 0.00018934844604223835, + "loss": 3.3512, + "step": 1025 + }, + { + "epoch": 1.51, + "grad_norm": 6.209192752838135, + "learning_rate": 0.00018932737215474714, + "loss": 3.388, + "step": 1026 + }, + { + "epoch": 1.51, + "grad_norm": 7.953395843505859, + "learning_rate": 0.00018930627861583275, + "loss": 3.402, + "step": 1027 + }, + { + "epoch": 1.51, + "grad_norm": 4.540157794952393, + "learning_rate": 0.00018928516543013552, + "loss": 3.3162, + "step": 1028 + }, + { + "epoch": 1.51, + "grad_norm": 6.285725116729736, + "learning_rate": 0.0001892640326023003, + "loss": 3.4329, + "step": 1029 + }, + { + "epoch": 1.51, + "grad_norm": 9.266477584838867, + "learning_rate": 0.0001892428801369761, + "loss": 3.381, + "step": 1030 + }, + { + "epoch": 1.52, + "grad_norm": 4.977963924407959, + "learning_rate": 0.00018922170803881635, + "loss": 3.3758, + "step": 1031 + }, + { + "epoch": 1.52, + "grad_norm": 5.400416374206543, + "learning_rate": 0.0001892005163124788, + "loss": 3.3279, + "step": 1032 + }, + { + "epoch": 1.52, + "grad_norm": 6.933363914489746, + "learning_rate": 0.00018917930496262542, + "loss": 3.3655, + "step": 1033 + }, + { + "epoch": 1.52, + "grad_norm": 3.6254875659942627, + "learning_rate": 0.00018915807399392262, + "loss": 3.3451, + "step": 1034 + }, + { + "epoch": 1.52, + "grad_norm": 7.3330841064453125, + "learning_rate": 0.00018913682341104105, + "loss": 3.4291, + "step": 1035 + }, + { + "epoch": 1.52, + "grad_norm": 10.801877975463867, + "learning_rate": 0.0001891155532186557, + "loss": 3.4253, + "step": 1036 + }, + { + "epoch": 1.53, + "grad_norm": 5.197689533233643, + "learning_rate": 0.00018909426342144587, + "loss": 3.3764, + "step": 1037 + }, + { + "epoch": 1.53, + "grad_norm": 5.4762797355651855, + "learning_rate": 0.00018907295402409514, + "loss": 3.3669, + "step": 1038 + }, + { + "epoch": 1.53, + "grad_norm": 9.155226707458496, + "learning_rate": 0.0001890516250312915, + "loss": 3.3748, + "step": 1039 + }, + { + "epoch": 1.53, + "grad_norm": 4.974143028259277, + "learning_rate": 0.00018903027644772715, + "loss": 3.3812, + "step": 1040 + }, + { + "epoch": 1.53, + "grad_norm": 3.935563802719116, + "learning_rate": 0.00018900890827809862, + "loss": 3.3748, + "step": 1041 + }, + { + "epoch": 1.53, + "grad_norm": 5.06904935836792, + "learning_rate": 0.0001889875205271068, + "loss": 3.4328, + "step": 1042 + }, + { + "epoch": 1.53, + "grad_norm": 1.7735739946365356, + "learning_rate": 0.0001889661131994568, + "loss": 3.3331, + "step": 1043 + }, + { + "epoch": 1.54, + "grad_norm": 3.266242504119873, + "learning_rate": 0.00018894468629985814, + "loss": 3.3797, + "step": 1044 + }, + { + "epoch": 1.54, + "grad_norm": 1.9417365789413452, + "learning_rate": 0.0001889232398330246, + "loss": 3.3556, + "step": 1045 + }, + { + "epoch": 1.54, + "grad_norm": 2.2055392265319824, + "learning_rate": 0.00018890177380367419, + "loss": 3.3695, + "step": 1046 + }, + { + "epoch": 1.54, + "grad_norm": 3.9984004497528076, + "learning_rate": 0.00018888028821652936, + "loss": 3.4158, + "step": 1047 + }, + { + "epoch": 1.54, + "grad_norm": 2.413911819458008, + "learning_rate": 0.0001888587830763168, + "loss": 3.4492, + "step": 1048 + }, + { + "epoch": 1.54, + "grad_norm": 4.597828388214111, + "learning_rate": 0.00018883725838776742, + "loss": 3.3518, + "step": 1049 + }, + { + "epoch": 1.54, + "grad_norm": 4.952722549438477, + "learning_rate": 0.00018881571415561656, + "loss": 3.371, + "step": 1050 + }, + { + "epoch": 1.55, + "grad_norm": 2.318856954574585, + "learning_rate": 0.00018879415038460383, + "loss": 3.3974, + "step": 1051 + }, + { + "epoch": 1.55, + "grad_norm": 2.227694034576416, + "learning_rate": 0.00018877256707947306, + "loss": 3.3784, + "step": 1052 + }, + { + "epoch": 1.55, + "grad_norm": 4.430450916290283, + "learning_rate": 0.00018875096424497248, + "loss": 3.3792, + "step": 1053 + }, + { + "epoch": 1.55, + "grad_norm": 2.9697399139404297, + "learning_rate": 0.00018872934188585454, + "loss": 3.4059, + "step": 1054 + }, + { + "epoch": 1.55, + "grad_norm": 4.346686840057373, + "learning_rate": 0.000188707700006876, + "loss": 3.4155, + "step": 1055 + }, + { + "epoch": 1.55, + "grad_norm": 4.2472381591796875, + "learning_rate": 0.000188686038612798, + "loss": 3.4242, + "step": 1056 + }, + { + "epoch": 1.55, + "grad_norm": 3.157696008682251, + "learning_rate": 0.0001886643577083858, + "loss": 3.4913, + "step": 1057 + }, + { + "epoch": 1.56, + "grad_norm": 2.940396308898926, + "learning_rate": 0.00018864265729840913, + "loss": 3.4734, + "step": 1058 + }, + { + "epoch": 1.56, + "grad_norm": 4.0289154052734375, + "learning_rate": 0.00018862093738764192, + "loss": 3.443, + "step": 1059 + }, + { + "epoch": 1.56, + "grad_norm": 3.4877893924713135, + "learning_rate": 0.00018859919798086238, + "loss": 3.5322, + "step": 1060 + }, + { + "epoch": 1.56, + "grad_norm": 4.22461462020874, + "learning_rate": 0.00018857743908285305, + "loss": 3.4562, + "step": 1061 + }, + { + "epoch": 1.56, + "grad_norm": 4.031467437744141, + "learning_rate": 0.00018855566069840077, + "loss": 3.4653, + "step": 1062 + }, + { + "epoch": 1.56, + "grad_norm": 3.602692127227783, + "learning_rate": 0.00018853386283229658, + "loss": 3.4555, + "step": 1063 + }, + { + "epoch": 1.57, + "grad_norm": 2.7948570251464844, + "learning_rate": 0.00018851204548933593, + "loss": 3.4255, + "step": 1064 + }, + { + "epoch": 1.57, + "grad_norm": 4.970945358276367, + "learning_rate": 0.00018849020867431842, + "loss": 3.4463, + "step": 1065 + }, + { + "epoch": 1.57, + "grad_norm": 4.409501552581787, + "learning_rate": 0.00018846835239204804, + "loss": 3.3992, + "step": 1066 + }, + { + "epoch": 1.57, + "grad_norm": 2.8853704929351807, + "learning_rate": 0.000188446476647333, + "loss": 3.4709, + "step": 1067 + }, + { + "epoch": 1.57, + "grad_norm": 2.3841402530670166, + "learning_rate": 0.00018842458144498588, + "loss": 3.4342, + "step": 1068 + }, + { + "epoch": 1.57, + "grad_norm": 4.642658710479736, + "learning_rate": 0.00018840266678982342, + "loss": 3.4102, + "step": 1069 + }, + { + "epoch": 1.57, + "grad_norm": 4.265832424163818, + "learning_rate": 0.00018838073268666672, + "loss": 3.4367, + "step": 1070 + }, + { + "epoch": 1.58, + "grad_norm": 2.900568723678589, + "learning_rate": 0.00018835877914034113, + "loss": 3.4052, + "step": 1071 + }, + { + "epoch": 1.58, + "grad_norm": 2.4887197017669678, + "learning_rate": 0.00018833680615567628, + "loss": 3.4175, + "step": 1072 + }, + { + "epoch": 1.58, + "grad_norm": 5.449963092803955, + "learning_rate": 0.0001883148137375061, + "loss": 3.4521, + "step": 1073 + }, + { + "epoch": 1.58, + "grad_norm": 5.6257643699646, + "learning_rate": 0.00018829280189066873, + "loss": 3.4741, + "step": 1074 + }, + { + "epoch": 1.58, + "grad_norm": 2.799276828765869, + "learning_rate": 0.00018827077062000668, + "loss": 3.4431, + "step": 1075 + }, + { + "epoch": 1.58, + "grad_norm": 2.863265037536621, + "learning_rate": 0.00018824871993036664, + "loss": 3.3978, + "step": 1076 + }, + { + "epoch": 1.58, + "grad_norm": 4.204624176025391, + "learning_rate": 0.00018822664982659964, + "loss": 3.4559, + "step": 1077 + }, + { + "epoch": 1.59, + "grad_norm": 3.0599143505096436, + "learning_rate": 0.00018820456031356093, + "loss": 3.3883, + "step": 1078 + }, + { + "epoch": 1.59, + "grad_norm": 4.109722137451172, + "learning_rate": 0.00018818245139611011, + "loss": 3.4014, + "step": 1079 + }, + { + "epoch": 1.59, + "grad_norm": 3.4017956256866455, + "learning_rate": 0.00018816032307911092, + "loss": 3.4348, + "step": 1080 + }, + { + "epoch": 1.59, + "grad_norm": 4.216831207275391, + "learning_rate": 0.00018813817536743148, + "loss": 3.452, + "step": 1081 + }, + { + "epoch": 1.59, + "grad_norm": 3.819659471511841, + "learning_rate": 0.00018811600826594416, + "loss": 3.3937, + "step": 1082 + }, + { + "epoch": 1.59, + "grad_norm": 3.5902295112609863, + "learning_rate": 0.00018809382177952552, + "loss": 3.3992, + "step": 1083 + }, + { + "epoch": 1.59, + "grad_norm": 2.6973037719726562, + "learning_rate": 0.0001880716159130565, + "loss": 3.4278, + "step": 1084 + }, + { + "epoch": 1.6, + "grad_norm": 5.055513858795166, + "learning_rate": 0.0001880493906714222, + "loss": 3.3913, + "step": 1085 + }, + { + "epoch": 1.6, + "grad_norm": 4.533970832824707, + "learning_rate": 0.00018802714605951199, + "loss": 3.4047, + "step": 1086 + }, + { + "epoch": 1.6, + "grad_norm": 2.5329983234405518, + "learning_rate": 0.00018800488208221963, + "loss": 3.4291, + "step": 1087 + }, + { + "epoch": 1.6, + "grad_norm": 1.993481993675232, + "learning_rate": 0.00018798259874444297, + "loss": 3.4334, + "step": 1088 + }, + { + "epoch": 1.6, + "grad_norm": 6.799105167388916, + "learning_rate": 0.00018796029605108424, + "loss": 3.4107, + "step": 1089 + }, + { + "epoch": 1.6, + "grad_norm": 6.37121057510376, + "learning_rate": 0.00018793797400704985, + "loss": 3.4237, + "step": 1090 + }, + { + "epoch": 1.61, + "grad_norm": 1.2064040899276733, + "learning_rate": 0.0001879156326172505, + "loss": 3.4487, + "step": 1091 + }, + { + "epoch": 1.61, + "grad_norm": 0.6396639943122864, + "learning_rate": 0.00018789327188660116, + "loss": 3.3509, + "step": 1092 + }, + { + "epoch": 1.61, + "grad_norm": 2.529014825820923, + "learning_rate": 0.00018787089182002105, + "loss": 3.4014, + "step": 1093 + }, + { + "epoch": 1.61, + "grad_norm": 2.6367127895355225, + "learning_rate": 0.00018784849242243362, + "loss": 3.3676, + "step": 1094 + }, + { + "epoch": 1.61, + "grad_norm": 0.968987226486206, + "learning_rate": 0.00018782607369876659, + "loss": 3.4156, + "step": 1095 + }, + { + "epoch": 1.61, + "grad_norm": 5.669402599334717, + "learning_rate": 0.00018780363565395192, + "loss": 3.4195, + "step": 1096 + }, + { + "epoch": 1.61, + "grad_norm": 4.468107223510742, + "learning_rate": 0.00018778117829292587, + "loss": 3.4453, + "step": 1097 + }, + { + "epoch": 1.62, + "grad_norm": 3.340261220932007, + "learning_rate": 0.00018775870162062886, + "loss": 3.4024, + "step": 1098 + }, + { + "epoch": 1.62, + "grad_norm": 2.798548698425293, + "learning_rate": 0.00018773620564200565, + "loss": 3.503, + "step": 1099 + }, + { + "epoch": 1.62, + "grad_norm": 5.848763465881348, + "learning_rate": 0.00018771369036200513, + "loss": 3.3962, + "step": 1100 + }, + { + "epoch": 1.62, + "grad_norm": 6.020981788635254, + "learning_rate": 0.00018769115578558062, + "loss": 3.4708, + "step": 1101 + }, + { + "epoch": 1.62, + "grad_norm": 2.3334991931915283, + "learning_rate": 0.00018766860191768946, + "loss": 3.3861, + "step": 1102 + }, + { + "epoch": 1.62, + "grad_norm": 3.1461942195892334, + "learning_rate": 0.00018764602876329345, + "loss": 3.3828, + "step": 1103 + }, + { + "epoch": 1.62, + "grad_norm": 2.833292007446289, + "learning_rate": 0.00018762343632735848, + "loss": 3.4628, + "step": 1104 + }, + { + "epoch": 1.63, + "grad_norm": 1.1478309631347656, + "learning_rate": 0.00018760082461485476, + "loss": 3.3512, + "step": 1105 + }, + { + "epoch": 1.63, + "grad_norm": 4.6011199951171875, + "learning_rate": 0.00018757819363075665, + "loss": 3.4301, + "step": 1106 + }, + { + "epoch": 1.63, + "grad_norm": 3.997016191482544, + "learning_rate": 0.0001875555433800429, + "loss": 3.4099, + "step": 1107 + }, + { + "epoch": 1.63, + "grad_norm": 4.506150722503662, + "learning_rate": 0.00018753287386769638, + "loss": 3.4169, + "step": 1108 + }, + { + "epoch": 1.63, + "grad_norm": 4.197047710418701, + "learning_rate": 0.0001875101850987042, + "loss": 3.3518, + "step": 1109 + }, + { + "epoch": 1.63, + "grad_norm": 3.504765033721924, + "learning_rate": 0.0001874874770780578, + "loss": 3.3838, + "step": 1110 + }, + { + "epoch": 1.64, + "grad_norm": 2.7405476570129395, + "learning_rate": 0.00018746474981075272, + "loss": 3.376, + "step": 1111 + }, + { + "epoch": 1.64, + "grad_norm": 4.3409037590026855, + "learning_rate": 0.00018744200330178887, + "loss": 3.4184, + "step": 1112 + }, + { + "epoch": 1.64, + "grad_norm": 4.318120002746582, + "learning_rate": 0.0001874192375561703, + "loss": 3.4744, + "step": 1113 + }, + { + "epoch": 1.64, + "grad_norm": 3.5253307819366455, + "learning_rate": 0.00018739645257890533, + "loss": 3.4058, + "step": 1114 + }, + { + "epoch": 1.64, + "grad_norm": 3.368002414703369, + "learning_rate": 0.0001873736483750065, + "loss": 3.4013, + "step": 1115 + }, + { + "epoch": 1.64, + "grad_norm": 3.804229736328125, + "learning_rate": 0.00018735082494949058, + "loss": 3.3625, + "step": 1116 + }, + { + "epoch": 1.64, + "grad_norm": 3.6312084197998047, + "learning_rate": 0.00018732798230737855, + "loss": 3.3568, + "step": 1117 + }, + { + "epoch": 1.65, + "grad_norm": 3.4278273582458496, + "learning_rate": 0.00018730512045369568, + "loss": 3.4077, + "step": 1118 + }, + { + "epoch": 1.65, + "grad_norm": 3.177262306213379, + "learning_rate": 0.0001872822393934714, + "loss": 3.41, + "step": 1119 + }, + { + "epoch": 1.65, + "grad_norm": 3.8335347175598145, + "learning_rate": 0.00018725933913173938, + "loss": 3.3968, + "step": 1120 + }, + { + "epoch": 1.65, + "grad_norm": 3.43869948387146, + "learning_rate": 0.00018723641967353756, + "loss": 3.4295, + "step": 1121 + }, + { + "epoch": 1.65, + "grad_norm": 4.323538780212402, + "learning_rate": 0.00018721348102390803, + "loss": 3.3734, + "step": 1122 + }, + { + "epoch": 1.65, + "grad_norm": 3.6614975929260254, + "learning_rate": 0.00018719052318789717, + "loss": 3.3522, + "step": 1123 + }, + { + "epoch": 1.65, + "grad_norm": 3.442986249923706, + "learning_rate": 0.0001871675461705555, + "loss": 3.3743, + "step": 1124 + }, + { + "epoch": 1.66, + "grad_norm": 3.7068862915039062, + "learning_rate": 0.00018714454997693783, + "loss": 3.3611, + "step": 1125 + }, + { + "epoch": 1.66, + "grad_norm": 3.6128504276275635, + "learning_rate": 0.00018712153461210326, + "loss": 3.3678, + "step": 1126 + }, + { + "epoch": 1.66, + "grad_norm": 3.2057278156280518, + "learning_rate": 0.00018709850008111488, + "loss": 3.4118, + "step": 1127 + }, + { + "epoch": 1.66, + "grad_norm": 4.726739406585693, + "learning_rate": 0.0001870754463890402, + "loss": 3.3669, + "step": 1128 + }, + { + "epoch": 1.66, + "grad_norm": 4.606038570404053, + "learning_rate": 0.00018705237354095085, + "loss": 3.3316, + "step": 1129 + }, + { + "epoch": 1.66, + "grad_norm": 2.9650778770446777, + "learning_rate": 0.00018702928154192274, + "loss": 3.3614, + "step": 1130 + }, + { + "epoch": 1.66, + "grad_norm": 2.7509312629699707, + "learning_rate": 0.00018700617039703595, + "loss": 3.3797, + "step": 1131 + }, + { + "epoch": 1.67, + "grad_norm": 5.575433731079102, + "learning_rate": 0.00018698304011137475, + "loss": 3.384, + "step": 1132 + }, + { + "epoch": 1.67, + "grad_norm": 4.589602947235107, + "learning_rate": 0.00018695989069002764, + "loss": 3.3729, + "step": 1133 + }, + { + "epoch": 1.67, + "grad_norm": 2.780677318572998, + "learning_rate": 0.00018693672213808738, + "loss": 3.3704, + "step": 1134 + }, + { + "epoch": 1.67, + "grad_norm": 1.9864367246627808, + "learning_rate": 0.00018691353446065087, + "loss": 3.3718, + "step": 1135 + }, + { + "epoch": 1.67, + "grad_norm": 5.260591983795166, + "learning_rate": 0.00018689032766281924, + "loss": 3.3856, + "step": 1136 + }, + { + "epoch": 1.67, + "grad_norm": 4.4715447425842285, + "learning_rate": 0.00018686710174969787, + "loss": 3.3433, + "step": 1137 + }, + { + "epoch": 1.68, + "grad_norm": 3.0349228382110596, + "learning_rate": 0.00018684385672639622, + "loss": 3.3361, + "step": 1138 + }, + { + "epoch": 1.68, + "grad_norm": 2.5944900512695312, + "learning_rate": 0.0001868205925980281, + "loss": 3.3379, + "step": 1139 + }, + { + "epoch": 1.68, + "grad_norm": 4.727088928222656, + "learning_rate": 0.00018679730936971148, + "loss": 3.3889, + "step": 1140 + }, + { + "epoch": 1.68, + "grad_norm": 5.072381019592285, + "learning_rate": 0.0001867740070465685, + "loss": 3.4185, + "step": 1141 + }, + { + "epoch": 1.68, + "grad_norm": 2.782468318939209, + "learning_rate": 0.00018675068563372545, + "loss": 3.3635, + "step": 1142 + }, + { + "epoch": 1.68, + "grad_norm": 2.7385001182556152, + "learning_rate": 0.000186727345136313, + "loss": 3.4115, + "step": 1143 + }, + { + "epoch": 1.68, + "grad_norm": 4.811634540557861, + "learning_rate": 0.0001867039855594658, + "loss": 3.4155, + "step": 1144 + }, + { + "epoch": 1.69, + "grad_norm": 4.225504398345947, + "learning_rate": 0.00018668060690832285, + "loss": 3.3729, + "step": 1145 + }, + { + "epoch": 1.69, + "grad_norm": 3.26381254196167, + "learning_rate": 0.0001866572091880273, + "loss": 3.3982, + "step": 1146 + }, + { + "epoch": 1.69, + "grad_norm": 2.842423915863037, + "learning_rate": 0.00018663379240372647, + "loss": 3.387, + "step": 1147 + }, + { + "epoch": 1.69, + "grad_norm": 4.7502217292785645, + "learning_rate": 0.0001866103565605719, + "loss": 3.377, + "step": 1148 + }, + { + "epoch": 1.69, + "grad_norm": 4.280509948730469, + "learning_rate": 0.00018658690166371933, + "loss": 3.3774, + "step": 1149 + }, + { + "epoch": 1.69, + "grad_norm": 3.5164103507995605, + "learning_rate": 0.00018656342771832866, + "loss": 3.357, + "step": 1150 + }, + { + "epoch": 1.69, + "grad_norm": 2.891474485397339, + "learning_rate": 0.00018653993472956404, + "loss": 3.3755, + "step": 1151 + }, + { + "epoch": 1.7, + "grad_norm": 4.281400203704834, + "learning_rate": 0.00018651642270259373, + "loss": 3.3511, + "step": 1152 + }, + { + "epoch": 1.7, + "grad_norm": 4.510589122772217, + "learning_rate": 0.00018649289164259024, + "loss": 3.4113, + "step": 1153 + }, + { + "epoch": 1.7, + "grad_norm": 2.6343142986297607, + "learning_rate": 0.00018646934155473022, + "loss": 3.3291, + "step": 1154 + }, + { + "epoch": 1.7, + "grad_norm": 2.3589959144592285, + "learning_rate": 0.00018644577244419455, + "loss": 3.3638, + "step": 1155 + }, + { + "epoch": 1.7, + "grad_norm": 5.584714889526367, + "learning_rate": 0.00018642218431616828, + "loss": 3.32, + "step": 1156 + }, + { + "epoch": 1.7, + "grad_norm": 4.909949779510498, + "learning_rate": 0.00018639857717584065, + "loss": 3.3947, + "step": 1157 + }, + { + "epoch": 1.7, + "grad_norm": 2.9206135272979736, + "learning_rate": 0.00018637495102840506, + "loss": 3.3826, + "step": 1158 + }, + { + "epoch": 1.71, + "grad_norm": 2.58848237991333, + "learning_rate": 0.0001863513058790591, + "loss": 3.3237, + "step": 1159 + }, + { + "epoch": 1.71, + "grad_norm": 5.167590141296387, + "learning_rate": 0.00018632764173300457, + "loss": 3.3972, + "step": 1160 + }, + { + "epoch": 1.71, + "grad_norm": 5.174654483795166, + "learning_rate": 0.00018630395859544735, + "loss": 3.357, + "step": 1161 + }, + { + "epoch": 1.71, + "grad_norm": 2.867471218109131, + "learning_rate": 0.00018628025647159766, + "loss": 3.3406, + "step": 1162 + }, + { + "epoch": 1.71, + "grad_norm": 2.780794620513916, + "learning_rate": 0.00018625653536666974, + "loss": 3.3527, + "step": 1163 + }, + { + "epoch": 1.71, + "grad_norm": 4.502468585968018, + "learning_rate": 0.00018623279528588214, + "loss": 3.3216, + "step": 1164 + }, + { + "epoch": 1.72, + "grad_norm": 3.549867868423462, + "learning_rate": 0.00018620903623445747, + "loss": 3.3949, + "step": 1165 + }, + { + "epoch": 1.72, + "grad_norm": 3.7050774097442627, + "learning_rate": 0.00018618525821762256, + "loss": 3.395, + "step": 1166 + }, + { + "epoch": 1.72, + "grad_norm": 3.520019769668579, + "learning_rate": 0.00018616146124060846, + "loss": 3.3757, + "step": 1167 + }, + { + "epoch": 1.72, + "grad_norm": 3.870093822479248, + "learning_rate": 0.00018613764530865028, + "loss": 3.3712, + "step": 1168 + }, + { + "epoch": 1.72, + "grad_norm": 3.196948766708374, + "learning_rate": 0.0001861138104269874, + "loss": 3.3337, + "step": 1169 + }, + { + "epoch": 1.72, + "grad_norm": 4.416697025299072, + "learning_rate": 0.00018608995660086337, + "loss": 3.4137, + "step": 1170 + }, + { + "epoch": 1.72, + "grad_norm": 3.9429569244384766, + "learning_rate": 0.00018606608383552582, + "loss": 3.389, + "step": 1171 + }, + { + "epoch": 1.73, + "grad_norm": 3.530494213104248, + "learning_rate": 0.0001860421921362266, + "loss": 3.3752, + "step": 1172 + }, + { + "epoch": 1.73, + "grad_norm": 3.766986608505249, + "learning_rate": 0.00018601828150822177, + "loss": 3.3386, + "step": 1173 + }, + { + "epoch": 1.73, + "grad_norm": 3.8910138607025146, + "learning_rate": 0.00018599435195677146, + "loss": 3.348, + "step": 1174 + }, + { + "epoch": 1.73, + "grad_norm": 3.5032382011413574, + "learning_rate": 0.00018597040348714003, + "loss": 3.408, + "step": 1175 + }, + { + "epoch": 1.73, + "grad_norm": 3.5396668910980225, + "learning_rate": 0.000185946436104596, + "loss": 3.317, + "step": 1176 + }, + { + "epoch": 1.73, + "grad_norm": 3.7018423080444336, + "learning_rate": 0.00018592244981441197, + "loss": 3.3456, + "step": 1177 + }, + { + "epoch": 1.73, + "grad_norm": 3.41996169090271, + "learning_rate": 0.00018589844462186484, + "loss": 3.4147, + "step": 1178 + }, + { + "epoch": 1.74, + "grad_norm": 2.907388210296631, + "learning_rate": 0.00018587442053223557, + "loss": 3.3377, + "step": 1179 + }, + { + "epoch": 1.74, + "grad_norm": 4.651727676391602, + "learning_rate": 0.00018585037755080926, + "loss": 3.3401, + "step": 1180 + }, + { + "epoch": 1.74, + "grad_norm": 4.285563945770264, + "learning_rate": 0.00018582631568287523, + "loss": 3.3671, + "step": 1181 + }, + { + "epoch": 1.74, + "grad_norm": 3.3015687465667725, + "learning_rate": 0.00018580223493372695, + "loss": 3.3049, + "step": 1182 + }, + { + "epoch": 1.74, + "grad_norm": 2.8395988941192627, + "learning_rate": 0.000185778135308662, + "loss": 3.3588, + "step": 1183 + }, + { + "epoch": 1.74, + "eval_loss": 3.1614086627960205, + "eval_runtime": 1168.1043, + "eval_samples_per_second": 14.675, + "eval_steps_per_second": 0.918, + "step": 1183 + }, + { + "epoch": 1.74, + "grad_norm": 4.551041603088379, + "learning_rate": 0.00018575401681298216, + "loss": 3.3579, + "step": 1184 + }, + { + "epoch": 1.74, + "grad_norm": 4.050691604614258, + "learning_rate": 0.00018572987945199328, + "loss": 3.3174, + "step": 1185 + }, + { + "epoch": 1.75, + "grad_norm": 3.483586072921753, + "learning_rate": 0.0001857057232310055, + "loss": 3.3689, + "step": 1186 + }, + { + "epoch": 1.75, + "grad_norm": 3.430351734161377, + "learning_rate": 0.00018568154815533293, + "loss": 3.3624, + "step": 1187 + }, + { + "epoch": 1.75, + "grad_norm": 4.206818103790283, + "learning_rate": 0.00018565735423029404, + "loss": 3.3365, + "step": 1188 + }, + { + "epoch": 1.75, + "grad_norm": 3.6076526641845703, + "learning_rate": 0.00018563314146121122, + "loss": 3.3728, + "step": 1189 + }, + { + "epoch": 1.75, + "grad_norm": 3.844933271408081, + "learning_rate": 0.0001856089098534112, + "loss": 3.3577, + "step": 1190 + }, + { + "epoch": 1.75, + "grad_norm": 3.1116933822631836, + "learning_rate": 0.00018558465941222472, + "loss": 3.3242, + "step": 1191 + }, + { + "epoch": 1.76, + "grad_norm": 4.367002010345459, + "learning_rate": 0.00018556039014298672, + "loss": 3.3382, + "step": 1192 + }, + { + "epoch": 1.76, + "grad_norm": 4.229283332824707, + "learning_rate": 0.0001855361020510363, + "loss": 3.3318, + "step": 1193 + }, + { + "epoch": 1.76, + "grad_norm": 3.6189987659454346, + "learning_rate": 0.00018551179514171668, + "loss": 3.3879, + "step": 1194 + }, + { + "epoch": 1.76, + "grad_norm": 3.045738935470581, + "learning_rate": 0.0001854874694203752, + "loss": 3.3407, + "step": 1195 + }, + { + "epoch": 1.76, + "grad_norm": 4.931478977203369, + "learning_rate": 0.00018546312489236335, + "loss": 3.3428, + "step": 1196 + }, + { + "epoch": 1.76, + "grad_norm": 4.312923908233643, + "learning_rate": 0.00018543876156303676, + "loss": 3.3768, + "step": 1197 + }, + { + "epoch": 1.76, + "grad_norm": 3.236755132675171, + "learning_rate": 0.00018541437943775523, + "loss": 3.3637, + "step": 1198 + }, + { + "epoch": 1.77, + "grad_norm": 2.9794223308563232, + "learning_rate": 0.00018538997852188266, + "loss": 3.2961, + "step": 1199 + }, + { + "epoch": 1.77, + "grad_norm": 4.777314186096191, + "learning_rate": 0.00018536555882078706, + "loss": 3.3838, + "step": 1200 + }, + { + "epoch": 1.77, + "grad_norm": 4.084963321685791, + "learning_rate": 0.0001853411203398406, + "loss": 3.3525, + "step": 1201 + }, + { + "epoch": 1.77, + "grad_norm": 3.598912477493286, + "learning_rate": 0.0001853166630844196, + "loss": 3.3362, + "step": 1202 + }, + { + "epoch": 1.77, + "grad_norm": 3.44163179397583, + "learning_rate": 0.0001852921870599045, + "loss": 3.4129, + "step": 1203 + }, + { + "epoch": 1.77, + "grad_norm": 3.8749027252197266, + "learning_rate": 0.00018526769227167984, + "loss": 3.3069, + "step": 1204 + }, + { + "epoch": 1.77, + "grad_norm": 3.8452203273773193, + "learning_rate": 0.00018524317872513433, + "loss": 3.3084, + "step": 1205 + }, + { + "epoch": 1.78, + "grad_norm": 3.8901963233947754, + "learning_rate": 0.00018521864642566076, + "loss": 3.322, + "step": 1206 + }, + { + "epoch": 1.78, + "grad_norm": 3.340991973876953, + "learning_rate": 0.00018519409537865608, + "loss": 3.3373, + "step": 1207 + }, + { + "epoch": 1.78, + "grad_norm": 4.287864685058594, + "learning_rate": 0.0001851695255895214, + "loss": 3.2985, + "step": 1208 + }, + { + "epoch": 1.78, + "grad_norm": 3.954977512359619, + "learning_rate": 0.00018514493706366184, + "loss": 3.3286, + "step": 1209 + }, + { + "epoch": 1.78, + "grad_norm": 3.5455312728881836, + "learning_rate": 0.00018512032980648674, + "loss": 3.3191, + "step": 1210 + }, + { + "epoch": 1.78, + "grad_norm": 3.2507898807525635, + "learning_rate": 0.00018509570382340955, + "loss": 3.3356, + "step": 1211 + }, + { + "epoch": 1.79, + "grad_norm": 4.234015464782715, + "learning_rate": 0.0001850710591198478, + "loss": 3.365, + "step": 1212 + }, + { + "epoch": 1.79, + "grad_norm": 3.4031906127929688, + "learning_rate": 0.00018504639570122316, + "loss": 3.3405, + "step": 1213 + }, + { + "epoch": 1.79, + "grad_norm": 3.9651551246643066, + "learning_rate": 0.00018502171357296144, + "loss": 3.3953, + "step": 1214 + }, + { + "epoch": 1.79, + "grad_norm": 3.6755764484405518, + "learning_rate": 0.00018499701274049248, + "loss": 3.3102, + "step": 1215 + }, + { + "epoch": 1.79, + "grad_norm": 3.6327264308929443, + "learning_rate": 0.00018497229320925038, + "loss": 3.318, + "step": 1216 + }, + { + "epoch": 1.79, + "grad_norm": 3.4605491161346436, + "learning_rate": 0.0001849475549846732, + "loss": 3.3059, + "step": 1217 + }, + { + "epoch": 1.79, + "grad_norm": 3.7450873851776123, + "learning_rate": 0.00018492279807220326, + "loss": 3.3303, + "step": 1218 + }, + { + "epoch": 1.8, + "grad_norm": 3.5053107738494873, + "learning_rate": 0.00018489802247728684, + "loss": 3.3211, + "step": 1219 + }, + { + "epoch": 1.8, + "grad_norm": 3.7314960956573486, + "learning_rate": 0.00018487322820537444, + "loss": 3.3359, + "step": 1220 + }, + { + "epoch": 1.8, + "grad_norm": 3.397026777267456, + "learning_rate": 0.00018484841526192065, + "loss": 3.3199, + "step": 1221 + }, + { + "epoch": 1.8, + "grad_norm": 3.9689042568206787, + "learning_rate": 0.00018482358365238413, + "loss": 3.3249, + "step": 1222 + }, + { + "epoch": 1.8, + "grad_norm": 3.4028615951538086, + "learning_rate": 0.00018479873338222763, + "loss": 3.3234, + "step": 1223 + }, + { + "epoch": 1.8, + "grad_norm": 4.0719780921936035, + "learning_rate": 0.00018477386445691813, + "loss": 3.3755, + "step": 1224 + }, + { + "epoch": 1.8, + "grad_norm": 3.8761250972747803, + "learning_rate": 0.00018474897688192655, + "loss": 3.3313, + "step": 1225 + }, + { + "epoch": 1.81, + "grad_norm": 3.763307809829712, + "learning_rate": 0.00018472407066272804, + "loss": 3.3112, + "step": 1226 + }, + { + "epoch": 1.81, + "grad_norm": 3.3675754070281982, + "learning_rate": 0.00018469914580480177, + "loss": 3.373, + "step": 1227 + }, + { + "epoch": 1.81, + "grad_norm": 3.7994701862335205, + "learning_rate": 0.00018467420231363103, + "loss": 3.3034, + "step": 1228 + }, + { + "epoch": 1.81, + "grad_norm": 3.433627128601074, + "learning_rate": 0.00018464924019470327, + "loss": 3.2695, + "step": 1229 + }, + { + "epoch": 1.81, + "grad_norm": 3.7266974449157715, + "learning_rate": 0.00018462425945350996, + "loss": 3.3498, + "step": 1230 + }, + { + "epoch": 1.81, + "grad_norm": 3.44728422164917, + "learning_rate": 0.00018459926009554666, + "loss": 3.3682, + "step": 1231 + }, + { + "epoch": 1.81, + "grad_norm": 4.175528049468994, + "learning_rate": 0.00018457424212631312, + "loss": 3.3374, + "step": 1232 + }, + { + "epoch": 1.82, + "grad_norm": 4.267003059387207, + "learning_rate": 0.0001845492055513131, + "loss": 3.2985, + "step": 1233 + }, + { + "epoch": 1.82, + "grad_norm": 3.53112530708313, + "learning_rate": 0.0001845241503760545, + "loss": 3.3654, + "step": 1234 + }, + { + "epoch": 1.82, + "grad_norm": 3.1011736392974854, + "learning_rate": 0.00018449907660604927, + "loss": 3.3169, + "step": 1235 + }, + { + "epoch": 1.82, + "grad_norm": 4.862052917480469, + "learning_rate": 0.0001844739842468135, + "loss": 3.329, + "step": 1236 + }, + { + "epoch": 1.82, + "grad_norm": 4.767108917236328, + "learning_rate": 0.00018444887330386727, + "loss": 3.3546, + "step": 1237 + }, + { + "epoch": 1.82, + "grad_norm": 3.3005921840667725, + "learning_rate": 0.00018442374378273488, + "loss": 3.338, + "step": 1238 + }, + { + "epoch": 1.83, + "grad_norm": 3.2703089714050293, + "learning_rate": 0.00018439859568894463, + "loss": 3.281, + "step": 1239 + }, + { + "epoch": 1.83, + "grad_norm": 4.4580888748168945, + "learning_rate": 0.00018437342902802897, + "loss": 3.3748, + "step": 1240 + }, + { + "epoch": 1.83, + "grad_norm": 3.4954566955566406, + "learning_rate": 0.00018434824380552433, + "loss": 3.3291, + "step": 1241 + }, + { + "epoch": 1.83, + "grad_norm": 3.5984458923339844, + "learning_rate": 0.00018432304002697137, + "loss": 3.3834, + "step": 1242 + }, + { + "epoch": 1.83, + "grad_norm": 3.217467784881592, + "learning_rate": 0.00018429781769791468, + "loss": 3.3814, + "step": 1243 + }, + { + "epoch": 1.83, + "grad_norm": 4.149362087249756, + "learning_rate": 0.00018427257682390304, + "loss": 3.3808, + "step": 1244 + }, + { + "epoch": 1.83, + "grad_norm": 3.7322371006011963, + "learning_rate": 0.0001842473174104893, + "loss": 3.3546, + "step": 1245 + }, + { + "epoch": 1.84, + "grad_norm": 3.667970895767212, + "learning_rate": 0.0001842220394632303, + "loss": 3.3144, + "step": 1246 + }, + { + "epoch": 1.84, + "grad_norm": 3.353289842605591, + "learning_rate": 0.00018419674298768705, + "loss": 3.367, + "step": 1247 + }, + { + "epoch": 1.84, + "grad_norm": 4.078193664550781, + "learning_rate": 0.00018417142798942462, + "loss": 3.3284, + "step": 1248 + }, + { + "epoch": 1.84, + "grad_norm": 3.791130304336548, + "learning_rate": 0.0001841460944740121, + "loss": 3.3223, + "step": 1249 + }, + { + "epoch": 1.84, + "grad_norm": 3.9178853034973145, + "learning_rate": 0.00018412074244702274, + "loss": 3.4204, + "step": 1250 + }, + { + "epoch": 1.84, + "grad_norm": 3.1923162937164307, + "learning_rate": 0.00018409537191403378, + "loss": 3.2976, + "step": 1251 + }, + { + "epoch": 1.84, + "grad_norm": 4.827831745147705, + "learning_rate": 0.00018406998288062658, + "loss": 3.3697, + "step": 1252 + }, + { + "epoch": 1.85, + "grad_norm": 5.54097843170166, + "learning_rate": 0.00018404457535238657, + "loss": 3.3319, + "step": 1253 + }, + { + "epoch": 1.85, + "grad_norm": 2.249396562576294, + "learning_rate": 0.00018401914933490322, + "loss": 3.3106, + "step": 1254 + }, + { + "epoch": 1.85, + "grad_norm": 1.4117510318756104, + "learning_rate": 0.0001839937048337701, + "loss": 3.3664, + "step": 1255 + }, + { + "epoch": 1.85, + "grad_norm": 5.818370819091797, + "learning_rate": 0.0001839682418545848, + "loss": 3.3226, + "step": 1256 + }, + { + "epoch": 1.85, + "grad_norm": 4.127283573150635, + "learning_rate": 0.00018394276040294903, + "loss": 3.2937, + "step": 1257 + }, + { + "epoch": 1.85, + "grad_norm": 3.759086847305298, + "learning_rate": 0.00018391726048446852, + "loss": 3.3379, + "step": 1258 + }, + { + "epoch": 1.85, + "grad_norm": 4.152595520019531, + "learning_rate": 0.0001838917421047531, + "loss": 3.3433, + "step": 1259 + }, + { + "epoch": 1.86, + "grad_norm": 2.377365827560425, + "learning_rate": 0.00018386620526941662, + "loss": 3.3103, + "step": 1260 + }, + { + "epoch": 1.86, + "grad_norm": 1.4664210081100464, + "learning_rate": 0.00018384064998407706, + "loss": 3.3038, + "step": 1261 + }, + { + "epoch": 1.86, + "grad_norm": 5.424790859222412, + "learning_rate": 0.00018381507625435633, + "loss": 3.3518, + "step": 1262 + }, + { + "epoch": 1.86, + "grad_norm": 4.665937900543213, + "learning_rate": 0.00018378948408588058, + "loss": 3.3437, + "step": 1263 + }, + { + "epoch": 1.86, + "grad_norm": 2.893569231033325, + "learning_rate": 0.00018376387348427984, + "loss": 3.2693, + "step": 1264 + }, + { + "epoch": 1.86, + "grad_norm": 3.674450159072876, + "learning_rate": 0.0001837382444551883, + "loss": 3.3324, + "step": 1265 + }, + { + "epoch": 1.87, + "grad_norm": 2.98838210105896, + "learning_rate": 0.00018371259700424416, + "loss": 3.3242, + "step": 1266 + }, + { + "epoch": 1.87, + "grad_norm": 2.417462110519409, + "learning_rate": 0.00018368693113708968, + "loss": 3.3771, + "step": 1267 + }, + { + "epoch": 1.87, + "grad_norm": 4.9014458656311035, + "learning_rate": 0.00018366124685937122, + "loss": 3.3235, + "step": 1268 + }, + { + "epoch": 1.87, + "grad_norm": 4.8541693687438965, + "learning_rate": 0.00018363554417673915, + "loss": 3.3149, + "step": 1269 + }, + { + "epoch": 1.87, + "grad_norm": 3.2948014736175537, + "learning_rate": 0.0001836098230948478, + "loss": 3.3211, + "step": 1270 + }, + { + "epoch": 1.87, + "grad_norm": 3.0249054431915283, + "learning_rate": 0.00018358408361935573, + "loss": 3.3073, + "step": 1271 + }, + { + "epoch": 1.87, + "grad_norm": 4.5272722244262695, + "learning_rate": 0.00018355832575592541, + "loss": 3.2809, + "step": 1272 + }, + { + "epoch": 1.88, + "grad_norm": 3.790753126144409, + "learning_rate": 0.0001835325495102234, + "loss": 3.3502, + "step": 1273 + }, + { + "epoch": 1.88, + "grad_norm": 3.7797255516052246, + "learning_rate": 0.00018350675488792035, + "loss": 3.2818, + "step": 1274 + }, + { + "epoch": 1.88, + "grad_norm": 3.5615875720977783, + "learning_rate": 0.00018348094189469082, + "loss": 3.3617, + "step": 1275 + }, + { + "epoch": 1.88, + "grad_norm": 3.750861167907715, + "learning_rate": 0.00018345511053621355, + "loss": 3.3649, + "step": 1276 + }, + { + "epoch": 1.88, + "grad_norm": 3.202655076980591, + "learning_rate": 0.00018342926081817127, + "loss": 3.3013, + "step": 1277 + }, + { + "epoch": 1.88, + "grad_norm": 3.6577436923980713, + "learning_rate": 0.0001834033927462507, + "loss": 3.3074, + "step": 1278 + }, + { + "epoch": 1.88, + "grad_norm": 3.164114236831665, + "learning_rate": 0.0001833775063261427, + "loss": 3.3211, + "step": 1279 + }, + { + "epoch": 1.89, + "grad_norm": 3.7693588733673096, + "learning_rate": 0.00018335160156354206, + "loss": 3.3229, + "step": 1280 + }, + { + "epoch": 1.89, + "grad_norm": 3.5220518112182617, + "learning_rate": 0.00018332567846414768, + "loss": 3.304, + "step": 1281 + }, + { + "epoch": 1.89, + "grad_norm": 3.884568691253662, + "learning_rate": 0.00018329973703366242, + "loss": 3.2905, + "step": 1282 + }, + { + "epoch": 1.89, + "grad_norm": 3.471947431564331, + "learning_rate": 0.0001832737772777933, + "loss": 3.3356, + "step": 1283 + }, + { + "epoch": 1.89, + "grad_norm": 3.8152236938476562, + "learning_rate": 0.00018324779920225127, + "loss": 3.3036, + "step": 1284 + }, + { + "epoch": 1.89, + "grad_norm": 3.813997507095337, + "learning_rate": 0.0001832218028127513, + "loss": 3.3465, + "step": 1285 + }, + { + "epoch": 1.89, + "grad_norm": 3.5465102195739746, + "learning_rate": 0.00018319578811501244, + "loss": 3.3241, + "step": 1286 + }, + { + "epoch": 1.9, + "grad_norm": 3.25008487701416, + "learning_rate": 0.00018316975511475776, + "loss": 3.3111, + "step": 1287 + }, + { + "epoch": 1.9, + "grad_norm": 4.218858242034912, + "learning_rate": 0.00018314370381771432, + "loss": 3.2914, + "step": 1288 + }, + { + "epoch": 1.9, + "grad_norm": 3.7324070930480957, + "learning_rate": 0.00018311763422961325, + "loss": 3.3188, + "step": 1289 + }, + { + "epoch": 1.9, + "grad_norm": 3.4570472240448, + "learning_rate": 0.00018309154635618965, + "loss": 3.3034, + "step": 1290 + }, + { + "epoch": 1.9, + "grad_norm": 2.9619879722595215, + "learning_rate": 0.00018306544020318274, + "loss": 3.3231, + "step": 1291 + }, + { + "epoch": 1.9, + "grad_norm": 4.235932350158691, + "learning_rate": 0.00018303931577633566, + "loss": 3.3226, + "step": 1292 + }, + { + "epoch": 1.91, + "grad_norm": 3.5511438846588135, + "learning_rate": 0.00018301317308139561, + "loss": 3.3305, + "step": 1293 + }, + { + "epoch": 1.91, + "grad_norm": 3.656703233718872, + "learning_rate": 0.00018298701212411382, + "loss": 3.3261, + "step": 1294 + }, + { + "epoch": 1.91, + "grad_norm": 3.338941812515259, + "learning_rate": 0.00018296083291024552, + "loss": 3.3564, + "step": 1295 + }, + { + "epoch": 1.91, + "grad_norm": 3.9150142669677734, + "learning_rate": 0.00018293463544554993, + "loss": 3.3292, + "step": 1296 + }, + { + "epoch": 1.91, + "grad_norm": 10.699624061584473, + "learning_rate": 0.00018290841973579039, + "loss": 3.3042, + "step": 1297 + }, + { + "epoch": 1.91, + "grad_norm": 1.4264583587646484, + "learning_rate": 0.0001828821857867341, + "loss": 3.3268, + "step": 1298 + }, + { + "epoch": 1.91, + "grad_norm": 6.050600528717041, + "learning_rate": 0.0001828559336041524, + "loss": 3.3629, + "step": 1299 + }, + { + "epoch": 1.92, + "grad_norm": 6.324507236480713, + "learning_rate": 0.00018282966319382057, + "loss": 3.3402, + "step": 1300 + }, + { + "epoch": 1.92, + "grad_norm": 1.820309042930603, + "learning_rate": 0.00018280337456151797, + "loss": 3.2955, + "step": 1301 + }, + { + "epoch": 1.92, + "grad_norm": 7.433422565460205, + "learning_rate": 0.00018277706771302788, + "loss": 3.3309, + "step": 1302 + }, + { + "epoch": 1.92, + "grad_norm": 9.409677505493164, + "learning_rate": 0.00018275074265413762, + "loss": 3.4307, + "step": 1303 + }, + { + "epoch": 1.92, + "grad_norm": 5.268000602722168, + "learning_rate": 0.00018272439939063856, + "loss": 3.4344, + "step": 1304 + }, + { + "epoch": 1.92, + "grad_norm": 5.51171875, + "learning_rate": 0.00018269803792832603, + "loss": 3.3385, + "step": 1305 + }, + { + "epoch": 1.92, + "grad_norm": 5.903563976287842, + "learning_rate": 0.00018267165827299934, + "loss": 3.3243, + "step": 1306 + }, + { + "epoch": 1.93, + "grad_norm": 3.7264065742492676, + "learning_rate": 0.00018264526043046192, + "loss": 3.3759, + "step": 1307 + }, + { + "epoch": 1.93, + "grad_norm": 5.053271293640137, + "learning_rate": 0.00018261884440652106, + "loss": 3.3202, + "step": 1308 + }, + { + "epoch": 1.93, + "grad_norm": 2.33587646484375, + "learning_rate": 0.0001825924102069881, + "loss": 3.2899, + "step": 1309 + }, + { + "epoch": 1.93, + "grad_norm": 4.696453094482422, + "learning_rate": 0.00018256595783767843, + "loss": 3.359, + "step": 1310 + }, + { + "epoch": 1.93, + "grad_norm": 4.715218544006348, + "learning_rate": 0.00018253948730441132, + "loss": 3.3466, + "step": 1311 + }, + { + "epoch": 1.93, + "grad_norm": 2.377490997314453, + "learning_rate": 0.0001825129986130102, + "loss": 3.3496, + "step": 1312 + }, + { + "epoch": 1.94, + "grad_norm": 2.269477367401123, + "learning_rate": 0.00018248649176930235, + "loss": 3.3404, + "step": 1313 + }, + { + "epoch": 1.94, + "grad_norm": 4.577181816101074, + "learning_rate": 0.0001824599667791191, + "loss": 3.3864, + "step": 1314 + }, + { + "epoch": 1.94, + "grad_norm": 4.408148288726807, + "learning_rate": 0.0001824334236482958, + "loss": 3.3405, + "step": 1315 + }, + { + "epoch": 1.94, + "grad_norm": 2.9572770595550537, + "learning_rate": 0.00018240686238267173, + "loss": 3.323, + "step": 1316 + }, + { + "epoch": 1.94, + "grad_norm": 4.4048380851745605, + "learning_rate": 0.0001823802829880902, + "loss": 3.2676, + "step": 1317 + }, + { + "epoch": 1.94, + "grad_norm": 1.4889674186706543, + "learning_rate": 0.0001823536854703985, + "loss": 3.2902, + "step": 1318 + }, + { + "epoch": 1.94, + "grad_norm": 1.6742371320724487, + "learning_rate": 0.00018232706983544792, + "loss": 3.2945, + "step": 1319 + }, + { + "epoch": 1.95, + "grad_norm": 3.9357426166534424, + "learning_rate": 0.0001823004360890937, + "loss": 3.3486, + "step": 1320 + }, + { + "epoch": 1.95, + "grad_norm": 2.596331834793091, + "learning_rate": 0.0001822737842371951, + "loss": 3.3539, + "step": 1321 + }, + { + "epoch": 1.95, + "grad_norm": 4.866861820220947, + "learning_rate": 0.00018224711428561533, + "loss": 3.3449, + "step": 1322 + }, + { + "epoch": 1.95, + "grad_norm": 5.0658955574035645, + "learning_rate": 0.0001822204262402216, + "loss": 3.3107, + "step": 1323 + }, + { + "epoch": 1.95, + "grad_norm": 1.9593414068222046, + "learning_rate": 0.00018219372010688515, + "loss": 3.348, + "step": 1324 + }, + { + "epoch": 1.95, + "grad_norm": 2.256303310394287, + "learning_rate": 0.0001821669958914811, + "loss": 3.3012, + "step": 1325 + }, + { + "epoch": 1.95, + "grad_norm": 3.2346420288085938, + "learning_rate": 0.0001821402535998886, + "loss": 3.3372, + "step": 1326 + }, + { + "epoch": 1.96, + "grad_norm": 1.7795019149780273, + "learning_rate": 0.00018211349323799078, + "loss": 3.3459, + "step": 1327 + }, + { + "epoch": 1.96, + "grad_norm": 7.262651443481445, + "learning_rate": 0.00018208671481167478, + "loss": 3.3454, + "step": 1328 + }, + { + "epoch": 1.96, + "grad_norm": 8.501165390014648, + "learning_rate": 0.00018205991832683164, + "loss": 3.3952, + "step": 1329 + }, + { + "epoch": 1.96, + "grad_norm": 2.0047447681427, + "learning_rate": 0.00018203310378935638, + "loss": 3.3304, + "step": 1330 + }, + { + "epoch": 1.96, + "grad_norm": 6.586772918701172, + "learning_rate": 0.00018200627120514807, + "loss": 3.3235, + "step": 1331 + }, + { + "epoch": 1.96, + "grad_norm": 6.3678083419799805, + "learning_rate": 0.00018197942058010967, + "loss": 3.4105, + "step": 1332 + }, + { + "epoch": 1.96, + "grad_norm": 3.969449520111084, + "learning_rate": 0.00018195255192014816, + "loss": 3.3689, + "step": 1333 + }, + { + "epoch": 1.97, + "grad_norm": 5.26508092880249, + "learning_rate": 0.0001819256652311744, + "loss": 3.3417, + "step": 1334 + }, + { + "epoch": 1.97, + "grad_norm": 5.3484673500061035, + "learning_rate": 0.00018189876051910337, + "loss": 3.32, + "step": 1335 + }, + { + "epoch": 1.97, + "grad_norm": 3.664024829864502, + "learning_rate": 0.00018187183778985388, + "loss": 3.4004, + "step": 1336 + }, + { + "epoch": 1.97, + "grad_norm": 5.414148807525635, + "learning_rate": 0.00018184489704934877, + "loss": 3.2964, + "step": 1337 + }, + { + "epoch": 1.97, + "grad_norm": 1.5016615390777588, + "learning_rate": 0.0001818179383035148, + "loss": 3.3444, + "step": 1338 + }, + { + "epoch": 1.97, + "grad_norm": 7.925120830535889, + "learning_rate": 0.00018179096155828274, + "loss": 3.3881, + "step": 1339 + }, + { + "epoch": 1.98, + "grad_norm": 9.725903511047363, + "learning_rate": 0.00018176396681958724, + "loss": 3.3698, + "step": 1340 + }, + { + "epoch": 1.98, + "grad_norm": 5.562051773071289, + "learning_rate": 0.00018173695409336702, + "loss": 3.3285, + "step": 1341 + }, + { + "epoch": 1.98, + "grad_norm": 4.413424491882324, + "learning_rate": 0.00018170992338556468, + "loss": 3.3209, + "step": 1342 + }, + { + "epoch": 1.98, + "grad_norm": 5.979745864868164, + "learning_rate": 0.00018168287470212677, + "loss": 3.3173, + "step": 1343 + }, + { + "epoch": 1.98, + "grad_norm": 2.968111753463745, + "learning_rate": 0.00018165580804900384, + "loss": 3.3304, + "step": 1344 + }, + { + "epoch": 1.98, + "grad_norm": 6.121416091918945, + "learning_rate": 0.00018162872343215037, + "loss": 3.325, + "step": 1345 + }, + { + "epoch": 1.98, + "grad_norm": 5.006643772125244, + "learning_rate": 0.0001816016208575248, + "loss": 3.3311, + "step": 1346 + }, + { + "epoch": 1.99, + "grad_norm": 2.833677291870117, + "learning_rate": 0.00018157450033108954, + "loss": 3.3484, + "step": 1347 + }, + { + "epoch": 1.99, + "grad_norm": 3.661172389984131, + "learning_rate": 0.00018154736185881088, + "loss": 3.3119, + "step": 1348 + }, + { + "epoch": 1.99, + "grad_norm": 1.4361642599105835, + "learning_rate": 0.0001815202054466591, + "loss": 3.3744, + "step": 1349 + }, + { + "epoch": 1.99, + "grad_norm": 2.236370801925659, + "learning_rate": 0.00018149303110060846, + "loss": 3.285, + "step": 1350 + }, + { + "epoch": 1.99, + "grad_norm": 3.9948387145996094, + "learning_rate": 0.00018146583882663712, + "loss": 3.3107, + "step": 1351 + }, + { + "epoch": 1.99, + "grad_norm": 2.2754180431365967, + "learning_rate": 0.0001814386286307272, + "loss": 3.3646, + "step": 1352 + }, + { + "epoch": 1.99, + "eval_loss": 3.131267786026001, + "eval_runtime": 1163.1, + "eval_samples_per_second": 14.738, + "eval_steps_per_second": 0.922, + "step": 1352 + }, + { + "epoch": 1.99, + "grad_norm": 4.385812282562256, + "learning_rate": 0.00018141140051886474, + "loss": 3.3364, + "step": 1353 + }, + { + "epoch": 2.0, + "grad_norm": 2.7103238105773926, + "learning_rate": 0.0001813841544970398, + "loss": 3.3096, + "step": 1354 + }, + { + "epoch": 2.0, + "grad_norm": 4.108714580535889, + "learning_rate": 0.00018135689057124625, + "loss": 3.3506, + "step": 1355 + }, + { + "epoch": 2.0, + "grad_norm": 3.6100189685821533, + "learning_rate": 0.000181329608747482, + "loss": 3.2999, + "step": 1356 + }, + { + "epoch": 2.0, + "grad_norm": 3.728175640106201, + "learning_rate": 0.0001813023090317489, + "loss": 3.2945, + "step": 1357 + }, + { + "epoch": 2.0, + "grad_norm": 3.231055974960327, + "learning_rate": 0.00018127499143005268, + "loss": 3.3229, + "step": 1358 + }, + { + "epoch": 2.0, + "grad_norm": 3.6905510425567627, + "learning_rate": 0.00018124765594840298, + "loss": 3.3186, + "step": 1359 + }, + { + "epoch": 2.0, + "grad_norm": 3.330751657485962, + "learning_rate": 0.0001812203025928135, + "loss": 3.3526, + "step": 1360 + }, + { + "epoch": 2.01, + "grad_norm": 3.546685218811035, + "learning_rate": 0.00018119293136930175, + "loss": 3.2914, + "step": 1361 + }, + { + "epoch": 2.01, + "grad_norm": 3.202545404434204, + "learning_rate": 0.00018116554228388924, + "loss": 3.3155, + "step": 1362 + }, + { + "epoch": 2.01, + "grad_norm": 3.762091875076294, + "learning_rate": 0.00018113813534260137, + "loss": 3.2653, + "step": 1363 + }, + { + "epoch": 2.01, + "grad_norm": 3.2883477210998535, + "learning_rate": 0.00018111071055146752, + "loss": 3.2802, + "step": 1364 + }, + { + "epoch": 2.01, + "grad_norm": 4.180018424987793, + "learning_rate": 0.0001810832679165209, + "loss": 3.2743, + "step": 1365 + }, + { + "epoch": 2.01, + "grad_norm": 3.009850025177002, + "learning_rate": 0.00018105580744379873, + "loss": 3.3396, + "step": 1366 + }, + { + "epoch": 2.02, + "grad_norm": 3.3239614963531494, + "learning_rate": 0.00018102832913934215, + "loss": 3.302, + "step": 1367 + }, + { + "epoch": 2.0, + "grad_norm": 2.7158830165863037, + "learning_rate": 0.00018100083300919616, + "loss": 3.2927, + "step": 1368 + }, + { + "epoch": 2.0, + "grad_norm": 4.113391399383545, + "learning_rate": 0.00018097331905940977, + "loss": 3.258, + "step": 1369 + }, + { + "epoch": 2.0, + "grad_norm": 4.452334880828857, + "learning_rate": 0.00018094578729603583, + "loss": 3.2988, + "step": 1370 + }, + { + "epoch": 2.0, + "grad_norm": 2.8618204593658447, + "learning_rate": 0.00018091823772513118, + "loss": 3.2938, + "step": 1371 + }, + { + "epoch": 2.01, + "grad_norm": 2.671171188354492, + "learning_rate": 0.00018089067035275652, + "loss": 3.2756, + "step": 1372 + }, + { + "epoch": 2.01, + "grad_norm": 4.292674541473389, + "learning_rate": 0.0001808630851849765, + "loss": 3.2655, + "step": 1373 + }, + { + "epoch": 2.01, + "grad_norm": 4.163233757019043, + "learning_rate": 0.00018083548222785965, + "loss": 3.2825, + "step": 1374 + }, + { + "epoch": 2.01, + "grad_norm": 2.8590521812438965, + "learning_rate": 0.00018080786148747842, + "loss": 3.3127, + "step": 1375 + }, + { + "epoch": 2.01, + "grad_norm": 2.4040653705596924, + "learning_rate": 0.00018078022296990923, + "loss": 3.348, + "step": 1376 + }, + { + "epoch": 2.01, + "grad_norm": 5.037558555603027, + "learning_rate": 0.00018075256668123238, + "loss": 3.275, + "step": 1377 + }, + { + "epoch": 2.02, + "grad_norm": 4.536971569061279, + "learning_rate": 0.000180724892627532, + "loss": 3.2533, + "step": 1378 + }, + { + "epoch": 2.02, + "grad_norm": 2.9320995807647705, + "learning_rate": 0.00018069720081489624, + "loss": 3.3223, + "step": 1379 + }, + { + "epoch": 2.02, + "grad_norm": 2.6839616298675537, + "learning_rate": 0.00018066949124941712, + "loss": 3.3313, + "step": 1380 + }, + { + "epoch": 2.02, + "grad_norm": 3.9813764095306396, + "learning_rate": 0.00018064176393719055, + "loss": 3.2821, + "step": 1381 + }, + { + "epoch": 2.02, + "grad_norm": 3.4426040649414062, + "learning_rate": 0.00018061401888431634, + "loss": 3.3005, + "step": 1382 + }, + { + "epoch": 2.02, + "grad_norm": 3.2250571250915527, + "learning_rate": 0.00018058625609689824, + "loss": 3.2879, + "step": 1383 + }, + { + "epoch": 2.02, + "grad_norm": 2.6666173934936523, + "learning_rate": 0.00018055847558104383, + "loss": 3.2522, + "step": 1384 + }, + { + "epoch": 2.03, + "grad_norm": 4.8121843338012695, + "learning_rate": 0.00018053067734286472, + "loss": 3.3785, + "step": 1385 + }, + { + "epoch": 2.03, + "grad_norm": 4.593588352203369, + "learning_rate": 0.00018050286138847628, + "loss": 3.2831, + "step": 1386 + }, + { + "epoch": 2.03, + "grad_norm": 3.0059878826141357, + "learning_rate": 0.0001804750277239978, + "loss": 3.2606, + "step": 1387 + }, + { + "epoch": 2.03, + "grad_norm": 2.9211039543151855, + "learning_rate": 0.00018044717635555259, + "loss": 3.2735, + "step": 1388 + }, + { + "epoch": 2.03, + "grad_norm": 3.4243338108062744, + "learning_rate": 0.00018041930728926768, + "loss": 3.2794, + "step": 1389 + }, + { + "epoch": 2.03, + "grad_norm": 3.398374557495117, + "learning_rate": 0.00018039142053127412, + "loss": 3.3066, + "step": 1390 + }, + { + "epoch": 2.03, + "grad_norm": 3.260343074798584, + "learning_rate": 0.0001803635160877068, + "loss": 3.3, + "step": 1391 + }, + { + "epoch": 2.04, + "grad_norm": 2.542720079421997, + "learning_rate": 0.00018033559396470454, + "loss": 3.2654, + "step": 1392 + }, + { + "epoch": 2.04, + "grad_norm": 4.721008777618408, + "learning_rate": 0.00018030765416840997, + "loss": 3.3286, + "step": 1393 + }, + { + "epoch": 2.04, + "grad_norm": 4.219196319580078, + "learning_rate": 0.00018027969670496966, + "loss": 3.2769, + "step": 1394 + }, + { + "epoch": 2.04, + "grad_norm": 2.8509461879730225, + "learning_rate": 0.00018025172158053413, + "loss": 3.2763, + "step": 1395 + }, + { + "epoch": 2.04, + "grad_norm": 3.1871495246887207, + "learning_rate": 0.00018022372880125765, + "loss": 3.3038, + "step": 1396 + }, + { + "epoch": 2.04, + "grad_norm": 4.717142105102539, + "learning_rate": 0.00018019571837329847, + "loss": 3.2649, + "step": 1397 + }, + { + "epoch": 2.04, + "grad_norm": 3.9490749835968018, + "learning_rate": 0.00018016769030281872, + "loss": 3.2709, + "step": 1398 + }, + { + "epoch": 2.05, + "grad_norm": 3.04848051071167, + "learning_rate": 0.00018013964459598434, + "loss": 3.2638, + "step": 1399 + }, + { + "epoch": 2.05, + "grad_norm": 2.971536636352539, + "learning_rate": 0.00018011158125896522, + "loss": 3.3796, + "step": 1400 + }, + { + "epoch": 2.05, + "grad_norm": 3.643239736557007, + "learning_rate": 0.0001800835002979351, + "loss": 3.2766, + "step": 1401 + }, + { + "epoch": 2.05, + "grad_norm": 3.0714406967163086, + "learning_rate": 0.00018005540171907162, + "loss": 3.2691, + "step": 1402 + }, + { + "epoch": 2.05, + "grad_norm": 3.928359031677246, + "learning_rate": 0.00018002728552855629, + "loss": 3.2842, + "step": 1403 + }, + { + "epoch": 2.05, + "grad_norm": 3.7628214359283447, + "learning_rate": 0.00017999915173257444, + "loss": 3.3371, + "step": 1404 + }, + { + "epoch": 2.06, + "grad_norm": 3.1093904972076416, + "learning_rate": 0.00017997100033731534, + "loss": 3.2866, + "step": 1405 + }, + { + "epoch": 2.06, + "grad_norm": 2.735902786254883, + "learning_rate": 0.0001799428313489721, + "loss": 3.2598, + "step": 1406 + }, + { + "epoch": 2.06, + "grad_norm": 3.890085458755493, + "learning_rate": 0.00017991464477374173, + "loss": 3.2565, + "step": 1407 + }, + { + "epoch": 2.06, + "grad_norm": 3.5979950428009033, + "learning_rate": 0.00017988644061782508, + "loss": 3.3153, + "step": 1408 + }, + { + "epoch": 2.06, + "grad_norm": 3.224681854248047, + "learning_rate": 0.00017985821888742685, + "loss": 3.2969, + "step": 1409 + }, + { + "epoch": 2.06, + "grad_norm": 3.645235776901245, + "learning_rate": 0.00017982997958875565, + "loss": 3.2823, + "step": 1410 + }, + { + "epoch": 2.06, + "grad_norm": 3.3547754287719727, + "learning_rate": 0.000179801722728024, + "loss": 3.3546, + "step": 1411 + }, + { + "epoch": 2.07, + "grad_norm": 2.8428351879119873, + "learning_rate": 0.00017977344831144807, + "loss": 3.3096, + "step": 1412 + }, + { + "epoch": 2.07, + "grad_norm": 4.488086223602295, + "learning_rate": 0.00017974515634524817, + "loss": 3.2783, + "step": 1413 + }, + { + "epoch": 2.07, + "grad_norm": 4.082129955291748, + "learning_rate": 0.0001797168468356483, + "loss": 3.3131, + "step": 1414 + }, + { + "epoch": 2.07, + "grad_norm": 3.037198305130005, + "learning_rate": 0.00017968851978887636, + "loss": 3.3435, + "step": 1415 + }, + { + "epoch": 2.07, + "grad_norm": 2.8374056816101074, + "learning_rate": 0.0001796601752111641, + "loss": 3.3026, + "step": 1416 + }, + { + "epoch": 2.07, + "grad_norm": 4.270615577697754, + "learning_rate": 0.00017963181310874716, + "loss": 3.3384, + "step": 1417 + }, + { + "epoch": 2.07, + "grad_norm": 4.1699323654174805, + "learning_rate": 0.00017960343348786504, + "loss": 3.2869, + "step": 1418 + }, + { + "epoch": 2.08, + "grad_norm": 2.7383670806884766, + "learning_rate": 0.00017957503635476099, + "loss": 3.2882, + "step": 1419 + }, + { + "epoch": 2.08, + "grad_norm": 2.707303762435913, + "learning_rate": 0.00017954662171568226, + "loss": 3.3616, + "step": 1420 + }, + { + "epoch": 2.08, + "grad_norm": 5.056338787078857, + "learning_rate": 0.00017951818957687981, + "loss": 3.381, + "step": 1421 + }, + { + "epoch": 2.08, + "grad_norm": 4.49586820602417, + "learning_rate": 0.0001794897399446086, + "loss": 3.2837, + "step": 1422 + }, + { + "epoch": 2.08, + "grad_norm": 2.5468156337738037, + "learning_rate": 0.0001794612728251273, + "loss": 3.2547, + "step": 1423 + }, + { + "epoch": 2.08, + "grad_norm": 2.2053098678588867, + "learning_rate": 0.00017943278822469852, + "loss": 3.3216, + "step": 1424 + }, + { + "epoch": 2.09, + "grad_norm": 4.875980854034424, + "learning_rate": 0.00017940428614958865, + "loss": 3.3319, + "step": 1425 + }, + { + "epoch": 2.09, + "grad_norm": 4.466098308563232, + "learning_rate": 0.000179375766606068, + "loss": 3.276, + "step": 1426 + }, + { + "epoch": 2.09, + "grad_norm": 2.8682076930999756, + "learning_rate": 0.0001793472296004106, + "loss": 3.2971, + "step": 1427 + }, + { + "epoch": 2.09, + "grad_norm": 2.2296552658081055, + "learning_rate": 0.00017931867513889448, + "loss": 3.2946, + "step": 1428 + }, + { + "epoch": 2.09, + "grad_norm": 4.954530239105225, + "learning_rate": 0.0001792901032278014, + "loss": 3.269, + "step": 1429 + }, + { + "epoch": 2.09, + "grad_norm": 4.960486888885498, + "learning_rate": 0.00017926151387341697, + "loss": 3.2616, + "step": 1430 + }, + { + "epoch": 2.09, + "grad_norm": 2.2952184677124023, + "learning_rate": 0.0001792329070820307, + "loss": 3.3348, + "step": 1431 + }, + { + "epoch": 2.1, + "grad_norm": 1.9920496940612793, + "learning_rate": 0.00017920428285993583, + "loss": 3.3137, + "step": 1432 + }, + { + "epoch": 2.1, + "grad_norm": 4.211342811584473, + "learning_rate": 0.00017917564121342958, + "loss": 3.3246, + "step": 1433 + }, + { + "epoch": 2.1, + "grad_norm": 3.4940345287323, + "learning_rate": 0.00017914698214881284, + "loss": 3.3421, + "step": 1434 + }, + { + "epoch": 2.1, + "grad_norm": 3.7058966159820557, + "learning_rate": 0.00017911830567239048, + "loss": 3.3377, + "step": 1435 + }, + { + "epoch": 2.1, + "grad_norm": 3.2200000286102295, + "learning_rate": 0.00017908961179047105, + "loss": 3.3125, + "step": 1436 + }, + { + "epoch": 2.1, + "grad_norm": 3.529256582260132, + "learning_rate": 0.00017906090050936712, + "loss": 3.2833, + "step": 1437 + }, + { + "epoch": 2.1, + "grad_norm": 3.189732313156128, + "learning_rate": 0.0001790321718353949, + "loss": 3.2662, + "step": 1438 + }, + { + "epoch": 2.11, + "grad_norm": 3.6067655086517334, + "learning_rate": 0.0001790034257748745, + "loss": 3.2795, + "step": 1439 + }, + { + "epoch": 2.11, + "grad_norm": 3.4113893508911133, + "learning_rate": 0.0001789746623341299, + "loss": 3.2666, + "step": 1440 + }, + { + "epoch": 2.11, + "grad_norm": 3.466991662979126, + "learning_rate": 0.00017894588151948885, + "loss": 3.325, + "step": 1441 + }, + { + "epoch": 2.11, + "grad_norm": 3.057206153869629, + "learning_rate": 0.00017891708333728296, + "loss": 3.2636, + "step": 1442 + }, + { + "epoch": 2.11, + "grad_norm": 3.8873162269592285, + "learning_rate": 0.0001788882677938476, + "loss": 3.3139, + "step": 1443 + }, + { + "epoch": 2.11, + "grad_norm": 3.4515161514282227, + "learning_rate": 0.000178859434895522, + "loss": 3.2884, + "step": 1444 + }, + { + "epoch": 2.11, + "grad_norm": 14.146556854248047, + "learning_rate": 0.00017883058464864923, + "loss": 3.2571, + "step": 1445 + }, + { + "epoch": 2.12, + "grad_norm": 9.11373519897461, + "learning_rate": 0.00017880171705957613, + "loss": 3.295, + "step": 1446 + }, + { + "epoch": 2.12, + "grad_norm": 5.7905049324035645, + "learning_rate": 0.0001787728321346534, + "loss": 3.2532, + "step": 1447 + }, + { + "epoch": 2.12, + "grad_norm": 3.7080447673797607, + "learning_rate": 0.0001787439298802355, + "loss": 3.3201, + "step": 1448 + }, + { + "epoch": 2.12, + "grad_norm": 4.587584018707275, + "learning_rate": 0.00017871501030268072, + "loss": 3.2702, + "step": 1449 + }, + { + "epoch": 2.12, + "grad_norm": 0.9951494932174683, + "learning_rate": 0.0001786860734083512, + "loss": 3.2862, + "step": 1450 + }, + { + "epoch": 2.12, + "grad_norm": 0.8612810373306274, + "learning_rate": 0.00017865711920361292, + "loss": 3.2968, + "step": 1451 + }, + { + "epoch": 2.13, + "grad_norm": 2.446154832839966, + "learning_rate": 0.00017862814769483549, + "loss": 3.2937, + "step": 1452 + }, + { + "epoch": 2.13, + "grad_norm": 1.2420624494552612, + "learning_rate": 0.00017859915888839253, + "loss": 3.2922, + "step": 1453 + }, + { + "epoch": 2.13, + "grad_norm": 1.9885892868041992, + "learning_rate": 0.00017857015279066135, + "loss": 3.3043, + "step": 1454 + }, + { + "epoch": 2.13, + "grad_norm": 2.127197504043579, + "learning_rate": 0.00017854112940802314, + "loss": 3.3503, + "step": 1455 + }, + { + "epoch": 2.13, + "grad_norm": 1.1510310173034668, + "learning_rate": 0.0001785120887468628, + "loss": 3.2314, + "step": 1456 + }, + { + "epoch": 2.13, + "grad_norm": 2.6665754318237305, + "learning_rate": 0.00017848303081356912, + "loss": 3.2788, + "step": 1457 + }, + { + "epoch": 2.13, + "grad_norm": 1.0793408155441284, + "learning_rate": 0.0001784539556145346, + "loss": 3.2623, + "step": 1458 + }, + { + "epoch": 2.14, + "grad_norm": 2.212766647338867, + "learning_rate": 0.00017842486315615563, + "loss": 3.3542, + "step": 1459 + }, + { + "epoch": 2.14, + "grad_norm": 1.8812615871429443, + "learning_rate": 0.00017839575344483238, + "loss": 3.2481, + "step": 1460 + }, + { + "epoch": 2.14, + "grad_norm": 1.537009358406067, + "learning_rate": 0.0001783666264869687, + "loss": 3.2492, + "step": 1461 + }, + { + "epoch": 2.14, + "grad_norm": 1.0844135284423828, + "learning_rate": 0.00017833748228897241, + "loss": 3.3093, + "step": 1462 + }, + { + "epoch": 2.14, + "grad_norm": 1.4660224914550781, + "learning_rate": 0.000178308320857255, + "loss": 3.2598, + "step": 1463 + }, + { + "epoch": 2.14, + "grad_norm": 0.9054508209228516, + "learning_rate": 0.00017827914219823183, + "loss": 3.3132, + "step": 1464 + }, + { + "epoch": 2.14, + "grad_norm": 5.584422588348389, + "learning_rate": 0.00017824994631832197, + "loss": 3.2961, + "step": 1465 + }, + { + "epoch": 2.15, + "grad_norm": 1.9765104055404663, + "learning_rate": 0.0001782207332239483, + "loss": 3.2963, + "step": 1466 + }, + { + "epoch": 2.15, + "grad_norm": 1.4504390954971313, + "learning_rate": 0.00017819150292153757, + "loss": 3.3335, + "step": 1467 + }, + { + "epoch": 2.15, + "grad_norm": 1.2761657238006592, + "learning_rate": 0.0001781622554175202, + "loss": 3.2779, + "step": 1468 + }, + { + "epoch": 2.15, + "grad_norm": 2.1837310791015625, + "learning_rate": 0.00017813299071833046, + "loss": 3.2791, + "step": 1469 + }, + { + "epoch": 2.15, + "grad_norm": 2.7105600833892822, + "learning_rate": 0.00017810370883040636, + "loss": 3.2983, + "step": 1470 + }, + { + "epoch": 2.15, + "grad_norm": 1.1672013998031616, + "learning_rate": 0.00017807440976018977, + "loss": 3.2989, + "step": 1471 + }, + { + "epoch": 2.15, + "grad_norm": 4.486381530761719, + "learning_rate": 0.00017804509351412629, + "loss": 3.3132, + "step": 1472 + }, + { + "epoch": 2.16, + "grad_norm": 3.2585389614105225, + "learning_rate": 0.0001780157600986652, + "loss": 3.3249, + "step": 1473 + }, + { + "epoch": 2.16, + "grad_norm": 3.1671297550201416, + "learning_rate": 0.00017798640952025978, + "loss": 3.31, + "step": 1474 + }, + { + "epoch": 2.16, + "grad_norm": 2.345313310623169, + "learning_rate": 0.0001779570417853669, + "loss": 3.2382, + "step": 1475 + }, + { + "epoch": 2.16, + "grad_norm": 5.773190498352051, + "learning_rate": 0.0001779276569004473, + "loss": 3.2903, + "step": 1476 + }, + { + "epoch": 2.16, + "grad_norm": 4.879696846008301, + "learning_rate": 0.00017789825487196537, + "loss": 3.3107, + "step": 1477 + }, + { + "epoch": 2.16, + "grad_norm": 4.470544815063477, + "learning_rate": 0.00017786883570638942, + "loss": 3.3, + "step": 1478 + }, + { + "epoch": 2.17, + "grad_norm": 5.051728248596191, + "learning_rate": 0.00017783939941019153, + "loss": 3.2964, + "step": 1479 + }, + { + "epoch": 2.17, + "grad_norm": 1.2672648429870605, + "learning_rate": 0.00017780994598984737, + "loss": 3.297, + "step": 1480 + }, + { + "epoch": 2.17, + "grad_norm": 5.323600769042969, + "learning_rate": 0.0001777804754518366, + "loss": 3.2878, + "step": 1481 + }, + { + "epoch": 2.17, + "grad_norm": 4.052299976348877, + "learning_rate": 0.00017775098780264243, + "loss": 3.3135, + "step": 1482 + }, + { + "epoch": 2.17, + "grad_norm": 3.538057565689087, + "learning_rate": 0.00017772148304875205, + "loss": 3.3209, + "step": 1483 + }, + { + "epoch": 2.17, + "grad_norm": 2.5743610858917236, + "learning_rate": 0.0001776919611966563, + "loss": 3.2468, + "step": 1484 + }, + { + "epoch": 2.17, + "grad_norm": 6.518936634063721, + "learning_rate": 0.00017766242225284966, + "loss": 3.2914, + "step": 1485 + }, + { + "epoch": 2.18, + "grad_norm": 5.448997497558594, + "learning_rate": 0.00017763286622383067, + "loss": 3.3127, + "step": 1486 + }, + { + "epoch": 2.18, + "grad_norm": 2.5418717861175537, + "learning_rate": 0.00017760329311610135, + "loss": 3.2913, + "step": 1487 + }, + { + "epoch": 2.18, + "grad_norm": 4.122663974761963, + "learning_rate": 0.00017757370293616762, + "loss": 3.3257, + "step": 1488 + }, + { + "epoch": 2.18, + "grad_norm": 3.3550655841827393, + "learning_rate": 0.00017754409569053917, + "loss": 3.271, + "step": 1489 + }, + { + "epoch": 2.18, + "grad_norm": 2.8279335498809814, + "learning_rate": 0.0001775144713857293, + "loss": 3.2989, + "step": 1490 + }, + { + "epoch": 2.18, + "grad_norm": 2.4097659587860107, + "learning_rate": 0.0001774848300282552, + "loss": 3.2557, + "step": 1491 + }, + { + "epoch": 2.18, + "grad_norm": 2.702836275100708, + "learning_rate": 0.00017745517162463776, + "loss": 3.2719, + "step": 1492 + }, + { + "epoch": 2.19, + "grad_norm": 1.937203288078308, + "learning_rate": 0.00017742549618140167, + "loss": 3.2801, + "step": 1493 + }, + { + "epoch": 2.19, + "grad_norm": 2.2853002548217773, + "learning_rate": 0.00017739580370507532, + "loss": 3.4001, + "step": 1494 + }, + { + "epoch": 2.19, + "grad_norm": 3.39026141166687, + "learning_rate": 0.00017736609420219082, + "loss": 3.2562, + "step": 1495 + }, + { + "epoch": 2.19, + "grad_norm": 2.192948341369629, + "learning_rate": 0.00017733636767928407, + "loss": 3.2548, + "step": 1496 + }, + { + "epoch": 2.19, + "grad_norm": 3.966799259185791, + "learning_rate": 0.00017730662414289472, + "loss": 3.2744, + "step": 1497 + }, + { + "epoch": 2.19, + "grad_norm": 2.6247634887695312, + "learning_rate": 0.00017727686359956615, + "loss": 3.287, + "step": 1498 + }, + { + "epoch": 2.19, + "grad_norm": 2.959963083267212, + "learning_rate": 0.00017724708605584545, + "loss": 3.2864, + "step": 1499 + }, + { + "epoch": 2.2, + "grad_norm": 1.7757418155670166, + "learning_rate": 0.00017721729151828355, + "loss": 3.3003, + "step": 1500 + }, + { + "epoch": 2.2, + "grad_norm": 3.2801480293273926, + "learning_rate": 0.00017718747999343498, + "loss": 3.2662, + "step": 1501 + }, + { + "epoch": 2.2, + "grad_norm": 2.6395089626312256, + "learning_rate": 0.0001771576514878581, + "loss": 3.2991, + "step": 1502 + }, + { + "epoch": 2.2, + "grad_norm": 2.687575340270996, + "learning_rate": 0.00017712780600811497, + "loss": 3.2676, + "step": 1503 + }, + { + "epoch": 2.2, + "grad_norm": 2.702707052230835, + "learning_rate": 0.00017709794356077143, + "loss": 3.2862, + "step": 1504 + }, + { + "epoch": 2.2, + "grad_norm": 2.10098934173584, + "learning_rate": 0.000177068064152397, + "loss": 3.2425, + "step": 1505 + }, + { + "epoch": 2.21, + "grad_norm": 1.3301888704299927, + "learning_rate": 0.0001770381677895649, + "loss": 3.2325, + "step": 1506 + }, + { + "epoch": 2.21, + "grad_norm": 3.560911178588867, + "learning_rate": 0.0001770082544788522, + "loss": 3.2914, + "step": 1507 + }, + { + "epoch": 2.21, + "grad_norm": 2.01891827583313, + "learning_rate": 0.0001769783242268396, + "loss": 3.2213, + "step": 1508 + }, + { + "epoch": 2.21, + "grad_norm": 3.6540088653564453, + "learning_rate": 0.00017694837704011155, + "loss": 3.3066, + "step": 1509 + }, + { + "epoch": 2.21, + "grad_norm": 3.1314940452575684, + "learning_rate": 0.00017691841292525625, + "loss": 3.2576, + "step": 1510 + }, + { + "epoch": 2.21, + "grad_norm": 3.816842555999756, + "learning_rate": 0.0001768884318888656, + "loss": 3.2784, + "step": 1511 + }, + { + "epoch": 2.21, + "grad_norm": 3.5961854457855225, + "learning_rate": 0.0001768584339375352, + "loss": 3.2562, + "step": 1512 + }, + { + "epoch": 2.22, + "grad_norm": 3.9628171920776367, + "learning_rate": 0.0001768284190778644, + "loss": 3.2561, + "step": 1513 + }, + { + "epoch": 2.22, + "grad_norm": 4.071855068206787, + "learning_rate": 0.00017679838731645632, + "loss": 3.2924, + "step": 1514 + }, + { + "epoch": 2.22, + "grad_norm": 2.176236152648926, + "learning_rate": 0.00017676833865991765, + "loss": 3.309, + "step": 1515 + }, + { + "epoch": 2.22, + "grad_norm": 1.6997188329696655, + "learning_rate": 0.000176738273114859, + "loss": 3.2835, + "step": 1516 + }, + { + "epoch": 2.22, + "grad_norm": 3.275249719619751, + "learning_rate": 0.00017670819068789454, + "loss": 3.2506, + "step": 1517 + }, + { + "epoch": 2.22, + "grad_norm": 1.3623771667480469, + "learning_rate": 0.0001766780913856422, + "loss": 3.2054, + "step": 1518 + }, + { + "epoch": 2.22, + "grad_norm": 5.944447994232178, + "learning_rate": 0.00017664797521472364, + "loss": 3.2794, + "step": 1519 + }, + { + "epoch": 2.23, + "grad_norm": 5.1751484870910645, + "learning_rate": 0.0001766178421817642, + "loss": 3.2498, + "step": 1520 + }, + { + "epoch": 2.23, + "grad_norm": 2.410102128982544, + "learning_rate": 0.00017658769229339294, + "loss": 3.2672, + "step": 1521 + }, + { + "epoch": 2.23, + "eval_loss": 3.0885374546051025, + "eval_runtime": 1164.3656, + "eval_samples_per_second": 14.722, + "eval_steps_per_second": 0.921, + "step": 1521 + }, + { + "epoch": 2.23, + "grad_norm": 2.7669081687927246, + "learning_rate": 0.00017655752555624263, + "loss": 3.2525, + "step": 1522 + }, + { + "epoch": 2.23, + "grad_norm": 4.010193347930908, + "learning_rate": 0.00017652734197694977, + "loss": 3.2827, + "step": 1523 + }, + { + "epoch": 2.23, + "grad_norm": 3.233734130859375, + "learning_rate": 0.00017649714156215456, + "loss": 3.2372, + "step": 1524 + }, + { + "epoch": 2.23, + "grad_norm": 3.2297024726867676, + "learning_rate": 0.00017646692431850085, + "loss": 3.2763, + "step": 1525 + }, + { + "epoch": 2.24, + "grad_norm": 2.7911462783813477, + "learning_rate": 0.00017643669025263625, + "loss": 3.2702, + "step": 1526 + }, + { + "epoch": 2.24, + "grad_norm": 2.8759915828704834, + "learning_rate": 0.00017640643937121207, + "loss": 3.2338, + "step": 1527 + }, + { + "epoch": 2.24, + "grad_norm": 2.3804805278778076, + "learning_rate": 0.00017637617168088325, + "loss": 3.2646, + "step": 1528 + }, + { + "epoch": 2.24, + "grad_norm": 4.322963237762451, + "learning_rate": 0.00017634588718830853, + "loss": 3.2743, + "step": 1529 + }, + { + "epoch": 2.24, + "grad_norm": 3.6001713275909424, + "learning_rate": 0.00017631558590015028, + "loss": 3.3172, + "step": 1530 + }, + { + "epoch": 2.24, + "grad_norm": 3.2861557006835938, + "learning_rate": 0.00017628526782307455, + "loss": 3.2723, + "step": 1531 + }, + { + "epoch": 2.24, + "grad_norm": 3.6503138542175293, + "learning_rate": 0.0001762549329637512, + "loss": 3.234, + "step": 1532 + }, + { + "epoch": 2.25, + "grad_norm": 3.1032047271728516, + "learning_rate": 0.00017622458132885357, + "loss": 3.2845, + "step": 1533 + }, + { + "epoch": 2.25, + "grad_norm": 2.5854177474975586, + "learning_rate": 0.00017619421292505892, + "loss": 3.2251, + "step": 1534 + }, + { + "epoch": 2.25, + "grad_norm": 4.162723064422607, + "learning_rate": 0.00017616382775904806, + "loss": 3.2519, + "step": 1535 + }, + { + "epoch": 2.25, + "grad_norm": 3.7609925270080566, + "learning_rate": 0.0001761334258375055, + "loss": 3.334, + "step": 1536 + }, + { + "epoch": 2.25, + "grad_norm": 3.426297664642334, + "learning_rate": 0.00017610300716711948, + "loss": 3.2644, + "step": 1537 + }, + { + "epoch": 2.25, + "grad_norm": 3.2877280712127686, + "learning_rate": 0.0001760725717545819, + "loss": 3.2466, + "step": 1538 + }, + { + "epoch": 2.25, + "grad_norm": 3.2899694442749023, + "learning_rate": 0.00017604211960658837, + "loss": 3.2245, + "step": 1539 + }, + { + "epoch": 2.26, + "grad_norm": 3.202381134033203, + "learning_rate": 0.0001760116507298381, + "loss": 3.2512, + "step": 1540 + }, + { + "epoch": 2.26, + "grad_norm": 9.299320220947266, + "learning_rate": 0.0001759811651310341, + "loss": 3.2295, + "step": 1541 + }, + { + "epoch": 2.26, + "grad_norm": 6.585042476654053, + "learning_rate": 0.000175950662816883, + "loss": 3.2845, + "step": 1542 + }, + { + "epoch": 2.26, + "grad_norm": 3.786369562149048, + "learning_rate": 0.00017592014379409502, + "loss": 3.2796, + "step": 1543 + }, + { + "epoch": 2.26, + "grad_norm": 5.25980281829834, + "learning_rate": 0.00017588960806938424, + "loss": 3.2427, + "step": 1544 + }, + { + "epoch": 2.26, + "grad_norm": 7.450433254241943, + "learning_rate": 0.00017585905564946827, + "loss": 3.3147, + "step": 1545 + }, + { + "epoch": 2.26, + "grad_norm": 3.1463863849639893, + "learning_rate": 0.00017582848654106845, + "loss": 3.2311, + "step": 1546 + }, + { + "epoch": 2.27, + "grad_norm": 5.310190200805664, + "learning_rate": 0.00017579790075090976, + "loss": 3.3092, + "step": 1547 + }, + { + "epoch": 2.27, + "grad_norm": 5.930779933929443, + "learning_rate": 0.0001757672982857209, + "loss": 3.2788, + "step": 1548 + }, + { + "epoch": 2.27, + "grad_norm": 2.841097831726074, + "learning_rate": 0.00017573667915223414, + "loss": 3.1998, + "step": 1549 + }, + { + "epoch": 2.27, + "grad_norm": 7.447371959686279, + "learning_rate": 0.0001757060433571856, + "loss": 3.2633, + "step": 1550 + }, + { + "epoch": 2.27, + "grad_norm": 7.628240585327148, + "learning_rate": 0.00017567539090731482, + "loss": 3.2626, + "step": 1551 + }, + { + "epoch": 2.27, + "grad_norm": 1.3023099899291992, + "learning_rate": 0.0001756447218093652, + "loss": 3.2525, + "step": 1552 + }, + { + "epoch": 2.28, + "grad_norm": 8.924667358398438, + "learning_rate": 0.00017561403607008376, + "loss": 3.2892, + "step": 1553 + }, + { + "epoch": 2.28, + "grad_norm": 11.807168006896973, + "learning_rate": 0.00017558333369622112, + "loss": 3.3097, + "step": 1554 + }, + { + "epoch": 2.28, + "grad_norm": 9.348532676696777, + "learning_rate": 0.0001755526146945316, + "loss": 3.3024, + "step": 1555 + }, + { + "epoch": 2.28, + "grad_norm": 1.3859421014785767, + "learning_rate": 0.00017552187907177315, + "loss": 3.2629, + "step": 1556 + }, + { + "epoch": 2.28, + "grad_norm": 12.93152141571045, + "learning_rate": 0.00017549112683470748, + "loss": 3.3065, + "step": 1557 + }, + { + "epoch": 2.28, + "grad_norm": 18.00539779663086, + "learning_rate": 0.00017546035799009978, + "loss": 3.3714, + "step": 1558 + }, + { + "epoch": 2.28, + "grad_norm": 13.751264572143555, + "learning_rate": 0.00017542957254471908, + "loss": 3.3417, + "step": 1559 + }, + { + "epoch": 2.29, + "grad_norm": 4.594606876373291, + "learning_rate": 0.0001753987705053379, + "loss": 3.343, + "step": 1560 + }, + { + "epoch": 2.29, + "grad_norm": 7.840448379516602, + "learning_rate": 0.0001753679518787325, + "loss": 3.3199, + "step": 1561 + }, + { + "epoch": 2.29, + "grad_norm": 11.789816856384277, + "learning_rate": 0.0001753371166716828, + "loss": 3.3301, + "step": 1562 + }, + { + "epoch": 2.29, + "grad_norm": 11.082632064819336, + "learning_rate": 0.0001753062648909723, + "loss": 3.3078, + "step": 1563 + }, + { + "epoch": 2.29, + "grad_norm": 5.516722679138184, + "learning_rate": 0.0001752753965433882, + "loss": 3.2917, + "step": 1564 + }, + { + "epoch": 2.29, + "grad_norm": 8.504383087158203, + "learning_rate": 0.00017524451163572136, + "loss": 3.3195, + "step": 1565 + }, + { + "epoch": 2.29, + "grad_norm": 14.364538192749023, + "learning_rate": 0.0001752136101747662, + "loss": 3.3471, + "step": 1566 + }, + { + "epoch": 2.3, + "grad_norm": 12.66904067993164, + "learning_rate": 0.00017518269216732087, + "loss": 3.3574, + "step": 1567 + }, + { + "epoch": 2.3, + "grad_norm": 4.23660945892334, + "learning_rate": 0.00017515175762018714, + "loss": 3.3283, + "step": 1568 + }, + { + "epoch": 2.3, + "grad_norm": 6.301276206970215, + "learning_rate": 0.00017512080654017036, + "loss": 3.3064, + "step": 1569 + }, + { + "epoch": 2.3, + "grad_norm": 8.942106246948242, + "learning_rate": 0.00017508983893407958, + "loss": 3.2761, + "step": 1570 + }, + { + "epoch": 2.3, + "grad_norm": 6.445720672607422, + "learning_rate": 0.00017505885480872745, + "loss": 3.3033, + "step": 1571 + }, + { + "epoch": 2.3, + "grad_norm": 3.0577292442321777, + "learning_rate": 0.0001750278541709303, + "loss": 3.2916, + "step": 1572 + }, + { + "epoch": 2.3, + "grad_norm": 5.473748683929443, + "learning_rate": 0.00017499683702750803, + "loss": 3.2609, + "step": 1573 + }, + { + "epoch": 2.31, + "grad_norm": 2.3021624088287354, + "learning_rate": 0.00017496580338528426, + "loss": 3.2478, + "step": 1574 + }, + { + "epoch": 2.31, + "grad_norm": 6.796987056732178, + "learning_rate": 0.00017493475325108611, + "loss": 3.3177, + "step": 1575 + }, + { + "epoch": 2.31, + "grad_norm": 8.425848007202148, + "learning_rate": 0.00017490368663174443, + "loss": 3.2999, + "step": 1576 + }, + { + "epoch": 2.31, + "grad_norm": 4.728596210479736, + "learning_rate": 0.0001748726035340937, + "loss": 3.2693, + "step": 1577 + }, + { + "epoch": 2.31, + "grad_norm": 6.588244915008545, + "learning_rate": 0.00017484150396497193, + "loss": 3.2474, + "step": 1578 + }, + { + "epoch": 2.31, + "grad_norm": 10.9650239944458, + "learning_rate": 0.00017481038793122088, + "loss": 3.3038, + "step": 1579 + }, + { + "epoch": 2.32, + "grad_norm": 5.750149726867676, + "learning_rate": 0.00017477925543968586, + "loss": 3.315, + "step": 1580 + }, + { + "epoch": 2.32, + "grad_norm": 4.661263465881348, + "learning_rate": 0.00017474810649721576, + "loss": 3.2819, + "step": 1581 + }, + { + "epoch": 2.32, + "grad_norm": 6.578607082366943, + "learning_rate": 0.0001747169411106632, + "loss": 3.268, + "step": 1582 + }, + { + "epoch": 2.32, + "grad_norm": 3.776803970336914, + "learning_rate": 0.0001746857592868843, + "loss": 3.2414, + "step": 1583 + }, + { + "epoch": 2.32, + "grad_norm": 5.648257732391357, + "learning_rate": 0.00017465456103273885, + "loss": 3.2601, + "step": 1584 + }, + { + "epoch": 2.32, + "grad_norm": 7.4345011711120605, + "learning_rate": 0.00017462334635509033, + "loss": 3.3107, + "step": 1585 + }, + { + "epoch": 2.32, + "grad_norm": 2.6171629428863525, + "learning_rate": 0.0001745921152608057, + "loss": 3.226, + "step": 1586 + }, + { + "epoch": 2.33, + "grad_norm": 6.692351341247559, + "learning_rate": 0.00017456086775675562, + "loss": 3.283, + "step": 1587 + }, + { + "epoch": 2.33, + "grad_norm": 7.625064373016357, + "learning_rate": 0.00017452960384981432, + "loss": 3.2796, + "step": 1588 + }, + { + "epoch": 2.33, + "grad_norm": 4.043941974639893, + "learning_rate": 0.00017449832354685961, + "loss": 3.2948, + "step": 1589 + }, + { + "epoch": 2.33, + "grad_norm": 6.564191818237305, + "learning_rate": 0.00017446702685477303, + "loss": 3.2824, + "step": 1590 + }, + { + "epoch": 2.33, + "grad_norm": 10.199933052062988, + "learning_rate": 0.00017443571378043959, + "loss": 3.3447, + "step": 1591 + }, + { + "epoch": 2.33, + "grad_norm": 6.0920729637146, + "learning_rate": 0.00017440438433074794, + "loss": 3.2682, + "step": 1592 + }, + { + "epoch": 2.33, + "grad_norm": 3.95531964302063, + "learning_rate": 0.00017437303851259038, + "loss": 3.2439, + "step": 1593 + }, + { + "epoch": 2.34, + "grad_norm": 6.185577869415283, + "learning_rate": 0.0001743416763328628, + "loss": 3.3002, + "step": 1594 + }, + { + "epoch": 2.34, + "grad_norm": 2.757133960723877, + "learning_rate": 0.00017431029779846463, + "loss": 3.2716, + "step": 1595 + }, + { + "epoch": 2.34, + "grad_norm": 8.002574920654297, + "learning_rate": 0.00017427890291629893, + "loss": 3.2844, + "step": 1596 + }, + { + "epoch": 2.34, + "grad_norm": 10.051774024963379, + "learning_rate": 0.0001742474916932724, + "loss": 3.2388, + "step": 1597 + }, + { + "epoch": 2.34, + "grad_norm": 5.608651638031006, + "learning_rate": 0.00017421606413629527, + "loss": 3.226, + "step": 1598 + }, + { + "epoch": 2.34, + "grad_norm": 4.5539231300354, + "learning_rate": 0.00017418462025228145, + "loss": 3.2424, + "step": 1599 + }, + { + "epoch": 2.34, + "grad_norm": 6.649083614349365, + "learning_rate": 0.0001741531600481483, + "loss": 3.2537, + "step": 1600 + }, + { + "epoch": 2.35, + "grad_norm": 3.6664934158325195, + "learning_rate": 0.00017412168353081686, + "loss": 3.243, + "step": 1601 + }, + { + "epoch": 2.35, + "grad_norm": 6.799222469329834, + "learning_rate": 0.00017409019070721183, + "loss": 3.2853, + "step": 1602 + }, + { + "epoch": 2.35, + "grad_norm": 9.875252723693848, + "learning_rate": 0.00017405868158426135, + "loss": 3.2505, + "step": 1603 + }, + { + "epoch": 2.35, + "grad_norm": 4.99692440032959, + "learning_rate": 0.00017402715616889728, + "loss": 3.3176, + "step": 1604 + }, + { + "epoch": 2.35, + "grad_norm": 6.484687328338623, + "learning_rate": 0.00017399561446805492, + "loss": 3.2547, + "step": 1605 + }, + { + "epoch": 2.35, + "grad_norm": 8.336297988891602, + "learning_rate": 0.00017396405648867328, + "loss": 3.2543, + "step": 1606 + }, + { + "epoch": 2.36, + "grad_norm": 5.791290283203125, + "learning_rate": 0.00017393248223769491, + "loss": 3.2538, + "step": 1607 + }, + { + "epoch": 2.36, + "grad_norm": 2.4937000274658203, + "learning_rate": 0.00017390089172206592, + "loss": 3.2665, + "step": 1608 + }, + { + "epoch": 2.36, + "grad_norm": 5.263038635253906, + "learning_rate": 0.00017386928494873602, + "loss": 3.2899, + "step": 1609 + }, + { + "epoch": 2.36, + "grad_norm": 0.9264150857925415, + "learning_rate": 0.00017383766192465846, + "loss": 3.2284, + "step": 1610 + }, + { + "epoch": 2.36, + "grad_norm": 7.343024730682373, + "learning_rate": 0.00017380602265679014, + "loss": 3.3191, + "step": 1611 + }, + { + "epoch": 2.36, + "grad_norm": 8.518827438354492, + "learning_rate": 0.00017377436715209144, + "loss": 3.3296, + "step": 1612 + }, + { + "epoch": 2.36, + "grad_norm": 5.042328357696533, + "learning_rate": 0.00017374269541752638, + "loss": 3.2711, + "step": 1613 + }, + { + "epoch": 2.37, + "grad_norm": 5.329464435577393, + "learning_rate": 0.00017371100746006252, + "loss": 3.2886, + "step": 1614 + }, + { + "epoch": 2.37, + "grad_norm": 8.302837371826172, + "learning_rate": 0.00017367930328667105, + "loss": 3.2565, + "step": 1615 + }, + { + "epoch": 2.37, + "grad_norm": 4.5046539306640625, + "learning_rate": 0.0001736475829043266, + "loss": 3.2309, + "step": 1616 + }, + { + "epoch": 2.37, + "grad_norm": 5.1940789222717285, + "learning_rate": 0.0001736158463200075, + "loss": 3.2687, + "step": 1617 + }, + { + "epoch": 2.37, + "grad_norm": 7.008042335510254, + "learning_rate": 0.00017358409354069554, + "loss": 3.2845, + "step": 1618 + }, + { + "epoch": 2.37, + "grad_norm": 3.637481689453125, + "learning_rate": 0.00017355232457337615, + "loss": 3.2221, + "step": 1619 + }, + { + "epoch": 2.37, + "grad_norm": 6.158731937408447, + "learning_rate": 0.00017352053942503828, + "loss": 3.2727, + "step": 1620 + }, + { + "epoch": 2.38, + "grad_norm": 9.95582103729248, + "learning_rate": 0.00017348873810267443, + "loss": 3.2826, + "step": 1621 + }, + { + "epoch": 2.38, + "grad_norm": 5.530641555786133, + "learning_rate": 0.0001734569206132807, + "loss": 3.291, + "step": 1622 + }, + { + "epoch": 2.38, + "grad_norm": 3.868178367614746, + "learning_rate": 0.00017342508696385674, + "loss": 3.2794, + "step": 1623 + }, + { + "epoch": 2.38, + "grad_norm": 6.0412516593933105, + "learning_rate": 0.00017339323716140573, + "loss": 3.2546, + "step": 1624 + }, + { + "epoch": 2.38, + "grad_norm": 2.217486619949341, + "learning_rate": 0.00017336137121293436, + "loss": 3.2526, + "step": 1625 + }, + { + "epoch": 2.38, + "grad_norm": 8.725723266601562, + "learning_rate": 0.000173329489125453, + "loss": 3.253, + "step": 1626 + }, + { + "epoch": 2.39, + "grad_norm": 12.251547813415527, + "learning_rate": 0.00017329759090597548, + "loss": 3.2593, + "step": 1627 + }, + { + "epoch": 2.39, + "grad_norm": 7.2429070472717285, + "learning_rate": 0.00017326567656151916, + "loss": 3.2555, + "step": 1628 + }, + { + "epoch": 2.39, + "grad_norm": 2.993010997772217, + "learning_rate": 0.000173233746099105, + "loss": 3.2165, + "step": 1629 + }, + { + "epoch": 2.39, + "grad_norm": 5.157336711883545, + "learning_rate": 0.0001732017995257575, + "loss": 3.3013, + "step": 1630 + }, + { + "epoch": 2.39, + "grad_norm": 2.0820674896240234, + "learning_rate": 0.00017316983684850467, + "loss": 3.2088, + "step": 1631 + }, + { + "epoch": 2.39, + "grad_norm": 8.305849075317383, + "learning_rate": 0.0001731378580743781, + "loss": 3.2902, + "step": 1632 + }, + { + "epoch": 2.39, + "grad_norm": 10.2841157913208, + "learning_rate": 0.00017310586321041288, + "loss": 3.2437, + "step": 1633 + }, + { + "epoch": 2.4, + "grad_norm": 4.831589698791504, + "learning_rate": 0.00017307385226364772, + "loss": 3.2177, + "step": 1634 + }, + { + "epoch": 2.4, + "grad_norm": 4.824088096618652, + "learning_rate": 0.00017304182524112478, + "loss": 3.2779, + "step": 1635 + }, + { + "epoch": 2.4, + "grad_norm": 7.442866802215576, + "learning_rate": 0.00017300978214988976, + "loss": 3.2716, + "step": 1636 + }, + { + "epoch": 2.4, + "grad_norm": 4.1812968254089355, + "learning_rate": 0.000172977722996992, + "loss": 3.2746, + "step": 1637 + }, + { + "epoch": 2.4, + "grad_norm": 5.722605228424072, + "learning_rate": 0.00017294564778948422, + "loss": 3.2238, + "step": 1638 + }, + { + "epoch": 2.4, + "grad_norm": 8.271087646484375, + "learning_rate": 0.00017291355653442278, + "loss": 3.3323, + "step": 1639 + }, + { + "epoch": 2.4, + "grad_norm": 3.2383551597595215, + "learning_rate": 0.00017288144923886757, + "loss": 3.2318, + "step": 1640 + }, + { + "epoch": 2.41, + "grad_norm": 6.346744537353516, + "learning_rate": 0.00017284932590988197, + "loss": 3.2569, + "step": 1641 + }, + { + "epoch": 2.41, + "grad_norm": 8.613646507263184, + "learning_rate": 0.00017281718655453286, + "loss": 3.2684, + "step": 1642 + }, + { + "epoch": 2.41, + "grad_norm": 5.7413129806518555, + "learning_rate": 0.00017278503117989073, + "loss": 3.2374, + "step": 1643 + }, + { + "epoch": 2.41, + "grad_norm": 3.2419638633728027, + "learning_rate": 0.0001727528597930295, + "loss": 3.2127, + "step": 1644 + }, + { + "epoch": 2.41, + "grad_norm": 6.177962779998779, + "learning_rate": 0.0001727206724010267, + "loss": 3.2859, + "step": 1645 + }, + { + "epoch": 2.41, + "grad_norm": 1.907907247543335, + "learning_rate": 0.0001726884690109633, + "loss": 3.2221, + "step": 1646 + }, + { + "epoch": 2.41, + "grad_norm": 6.6156415939331055, + "learning_rate": 0.0001726562496299239, + "loss": 3.2339, + "step": 1647 + }, + { + "epoch": 2.42, + "grad_norm": 8.058192253112793, + "learning_rate": 0.00017262401426499646, + "loss": 3.2718, + "step": 1648 + }, + { + "epoch": 2.42, + "grad_norm": 4.197020530700684, + "learning_rate": 0.00017259176292327259, + "loss": 3.3056, + "step": 1649 + }, + { + "epoch": 2.42, + "grad_norm": 5.8328166007995605, + "learning_rate": 0.00017255949561184738, + "loss": 3.2645, + "step": 1650 + }, + { + "epoch": 2.42, + "grad_norm": 9.331326484680176, + "learning_rate": 0.00017252721233781938, + "loss": 3.3117, + "step": 1651 + }, + { + "epoch": 2.42, + "grad_norm": 5.499820232391357, + "learning_rate": 0.00017249491310829078, + "loss": 3.278, + "step": 1652 + }, + { + "epoch": 2.42, + "grad_norm": 4.50565242767334, + "learning_rate": 0.00017246259793036707, + "loss": 3.2662, + "step": 1653 + }, + { + "epoch": 2.43, + "grad_norm": 6.363594055175781, + "learning_rate": 0.00017243026681115748, + "loss": 3.2784, + "step": 1654 + }, + { + "epoch": 2.43, + "grad_norm": 2.7347965240478516, + "learning_rate": 0.00017239791975777458, + "loss": 3.2907, + "step": 1655 + }, + { + "epoch": 2.43, + "grad_norm": 7.49398946762085, + "learning_rate": 0.00017236555677733453, + "loss": 3.3239, + "step": 1656 + }, + { + "epoch": 2.43, + "grad_norm": 10.45259952545166, + "learning_rate": 0.00017233317787695697, + "loss": 3.316, + "step": 1657 + }, + { + "epoch": 2.43, + "grad_norm": 5.672200679779053, + "learning_rate": 0.00017230078306376502, + "loss": 3.2043, + "step": 1658 + }, + { + "epoch": 2.43, + "grad_norm": 4.071951866149902, + "learning_rate": 0.00017226837234488534, + "loss": 3.2218, + "step": 1659 + }, + { + "epoch": 2.43, + "grad_norm": 5.800225257873535, + "learning_rate": 0.00017223594572744806, + "loss": 3.2586, + "step": 1660 + }, + { + "epoch": 2.44, + "grad_norm": 2.592320442199707, + "learning_rate": 0.0001722035032185868, + "loss": 3.244, + "step": 1661 + }, + { + "epoch": 2.44, + "grad_norm": 7.940516471862793, + "learning_rate": 0.00017217104482543876, + "loss": 3.2501, + "step": 1662 + }, + { + "epoch": 2.44, + "grad_norm": 10.571852684020996, + "learning_rate": 0.0001721385705551445, + "loss": 3.2631, + "step": 1663 + }, + { + "epoch": 2.44, + "grad_norm": 5.3643574714660645, + "learning_rate": 0.0001721060804148482, + "loss": 3.2116, + "step": 1664 + }, + { + "epoch": 2.44, + "grad_norm": 4.0979905128479, + "learning_rate": 0.00017207357441169742, + "loss": 3.1886, + "step": 1665 + }, + { + "epoch": 2.44, + "grad_norm": 6.099160194396973, + "learning_rate": 0.0001720410525528433, + "loss": 3.2502, + "step": 1666 + }, + { + "epoch": 2.44, + "grad_norm": 3.1805484294891357, + "learning_rate": 0.00017200851484544037, + "loss": 3.213, + "step": 1667 + }, + { + "epoch": 2.45, + "grad_norm": 7.588128089904785, + "learning_rate": 0.00017197596129664682, + "loss": 3.2935, + "step": 1668 + }, + { + "epoch": 2.45, + "grad_norm": 10.550485610961914, + "learning_rate": 0.0001719433919136241, + "loss": 3.2823, + "step": 1669 + }, + { + "epoch": 2.45, + "grad_norm": 5.52534818649292, + "learning_rate": 0.00017191080670353733, + "loss": 3.2452, + "step": 1670 + }, + { + "epoch": 2.45, + "grad_norm": 4.016864776611328, + "learning_rate": 0.00017187820567355497, + "loss": 3.2412, + "step": 1671 + }, + { + "epoch": 2.45, + "grad_norm": 6.307382583618164, + "learning_rate": 0.0001718455888308491, + "loss": 3.2699, + "step": 1672 + }, + { + "epoch": 2.45, + "grad_norm": 2.715496301651001, + "learning_rate": 0.00017181295618259515, + "loss": 3.2695, + "step": 1673 + }, + { + "epoch": 2.45, + "grad_norm": 7.303908348083496, + "learning_rate": 0.00017178030773597213, + "loss": 3.2392, + "step": 1674 + }, + { + "epoch": 2.46, + "grad_norm": 10.16519546508789, + "learning_rate": 0.00017174764349816245, + "loss": 3.2635, + "step": 1675 + }, + { + "epoch": 2.46, + "grad_norm": 5.080345630645752, + "learning_rate": 0.00017171496347635206, + "loss": 3.2302, + "step": 1676 + }, + { + "epoch": 2.46, + "grad_norm": 5.1449408531188965, + "learning_rate": 0.00017168226767773028, + "loss": 3.2135, + "step": 1677 + }, + { + "epoch": 2.46, + "grad_norm": 7.340461254119873, + "learning_rate": 0.00017164955610949004, + "loss": 3.2561, + "step": 1678 + }, + { + "epoch": 2.46, + "grad_norm": 3.821507453918457, + "learning_rate": 0.00017161682877882758, + "loss": 3.2111, + "step": 1679 + }, + { + "epoch": 2.46, + "grad_norm": 6.017604351043701, + "learning_rate": 0.0001715840856929428, + "loss": 3.3209, + "step": 1680 + }, + { + "epoch": 2.47, + "grad_norm": 8.912531852722168, + "learning_rate": 0.00017155132685903887, + "loss": 3.2441, + "step": 1681 + }, + { + "epoch": 2.47, + "grad_norm": 4.574288368225098, + "learning_rate": 0.00017151855228432253, + "loss": 3.2489, + "step": 1682 + }, + { + "epoch": 2.47, + "grad_norm": 4.964592456817627, + "learning_rate": 0.00017148576197600402, + "loss": 3.2607, + "step": 1683 + }, + { + "epoch": 2.47, + "grad_norm": 7.388319969177246, + "learning_rate": 0.00017145295594129693, + "loss": 3.2537, + "step": 1684 + }, + { + "epoch": 2.47, + "grad_norm": 4.345710277557373, + "learning_rate": 0.0001714201341874184, + "loss": 3.2435, + "step": 1685 + }, + { + "epoch": 2.47, + "grad_norm": 5.151455402374268, + "learning_rate": 0.000171387296721589, + "loss": 3.2198, + "step": 1686 + }, + { + "epoch": 2.47, + "grad_norm": 9.679162979125977, + "learning_rate": 0.0001713544435510327, + "loss": 3.3015, + "step": 1687 + }, + { + "epoch": 2.48, + "grad_norm": 4.5911383628845215, + "learning_rate": 0.00017132157468297704, + "loss": 3.2534, + "step": 1688 + }, + { + "epoch": 2.48, + "grad_norm": 3.5955042839050293, + "learning_rate": 0.00017128869012465292, + "loss": 3.2479, + "step": 1689 + }, + { + "epoch": 2.48, + "grad_norm": 4.769717693328857, + "learning_rate": 0.00017125578988329472, + "loss": 3.2706, + "step": 1690 + }, + { + "epoch": 2.48, + "eval_loss": 3.0677621364593506, + "eval_runtime": 1164.46, + "eval_samples_per_second": 14.721, + "eval_steps_per_second": 0.921, + "step": 1690 + }, + { + "epoch": 2.48, + "grad_norm": 0.9628227949142456, + "learning_rate": 0.0001712228739661403, + "loss": 3.2333, + "step": 1691 + }, + { + "epoch": 2.48, + "grad_norm": 9.899592399597168, + "learning_rate": 0.00017118994238043088, + "loss": 3.287, + "step": 1692 + }, + { + "epoch": 2.48, + "grad_norm": 13.610316276550293, + "learning_rate": 0.00017115699513341127, + "loss": 3.3172, + "step": 1693 + }, + { + "epoch": 2.48, + "grad_norm": 8.333012580871582, + "learning_rate": 0.00017112403223232964, + "loss": 3.2552, + "step": 1694 + }, + { + "epoch": 2.49, + "grad_norm": 1.985851764678955, + "learning_rate": 0.0001710910536844375, + "loss": 3.2494, + "step": 1695 + }, + { + "epoch": 2.49, + "grad_norm": 4.960620403289795, + "learning_rate": 0.00017105805949699, + "loss": 3.2411, + "step": 1696 + }, + { + "epoch": 2.49, + "grad_norm": 2.77993106842041, + "learning_rate": 0.00017102504967724563, + "loss": 3.264, + "step": 1697 + }, + { + "epoch": 2.49, + "grad_norm": 6.050425052642822, + "learning_rate": 0.0001709920242324663, + "loss": 3.2725, + "step": 1698 + }, + { + "epoch": 2.49, + "grad_norm": 6.0771307945251465, + "learning_rate": 0.00017095898316991744, + "loss": 3.199, + "step": 1699 + }, + { + "epoch": 2.49, + "grad_norm": 1.5405139923095703, + "learning_rate": 0.0001709259264968678, + "loss": 3.2523, + "step": 1700 + }, + { + "epoch": 2.49, + "grad_norm": 5.121715068817139, + "learning_rate": 0.00017089285422058968, + "loss": 3.2944, + "step": 1701 + }, + { + "epoch": 2.5, + "grad_norm": 4.806593894958496, + "learning_rate": 0.00017085976634835873, + "loss": 3.2461, + "step": 1702 + }, + { + "epoch": 2.5, + "grad_norm": 3.7719340324401855, + "learning_rate": 0.00017082666288745403, + "loss": 3.2067, + "step": 1703 + }, + { + "epoch": 2.5, + "grad_norm": 3.7723586559295654, + "learning_rate": 0.0001707935438451582, + "loss": 3.2715, + "step": 1704 + }, + { + "epoch": 2.5, + "grad_norm": 2.5075759887695312, + "learning_rate": 0.00017076040922875713, + "loss": 3.2053, + "step": 1705 + }, + { + "epoch": 2.5, + "grad_norm": 2.315262794494629, + "learning_rate": 0.00017072725904554025, + "loss": 3.2529, + "step": 1706 + }, + { + "epoch": 2.5, + "grad_norm": 3.849294900894165, + "learning_rate": 0.0001706940933028004, + "loss": 3.2501, + "step": 1707 + }, + { + "epoch": 2.51, + "grad_norm": 3.2091238498687744, + "learning_rate": 0.00017066091200783375, + "loss": 3.1893, + "step": 1708 + }, + { + "epoch": 2.51, + "grad_norm": 3.587019443511963, + "learning_rate": 0.00017062771516794002, + "loss": 3.2435, + "step": 1709 + }, + { + "epoch": 2.51, + "grad_norm": 2.367382526397705, + "learning_rate": 0.00017059450279042232, + "loss": 3.2453, + "step": 1710 + }, + { + "epoch": 2.51, + "grad_norm": 4.0962700843811035, + "learning_rate": 0.00017056127488258705, + "loss": 3.2043, + "step": 1711 + }, + { + "epoch": 2.51, + "grad_norm": 4.637650966644287, + "learning_rate": 0.0001705280314517442, + "loss": 3.2506, + "step": 1712 + }, + { + "epoch": 2.51, + "grad_norm": 1.244682788848877, + "learning_rate": 0.00017049477250520714, + "loss": 3.2169, + "step": 1713 + }, + { + "epoch": 2.51, + "grad_norm": 0.7056719660758972, + "learning_rate": 0.00017046149805029253, + "loss": 3.212, + "step": 1714 + }, + { + "epoch": 2.52, + "grad_norm": 2.6713907718658447, + "learning_rate": 0.00017042820809432056, + "loss": 3.2665, + "step": 1715 + }, + { + "epoch": 2.52, + "grad_norm": 1.15023934841156, + "learning_rate": 0.00017039490264461477, + "loss": 3.2735, + "step": 1716 + }, + { + "epoch": 2.52, + "grad_norm": 3.00900936126709, + "learning_rate": 0.0001703615817085022, + "loss": 3.2198, + "step": 1717 + }, + { + "epoch": 2.52, + "grad_norm": 3.382124423980713, + "learning_rate": 0.0001703282452933132, + "loss": 3.2279, + "step": 1718 + }, + { + "epoch": 2.52, + "grad_norm": 1.6238938570022583, + "learning_rate": 0.00017029489340638154, + "loss": 3.2521, + "step": 1719 + }, + { + "epoch": 2.52, + "grad_norm": 4.034934997558594, + "learning_rate": 0.00017026152605504442, + "loss": 3.2477, + "step": 1720 + }, + { + "epoch": 2.52, + "grad_norm": 4.463752269744873, + "learning_rate": 0.00017022814324664244, + "loss": 3.2478, + "step": 1721 + }, + { + "epoch": 2.53, + "grad_norm": 2.6497182846069336, + "learning_rate": 0.00017019474498851958, + "loss": 3.2774, + "step": 1722 + }, + { + "epoch": 2.53, + "grad_norm": 2.8634886741638184, + "learning_rate": 0.0001701613312880233, + "loss": 3.2704, + "step": 1723 + }, + { + "epoch": 2.53, + "grad_norm": 3.318505048751831, + "learning_rate": 0.00017012790215250432, + "loss": 3.2681, + "step": 1724 + }, + { + "epoch": 2.53, + "grad_norm": 2.9322783946990967, + "learning_rate": 0.0001700944575893168, + "loss": 3.2489, + "step": 1725 + }, + { + "epoch": 2.53, + "grad_norm": 1.9499863386154175, + "learning_rate": 0.00017006099760581838, + "loss": 3.2197, + "step": 1726 + }, + { + "epoch": 2.53, + "grad_norm": 2.5505218505859375, + "learning_rate": 0.00017002752220937004, + "loss": 3.2399, + "step": 1727 + }, + { + "epoch": 2.53, + "grad_norm": 2.46716570854187, + "learning_rate": 0.0001699940314073361, + "loss": 3.2078, + "step": 1728 + }, + { + "epoch": 2.54, + "grad_norm": 1.8553292751312256, + "learning_rate": 0.00016996052520708434, + "loss": 3.292, + "step": 1729 + }, + { + "epoch": 2.54, + "grad_norm": 2.041597366333008, + "learning_rate": 0.00016992700361598585, + "loss": 3.201, + "step": 1730 + }, + { + "epoch": 2.54, + "grad_norm": 2.173717737197876, + "learning_rate": 0.00016989346664141524, + "loss": 3.2707, + "step": 1731 + }, + { + "epoch": 2.54, + "grad_norm": 1.5214920043945312, + "learning_rate": 0.00016985991429075036, + "loss": 3.2218, + "step": 1732 + }, + { + "epoch": 2.54, + "grad_norm": 6.029373645782471, + "learning_rate": 0.00016982634657137252, + "loss": 3.3117, + "step": 1733 + }, + { + "epoch": 2.54, + "grad_norm": 5.91231107711792, + "learning_rate": 0.00016979276349066638, + "loss": 3.2202, + "step": 1734 + }, + { + "epoch": 2.55, + "grad_norm": 2.8163087368011475, + "learning_rate": 0.00016975916505602, + "loss": 3.2803, + "step": 1735 + }, + { + "epoch": 2.55, + "grad_norm": 3.7836880683898926, + "learning_rate": 0.00016972555127482483, + "loss": 3.2399, + "step": 1736 + }, + { + "epoch": 2.55, + "grad_norm": 3.339012622833252, + "learning_rate": 0.00016969192215447566, + "loss": 3.2619, + "step": 1737 + }, + { + "epoch": 2.55, + "grad_norm": 3.779021739959717, + "learning_rate": 0.00016965827770237067, + "loss": 3.2205, + "step": 1738 + }, + { + "epoch": 2.55, + "grad_norm": 0.7899141311645508, + "learning_rate": 0.00016962461792591137, + "loss": 3.2228, + "step": 1739 + }, + { + "epoch": 2.55, + "grad_norm": 2.307349443435669, + "learning_rate": 0.00016959094283250278, + "loss": 3.2412, + "step": 1740 + }, + { + "epoch": 2.55, + "grad_norm": 1.5662769079208374, + "learning_rate": 0.00016955725242955315, + "loss": 3.257, + "step": 1741 + }, + { + "epoch": 2.56, + "grad_norm": 1.4268426895141602, + "learning_rate": 0.0001695235467244741, + "loss": 3.2718, + "step": 1742 + }, + { + "epoch": 2.56, + "grad_norm": 1.7694865465164185, + "learning_rate": 0.0001694898257246807, + "loss": 3.2414, + "step": 1743 + }, + { + "epoch": 2.56, + "grad_norm": 2.555623769760132, + "learning_rate": 0.00016945608943759138, + "loss": 3.2384, + "step": 1744 + }, + { + "epoch": 2.56, + "grad_norm": 1.4633605480194092, + "learning_rate": 0.0001694223378706278, + "loss": 3.2164, + "step": 1745 + }, + { + "epoch": 2.56, + "grad_norm": 2.0434815883636475, + "learning_rate": 0.0001693885710312152, + "loss": 3.2547, + "step": 1746 + }, + { + "epoch": 2.56, + "grad_norm": 2.4210758209228516, + "learning_rate": 0.00016935478892678195, + "loss": 3.2367, + "step": 1747 + }, + { + "epoch": 2.56, + "grad_norm": 0.8179725408554077, + "learning_rate": 0.00016932099156475995, + "loss": 3.2619, + "step": 1748 + }, + { + "epoch": 2.57, + "grad_norm": 3.3028626441955566, + "learning_rate": 0.00016928717895258437, + "loss": 3.2256, + "step": 1749 + }, + { + "epoch": 2.57, + "grad_norm": 1.852142572402954, + "learning_rate": 0.00016925335109769374, + "loss": 3.2402, + "step": 1750 + }, + { + "epoch": 2.57, + "grad_norm": 2.416368007659912, + "learning_rate": 0.00016921950800753002, + "loss": 3.234, + "step": 1751 + }, + { + "epoch": 2.57, + "grad_norm": 1.5477232933044434, + "learning_rate": 0.00016918564968953842, + "loss": 3.2142, + "step": 1752 + }, + { + "epoch": 2.57, + "grad_norm": 1.5452466011047363, + "learning_rate": 0.00016915177615116754, + "loss": 3.1941, + "step": 1753 + }, + { + "epoch": 2.57, + "grad_norm": 3.213561773300171, + "learning_rate": 0.00016911788739986934, + "loss": 3.2154, + "step": 1754 + }, + { + "epoch": 2.58, + "grad_norm": 1.2881815433502197, + "learning_rate": 0.0001690839834430991, + "loss": 3.2909, + "step": 1755 + }, + { + "epoch": 2.58, + "grad_norm": 3.375774383544922, + "learning_rate": 0.00016905006428831552, + "loss": 3.2096, + "step": 1756 + }, + { + "epoch": 2.58, + "grad_norm": 2.6476569175720215, + "learning_rate": 0.00016901612994298053, + "loss": 3.2082, + "step": 1757 + }, + { + "epoch": 2.58, + "grad_norm": 3.5997047424316406, + "learning_rate": 0.0001689821804145595, + "loss": 3.1989, + "step": 1758 + }, + { + "epoch": 2.58, + "grad_norm": 2.435394763946533, + "learning_rate": 0.00016894821571052104, + "loss": 3.2109, + "step": 1759 + }, + { + "epoch": 2.58, + "grad_norm": 5.273232936859131, + "learning_rate": 0.00016891423583833724, + "loss": 3.2358, + "step": 1760 + }, + { + "epoch": 2.58, + "grad_norm": 4.958769798278809, + "learning_rate": 0.00016888024080548335, + "loss": 3.2315, + "step": 1761 + }, + { + "epoch": 2.59, + "grad_norm": 2.171224594116211, + "learning_rate": 0.00016884623061943813, + "loss": 3.2587, + "step": 1762 + }, + { + "epoch": 2.59, + "grad_norm": 1.9900020360946655, + "learning_rate": 0.00016881220528768356, + "loss": 3.2082, + "step": 1763 + }, + { + "epoch": 2.59, + "grad_norm": 2.3555827140808105, + "learning_rate": 0.00016877816481770496, + "loss": 3.2356, + "step": 1764 + }, + { + "epoch": 2.59, + "grad_norm": 1.4140695333480835, + "learning_rate": 0.00016874410921699105, + "loss": 3.1722, + "step": 1765 + }, + { + "epoch": 2.59, + "grad_norm": 4.323090076446533, + "learning_rate": 0.00016871003849303382, + "loss": 3.2533, + "step": 1766 + }, + { + "epoch": 2.59, + "grad_norm": 2.4270732402801514, + "learning_rate": 0.00016867595265332859, + "loss": 3.2092, + "step": 1767 + }, + { + "epoch": 2.59, + "grad_norm": 3.9084079265594482, + "learning_rate": 0.000168641851705374, + "loss": 3.2595, + "step": 1768 + }, + { + "epoch": 2.6, + "grad_norm": 3.986670970916748, + "learning_rate": 0.00016860773565667206, + "loss": 3.2302, + "step": 1769 + }, + { + "epoch": 2.6, + "grad_norm": 3.4050045013427734, + "learning_rate": 0.0001685736045147281, + "loss": 3.2589, + "step": 1770 + }, + { + "epoch": 2.6, + "grad_norm": 4.226673126220703, + "learning_rate": 0.0001685394582870507, + "loss": 3.235, + "step": 1771 + }, + { + "epoch": 2.6, + "grad_norm": 1.5907903909683228, + "learning_rate": 0.00016850529698115175, + "loss": 3.3035, + "step": 1772 + }, + { + "epoch": 2.6, + "grad_norm": 2.343108654022217, + "learning_rate": 0.00016847112060454665, + "loss": 3.1848, + "step": 1773 + }, + { + "epoch": 2.6, + "grad_norm": 3.3092055320739746, + "learning_rate": 0.00016843692916475384, + "loss": 3.2648, + "step": 1774 + }, + { + "epoch": 2.6, + "grad_norm": 1.7477455139160156, + "learning_rate": 0.00016840272266929523, + "loss": 3.206, + "step": 1775 + }, + { + "epoch": 2.61, + "grad_norm": 4.240764617919922, + "learning_rate": 0.00016836850112569613, + "loss": 3.2177, + "step": 1776 + }, + { + "epoch": 2.61, + "grad_norm": 4.2325758934021, + "learning_rate": 0.0001683342645414849, + "loss": 3.2154, + "step": 1777 + }, + { + "epoch": 2.61, + "grad_norm": 2.993154525756836, + "learning_rate": 0.00016830001292419345, + "loss": 3.1972, + "step": 1778 + }, + { + "epoch": 2.61, + "grad_norm": 3.2484753131866455, + "learning_rate": 0.00016826574628135688, + "loss": 3.232, + "step": 1779 + }, + { + "epoch": 2.61, + "grad_norm": 2.438793420791626, + "learning_rate": 0.00016823146462051364, + "loss": 3.2292, + "step": 1780 + }, + { + "epoch": 2.61, + "grad_norm": 2.0203206539154053, + "learning_rate": 0.0001681971679492054, + "loss": 3.1791, + "step": 1781 + }, + { + "epoch": 2.62, + "grad_norm": 4.8010125160217285, + "learning_rate": 0.0001681628562749773, + "loss": 3.2012, + "step": 1782 + }, + { + "epoch": 2.62, + "grad_norm": 4.596678256988525, + "learning_rate": 0.00016812852960537762, + "loss": 3.2204, + "step": 1783 + }, + { + "epoch": 2.62, + "grad_norm": 2.4350953102111816, + "learning_rate": 0.00016809418794795796, + "loss": 3.2175, + "step": 1784 + }, + { + "epoch": 2.62, + "grad_norm": 15.600502014160156, + "learning_rate": 0.00016805983131027333, + "loss": 3.256, + "step": 1785 + }, + { + "epoch": 2.62, + "grad_norm": 3.7827413082122803, + "learning_rate": 0.00016802545969988195, + "loss": 3.2165, + "step": 1786 + }, + { + "epoch": 2.62, + "grad_norm": 3.0212562084198, + "learning_rate": 0.0001679910731243453, + "loss": 3.2537, + "step": 1787 + }, + { + "epoch": 2.62, + "grad_norm": 2.496250629425049, + "learning_rate": 0.00016795667159122824, + "loss": 3.212, + "step": 1788 + }, + { + "epoch": 2.63, + "grad_norm": 3.222059488296509, + "learning_rate": 0.00016792225510809883, + "loss": 3.2665, + "step": 1789 + }, + { + "epoch": 2.63, + "grad_norm": 3.4902381896972656, + "learning_rate": 0.0001678878236825285, + "loss": 3.2326, + "step": 1790 + }, + { + "epoch": 2.63, + "grad_norm": 0.7913928627967834, + "learning_rate": 0.00016785337732209193, + "loss": 3.1662, + "step": 1791 + }, + { + "epoch": 2.63, + "grad_norm": 3.9567112922668457, + "learning_rate": 0.00016781891603436706, + "loss": 3.2217, + "step": 1792 + }, + { + "epoch": 2.63, + "grad_norm": 1.712324619293213, + "learning_rate": 0.00016778443982693519, + "loss": 3.1855, + "step": 1793 + }, + { + "epoch": 2.63, + "grad_norm": 1.2269927263259888, + "learning_rate": 0.0001677499487073808, + "loss": 3.2306, + "step": 1794 + }, + { + "epoch": 2.63, + "grad_norm": 3.1048622131347656, + "learning_rate": 0.00016771544268329177, + "loss": 3.1975, + "step": 1795 + }, + { + "epoch": 2.64, + "grad_norm": 1.18289315700531, + "learning_rate": 0.00016768092176225915, + "loss": 3.1904, + "step": 1796 + }, + { + "epoch": 2.64, + "grad_norm": 1.7925835847854614, + "learning_rate": 0.0001676463859518773, + "loss": 3.2143, + "step": 1797 + }, + { + "epoch": 2.64, + "grad_norm": 2.179051637649536, + "learning_rate": 0.0001676118352597439, + "loss": 3.2297, + "step": 1798 + }, + { + "epoch": 2.64, + "grad_norm": 1.2693228721618652, + "learning_rate": 0.00016757726969345984, + "loss": 3.2181, + "step": 1799 + }, + { + "epoch": 2.64, + "grad_norm": 5.075618743896484, + "learning_rate": 0.00016754268926062938, + "loss": 3.237, + "step": 1800 + }, + { + "epoch": 2.64, + "grad_norm": 5.278939723968506, + "learning_rate": 0.0001675080939688599, + "loss": 3.2266, + "step": 1801 + }, + { + "epoch": 2.64, + "grad_norm": 2.5977909564971924, + "learning_rate": 0.00016747348382576213, + "loss": 3.2409, + "step": 1802 + }, + { + "epoch": 2.65, + "grad_norm": 3.318488597869873, + "learning_rate": 0.00016743885883895013, + "loss": 3.231, + "step": 1803 + }, + { + "epoch": 2.65, + "grad_norm": 2.22575306892395, + "learning_rate": 0.00016740421901604118, + "loss": 3.2313, + "step": 1804 + }, + { + "epoch": 2.65, + "grad_norm": 2.436943769454956, + "learning_rate": 0.00016736956436465573, + "loss": 3.2268, + "step": 1805 + }, + { + "epoch": 2.65, + "grad_norm": 2.966697931289673, + "learning_rate": 0.0001673348948924176, + "loss": 3.2139, + "step": 1806 + }, + { + "epoch": 2.65, + "grad_norm": 2.458636999130249, + "learning_rate": 0.0001673002106069539, + "loss": 3.2228, + "step": 1807 + }, + { + "epoch": 2.65, + "grad_norm": 3.647672414779663, + "learning_rate": 0.00016726551151589485, + "loss": 3.256, + "step": 1808 + }, + { + "epoch": 2.66, + "grad_norm": 2.713827133178711, + "learning_rate": 0.0001672307976268741, + "loss": 3.2256, + "step": 1809 + }, + { + "epoch": 2.66, + "grad_norm": 3.452815294265747, + "learning_rate": 0.0001671960689475284, + "loss": 3.2875, + "step": 1810 + }, + { + "epoch": 2.66, + "grad_norm": 2.377556085586548, + "learning_rate": 0.00016716132548549786, + "loss": 3.2188, + "step": 1811 + }, + { + "epoch": 2.66, + "grad_norm": 3.791208028793335, + "learning_rate": 0.0001671265672484258, + "loss": 3.234, + "step": 1812 + }, + { + "epoch": 2.66, + "grad_norm": 3.7197184562683105, + "learning_rate": 0.00016709179424395884, + "loss": 3.1895, + "step": 1813 + }, + { + "epoch": 2.66, + "grad_norm": 2.4916012287139893, + "learning_rate": 0.00016705700647974675, + "loss": 3.2053, + "step": 1814 + }, + { + "epoch": 2.66, + "grad_norm": 2.569413185119629, + "learning_rate": 0.00016702220396344267, + "loss": 3.2358, + "step": 1815 + }, + { + "epoch": 2.67, + "grad_norm": 3.520432949066162, + "learning_rate": 0.00016698738670270285, + "loss": 3.2369, + "step": 1816 + }, + { + "epoch": 2.67, + "grad_norm": 2.880115032196045, + "learning_rate": 0.00016695255470518686, + "loss": 3.2232, + "step": 1817 + }, + { + "epoch": 2.67, + "grad_norm": 3.42354416847229, + "learning_rate": 0.00016691770797855758, + "loss": 3.1885, + "step": 1818 + }, + { + "epoch": 2.67, + "grad_norm": 3.31255841255188, + "learning_rate": 0.00016688284653048097, + "loss": 3.21, + "step": 1819 + }, + { + "epoch": 2.67, + "grad_norm": 3.157924175262451, + "learning_rate": 0.00016684797036862635, + "loss": 3.2055, + "step": 1820 + }, + { + "epoch": 2.67, + "grad_norm": 2.820952892303467, + "learning_rate": 0.00016681307950066628, + "loss": 3.1809, + "step": 1821 + }, + { + "epoch": 2.67, + "grad_norm": 3.921417236328125, + "learning_rate": 0.00016677817393427645, + "loss": 3.1902, + "step": 1822 + }, + { + "epoch": 2.68, + "grad_norm": 3.5899341106414795, + "learning_rate": 0.00016674325367713592, + "loss": 3.2222, + "step": 1823 + }, + { + "epoch": 2.68, + "grad_norm": 2.645939350128174, + "learning_rate": 0.00016670831873692683, + "loss": 3.1834, + "step": 1824 + }, + { + "epoch": 2.68, + "grad_norm": 2.4159257411956787, + "learning_rate": 0.00016667336912133468, + "loss": 3.2018, + "step": 1825 + }, + { + "epoch": 2.68, + "grad_norm": 3.56545090675354, + "learning_rate": 0.0001666384048380482, + "loss": 3.2096, + "step": 1826 + }, + { + "epoch": 2.68, + "grad_norm": 2.857130765914917, + "learning_rate": 0.0001666034258947592, + "loss": 3.2102, + "step": 1827 + }, + { + "epoch": 2.68, + "grad_norm": 3.4373629093170166, + "learning_rate": 0.00016656843229916285, + "loss": 3.2017, + "step": 1828 + }, + { + "epoch": 2.68, + "grad_norm": 2.5930848121643066, + "learning_rate": 0.00016653342405895756, + "loss": 3.2265, + "step": 1829 + }, + { + "epoch": 2.69, + "grad_norm": 4.415797710418701, + "learning_rate": 0.0001664984011818448, + "loss": 3.1589, + "step": 1830 + }, + { + "epoch": 2.69, + "grad_norm": 4.656808853149414, + "learning_rate": 0.0001664633636755295, + "loss": 3.2189, + "step": 1831 + }, + { + "epoch": 2.69, + "grad_norm": 1.3080215454101562, + "learning_rate": 0.00016642831154771957, + "loss": 3.2178, + "step": 1832 + }, + { + "epoch": 2.69, + "grad_norm": 1.491684913635254, + "learning_rate": 0.0001663932448061263, + "loss": 3.1922, + "step": 1833 + }, + { + "epoch": 2.69, + "grad_norm": 5.020033359527588, + "learning_rate": 0.0001663581634584641, + "loss": 3.2153, + "step": 1834 + }, + { + "epoch": 2.69, + "grad_norm": 4.078526973724365, + "learning_rate": 0.0001663230675124507, + "loss": 3.2165, + "step": 1835 + }, + { + "epoch": 2.7, + "grad_norm": 2.787774085998535, + "learning_rate": 0.00016628795697580686, + "loss": 3.1903, + "step": 1836 + }, + { + "epoch": 2.7, + "grad_norm": 2.4265925884246826, + "learning_rate": 0.0001662528318562568, + "loss": 3.1732, + "step": 1837 + }, + { + "epoch": 2.7, + "grad_norm": 3.579639434814453, + "learning_rate": 0.00016621769216152766, + "loss": 3.1891, + "step": 1838 + }, + { + "epoch": 2.7, + "grad_norm": 2.984905958175659, + "learning_rate": 0.00016618253789935005, + "loss": 3.2348, + "step": 1839 + }, + { + "epoch": 2.7, + "grad_norm": 4.036933898925781, + "learning_rate": 0.00016614736907745768, + "loss": 3.1677, + "step": 1840 + }, + { + "epoch": 2.7, + "grad_norm": 3.9433419704437256, + "learning_rate": 0.00016611218570358735, + "loss": 3.18, + "step": 1841 + }, + { + "epoch": 2.7, + "grad_norm": 2.285982370376587, + "learning_rate": 0.00016607698778547925, + "loss": 3.2298, + "step": 1842 + }, + { + "epoch": 2.71, + "grad_norm": 2.241262197494507, + "learning_rate": 0.00016604177533087667, + "loss": 3.1903, + "step": 1843 + }, + { + "epoch": 2.71, + "grad_norm": 3.7759506702423096, + "learning_rate": 0.00016600654834752611, + "loss": 3.2093, + "step": 1844 + }, + { + "epoch": 2.71, + "grad_norm": 3.2228505611419678, + "learning_rate": 0.00016597130684317725, + "loss": 3.1934, + "step": 1845 + }, + { + "epoch": 2.71, + "grad_norm": 3.3804237842559814, + "learning_rate": 0.00016593605082558305, + "loss": 3.2345, + "step": 1846 + }, + { + "epoch": 2.71, + "grad_norm": 2.6020681858062744, + "learning_rate": 0.0001659007803024995, + "loss": 3.1868, + "step": 1847 + }, + { + "epoch": 2.71, + "grad_norm": 3.3784382343292236, + "learning_rate": 0.00016586549528168596, + "loss": 3.1865, + "step": 1848 + }, + { + "epoch": 2.71, + "grad_norm": 2.881960391998291, + "learning_rate": 0.0001658301957709049, + "loss": 3.1906, + "step": 1849 + }, + { + "epoch": 2.72, + "grad_norm": 3.543879270553589, + "learning_rate": 0.00016579488177792187, + "loss": 3.2124, + "step": 1850 + }, + { + "epoch": 2.72, + "grad_norm": 3.8492488861083984, + "learning_rate": 0.00016575955331050584, + "loss": 3.3212, + "step": 1851 + }, + { + "epoch": 2.72, + "grad_norm": 2.2801854610443115, + "learning_rate": 0.00016572421037642875, + "loss": 3.2374, + "step": 1852 + }, + { + "epoch": 2.72, + "grad_norm": 2.0372071266174316, + "learning_rate": 0.00016568885298346586, + "loss": 3.2105, + "step": 1853 + }, + { + "epoch": 2.72, + "grad_norm": 4.137518405914307, + "learning_rate": 0.00016565348113939556, + "loss": 3.1718, + "step": 1854 + }, + { + "epoch": 2.72, + "grad_norm": 3.7536251544952393, + "learning_rate": 0.00016561809485199938, + "loss": 3.2118, + "step": 1855 + }, + { + "epoch": 2.73, + "grad_norm": 3.2339396476745605, + "learning_rate": 0.00016558269412906206, + "loss": 3.1746, + "step": 1856 + }, + { + "epoch": 2.73, + "grad_norm": 2.7833776473999023, + "learning_rate": 0.0001655472789783716, + "loss": 3.1876, + "step": 1857 + }, + { + "epoch": 2.73, + "grad_norm": 3.6448328495025635, + "learning_rate": 0.000165511849407719, + "loss": 3.2143, + "step": 1858 + }, + { + "epoch": 2.73, + "grad_norm": 3.4505040645599365, + "learning_rate": 0.0001654764054248986, + "loss": 3.173, + "step": 1859 + }, + { + "epoch": 2.73, + "eval_loss": 3.040992498397827, + "eval_runtime": 1164.9923, + "eval_samples_per_second": 14.714, + "eval_steps_per_second": 0.92, + "step": 1859 + }, + { + "epoch": 2.73, + "grad_norm": 2.8391804695129395, + "learning_rate": 0.0001654409470377078, + "loss": 3.2099, + "step": 1860 + }, + { + "epoch": 2.73, + "grad_norm": 2.7543411254882812, + "learning_rate": 0.0001654054742539472, + "loss": 3.1918, + "step": 1861 + }, + { + "epoch": 2.73, + "grad_norm": 3.5177597999572754, + "learning_rate": 0.00016536998708142062, + "loss": 3.245, + "step": 1862 + }, + { + "epoch": 2.74, + "grad_norm": 3.0664377212524414, + "learning_rate": 0.00016533448552793494, + "loss": 3.1616, + "step": 1863 + }, + { + "epoch": 2.74, + "grad_norm": 2.924795389175415, + "learning_rate": 0.00016529896960130032, + "loss": 3.229, + "step": 1864 + }, + { + "epoch": 2.74, + "grad_norm": 3.104602575302124, + "learning_rate": 0.00016526343930933, + "loss": 3.2294, + "step": 1865 + }, + { + "epoch": 2.74, + "grad_norm": 3.728653907775879, + "learning_rate": 0.0001652278946598404, + "loss": 3.2088, + "step": 1866 + }, + { + "epoch": 2.74, + "grad_norm": 3.854654312133789, + "learning_rate": 0.00016519233566065112, + "loss": 3.1829, + "step": 1867 + }, + { + "epoch": 2.74, + "grad_norm": 2.6493520736694336, + "learning_rate": 0.0001651567623195849, + "loss": 3.1807, + "step": 1868 + }, + { + "epoch": 2.74, + "grad_norm": 2.057978391647339, + "learning_rate": 0.00016512117464446762, + "loss": 3.141, + "step": 1869 + }, + { + "epoch": 2.75, + "grad_norm": 4.454994201660156, + "learning_rate": 0.00016508557264312835, + "loss": 3.1908, + "step": 1870 + }, + { + "epoch": 2.75, + "grad_norm": 4.243129253387451, + "learning_rate": 0.00016504995632339928, + "loss": 3.1819, + "step": 1871 + }, + { + "epoch": 2.75, + "grad_norm": 2.3739356994628906, + "learning_rate": 0.00016501432569311574, + "loss": 3.2002, + "step": 1872 + }, + { + "epoch": 2.75, + "grad_norm": 1.8129799365997314, + "learning_rate": 0.00016497868076011628, + "loss": 3.2084, + "step": 1873 + }, + { + "epoch": 2.75, + "grad_norm": 4.162952899932861, + "learning_rate": 0.0001649430215322425, + "loss": 3.1766, + "step": 1874 + }, + { + "epoch": 2.75, + "grad_norm": 3.205064535140991, + "learning_rate": 0.00016490734801733923, + "loss": 3.1601, + "step": 1875 + }, + { + "epoch": 2.75, + "grad_norm": 3.1646533012390137, + "learning_rate": 0.0001648716602232544, + "loss": 3.2347, + "step": 1876 + }, + { + "epoch": 2.76, + "grad_norm": 3.2153899669647217, + "learning_rate": 0.00016483595815783904, + "loss": 3.2166, + "step": 1877 + }, + { + "epoch": 2.76, + "grad_norm": 2.5029168128967285, + "learning_rate": 0.00016480024182894743, + "loss": 3.2366, + "step": 1878 + }, + { + "epoch": 2.76, + "grad_norm": 1.944775938987732, + "learning_rate": 0.0001647645112444369, + "loss": 3.1608, + "step": 1879 + }, + { + "epoch": 2.76, + "grad_norm": 4.4062700271606445, + "learning_rate": 0.00016472876641216786, + "loss": 3.226, + "step": 1880 + }, + { + "epoch": 2.76, + "grad_norm": 3.8181326389312744, + "learning_rate": 0.00016469300734000403, + "loss": 3.1934, + "step": 1881 + }, + { + "epoch": 2.76, + "grad_norm": 2.7566487789154053, + "learning_rate": 0.00016465723403581218, + "loss": 3.2031, + "step": 1882 + }, + { + "epoch": 2.77, + "grad_norm": 2.438339948654175, + "learning_rate": 0.00016462144650746213, + "loss": 3.2028, + "step": 1883 + }, + { + "epoch": 2.77, + "grad_norm": 3.6031992435455322, + "learning_rate": 0.0001645856447628269, + "loss": 3.2054, + "step": 1884 + }, + { + "epoch": 2.77, + "grad_norm": 2.86798095703125, + "learning_rate": 0.00016454982880978266, + "loss": 3.2189, + "step": 1885 + }, + { + "epoch": 2.77, + "grad_norm": 3.7704923152923584, + "learning_rate": 0.0001645139986562087, + "loss": 3.1854, + "step": 1886 + }, + { + "epoch": 2.77, + "grad_norm": 3.522306442260742, + "learning_rate": 0.0001644781543099874, + "loss": 3.2357, + "step": 1887 + }, + { + "epoch": 2.77, + "grad_norm": 2.8147850036621094, + "learning_rate": 0.00016444229577900423, + "loss": 3.2041, + "step": 1888 + }, + { + "epoch": 2.77, + "grad_norm": 4.882560729980469, + "learning_rate": 0.00016440642307114786, + "loss": 3.1773, + "step": 1889 + }, + { + "epoch": 2.78, + "grad_norm": 1.0316693782806396, + "learning_rate": 0.00016437053619431006, + "loss": 3.195, + "step": 1890 + }, + { + "epoch": 2.78, + "grad_norm": 1.5295097827911377, + "learning_rate": 0.0001643346351563857, + "loss": 3.2313, + "step": 1891 + }, + { + "epoch": 2.78, + "grad_norm": 1.8384897708892822, + "learning_rate": 0.00016429871996527276, + "loss": 3.1694, + "step": 1892 + }, + { + "epoch": 2.78, + "grad_norm": 1.1532273292541504, + "learning_rate": 0.00016426279062887233, + "loss": 3.1872, + "step": 1893 + }, + { + "epoch": 2.78, + "grad_norm": 0.97287917137146, + "learning_rate": 0.0001642268471550886, + "loss": 3.2283, + "step": 1894 + }, + { + "epoch": 2.78, + "grad_norm": 3.258812665939331, + "learning_rate": 0.00016419088955182897, + "loss": 3.2109, + "step": 1895 + }, + { + "epoch": 2.78, + "grad_norm": 3.9376001358032227, + "learning_rate": 0.00016415491782700384, + "loss": 3.213, + "step": 1896 + }, + { + "epoch": 2.79, + "grad_norm": 3.313781261444092, + "learning_rate": 0.00016411893198852668, + "loss": 3.2419, + "step": 1897 + }, + { + "epoch": 2.79, + "grad_norm": 1.2715352773666382, + "learning_rate": 0.00016408293204431425, + "loss": 3.1853, + "step": 1898 + }, + { + "epoch": 2.79, + "grad_norm": 4.407071113586426, + "learning_rate": 0.00016404691800228616, + "loss": 3.1633, + "step": 1899 + }, + { + "epoch": 2.79, + "grad_norm": 2.881059408187866, + "learning_rate": 0.00016401088987036538, + "loss": 3.1711, + "step": 1900 + }, + { + "epoch": 2.79, + "grad_norm": 213.681884765625, + "learning_rate": 0.0001639748476564778, + "loss": 3.219, + "step": 1901 + }, + { + "epoch": 2.79, + "grad_norm": 15.1211576461792, + "learning_rate": 0.00016393879136855248, + "loss": 3.3071, + "step": 1902 + }, + { + "epoch": 2.79, + "grad_norm": 227.103759765625, + "learning_rate": 0.00016390272101452156, + "loss": 3.3632, + "step": 1903 + }, + { + "epoch": 2.8, + "grad_norm": 94.93277740478516, + "learning_rate": 0.00016386663660232026, + "loss": 3.3738, + "step": 1904 + }, + { + "epoch": 2.8, + "grad_norm": 21.8365421295166, + "learning_rate": 0.00016383053813988694, + "loss": 3.4633, + "step": 1905 + }, + { + "epoch": 2.8, + "grad_norm": 461.8182373046875, + "learning_rate": 0.00016379442563516303, + "loss": 3.4446, + "step": 1906 + }, + { + "epoch": 2.8, + "grad_norm": 73.93539428710938, + "learning_rate": 0.00016375829909609296, + "loss": 3.511, + "step": 1907 + }, + { + "epoch": 2.8, + "grad_norm": 26.853099822998047, + "learning_rate": 0.00016372215853062438, + "loss": 3.4775, + "step": 1908 + }, + { + "epoch": 2.8, + "grad_norm": 449.6775207519531, + "learning_rate": 0.000163686003946708, + "loss": 3.5932, + "step": 1909 + }, + { + "epoch": 2.81, + "grad_norm": 1817.4749755859375, + "learning_rate": 0.00016364983535229754, + "loss": 3.6702, + "step": 1910 + }, + { + "epoch": 2.81, + "grad_norm": 431.15777587890625, + "learning_rate": 0.00016361365275534986, + "loss": 4.165, + "step": 1911 + }, + { + "epoch": 2.81, + "grad_norm": 1153.154296875, + "learning_rate": 0.0001635774561638249, + "loss": 4.4609, + "step": 1912 + }, + { + "epoch": 2.81, + "grad_norm": 378.61279296875, + "learning_rate": 0.00016354124558568567, + "loss": 4.9769, + "step": 1913 + }, + { + "epoch": 2.81, + "grad_norm": 228.8455352783203, + "learning_rate": 0.00016350502102889826, + "loss": 5.4494, + "step": 1914 + }, + { + "epoch": 2.81, + "grad_norm": 2911.531005859375, + "learning_rate": 0.0001634687825014318, + "loss": 5.2172, + "step": 1915 + }, + { + "epoch": 2.81, + "grad_norm": 8.003608703613281, + "learning_rate": 0.00016343253001125852, + "loss": 4.3254, + "step": 1916 + }, + { + "epoch": 2.82, + "grad_norm": 122.48676300048828, + "learning_rate": 0.00016339626356635372, + "loss": 3.9329, + "step": 1917 + }, + { + "epoch": 2.82, + "grad_norm": 165.21929931640625, + "learning_rate": 0.00016335998317469586, + "loss": 4.2273, + "step": 1918 + }, + { + "epoch": 2.82, + "grad_norm": 335.0215148925781, + "learning_rate": 0.00016332368884426626, + "loss": 4.668, + "step": 1919 + }, + { + "epoch": 2.82, + "grad_norm": 3772.19091796875, + "learning_rate": 0.00016328738058304956, + "loss": 5.2298, + "step": 1920 + }, + { + "epoch": 2.82, + "grad_norm": 34.06552505493164, + "learning_rate": 0.0001632510583990332, + "loss": 7.4421, + "step": 1921 + }, + { + "epoch": 2.82, + "grad_norm": 3216.72998046875, + "learning_rate": 0.00016321472230020794, + "loss": 7.2781, + "step": 1922 + }, + { + "epoch": 2.82, + "grad_norm": 9.185710906982422, + "learning_rate": 0.0001631783722945674, + "loss": 6.4889, + "step": 1923 + }, + { + "epoch": 2.83, + "grad_norm": 25.194772720336914, + "learning_rate": 0.00016314200839010834, + "loss": 6.3019, + "step": 1924 + }, + { + "epoch": 2.83, + "grad_norm": 35.877498626708984, + "learning_rate": 0.0001631056305948306, + "loss": 6.1576, + "step": 1925 + }, + { + "epoch": 2.83, + "grad_norm": 35.11639404296875, + "learning_rate": 0.00016306923891673707, + "loss": 5.7687, + "step": 1926 + }, + { + "epoch": 2.83, + "grad_norm": 7.56588077545166, + "learning_rate": 0.0001630328333638336, + "loss": 5.4341, + "step": 1927 + }, + { + "epoch": 2.83, + "grad_norm": 11.163580894470215, + "learning_rate": 0.00016299641394412927, + "loss": 5.2751, + "step": 1928 + }, + { + "epoch": 2.83, + "grad_norm": 9.635797500610352, + "learning_rate": 0.00016295998066563603, + "loss": 5.1351, + "step": 1929 + }, + { + "epoch": 2.83, + "grad_norm": 7.929530143737793, + "learning_rate": 0.000162923533536369, + "loss": 4.9651, + "step": 1930 + }, + { + "epoch": 2.84, + "grad_norm": 12.908093452453613, + "learning_rate": 0.00016288707256434623, + "loss": 4.9926, + "step": 1931 + }, + { + "epoch": 2.84, + "grad_norm": 17.120487213134766, + "learning_rate": 0.000162850597757589, + "loss": 4.9808, + "step": 1932 + }, + { + "epoch": 2.84, + "grad_norm": 12.659987449645996, + "learning_rate": 0.00016281410912412144, + "loss": 4.8239, + "step": 1933 + }, + { + "epoch": 2.84, + "grad_norm": 6.449459552764893, + "learning_rate": 0.00016277760667197087, + "loss": 4.7229, + "step": 1934 + }, + { + "epoch": 2.84, + "grad_norm": 11.50796127319336, + "learning_rate": 0.00016274109040916753, + "loss": 4.5966, + "step": 1935 + }, + { + "epoch": 2.84, + "grad_norm": 10.945295333862305, + "learning_rate": 0.00016270456034374474, + "loss": 4.5114, + "step": 1936 + }, + { + "epoch": 2.85, + "grad_norm": 6.565453052520752, + "learning_rate": 0.00016266801648373894, + "loss": 4.4361, + "step": 1937 + }, + { + "epoch": 2.85, + "grad_norm": 6.896590709686279, + "learning_rate": 0.0001626314588371895, + "loss": 4.3267, + "step": 1938 + }, + { + "epoch": 2.85, + "grad_norm": 6.761734485626221, + "learning_rate": 0.00016259488741213884, + "loss": 4.336, + "step": 1939 + }, + { + "epoch": 2.85, + "grad_norm": 7.6577301025390625, + "learning_rate": 0.00016255830221663243, + "loss": 4.2871, + "step": 1940 + }, + { + "epoch": 2.85, + "grad_norm": 5.5717620849609375, + "learning_rate": 0.00016252170325871878, + "loss": 4.2744, + "step": 1941 + }, + { + "epoch": 2.85, + "grad_norm": 6.425532341003418, + "learning_rate": 0.00016248509054644943, + "loss": 4.2368, + "step": 1942 + }, + { + "epoch": 2.85, + "grad_norm": 6.052181720733643, + "learning_rate": 0.00016244846408787895, + "loss": 4.1986, + "step": 1943 + }, + { + "epoch": 2.86, + "grad_norm": 4.308633804321289, + "learning_rate": 0.00016241182389106483, + "loss": 4.2131, + "step": 1944 + }, + { + "epoch": 2.86, + "grad_norm": 3.014502763748169, + "learning_rate": 0.00016237516996406775, + "loss": 4.1368, + "step": 1945 + }, + { + "epoch": 2.86, + "grad_norm": 3.824622869491577, + "learning_rate": 0.00016233850231495126, + "loss": 4.1542, + "step": 1946 + }, + { + "epoch": 2.86, + "grad_norm": 4.700507640838623, + "learning_rate": 0.00016230182095178206, + "loss": 4.032, + "step": 1947 + }, + { + "epoch": 2.86, + "grad_norm": 2.4007976055145264, + "learning_rate": 0.0001622651258826298, + "loss": 4.043, + "step": 1948 + }, + { + "epoch": 2.86, + "grad_norm": 2.202371835708618, + "learning_rate": 0.0001622284171155671, + "loss": 4.0561, + "step": 1949 + }, + { + "epoch": 2.86, + "grad_norm": 3.510922431945801, + "learning_rate": 0.00016219169465866965, + "loss": 4.0186, + "step": 1950 + }, + { + "epoch": 2.87, + "grad_norm": 3.223702907562256, + "learning_rate": 0.0001621549585200162, + "loss": 4.0432, + "step": 1951 + }, + { + "epoch": 2.87, + "grad_norm": 2.4226481914520264, + "learning_rate": 0.00016211820870768842, + "loss": 4.0018, + "step": 1952 + }, + { + "epoch": 2.87, + "grad_norm": 4.14686393737793, + "learning_rate": 0.000162081445229771, + "loss": 3.9202, + "step": 1953 + }, + { + "epoch": 2.87, + "grad_norm": 4.985760688781738, + "learning_rate": 0.0001620446680943517, + "loss": 4.0172, + "step": 1954 + }, + { + "epoch": 2.87, + "grad_norm": 4.1533427238464355, + "learning_rate": 0.00016200787730952124, + "loss": 3.9214, + "step": 1955 + }, + { + "epoch": 2.87, + "grad_norm": 2.108228921890259, + "learning_rate": 0.00016197107288337328, + "loss": 3.9857, + "step": 1956 + }, + { + "epoch": 2.88, + "grad_norm": 2.6513140201568604, + "learning_rate": 0.00016193425482400464, + "loss": 3.9326, + "step": 1957 + }, + { + "epoch": 2.88, + "grad_norm": 3.7441959381103516, + "learning_rate": 0.00016189742313951498, + "loss": 3.8733, + "step": 1958 + }, + { + "epoch": 2.88, + "grad_norm": 2.4456493854522705, + "learning_rate": 0.00016186057783800704, + "loss": 3.9193, + "step": 1959 + }, + { + "epoch": 2.88, + "grad_norm": 2.1966073513031006, + "learning_rate": 0.00016182371892758658, + "loss": 3.9049, + "step": 1960 + }, + { + "epoch": 2.88, + "grad_norm": 3.958984613418579, + "learning_rate": 0.00016178684641636226, + "loss": 3.8627, + "step": 1961 + }, + { + "epoch": 2.88, + "grad_norm": 2.8413755893707275, + "learning_rate": 0.00016174996031244581, + "loss": 3.9005, + "step": 1962 + }, + { + "epoch": 2.88, + "grad_norm": 2.8592469692230225, + "learning_rate": 0.00016171306062395192, + "loss": 3.8878, + "step": 1963 + }, + { + "epoch": 2.89, + "grad_norm": 5.321715831756592, + "learning_rate": 0.00016167614735899827, + "loss": 3.8257, + "step": 1964 + }, + { + "epoch": 2.89, + "grad_norm": 2.1376495361328125, + "learning_rate": 0.00016163922052570554, + "loss": 3.8189, + "step": 1965 + }, + { + "epoch": 2.89, + "grad_norm": 6.535236835479736, + "learning_rate": 0.0001616022801321974, + "loss": 3.8569, + "step": 1966 + }, + { + "epoch": 2.89, + "grad_norm": 4.882155895233154, + "learning_rate": 0.00016156532618660045, + "loss": 3.8699, + "step": 1967 + }, + { + "epoch": 2.89, + "grad_norm": 5.889296531677246, + "learning_rate": 0.00016152835869704432, + "loss": 3.8201, + "step": 1968 + }, + { + "epoch": 2.89, + "grad_norm": 5.7923102378845215, + "learning_rate": 0.00016149137767166165, + "loss": 3.8467, + "step": 1969 + }, + { + "epoch": 2.89, + "grad_norm": 4.793107509613037, + "learning_rate": 0.000161454383118588, + "loss": 3.789, + "step": 1970 + }, + { + "epoch": 2.9, + "grad_norm": 5.585588455200195, + "learning_rate": 0.00016141737504596186, + "loss": 3.8359, + "step": 1971 + }, + { + "epoch": 2.9, + "grad_norm": 1.328306794166565, + "learning_rate": 0.00016138035346192483, + "loss": 3.781, + "step": 1972 + }, + { + "epoch": 2.9, + "grad_norm": 1.4749982357025146, + "learning_rate": 0.00016134331837462142, + "loss": 3.7788, + "step": 1973 + }, + { + "epoch": 2.9, + "grad_norm": 4.906754016876221, + "learning_rate": 0.00016130626979219906, + "loss": 3.8094, + "step": 1974 + }, + { + "epoch": 2.9, + "grad_norm": 2.958904266357422, + "learning_rate": 0.00016126920772280818, + "loss": 3.8173, + "step": 1975 + }, + { + "epoch": 2.9, + "grad_norm": 8.751588821411133, + "learning_rate": 0.00016123213217460224, + "loss": 3.7913, + "step": 1976 + }, + { + "epoch": 2.9, + "grad_norm": 10.918651580810547, + "learning_rate": 0.00016119504315573755, + "loss": 3.7906, + "step": 1977 + }, + { + "epoch": 2.91, + "grad_norm": 2.1029868125915527, + "learning_rate": 0.0001611579406743735, + "loss": 3.8032, + "step": 1978 + }, + { + "epoch": 2.91, + "grad_norm": 9.132828712463379, + "learning_rate": 0.00016112082473867235, + "loss": 3.8242, + "step": 1979 + }, + { + "epoch": 2.91, + "grad_norm": 10.694255828857422, + "learning_rate": 0.0001610836953567994, + "loss": 3.8359, + "step": 1980 + }, + { + "epoch": 2.91, + "grad_norm": 6.331207752227783, + "learning_rate": 0.0001610465525369228, + "loss": 3.7975, + "step": 1981 + }, + { + "epoch": 2.91, + "grad_norm": 7.4574055671691895, + "learning_rate": 0.00016100939628721375, + "loss": 3.7936, + "step": 1982 + }, + { + "epoch": 2.91, + "grad_norm": 12.698988914489746, + "learning_rate": 0.00016097222661584643, + "loss": 3.8415, + "step": 1983 + }, + { + "epoch": 2.92, + "grad_norm": 7.2185750007629395, + "learning_rate": 0.00016093504353099783, + "loss": 3.8274, + "step": 1984 + }, + { + "epoch": 2.92, + "grad_norm": 5.854779243469238, + "learning_rate": 0.00016089784704084804, + "loss": 3.8396, + "step": 1985 + }, + { + "epoch": 2.92, + "grad_norm": 8.324865341186523, + "learning_rate": 0.00016086063715358003, + "loss": 3.8204, + "step": 1986 + }, + { + "epoch": 2.92, + "grad_norm": 3.311805009841919, + "learning_rate": 0.0001608234138773797, + "loss": 3.7652, + "step": 1987 + }, + { + "epoch": 2.92, + "grad_norm": 11.10246467590332, + "learning_rate": 0.00016078617722043595, + "loss": 3.8391, + "step": 1988 + }, + { + "epoch": 2.92, + "grad_norm": 15.776145935058594, + "learning_rate": 0.00016074892719094056, + "loss": 3.799, + "step": 1989 + }, + { + "epoch": 2.92, + "grad_norm": 7.907688617706299, + "learning_rate": 0.00016071166379708833, + "loss": 3.7895, + "step": 1990 + }, + { + "epoch": 2.93, + "grad_norm": 5.6883649826049805, + "learning_rate": 0.00016067438704707695, + "loss": 3.794, + "step": 1991 + }, + { + "epoch": 2.93, + "grad_norm": 8.36189079284668, + "learning_rate": 0.00016063709694910701, + "loss": 3.8036, + "step": 1992 + }, + { + "epoch": 2.93, + "grad_norm": 4.989264488220215, + "learning_rate": 0.00016059979351138213, + "loss": 3.7555, + "step": 1993 + }, + { + "epoch": 2.93, + "grad_norm": 7.9405670166015625, + "learning_rate": 0.0001605624767421088, + "loss": 3.7995, + "step": 1994 + }, + { + "epoch": 2.93, + "grad_norm": 11.919158935546875, + "learning_rate": 0.00016052514664949646, + "loss": 3.8064, + "step": 1995 + }, + { + "epoch": 2.93, + "grad_norm": 15.537467002868652, + "learning_rate": 0.00016048780324175746, + "loss": 3.8284, + "step": 1996 + }, + { + "epoch": 2.93, + "grad_norm": 3.631410837173462, + "learning_rate": 0.00016045044652710712, + "loss": 3.838, + "step": 1997 + }, + { + "epoch": 2.94, + "grad_norm": 2.047104597091675, + "learning_rate": 0.0001604130765137637, + "loss": 3.7354, + "step": 1998 + }, + { + "epoch": 2.94, + "grad_norm": 8.390944480895996, + "learning_rate": 0.00016037569320994827, + "loss": 3.8143, + "step": 1999 + }, + { + "epoch": 2.94, + "grad_norm": 8.165762901306152, + "learning_rate": 0.00016033829662388498, + "loss": 3.7676, + "step": 2000 + }, + { + "epoch": 2.94, + "grad_norm": 2.4152541160583496, + "learning_rate": 0.0001603008867638008, + "loss": 3.7495, + "step": 2001 + }, + { + "epoch": 2.94, + "grad_norm": 5.799648761749268, + "learning_rate": 0.00016026346363792567, + "loss": 3.7996, + "step": 2002 + }, + { + "epoch": 2.94, + "grad_norm": 3.751105785369873, + "learning_rate": 0.00016022602725449238, + "loss": 3.7672, + "step": 2003 + }, + { + "epoch": 2.94, + "grad_norm": 7.3117780685424805, + "learning_rate": 0.0001601885776217367, + "loss": 3.7477, + "step": 2004 + }, + { + "epoch": 2.95, + "grad_norm": 7.469902992248535, + "learning_rate": 0.00016015111474789733, + "loss": 3.7555, + "step": 2005 + }, + { + "epoch": 2.95, + "grad_norm": 2.65592622756958, + "learning_rate": 0.00016011363864121582, + "loss": 3.7347, + "step": 2006 + }, + { + "epoch": 2.95, + "grad_norm": 3.7348546981811523, + "learning_rate": 0.0001600761493099367, + "loss": 3.7731, + "step": 2007 + }, + { + "epoch": 2.95, + "grad_norm": 2.6096608638763428, + "learning_rate": 0.0001600386467623073, + "loss": 3.7667, + "step": 2008 + }, + { + "epoch": 2.95, + "grad_norm": 1.8336734771728516, + "learning_rate": 0.000160001131006578, + "loss": 3.7475, + "step": 2009 + }, + { + "epoch": 2.95, + "grad_norm": 9.613265991210938, + "learning_rate": 0.000159963602051002, + "loss": 3.7442, + "step": 2010 + }, + { + "epoch": 2.96, + "grad_norm": 4.934391021728516, + "learning_rate": 0.00015992605990383538, + "loss": 3.7025, + "step": 2011 + }, + { + "epoch": 2.96, + "grad_norm": 2.538320779800415, + "learning_rate": 0.0001598885045733372, + "loss": 3.7681, + "step": 2012 + }, + { + "epoch": 2.96, + "grad_norm": 2.606849193572998, + "learning_rate": 0.00015985093606776937, + "loss": 3.7237, + "step": 2013 + }, + { + "epoch": 2.96, + "grad_norm": 3.6826469898223877, + "learning_rate": 0.0001598133543953967, + "loss": 3.7382, + "step": 2014 + }, + { + "epoch": 2.96, + "grad_norm": 1.016801118850708, + "learning_rate": 0.0001597757595644869, + "loss": 3.7059, + "step": 2015 + }, + { + "epoch": 2.96, + "grad_norm": 6.855797290802002, + "learning_rate": 0.00015973815158331064, + "loss": 3.7315, + "step": 2016 + }, + { + "epoch": 2.96, + "grad_norm": 4.787756443023682, + "learning_rate": 0.00015970053046014137, + "loss": 3.7273, + "step": 2017 + }, + { + "epoch": 2.97, + "grad_norm": 5.884622573852539, + "learning_rate": 0.00015966289620325546, + "loss": 3.7079, + "step": 2018 + }, + { + "epoch": 2.97, + "grad_norm": 7.18374490737915, + "learning_rate": 0.00015962524882093227, + "loss": 3.7161, + "step": 2019 + }, + { + "epoch": 2.97, + "grad_norm": 1.6866484880447388, + "learning_rate": 0.00015958758832145388, + "loss": 3.6481, + "step": 2020 + }, + { + "epoch": 2.97, + "grad_norm": 10.191508293151855, + "learning_rate": 0.00015954991471310543, + "loss": 3.7034, + "step": 2021 + }, + { + "epoch": 2.97, + "grad_norm": 14.30430793762207, + "learning_rate": 0.0001595122280041748, + "loss": 3.7654, + "step": 2022 + }, + { + "epoch": 2.97, + "grad_norm": 5.003686904907227, + "learning_rate": 0.00015947452820295292, + "loss": 3.7148, + "step": 2023 + }, + { + "epoch": 2.97, + "grad_norm": 6.288287162780762, + "learning_rate": 0.00015943681531773333, + "loss": 3.7528, + "step": 2024 + }, + { + "epoch": 2.98, + "grad_norm": 8.045740127563477, + "learning_rate": 0.00015939908935681273, + "loss": 3.6763, + "step": 2025 + }, + { + "epoch": 2.98, + "grad_norm": 3.2493703365325928, + "learning_rate": 0.0001593613503284906, + "loss": 3.6939, + "step": 2026 + }, + { + "epoch": 2.98, + "grad_norm": 10.84262752532959, + "learning_rate": 0.0001593235982410691, + "loss": 3.7516, + "step": 2027 + }, + { + "epoch": 2.98, + "grad_norm": 15.299811363220215, + "learning_rate": 0.00015928583310285366, + "loss": 3.7319, + "step": 2028 + }, + { + "epoch": 2.98, + "eval_loss": 3.5392327308654785, + "eval_runtime": 1153.2794, + "eval_samples_per_second": 14.864, + "eval_steps_per_second": 0.93, + "step": 2028 + }, + { + "epoch": 2.98, + "grad_norm": 8.231132507324219, + "learning_rate": 0.0001592480549221522, + "loss": 3.7177, + "step": 2029 + }, + { + "epoch": 2.98, + "grad_norm": 5.159106254577637, + "learning_rate": 0.00015921026370727573, + "loss": 3.7017, + "step": 2030 + }, + { + "epoch": 2.98, + "grad_norm": 22.314449310302734, + "learning_rate": 0.00015917245946653807, + "loss": 3.6842, + "step": 2031 + }, + { + "epoch": 2.99, + "grad_norm": 8.433747291564941, + "learning_rate": 0.00015913464220825588, + "loss": 3.6786, + "step": 2032 + }, + { + "epoch": 2.99, + "grad_norm": 3.3623647689819336, + "learning_rate": 0.00015909681194074867, + "loss": 3.678, + "step": 2033 + }, + { + "epoch": 2.99, + "grad_norm": 10.62174129486084, + "learning_rate": 0.00015905896867233894, + "loss": 3.7098, + "step": 2034 + }, + { + "epoch": 2.99, + "grad_norm": 15.299424171447754, + "learning_rate": 0.00015902111241135187, + "loss": 3.6926, + "step": 2035 + }, + { + "epoch": 2.99, + "grad_norm": 9.536619186401367, + "learning_rate": 0.00015898324316611563, + "loss": 3.673, + "step": 2036 + }, + { + "epoch": 2.99, + "grad_norm": 5.059504508972168, + "learning_rate": 0.00015894536094496117, + "loss": 3.6625, + "step": 2037 + }, + { + "epoch": 3.0, + "grad_norm": 7.704652786254883, + "learning_rate": 0.00015890746575622231, + "loss": 3.6632, + "step": 2038 + }, + { + "epoch": 3.0, + "grad_norm": 4.4129557609558105, + "learning_rate": 0.00015886955760823578, + "loss": 3.6315, + "step": 2039 + }, + { + "epoch": 3.0, + "grad_norm": 7.341033935546875, + "learning_rate": 0.0001588316365093411, + "loss": 3.6763, + "step": 2040 + }, + { + "epoch": 3.0, + "grad_norm": 9.556234359741211, + "learning_rate": 0.00015879370246788067, + "loss": 3.6774, + "step": 2041 + }, + { + "epoch": 3.0, + "grad_norm": 3.8044204711914062, + "learning_rate": 0.00015875575549219973, + "loss": 3.6249, + "step": 2042 + }, + { + "epoch": 3.0, + "grad_norm": 7.746762275695801, + "learning_rate": 0.0001587177955906463, + "loss": 3.6366, + "step": 2043 + }, + { + "epoch": 3.0, + "grad_norm": 9.851776123046875, + "learning_rate": 0.0001586798227715714, + "loss": 3.6885, + "step": 2044 + }, + { + "epoch": 3.01, + "grad_norm": 5.628739833831787, + "learning_rate": 0.00015864183704332872, + "loss": 3.6248, + "step": 2045 + }, + { + "epoch": 3.01, + "grad_norm": 6.932154178619385, + "learning_rate": 0.00015860383841427485, + "loss": 3.6499, + "step": 2046 + }, + { + "epoch": 3.01, + "grad_norm": 10.925968170166016, + "learning_rate": 0.00015856582689276934, + "loss": 3.6793, + "step": 2047 + }, + { + "epoch": 3.01, + "grad_norm": 5.439330577850342, + "learning_rate": 0.00015852780248717442, + "loss": 3.654, + "step": 2048 + }, + { + "epoch": 3.01, + "grad_norm": 6.066278457641602, + "learning_rate": 0.0001584897652058551, + "loss": 3.6738, + "step": 2049 + }, + { + "epoch": 3.01, + "grad_norm": 8.065271377563477, + "learning_rate": 0.00015845171505717947, + "loss": 3.659, + "step": 2050 + }, + { + "epoch": 3.01, + "grad_norm": 3.8582706451416016, + "learning_rate": 0.00015841365204951826, + "loss": 3.6458, + "step": 2051 + }, + { + "epoch": 3.02, + "grad_norm": 9.716731071472168, + "learning_rate": 0.00015837557619124507, + "loss": 3.6313, + "step": 2052 + }, + { + "epoch": 3.0, + "grad_norm": 14.822004318237305, + "learning_rate": 0.00015833748749073632, + "loss": 3.6512, + "step": 2053 + }, + { + "epoch": 3.0, + "grad_norm": 8.674092292785645, + "learning_rate": 0.00015829938595637128, + "loss": 3.6251, + "step": 2054 + }, + { + "epoch": 3.0, + "grad_norm": 4.23799467086792, + "learning_rate": 0.00015826127159653203, + "loss": 3.6496, + "step": 2055 + }, + { + "epoch": 3.01, + "grad_norm": 9.582878112792969, + "learning_rate": 0.00015822314441960348, + "loss": 3.5982, + "step": 2056 + }, + { + "epoch": 3.01, + "grad_norm": 5.006598949432373, + "learning_rate": 0.00015818500443397337, + "loss": 3.6347, + "step": 2057 + }, + { + "epoch": 3.01, + "grad_norm": 4.559854984283447, + "learning_rate": 0.0001581468516480322, + "loss": 3.6414, + "step": 2058 + }, + { + "epoch": 3.01, + "grad_norm": 5.798408508300781, + "learning_rate": 0.00015810868607017337, + "loss": 3.6176, + "step": 2059 + }, + { + "epoch": 3.01, + "grad_norm": 1.9246889352798462, + "learning_rate": 0.00015807050770879304, + "loss": 3.6134, + "step": 2060 + }, + { + "epoch": 3.01, + "grad_norm": 1.124200701713562, + "learning_rate": 0.00015803231657229017, + "loss": 3.6295, + "step": 2061 + }, + { + "epoch": 3.01, + "grad_norm": 7.120904445648193, + "learning_rate": 0.00015799411266906657, + "loss": 3.6101, + "step": 2062 + }, + { + "epoch": 3.02, + "grad_norm": 5.709414482116699, + "learning_rate": 0.00015795589600752682, + "loss": 3.6255, + "step": 2063 + }, + { + "epoch": 3.02, + "grad_norm": 3.2987709045410156, + "learning_rate": 0.0001579176665960784, + "loss": 3.6194, + "step": 2064 + }, + { + "epoch": 3.02, + "grad_norm": 4.208385944366455, + "learning_rate": 0.00015787942444313145, + "loss": 3.6357, + "step": 2065 + }, + { + "epoch": 3.02, + "grad_norm": 2.485320568084717, + "learning_rate": 0.00015784116955709897, + "loss": 3.6233, + "step": 2066 + }, + { + "epoch": 3.02, + "grad_norm": 0.9779897332191467, + "learning_rate": 0.0001578029019463969, + "loss": 3.6382, + "step": 2067 + }, + { + "epoch": 3.02, + "grad_norm": 5.236642837524414, + "learning_rate": 0.0001577646216194437, + "loss": 3.5985, + "step": 2068 + }, + { + "epoch": 3.03, + "grad_norm": 4.276678562164307, + "learning_rate": 0.00015772632858466093, + "loss": 3.614, + "step": 2069 + }, + { + "epoch": 3.03, + "grad_norm": 4.390791893005371, + "learning_rate": 0.00015768802285047272, + "loss": 3.6166, + "step": 2070 + }, + { + "epoch": 3.03, + "grad_norm": 4.886763572692871, + "learning_rate": 0.00015764970442530609, + "loss": 3.5902, + "step": 2071 + }, + { + "epoch": 3.03, + "grad_norm": 3.569462776184082, + "learning_rate": 0.00015761137331759084, + "loss": 3.5779, + "step": 2072 + }, + { + "epoch": 3.03, + "grad_norm": 2.935269832611084, + "learning_rate": 0.00015757302953575954, + "loss": 3.6369, + "step": 2073 + }, + { + "epoch": 3.03, + "grad_norm": 4.911141872406006, + "learning_rate": 0.0001575346730882476, + "loss": 3.6034, + "step": 2074 + }, + { + "epoch": 3.03, + "grad_norm": 4.09962797164917, + "learning_rate": 0.00015749630398349316, + "loss": 3.6431, + "step": 2075 + }, + { + "epoch": 3.04, + "grad_norm": 4.247868537902832, + "learning_rate": 0.00015745792222993715, + "loss": 3.6258, + "step": 2076 + }, + { + "epoch": 3.04, + "grad_norm": 3.7722268104553223, + "learning_rate": 0.00015741952783602332, + "loss": 3.597, + "step": 2077 + }, + { + "epoch": 3.04, + "grad_norm": 5.005308151245117, + "learning_rate": 0.00015738112081019817, + "loss": 3.6146, + "step": 2078 + }, + { + "epoch": 3.04, + "grad_norm": 4.696826934814453, + "learning_rate": 0.000157342701160911, + "loss": 3.6477, + "step": 2079 + }, + { + "epoch": 3.04, + "grad_norm": 3.856837034225464, + "learning_rate": 0.00015730426889661386, + "loss": 3.6594, + "step": 2080 + }, + { + "epoch": 3.04, + "grad_norm": 3.4349594116210938, + "learning_rate": 0.0001572658240257616, + "loss": 3.5798, + "step": 2081 + }, + { + "epoch": 3.04, + "grad_norm": 4.7815842628479, + "learning_rate": 0.00015722736655681186, + "loss": 3.5797, + "step": 2082 + }, + { + "epoch": 3.05, + "grad_norm": 3.9778714179992676, + "learning_rate": 0.00015718889649822494, + "loss": 3.6173, + "step": 2083 + }, + { + "epoch": 3.05, + "grad_norm": 3.863435745239258, + "learning_rate": 0.0001571504138584641, + "loss": 3.5669, + "step": 2084 + }, + { + "epoch": 3.05, + "grad_norm": 3.9693548679351807, + "learning_rate": 0.0001571119186459952, + "loss": 3.6072, + "step": 2085 + }, + { + "epoch": 3.05, + "grad_norm": 4.673881530761719, + "learning_rate": 0.00015707341086928698, + "loss": 3.5794, + "step": 2086 + }, + { + "epoch": 3.05, + "grad_norm": 3.680610179901123, + "learning_rate": 0.0001570348905368108, + "loss": 3.5927, + "step": 2087 + }, + { + "epoch": 3.05, + "grad_norm": 4.484524726867676, + "learning_rate": 0.00015699635765704096, + "loss": 3.5588, + "step": 2088 + }, + { + "epoch": 3.05, + "grad_norm": 4.283873558044434, + "learning_rate": 0.00015695781223845441, + "loss": 3.6385, + "step": 2089 + }, + { + "epoch": 3.06, + "grad_norm": 3.3641796112060547, + "learning_rate": 0.00015691925428953089, + "loss": 3.6163, + "step": 2090 + }, + { + "epoch": 3.06, + "grad_norm": 2.774048089981079, + "learning_rate": 0.00015688068381875287, + "loss": 3.6696, + "step": 2091 + }, + { + "epoch": 3.06, + "grad_norm": 5.591458320617676, + "learning_rate": 0.0001568421008346056, + "loss": 3.54, + "step": 2092 + }, + { + "epoch": 3.06, + "grad_norm": 5.354968547821045, + "learning_rate": 0.0001568035053455771, + "loss": 3.5762, + "step": 2093 + }, + { + "epoch": 3.06, + "grad_norm": 2.219583749771118, + "learning_rate": 0.0001567648973601581, + "loss": 3.5527, + "step": 2094 + }, + { + "epoch": 3.06, + "grad_norm": 1.655458688735962, + "learning_rate": 0.0001567262768868421, + "loss": 3.5702, + "step": 2095 + }, + { + "epoch": 3.07, + "grad_norm": 5.801334381103516, + "learning_rate": 0.00015668764393412537, + "loss": 3.5317, + "step": 2096 + }, + { + "epoch": 3.07, + "grad_norm": 5.561006546020508, + "learning_rate": 0.00015664899851050684, + "loss": 3.5875, + "step": 2097 + }, + { + "epoch": 3.07, + "grad_norm": 2.8261959552764893, + "learning_rate": 0.0001566103406244883, + "loss": 3.5284, + "step": 2098 + }, + { + "epoch": 3.07, + "grad_norm": 2.3607285022735596, + "learning_rate": 0.00015657167028457423, + "loss": 3.5596, + "step": 2099 + }, + { + "epoch": 3.07, + "grad_norm": 6.529599666595459, + "learning_rate": 0.00015653298749927178, + "loss": 3.5603, + "step": 2100 + }, + { + "epoch": 3.07, + "grad_norm": 5.787454605102539, + "learning_rate": 0.00015649429227709098, + "loss": 3.5419, + "step": 2101 + }, + { + "epoch": 3.07, + "grad_norm": 1.2445347309112549, + "learning_rate": 0.00015645558462654447, + "loss": 3.5599, + "step": 2102 + }, + { + "epoch": 3.08, + "grad_norm": 1.686257243156433, + "learning_rate": 0.0001564168645561477, + "loss": 3.5051, + "step": 2103 + }, + { + "epoch": 3.08, + "grad_norm": 5.125936985015869, + "learning_rate": 0.00015637813207441882, + "loss": 3.4685, + "step": 2104 + }, + { + "epoch": 3.08, + "grad_norm": 4.658946514129639, + "learning_rate": 0.0001563393871898787, + "loss": 3.5291, + "step": 2105 + }, + { + "epoch": 3.08, + "grad_norm": 4.34171199798584, + "learning_rate": 0.00015630062991105098, + "loss": 3.5044, + "step": 2106 + }, + { + "epoch": 3.08, + "grad_norm": 5.2388458251953125, + "learning_rate": 0.00015626186024646197, + "loss": 3.4543, + "step": 2107 + }, + { + "epoch": 3.08, + "grad_norm": 3.39254093170166, + "learning_rate": 0.00015622307820464077, + "loss": 3.4678, + "step": 2108 + }, + { + "epoch": 3.08, + "grad_norm": 2.4865806102752686, + "learning_rate": 0.00015618428379411917, + "loss": 3.4326, + "step": 2109 + }, + { + "epoch": 3.09, + "grad_norm": 7.326157093048096, + "learning_rate": 0.00015614547702343166, + "loss": 3.4709, + "step": 2110 + }, + { + "epoch": 3.09, + "grad_norm": 7.048543930053711, + "learning_rate": 0.00015610665790111546, + "loss": 3.4498, + "step": 2111 + }, + { + "epoch": 3.09, + "grad_norm": 10.954419136047363, + "learning_rate": 0.00015606782643571055, + "loss": 3.3944, + "step": 2112 + }, + { + "epoch": 3.09, + "grad_norm": 11.594827651977539, + "learning_rate": 0.00015602898263575952, + "loss": 3.4224, + "step": 2113 + }, + { + "epoch": 3.09, + "grad_norm": 6.556940078735352, + "learning_rate": 0.0001559901265098079, + "loss": 3.4079, + "step": 2114 + }, + { + "epoch": 3.09, + "grad_norm": 3.0502734184265137, + "learning_rate": 0.00015595125806640357, + "loss": 3.409, + "step": 2115 + }, + { + "epoch": 3.09, + "grad_norm": 6.4218220710754395, + "learning_rate": 0.0001559123773140975, + "loss": 3.4382, + "step": 2116 + }, + { + "epoch": 3.1, + "grad_norm": 6.8841753005981445, + "learning_rate": 0.00015587348426144312, + "loss": 3.463, + "step": 2117 + }, + { + "epoch": 3.1, + "grad_norm": 35.60420227050781, + "learning_rate": 0.00015583457891699666, + "loss": 3.3878, + "step": 2118 + }, + { + "epoch": 3.1, + "grad_norm": 2.3054988384246826, + "learning_rate": 0.000155795661289317, + "loss": 3.382, + "step": 2119 + }, + { + "epoch": 3.1, + "grad_norm": 6.13869571685791, + "learning_rate": 0.00015575673138696578, + "loss": 3.4213, + "step": 2120 + }, + { + "epoch": 3.1, + "grad_norm": 5.998043537139893, + "learning_rate": 0.00015571778921850732, + "loss": 3.382, + "step": 2121 + }, + { + "epoch": 3.1, + "grad_norm": 1.6560511589050293, + "learning_rate": 0.00015567883479250863, + "loss": 3.4691, + "step": 2122 + }, + { + "epoch": 3.11, + "grad_norm": 2.120764970779419, + "learning_rate": 0.00015563986811753945, + "loss": 3.3711, + "step": 2123 + }, + { + "epoch": 3.11, + "grad_norm": 1.6065012216567993, + "learning_rate": 0.00015560088920217217, + "loss": 3.4219, + "step": 2124 + }, + { + "epoch": 3.11, + "grad_norm": 6.458485126495361, + "learning_rate": 0.00015556189805498188, + "loss": 3.4184, + "step": 2125 + }, + { + "epoch": 3.11, + "grad_norm": 5.5630202293396, + "learning_rate": 0.0001555228946845464, + "loss": 3.4185, + "step": 2126 + }, + { + "epoch": 3.11, + "grad_norm": 4.821037769317627, + "learning_rate": 0.00015548387909944614, + "loss": 3.4003, + "step": 2127 + }, + { + "epoch": 3.11, + "grad_norm": 13.251823425292969, + "learning_rate": 0.00015544485130826433, + "loss": 3.4081, + "step": 2128 + }, + { + "epoch": 3.11, + "grad_norm": 5.775404930114746, + "learning_rate": 0.00015540581131958682, + "loss": 3.3918, + "step": 2129 + }, + { + "epoch": 3.12, + "grad_norm": 6.066825866699219, + "learning_rate": 0.00015536675914200213, + "loss": 3.3863, + "step": 2130 + }, + { + "epoch": 3.12, + "grad_norm": 11.889984130859375, + "learning_rate": 0.0001553276947841015, + "loss": 3.4055, + "step": 2131 + }, + { + "epoch": 3.12, + "grad_norm": 3.267805576324463, + "learning_rate": 0.00015528861825447876, + "loss": 3.4031, + "step": 2132 + }, + { + "epoch": 3.12, + "grad_norm": 11.77675724029541, + "learning_rate": 0.00015524952956173056, + "loss": 3.4159, + "step": 2133 + }, + { + "epoch": 3.12, + "grad_norm": 16.67833137512207, + "learning_rate": 0.0001552104287144561, + "loss": 3.4706, + "step": 2134 + }, + { + "epoch": 3.12, + "grad_norm": 9.038000106811523, + "learning_rate": 0.00015517131572125736, + "loss": 3.4391, + "step": 2135 + }, + { + "epoch": 3.12, + "grad_norm": 7.349591255187988, + "learning_rate": 0.0001551321905907389, + "loss": 3.4181, + "step": 2136 + }, + { + "epoch": 3.13, + "grad_norm": 11.440853118896484, + "learning_rate": 0.00015509305333150797, + "loss": 3.4324, + "step": 2137 + }, + { + "epoch": 3.13, + "grad_norm": 9.080293655395508, + "learning_rate": 0.00015505390395217453, + "loss": 3.4841, + "step": 2138 + }, + { + "epoch": 3.13, + "grad_norm": 2.2852790355682373, + "learning_rate": 0.0001550147424613512, + "loss": 3.35, + "step": 2139 + }, + { + "epoch": 3.13, + "grad_norm": 17.961963653564453, + "learning_rate": 0.00015497556886765316, + "loss": 3.4265, + "step": 2140 + }, + { + "epoch": 3.13, + "grad_norm": 26.195865631103516, + "learning_rate": 0.00015493638317969844, + "loss": 3.6051, + "step": 2141 + }, + { + "epoch": 3.13, + "grad_norm": 19.95352554321289, + "learning_rate": 0.0001548971854061076, + "loss": 3.6381, + "step": 2142 + }, + { + "epoch": 3.13, + "grad_norm": 5.972691535949707, + "learning_rate": 0.00015485797555550386, + "loss": 3.5607, + "step": 2143 + }, + { + "epoch": 3.14, + "grad_norm": 13.078893661499023, + "learning_rate": 0.0001548187536365131, + "loss": 3.4778, + "step": 2144 + }, + { + "epoch": 3.14, + "grad_norm": 15.685088157653809, + "learning_rate": 0.000154779519657764, + "loss": 3.5011, + "step": 2145 + }, + { + "epoch": 3.14, + "grad_norm": 14.251028060913086, + "learning_rate": 0.00015474027362788763, + "loss": 3.6155, + "step": 2146 + }, + { + "epoch": 3.14, + "grad_norm": 10.396077156066895, + "learning_rate": 0.00015470101555551796, + "loss": 3.5646, + "step": 2147 + }, + { + "epoch": 3.14, + "grad_norm": 5.34552526473999, + "learning_rate": 0.0001546617454492914, + "loss": 3.4643, + "step": 2148 + }, + { + "epoch": 3.14, + "grad_norm": 18.76544189453125, + "learning_rate": 0.0001546224633178472, + "loss": 3.4686, + "step": 2149 + }, + { + "epoch": 3.15, + "grad_norm": 32.796051025390625, + "learning_rate": 0.00015458316916982715, + "loss": 3.6171, + "step": 2150 + }, + { + "epoch": 3.15, + "grad_norm": 31.727258682250977, + "learning_rate": 0.00015454386301387565, + "loss": 3.751, + "step": 2151 + }, + { + "epoch": 3.15, + "grad_norm": 18.45448112487793, + "learning_rate": 0.00015450454485863984, + "loss": 3.6405, + "step": 2152 + }, + { + "epoch": 3.15, + "grad_norm": 8.118375778198242, + "learning_rate": 0.00015446521471276946, + "loss": 3.6001, + "step": 2153 + }, + { + "epoch": 3.15, + "grad_norm": 16.14570426940918, + "learning_rate": 0.0001544258725849168, + "loss": 3.559, + "step": 2154 + }, + { + "epoch": 3.15, + "grad_norm": 16.983943939208984, + "learning_rate": 0.00015438651848373696, + "loss": 3.5108, + "step": 2155 + }, + { + "epoch": 3.15, + "grad_norm": 13.206223487854004, + "learning_rate": 0.00015434715241788752, + "loss": 3.4883, + "step": 2156 + }, + { + "epoch": 3.16, + "grad_norm": 8.664280891418457, + "learning_rate": 0.00015430777439602876, + "loss": 3.4346, + "step": 2157 + }, + { + "epoch": 3.16, + "grad_norm": 5.633321762084961, + "learning_rate": 0.00015426838442682362, + "loss": 3.4285, + "step": 2158 + }, + { + "epoch": 3.16, + "grad_norm": 14.557170867919922, + "learning_rate": 0.00015422898251893756, + "loss": 3.4706, + "step": 2159 + }, + { + "epoch": 3.16, + "grad_norm": 19.554941177368164, + "learning_rate": 0.00015418956868103878, + "loss": 3.4929, + "step": 2160 + }, + { + "epoch": 3.16, + "grad_norm": 15.579254150390625, + "learning_rate": 0.00015415014292179805, + "loss": 3.4144, + "step": 2161 + }, + { + "epoch": 3.16, + "grad_norm": 5.079093933105469, + "learning_rate": 0.0001541107052498888, + "loss": 3.3886, + "step": 2162 + }, + { + "epoch": 3.16, + "grad_norm": 10.48103141784668, + "learning_rate": 0.00015407125567398698, + "loss": 3.4392, + "step": 2163 + }, + { + "epoch": 3.17, + "grad_norm": 14.062845230102539, + "learning_rate": 0.00015403179420277131, + "loss": 3.4527, + "step": 2164 + }, + { + "epoch": 3.17, + "grad_norm": 13.237971305847168, + "learning_rate": 0.00015399232084492304, + "loss": 3.4518, + "step": 2165 + }, + { + "epoch": 3.17, + "grad_norm": 7.261989593505859, + "learning_rate": 0.000153952835609126, + "loss": 3.4716, + "step": 2166 + }, + { + "epoch": 3.17, + "grad_norm": 5.766734600067139, + "learning_rate": 0.0001539133385040667, + "loss": 3.3591, + "step": 2167 + }, + { + "epoch": 3.17, + "grad_norm": 10.160755157470703, + "learning_rate": 0.0001538738295384343, + "loss": 3.3638, + "step": 2168 + }, + { + "epoch": 3.17, + "grad_norm": 7.740986347198486, + "learning_rate": 0.00015383430872092037, + "loss": 3.4132, + "step": 2169 + }, + { + "epoch": 3.18, + "grad_norm": 2.3649628162384033, + "learning_rate": 0.00015379477606021933, + "loss": 3.4459, + "step": 2170 + }, + { + "epoch": 3.18, + "grad_norm": 4.513500213623047, + "learning_rate": 0.00015375523156502805, + "loss": 3.4318, + "step": 2171 + }, + { + "epoch": 3.18, + "grad_norm": 3.214402198791504, + "learning_rate": 0.00015371567524404613, + "loss": 3.3651, + "step": 2172 + }, + { + "epoch": 3.18, + "grad_norm": 3.8382275104522705, + "learning_rate": 0.0001536761071059756, + "loss": 3.4077, + "step": 2173 + }, + { + "epoch": 3.18, + "grad_norm": 1.4553401470184326, + "learning_rate": 0.0001536365271595212, + "loss": 3.3301, + "step": 2174 + }, + { + "epoch": 3.18, + "grad_norm": 4.466412544250488, + "learning_rate": 0.00015359693541339037, + "loss": 3.3988, + "step": 2175 + }, + { + "epoch": 3.18, + "grad_norm": 3.6478259563446045, + "learning_rate": 0.00015355733187629288, + "loss": 3.3779, + "step": 2176 + }, + { + "epoch": 3.19, + "grad_norm": 5.013825416564941, + "learning_rate": 0.0001535177165569413, + "loss": 3.413, + "step": 2177 + }, + { + "epoch": 3.19, + "grad_norm": 4.778167247772217, + "learning_rate": 0.00015347808946405075, + "loss": 3.3425, + "step": 2178 + }, + { + "epoch": 3.19, + "grad_norm": 5.024134159088135, + "learning_rate": 0.00015343845060633893, + "loss": 3.3676, + "step": 2179 + }, + { + "epoch": 3.19, + "grad_norm": 5.648217678070068, + "learning_rate": 0.00015339879999252606, + "loss": 3.3165, + "step": 2180 + }, + { + "epoch": 3.19, + "grad_norm": 2.347975015640259, + "learning_rate": 0.00015335913763133511, + "loss": 3.3588, + "step": 2181 + }, + { + "epoch": 3.19, + "grad_norm": 5.947681427001953, + "learning_rate": 0.00015331946353149144, + "loss": 3.3811, + "step": 2182 + }, + { + "epoch": 3.19, + "grad_norm": 5.794849395751953, + "learning_rate": 0.00015327977770172315, + "loss": 3.3751, + "step": 2183 + }, + { + "epoch": 3.2, + "grad_norm": 2.177245855331421, + "learning_rate": 0.0001532400801507608, + "loss": 3.3472, + "step": 2184 + }, + { + "epoch": 3.2, + "grad_norm": 4.221487045288086, + "learning_rate": 0.00015320037088733764, + "loss": 3.3766, + "step": 2185 + }, + { + "epoch": 3.2, + "grad_norm": 3.649068832397461, + "learning_rate": 0.00015316064992018945, + "loss": 3.3633, + "step": 2186 + }, + { + "epoch": 3.2, + "grad_norm": 3.780611038208008, + "learning_rate": 0.00015312091725805448, + "loss": 3.3371, + "step": 2187 + }, + { + "epoch": 3.2, + "grad_norm": 1.3155494928359985, + "learning_rate": 0.00015308117290967373, + "loss": 3.3083, + "step": 2188 + }, + { + "epoch": 3.2, + "grad_norm": 1.961712121963501, + "learning_rate": 0.0001530414168837907, + "loss": 3.3465, + "step": 2189 + }, + { + "epoch": 3.2, + "grad_norm": 2.569746494293213, + "learning_rate": 0.00015300164918915138, + "loss": 3.3853, + "step": 2190 + }, + { + "epoch": 3.21, + "grad_norm": 1.9574949741363525, + "learning_rate": 0.00015296186983450446, + "loss": 3.3444, + "step": 2191 + }, + { + "epoch": 3.21, + "grad_norm": 2.3811464309692383, + "learning_rate": 0.00015292207882860114, + "loss": 3.316, + "step": 2192 + }, + { + "epoch": 3.21, + "grad_norm": 1.6529579162597656, + "learning_rate": 0.00015288227618019513, + "loss": 3.3269, + "step": 2193 + }, + { + "epoch": 3.21, + "grad_norm": 1.4792758226394653, + "learning_rate": 0.00015284246189804272, + "loss": 3.3406, + "step": 2194 + }, + { + "epoch": 3.21, + "grad_norm": 1.3875869512557983, + "learning_rate": 0.0001528026359909029, + "loss": 3.3492, + "step": 2195 + }, + { + "epoch": 3.21, + "grad_norm": 3.4284157752990723, + "learning_rate": 0.00015276279846753696, + "loss": 3.3176, + "step": 2196 + }, + { + "epoch": 3.22, + "grad_norm": 1.68557608127594, + "learning_rate": 0.000152722949336709, + "loss": 3.3142, + "step": 2197 + }, + { + "epoch": 3.22, + "eval_loss": 3.1609973907470703, + "eval_runtime": 1163.3297, + "eval_samples_per_second": 14.735, + "eval_steps_per_second": 0.921, + "step": 2197 + }, + { + "epoch": 3.22, + "grad_norm": 4.416086673736572, + "learning_rate": 0.0001526830886071855, + "loss": 3.3143, + "step": 2198 + }, + { + "epoch": 3.22, + "grad_norm": 2.9269638061523438, + "learning_rate": 0.0001526432162877356, + "loss": 3.3417, + "step": 2199 + }, + { + "epoch": 3.22, + "grad_norm": 3.496257781982422, + "learning_rate": 0.00015260333238713088, + "loss": 3.3356, + "step": 2200 + }, + { + "epoch": 3.22, + "grad_norm": 3.9582598209381104, + "learning_rate": 0.00015256343691414558, + "loss": 3.3688, + "step": 2201 + }, + { + "epoch": 3.22, + "grad_norm": 3.0187695026397705, + "learning_rate": 0.0001525235298775564, + "loss": 3.3381, + "step": 2202 + }, + { + "epoch": 3.22, + "grad_norm": 1.4246793985366821, + "learning_rate": 0.0001524836112861427, + "loss": 3.3228, + "step": 2203 + }, + { + "epoch": 3.23, + "grad_norm": 2.6875085830688477, + "learning_rate": 0.00015244368114868625, + "loss": 3.2918, + "step": 2204 + }, + { + "epoch": 3.23, + "grad_norm": 2.1740024089813232, + "learning_rate": 0.00015240373947397133, + "loss": 3.2933, + "step": 2205 + }, + { + "epoch": 3.23, + "grad_norm": 5.561516284942627, + "learning_rate": 0.000152363786270785, + "loss": 3.3344, + "step": 2206 + }, + { + "epoch": 3.23, + "grad_norm": 2.453502655029297, + "learning_rate": 0.00015232382154791655, + "loss": 3.3204, + "step": 2207 + }, + { + "epoch": 3.23, + "grad_norm": 3.4560389518737793, + "learning_rate": 0.0001522838453141581, + "loss": 3.3054, + "step": 2208 + }, + { + "epoch": 3.23, + "grad_norm": 1.4101018905639648, + "learning_rate": 0.000152243857578304, + "loss": 3.3298, + "step": 2209 + }, + { + "epoch": 3.23, + "grad_norm": 1.3143606185913086, + "learning_rate": 0.0001522038583491514, + "loss": 3.3294, + "step": 2210 + }, + { + "epoch": 3.24, + "grad_norm": 2.1382877826690674, + "learning_rate": 0.00015216384763549977, + "loss": 3.2964, + "step": 2211 + }, + { + "epoch": 3.24, + "grad_norm": 3.5953447818756104, + "learning_rate": 0.00015212382544615126, + "loss": 3.3138, + "step": 2212 + }, + { + "epoch": 3.24, + "grad_norm": 2.013397216796875, + "learning_rate": 0.0001520837917899105, + "loss": 3.3318, + "step": 2213 + }, + { + "epoch": 3.24, + "grad_norm": 5.66317892074585, + "learning_rate": 0.0001520437466755846, + "loss": 3.3226, + "step": 2214 + }, + { + "epoch": 3.24, + "grad_norm": 5.973227024078369, + "learning_rate": 0.00015200369011198319, + "loss": 3.2917, + "step": 2215 + }, + { + "epoch": 3.24, + "grad_norm": 1.0823651552200317, + "learning_rate": 0.00015196362210791845, + "loss": 3.2837, + "step": 2216 + }, + { + "epoch": 3.24, + "grad_norm": 2.2828619480133057, + "learning_rate": 0.00015192354267220512, + "loss": 3.2994, + "step": 2217 + }, + { + "epoch": 3.25, + "grad_norm": 3.3102993965148926, + "learning_rate": 0.00015188345181366036, + "loss": 3.3674, + "step": 2218 + }, + { + "epoch": 3.25, + "grad_norm": 5.152593612670898, + "learning_rate": 0.00015184334954110394, + "loss": 3.3411, + "step": 2219 + }, + { + "epoch": 3.25, + "grad_norm": 6.140212535858154, + "learning_rate": 0.00015180323586335806, + "loss": 3.3188, + "step": 2220 + }, + { + "epoch": 3.25, + "grad_norm": 2.0337228775024414, + "learning_rate": 0.0001517631107892474, + "loss": 3.2678, + "step": 2221 + }, + { + "epoch": 3.25, + "grad_norm": 4.024622917175293, + "learning_rate": 0.0001517229743275993, + "loss": 3.2708, + "step": 2222 + }, + { + "epoch": 3.25, + "grad_norm": 2.378216505050659, + "learning_rate": 0.00015168282648724351, + "loss": 3.3462, + "step": 2223 + }, + { + "epoch": 3.26, + "grad_norm": 4.922878742218018, + "learning_rate": 0.0001516426672770122, + "loss": 3.291, + "step": 2224 + }, + { + "epoch": 3.26, + "grad_norm": 4.67563533782959, + "learning_rate": 0.00015160249670574026, + "loss": 3.2711, + "step": 2225 + }, + { + "epoch": 3.26, + "grad_norm": 4.26483154296875, + "learning_rate": 0.0001515623147822648, + "loss": 3.2862, + "step": 2226 + }, + { + "epoch": 3.26, + "grad_norm": 6.510219097137451, + "learning_rate": 0.00015152212151542567, + "loss": 3.3276, + "step": 2227 + }, + { + "epoch": 3.26, + "grad_norm": 4.469378471374512, + "learning_rate": 0.0001514819169140651, + "loss": 3.3276, + "step": 2228 + }, + { + "epoch": 3.26, + "grad_norm": 4.4898681640625, + "learning_rate": 0.0001514417009870278, + "loss": 3.2907, + "step": 2229 + }, + { + "epoch": 3.26, + "grad_norm": 5.1355485916137695, + "learning_rate": 0.00015140147374316106, + "loss": 3.3146, + "step": 2230 + }, + { + "epoch": 3.27, + "grad_norm": 3.954402208328247, + "learning_rate": 0.00015136123519131456, + "loss": 3.2888, + "step": 2231 + }, + { + "epoch": 3.27, + "grad_norm": 8.745506286621094, + "learning_rate": 0.00015132098534034052, + "loss": 3.3684, + "step": 2232 + }, + { + "epoch": 3.27, + "grad_norm": 7.295302391052246, + "learning_rate": 0.00015128072419909368, + "loss": 3.3378, + "step": 2233 + }, + { + "epoch": 3.27, + "grad_norm": 1.574733853340149, + "learning_rate": 0.00015124045177643112, + "loss": 3.3003, + "step": 2234 + }, + { + "epoch": 3.27, + "grad_norm": 6.663206100463867, + "learning_rate": 0.0001512001680812126, + "loss": 3.3466, + "step": 2235 + }, + { + "epoch": 3.27, + "grad_norm": 7.278547286987305, + "learning_rate": 0.00015115987312230026, + "loss": 3.2945, + "step": 2236 + }, + { + "epoch": 3.27, + "grad_norm": 3.385538101196289, + "learning_rate": 0.00015111956690855863, + "loss": 3.2958, + "step": 2237 + }, + { + "epoch": 3.28, + "grad_norm": 7.496275424957275, + "learning_rate": 0.00015107924944885486, + "loss": 3.3214, + "step": 2238 + }, + { + "epoch": 3.28, + "grad_norm": 25.01158332824707, + "learning_rate": 0.00015103892075205862, + "loss": 3.4204, + "step": 2239 + }, + { + "epoch": 3.28, + "grad_norm": 9.22580337524414, + "learning_rate": 0.00015099858082704182, + "loss": 3.2895, + "step": 2240 + }, + { + "epoch": 3.28, + "grad_norm": 1.5853315591812134, + "learning_rate": 0.00015095822968267896, + "loss": 3.3388, + "step": 2241 + }, + { + "epoch": 3.28, + "grad_norm": 8.119154930114746, + "learning_rate": 0.00015091786732784716, + "loss": 3.35, + "step": 2242 + }, + { + "epoch": 3.28, + "grad_norm": 10.604775428771973, + "learning_rate": 0.00015087749377142574, + "loss": 3.3851, + "step": 2243 + }, + { + "epoch": 3.28, + "grad_norm": 7.424951076507568, + "learning_rate": 0.00015083710902229664, + "loss": 3.3619, + "step": 2244 + }, + { + "epoch": 3.29, + "grad_norm": 5.7252373695373535, + "learning_rate": 0.0001507967130893443, + "loss": 3.342, + "step": 2245 + }, + { + "epoch": 3.29, + "grad_norm": 8.934366226196289, + "learning_rate": 0.00015075630598145547, + "loss": 3.3553, + "step": 2246 + }, + { + "epoch": 3.29, + "grad_norm": 6.990628719329834, + "learning_rate": 0.00015071588770751947, + "loss": 3.3318, + "step": 2247 + }, + { + "epoch": 3.29, + "grad_norm": 19.209877014160156, + "learning_rate": 0.00015067545827642808, + "loss": 3.3181, + "step": 2248 + }, + { + "epoch": 3.29, + "grad_norm": 6.441047191619873, + "learning_rate": 0.00015063501769707548, + "loss": 3.3515, + "step": 2249 + }, + { + "epoch": 3.29, + "grad_norm": 5.99086332321167, + "learning_rate": 0.00015059456597835835, + "loss": 3.334, + "step": 2250 + }, + { + "epoch": 3.3, + "grad_norm": 2.6446921825408936, + "learning_rate": 0.00015055410312917573, + "loss": 3.332, + "step": 2251 + }, + { + "epoch": 3.3, + "grad_norm": 266.819091796875, + "learning_rate": 0.00015051362915842923, + "loss": 3.3605, + "step": 2252 + }, + { + "epoch": 3.3, + "grad_norm": 16.00495719909668, + "learning_rate": 0.00015047314407502283, + "loss": 3.3521, + "step": 2253 + }, + { + "epoch": 3.3, + "grad_norm": 8.721566200256348, + "learning_rate": 0.00015043264788786305, + "loss": 3.3846, + "step": 2254 + }, + { + "epoch": 3.3, + "grad_norm": 3.7059578895568848, + "learning_rate": 0.00015039214060585864, + "loss": 3.39, + "step": 2255 + }, + { + "epoch": 3.3, + "grad_norm": 5.986879825592041, + "learning_rate": 0.00015035162223792106, + "loss": 3.3791, + "step": 2256 + }, + { + "epoch": 3.3, + "grad_norm": 4.337857723236084, + "learning_rate": 0.00015031109279296402, + "loss": 3.3309, + "step": 2257 + }, + { + "epoch": 3.31, + "grad_norm": 4.526264190673828, + "learning_rate": 0.00015027055227990373, + "loss": 3.3768, + "step": 2258 + }, + { + "epoch": 3.31, + "grad_norm": 7.14890718460083, + "learning_rate": 0.00015023000070765884, + "loss": 3.3677, + "step": 2259 + }, + { + "epoch": 3.31, + "grad_norm": 3.9721760749816895, + "learning_rate": 0.0001501894380851504, + "loss": 3.3136, + "step": 2260 + }, + { + "epoch": 3.31, + "grad_norm": 5.246433734893799, + "learning_rate": 0.00015014886442130197, + "loss": 3.3263, + "step": 2261 + }, + { + "epoch": 3.31, + "grad_norm": 7.421927452087402, + "learning_rate": 0.00015010827972503944, + "loss": 3.3386, + "step": 2262 + }, + { + "epoch": 3.31, + "grad_norm": 3.354271173477173, + "learning_rate": 0.00015006768400529117, + "loss": 3.3509, + "step": 2263 + }, + { + "epoch": 3.31, + "grad_norm": 5.732051849365234, + "learning_rate": 0.00015002707727098798, + "loss": 3.3885, + "step": 2264 + }, + { + "epoch": 3.32, + "grad_norm": 6.248003005981445, + "learning_rate": 0.00014998645953106303, + "loss": 3.3291, + "step": 2265 + }, + { + "epoch": 3.32, + "grad_norm": 2.3988747596740723, + "learning_rate": 0.000149945830794452, + "loss": 3.2951, + "step": 2266 + }, + { + "epoch": 3.32, + "grad_norm": 4.345824241638184, + "learning_rate": 0.00014990519107009292, + "loss": 3.3256, + "step": 2267 + }, + { + "epoch": 3.32, + "grad_norm": 4.299257278442383, + "learning_rate": 0.0001498645403669262, + "loss": 3.3154, + "step": 2268 + }, + { + "epoch": 3.32, + "grad_norm": 4.104548931121826, + "learning_rate": 0.00014982387869389485, + "loss": 3.3248, + "step": 2269 + }, + { + "epoch": 3.32, + "grad_norm": 3.4004967212677, + "learning_rate": 0.00014978320605994405, + "loss": 3.3147, + "step": 2270 + }, + { + "epoch": 3.33, + "grad_norm": 3.225266456604004, + "learning_rate": 0.00014974252247402156, + "loss": 3.2917, + "step": 2271 + }, + { + "epoch": 3.33, + "grad_norm": 3.6412038803100586, + "learning_rate": 0.00014970182794507748, + "loss": 3.3557, + "step": 2272 + }, + { + "epoch": 3.33, + "grad_norm": 3.6593449115753174, + "learning_rate": 0.00014966112248206436, + "loss": 3.3048, + "step": 2273 + }, + { + "epoch": 3.33, + "grad_norm": 3.3542096614837646, + "learning_rate": 0.0001496204060939371, + "loss": 3.2807, + "step": 2274 + }, + { + "epoch": 3.33, + "grad_norm": 3.2932636737823486, + "learning_rate": 0.000149579678789653, + "loss": 3.3457, + "step": 2275 + }, + { + "epoch": 3.33, + "grad_norm": 2.769178867340088, + "learning_rate": 0.00014953894057817188, + "loss": 3.3023, + "step": 2276 + }, + { + "epoch": 3.33, + "grad_norm": 3.352829694747925, + "learning_rate": 0.0001494981914684558, + "loss": 3.2836, + "step": 2277 + }, + { + "epoch": 3.34, + "grad_norm": 1.4692974090576172, + "learning_rate": 0.00014945743146946937, + "loss": 3.354, + "step": 2278 + }, + { + "epoch": 3.34, + "grad_norm": 5.136416912078857, + "learning_rate": 0.0001494166605901794, + "loss": 3.2915, + "step": 2279 + }, + { + "epoch": 3.34, + "grad_norm": 4.615678310394287, + "learning_rate": 0.00014937587883955534, + "loss": 3.2766, + "step": 2280 + }, + { + "epoch": 3.34, + "grad_norm": 1.0332159996032715, + "learning_rate": 0.00014933508622656878, + "loss": 3.3291, + "step": 2281 + }, + { + "epoch": 3.34, + "grad_norm": 1.1297160387039185, + "learning_rate": 0.00014929428276019388, + "loss": 3.3465, + "step": 2282 + }, + { + "epoch": 3.34, + "grad_norm": 3.5306038856506348, + "learning_rate": 0.00014925346844940718, + "loss": 3.2554, + "step": 2283 + }, + { + "epoch": 3.34, + "grad_norm": 9.56828498840332, + "learning_rate": 0.00014921264330318746, + "loss": 3.3203, + "step": 2284 + }, + { + "epoch": 3.35, + "grad_norm": 2.0608508586883545, + "learning_rate": 0.00014917180733051602, + "loss": 3.2899, + "step": 2285 + }, + { + "epoch": 3.35, + "grad_norm": 1.6151994466781616, + "learning_rate": 0.00014913096054037654, + "loss": 3.3209, + "step": 2286 + }, + { + "epoch": 3.35, + "grad_norm": 2.8889124393463135, + "learning_rate": 0.00014909010294175497, + "loss": 3.3144, + "step": 2287 + }, + { + "epoch": 3.35, + "grad_norm": 0.650810182094574, + "learning_rate": 0.00014904923454363975, + "loss": 3.2888, + "step": 2288 + }, + { + "epoch": 3.35, + "grad_norm": 2.751948833465576, + "learning_rate": 0.00014900835535502165, + "loss": 3.3317, + "step": 2289 + }, + { + "epoch": 3.35, + "grad_norm": 1.9524742364883423, + "learning_rate": 0.0001489674653848938, + "loss": 3.2798, + "step": 2290 + }, + { + "epoch": 3.35, + "grad_norm": 0.6142035126686096, + "learning_rate": 0.00014892656464225175, + "loss": 3.2485, + "step": 2291 + }, + { + "epoch": 3.36, + "grad_norm": 1.9580166339874268, + "learning_rate": 0.00014888565313609333, + "loss": 3.3353, + "step": 2292 + }, + { + "epoch": 3.36, + "grad_norm": 4.1013898849487305, + "learning_rate": 0.0001488447308754189, + "loss": 3.3178, + "step": 2293 + }, + { + "epoch": 3.36, + "grad_norm": 1.7551053762435913, + "learning_rate": 0.00014880379786923098, + "loss": 3.2874, + "step": 2294 + }, + { + "epoch": 3.36, + "grad_norm": 5.447327136993408, + "learning_rate": 0.00014876285412653456, + "loss": 3.2928, + "step": 2295 + }, + { + "epoch": 3.36, + "grad_norm": 5.519541263580322, + "learning_rate": 0.00014872189965633707, + "loss": 3.3418, + "step": 2296 + }, + { + "epoch": 3.36, + "grad_norm": 2.044977903366089, + "learning_rate": 0.0001486809344676482, + "loss": 3.3401, + "step": 2297 + }, + { + "epoch": 3.37, + "grad_norm": 5.334053993225098, + "learning_rate": 0.00014863995856947995, + "loss": 3.3156, + "step": 2298 + }, + { + "epoch": 3.37, + "grad_norm": 3.4951860904693604, + "learning_rate": 0.0001485989719708468, + "loss": 3.2812, + "step": 2299 + }, + { + "epoch": 3.37, + "grad_norm": 5.83620548248291, + "learning_rate": 0.00014855797468076552, + "loss": 3.3016, + "step": 2300 + }, + { + "epoch": 3.37, + "grad_norm": 6.08860969543457, + "learning_rate": 0.00014851696670825523, + "loss": 3.2941, + "step": 2301 + }, + { + "epoch": 3.37, + "grad_norm": 7.569385051727295, + "learning_rate": 0.00014847594806233737, + "loss": 3.3478, + "step": 2302 + }, + { + "epoch": 3.37, + "grad_norm": 5.372185230255127, + "learning_rate": 0.00014843491875203584, + "loss": 3.2692, + "step": 2303 + }, + { + "epoch": 3.37, + "grad_norm": 4.030975341796875, + "learning_rate": 0.0001483938787863768, + "loss": 3.2826, + "step": 2304 + }, + { + "epoch": 3.38, + "grad_norm": 6.85429048538208, + "learning_rate": 0.00014835282817438875, + "loss": 3.2932, + "step": 2305 + }, + { + "epoch": 3.38, + "grad_norm": 6.535613059997559, + "learning_rate": 0.0001483117669251025, + "loss": 3.3158, + "step": 2306 + }, + { + "epoch": 3.38, + "grad_norm": 3.962482452392578, + "learning_rate": 0.00014827069504755132, + "loss": 3.308, + "step": 2307 + }, + { + "epoch": 3.38, + "grad_norm": 6.860799312591553, + "learning_rate": 0.00014822961255077073, + "loss": 3.3145, + "step": 2308 + }, + { + "epoch": 3.38, + "grad_norm": 9.554657936096191, + "learning_rate": 0.0001481885194437986, + "loss": 3.3366, + "step": 2309 + }, + { + "epoch": 3.38, + "grad_norm": 4.180676460266113, + "learning_rate": 0.00014814741573567514, + "loss": 3.3096, + "step": 2310 + }, + { + "epoch": 3.38, + "grad_norm": 6.105428218841553, + "learning_rate": 0.0001481063014354429, + "loss": 3.2919, + "step": 2311 + }, + { + "epoch": 3.39, + "grad_norm": 7.584985733032227, + "learning_rate": 0.00014806517655214672, + "loss": 3.2756, + "step": 2312 + }, + { + "epoch": 3.39, + "grad_norm": 5.3676300048828125, + "learning_rate": 0.00014802404109483383, + "loss": 3.2969, + "step": 2313 + }, + { + "epoch": 3.39, + "grad_norm": 3.991481065750122, + "learning_rate": 0.00014798289507255373, + "loss": 3.2808, + "step": 2314 + }, + { + "epoch": 3.39, + "grad_norm": 6.212815761566162, + "learning_rate": 0.00014794173849435826, + "loss": 3.2809, + "step": 2315 + }, + { + "epoch": 3.39, + "grad_norm": 1.0110559463500977, + "learning_rate": 0.00014790057136930166, + "loss": 3.3016, + "step": 2316 + }, + { + "epoch": 3.39, + "grad_norm": 9.187312126159668, + "learning_rate": 0.00014785939370644036, + "loss": 3.275, + "step": 2317 + }, + { + "epoch": 3.39, + "grad_norm": 12.635149955749512, + "learning_rate": 0.00014781820551483315, + "loss": 3.3793, + "step": 2318 + }, + { + "epoch": 3.4, + "grad_norm": 9.69156551361084, + "learning_rate": 0.00014777700680354124, + "loss": 3.3063, + "step": 2319 + }, + { + "epoch": 3.4, + "grad_norm": 1.8201349973678589, + "learning_rate": 0.000147735797581628, + "loss": 3.3122, + "step": 2320 + }, + { + "epoch": 3.4, + "grad_norm": 11.608341217041016, + "learning_rate": 0.0001476945778581592, + "loss": 3.341, + "step": 2321 + }, + { + "epoch": 3.4, + "grad_norm": 13.871359825134277, + "learning_rate": 0.0001476533476422029, + "loss": 3.3459, + "step": 2322 + }, + { + "epoch": 3.4, + "grad_norm": 9.090968132019043, + "learning_rate": 0.00014761210694282947, + "loss": 3.369, + "step": 2323 + }, + { + "epoch": 3.4, + "grad_norm": 4.659329414367676, + "learning_rate": 0.0001475708557691116, + "loss": 3.417, + "step": 2324 + }, + { + "epoch": 3.41, + "grad_norm": 6.619305610656738, + "learning_rate": 0.00014752959413012428, + "loss": 3.3524, + "step": 2325 + }, + { + "epoch": 3.41, + "grad_norm": 9.944043159484863, + "learning_rate": 0.00014748832203494478, + "loss": 3.3326, + "step": 2326 + }, + { + "epoch": 3.41, + "grad_norm": 5.2585062980651855, + "learning_rate": 0.00014744703949265268, + "loss": 3.3292, + "step": 2327 + }, + { + "epoch": 3.41, + "grad_norm": 5.1725311279296875, + "learning_rate": 0.00014740574651232987, + "loss": 3.3172, + "step": 2328 + }, + { + "epoch": 3.41, + "grad_norm": 30.300024032592773, + "learning_rate": 0.00014736444310306052, + "loss": 3.3064, + "step": 2329 + }, + { + "epoch": 3.41, + "grad_norm": 9.257378578186035, + "learning_rate": 0.00014732312927393112, + "loss": 3.2872, + "step": 2330 + }, + { + "epoch": 3.41, + "grad_norm": 2.5596399307250977, + "learning_rate": 0.0001472818050340304, + "loss": 3.2914, + "step": 2331 + }, + { + "epoch": 3.42, + "grad_norm": 6.8813934326171875, + "learning_rate": 0.00014724047039244947, + "loss": 3.3698, + "step": 2332 + }, + { + "epoch": 3.42, + "grad_norm": 10.194779396057129, + "learning_rate": 0.00014719912535828166, + "loss": 3.2775, + "step": 2333 + }, + { + "epoch": 3.42, + "grad_norm": 5.2471489906311035, + "learning_rate": 0.00014715776994062254, + "loss": 3.3149, + "step": 2334 + }, + { + "epoch": 3.42, + "grad_norm": 3.753305196762085, + "learning_rate": 0.00014711640414857008, + "loss": 3.3024, + "step": 2335 + }, + { + "epoch": 3.42, + "grad_norm": 5.786775588989258, + "learning_rate": 0.00014707502799122447, + "loss": 3.2799, + "step": 2336 + }, + { + "epoch": 3.42, + "grad_norm": 1.7370816469192505, + "learning_rate": 0.00014703364147768822, + "loss": 3.2881, + "step": 2337 + }, + { + "epoch": 3.42, + "grad_norm": 7.590539932250977, + "learning_rate": 0.000146992244617066, + "loss": 3.3002, + "step": 2338 + }, + { + "epoch": 3.43, + "grad_norm": 9.67332935333252, + "learning_rate": 0.00014695083741846492, + "loss": 3.3037, + "step": 2339 + }, + { + "epoch": 3.43, + "grad_norm": 5.325073719024658, + "learning_rate": 0.00014690941989099426, + "loss": 3.3241, + "step": 2340 + }, + { + "epoch": 3.43, + "grad_norm": 7.251718044281006, + "learning_rate": 0.0001468679920437656, + "loss": 3.316, + "step": 2341 + }, + { + "epoch": 3.43, + "grad_norm": 8.890804290771484, + "learning_rate": 0.00014682655388589274, + "loss": 3.3177, + "step": 2342 + }, + { + "epoch": 3.43, + "grad_norm": 3.9834563732147217, + "learning_rate": 0.0001467851054264919, + "loss": 3.3002, + "step": 2343 + }, + { + "epoch": 3.43, + "grad_norm": 5.180391788482666, + "learning_rate": 0.0001467436466746814, + "loss": 3.3166, + "step": 2344 + }, + { + "epoch": 3.43, + "grad_norm": 9.531121253967285, + "learning_rate": 0.00014670217763958188, + "loss": 3.2482, + "step": 2345 + }, + { + "epoch": 3.44, + "grad_norm": 4.674380302429199, + "learning_rate": 0.00014666069833031624, + "loss": 3.2703, + "step": 2346 + }, + { + "epoch": 3.44, + "grad_norm": 3.9839441776275635, + "learning_rate": 0.00014661920875600973, + "loss": 3.2701, + "step": 2347 + }, + { + "epoch": 3.44, + "grad_norm": 4.408015251159668, + "learning_rate": 0.00014657770892578968, + "loss": 3.2791, + "step": 2348 + }, + { + "epoch": 3.44, + "grad_norm": 2.449519395828247, + "learning_rate": 0.0001465361988487858, + "loss": 3.2654, + "step": 2349 + }, + { + "epoch": 3.44, + "grad_norm": 1.4051802158355713, + "learning_rate": 0.00014649467853413008, + "loss": 3.3101, + "step": 2350 + }, + { + "epoch": 3.44, + "grad_norm": 6.514969825744629, + "learning_rate": 0.00014645314799095665, + "loss": 3.2282, + "step": 2351 + }, + { + "epoch": 3.45, + "grad_norm": 5.476304054260254, + "learning_rate": 0.000146411607228402, + "loss": 3.3043, + "step": 2352 + }, + { + "epoch": 3.45, + "grad_norm": 2.443798780441284, + "learning_rate": 0.00014637005625560477, + "loss": 3.3045, + "step": 2353 + }, + { + "epoch": 3.45, + "grad_norm": 6.676492214202881, + "learning_rate": 0.00014632849508170595, + "loss": 3.2794, + "step": 2354 + }, + { + "epoch": 3.45, + "grad_norm": 2.0837247371673584, + "learning_rate": 0.0001462869237158487, + "loss": 3.3134, + "step": 2355 + }, + { + "epoch": 3.45, + "grad_norm": 7.334787368774414, + "learning_rate": 0.00014624534216717841, + "loss": 3.2919, + "step": 2356 + }, + { + "epoch": 3.45, + "grad_norm": 8.694489479064941, + "learning_rate": 0.00014620375044484277, + "loss": 3.2639, + "step": 2357 + }, + { + "epoch": 3.45, + "grad_norm": 2.46984601020813, + "learning_rate": 0.00014616214855799167, + "loss": 3.2571, + "step": 2358 + }, + { + "epoch": 3.46, + "grad_norm": 7.231738567352295, + "learning_rate": 0.00014612053651577727, + "loss": 3.3468, + "step": 2359 + }, + { + "epoch": 3.46, + "grad_norm": 7.998284816741943, + "learning_rate": 0.0001460789143273539, + "loss": 3.3764, + "step": 2360 + }, + { + "epoch": 3.46, + "grad_norm": 4.452177047729492, + "learning_rate": 0.00014603728200187824, + "loss": 3.3293, + "step": 2361 + }, + { + "epoch": 3.46, + "grad_norm": 9.913199424743652, + "learning_rate": 0.00014599563954850901, + "loss": 3.2631, + "step": 2362 + }, + { + "epoch": 3.46, + "grad_norm": 15.90157413482666, + "learning_rate": 0.00014595398697640733, + "loss": 3.3812, + "step": 2363 + }, + { + "epoch": 3.46, + "grad_norm": 12.660530090332031, + "learning_rate": 0.0001459123242947365, + "loss": 3.371, + "step": 2364 + }, + { + "epoch": 3.46, + "grad_norm": 5.289348125457764, + "learning_rate": 0.00014587065151266203, + "loss": 3.3892, + "step": 2365 + }, + { + "epoch": 3.47, + "grad_norm": 9.648274421691895, + "learning_rate": 0.00014582896863935162, + "loss": 3.2931, + "step": 2366 + }, + { + "epoch": 3.47, + "eval_loss": 3.133899211883545, + "eval_runtime": 1162.2679, + "eval_samples_per_second": 14.749, + "eval_steps_per_second": 0.922, + "step": 2366 + }, + { + "epoch": 3.47, + "grad_norm": 9.562708854675293, + "learning_rate": 0.00014578727568397522, + "loss": 3.3485, + "step": 2367 + }, + { + "epoch": 3.47, + "grad_norm": 19.97218894958496, + "learning_rate": 0.00014574557265570505, + "loss": 3.3403, + "step": 2368 + }, + { + "epoch": 3.47, + "grad_norm": 6.222564220428467, + "learning_rate": 0.0001457038595637155, + "loss": 3.3743, + "step": 2369 + }, + { + "epoch": 3.47, + "grad_norm": 7.945571422576904, + "learning_rate": 0.00014566213641718307, + "loss": 3.2679, + "step": 2370 + }, + { + "epoch": 3.47, + "grad_norm": 8.883635520935059, + "learning_rate": 0.0001456204032252867, + "loss": 3.2295, + "step": 2371 + }, + { + "epoch": 3.48, + "grad_norm": 5.227753639221191, + "learning_rate": 0.00014557865999720735, + "loss": 3.2934, + "step": 2372 + }, + { + "epoch": 3.48, + "grad_norm": 8.865499496459961, + "learning_rate": 0.0001455369067421282, + "loss": 3.3526, + "step": 2373 + }, + { + "epoch": 3.48, + "grad_norm": 10.483139991760254, + "learning_rate": 0.00014549514346923477, + "loss": 3.3058, + "step": 2374 + }, + { + "epoch": 3.48, + "grad_norm": 12.922052383422852, + "learning_rate": 0.00014545337018771472, + "loss": 3.2366, + "step": 2375 + }, + { + "epoch": 3.48, + "grad_norm": 2.131599187850952, + "learning_rate": 0.00014541158690675778, + "loss": 3.2613, + "step": 2376 + }, + { + "epoch": 3.48, + "grad_norm": 10.010235786437988, + "learning_rate": 0.0001453697936355561, + "loss": 3.2698, + "step": 2377 + }, + { + "epoch": 3.48, + "grad_norm": 14.034173011779785, + "learning_rate": 0.00014532799038330385, + "loss": 3.3691, + "step": 2378 + }, + { + "epoch": 3.49, + "grad_norm": 9.930380821228027, + "learning_rate": 0.0001452861771591975, + "loss": 3.304, + "step": 2379 + }, + { + "epoch": 3.49, + "grad_norm": 7.6726555824279785, + "learning_rate": 0.00014524435397243567, + "loss": 3.2806, + "step": 2380 + }, + { + "epoch": 3.49, + "grad_norm": 8.018184661865234, + "learning_rate": 0.00014520252083221916, + "loss": 3.2673, + "step": 2381 + }, + { + "epoch": 3.49, + "grad_norm": 8.354979515075684, + "learning_rate": 0.00014516067774775101, + "loss": 3.295, + "step": 2382 + }, + { + "epoch": 3.49, + "grad_norm": 3.118138313293457, + "learning_rate": 0.00014511882472823643, + "loss": 3.2593, + "step": 2383 + }, + { + "epoch": 3.49, + "grad_norm": 9.12159538269043, + "learning_rate": 0.00014507696178288278, + "loss": 3.2893, + "step": 2384 + }, + { + "epoch": 3.49, + "grad_norm": 13.92984390258789, + "learning_rate": 0.0001450350889208996, + "loss": 3.3191, + "step": 2385 + }, + { + "epoch": 3.5, + "grad_norm": 9.545883178710938, + "learning_rate": 0.0001449932061514987, + "loss": 3.2753, + "step": 2386 + }, + { + "epoch": 3.5, + "grad_norm": 1.6055418252944946, + "learning_rate": 0.00014495131348389394, + "loss": 3.2373, + "step": 2387 + }, + { + "epoch": 3.5, + "grad_norm": 8.169912338256836, + "learning_rate": 0.00014490941092730152, + "loss": 3.2748, + "step": 2388 + }, + { + "epoch": 3.5, + "grad_norm": 8.912496566772461, + "learning_rate": 0.00014486749849093963, + "loss": 3.302, + "step": 2389 + }, + { + "epoch": 3.5, + "grad_norm": 6.068753242492676, + "learning_rate": 0.00014482557618402875, + "loss": 3.2805, + "step": 2390 + }, + { + "epoch": 3.5, + "grad_norm": 4.4562578201293945, + "learning_rate": 0.00014478364401579154, + "loss": 3.2488, + "step": 2391 + }, + { + "epoch": 3.5, + "grad_norm": 4.690038204193115, + "learning_rate": 0.00014474170199545275, + "loss": 3.2896, + "step": 2392 + }, + { + "epoch": 3.51, + "grad_norm": 3.260173797607422, + "learning_rate": 0.0001446997501322394, + "loss": 3.2497, + "step": 2393 + }, + { + "epoch": 3.51, + "grad_norm": 3.311274528503418, + "learning_rate": 0.0001446577884353806, + "loss": 3.2359, + "step": 2394 + }, + { + "epoch": 3.51, + "grad_norm": 2.5797505378723145, + "learning_rate": 0.00014461581691410758, + "loss": 3.3372, + "step": 2395 + }, + { + "epoch": 3.51, + "grad_norm": 1.9033737182617188, + "learning_rate": 0.00014457383557765386, + "loss": 3.2308, + "step": 2396 + }, + { + "epoch": 3.51, + "grad_norm": 3.9633514881134033, + "learning_rate": 0.000144531844435255, + "loss": 3.2787, + "step": 2397 + }, + { + "epoch": 3.51, + "grad_norm": 3.756439685821533, + "learning_rate": 0.00014448984349614886, + "loss": 3.3238, + "step": 2398 + }, + { + "epoch": 3.52, + "grad_norm": 3.5643348693847656, + "learning_rate": 0.00014444783276957524, + "loss": 3.2315, + "step": 2399 + }, + { + "epoch": 3.52, + "grad_norm": 4.421493053436279, + "learning_rate": 0.00014440581226477634, + "loss": 3.277, + "step": 2400 + }, + { + "epoch": 3.52, + "grad_norm": 4.980103969573975, + "learning_rate": 0.00014436378199099634, + "loss": 3.248, + "step": 2401 + }, + { + "epoch": 3.52, + "grad_norm": 7.962924003601074, + "learning_rate": 0.0001443217419574816, + "loss": 3.2628, + "step": 2402 + }, + { + "epoch": 3.52, + "grad_norm": 3.279404401779175, + "learning_rate": 0.00014427969217348065, + "loss": 3.2662, + "step": 2403 + }, + { + "epoch": 3.52, + "grad_norm": 6.002499580383301, + "learning_rate": 0.00014423763264824417, + "loss": 3.2498, + "step": 2404 + }, + { + "epoch": 3.52, + "grad_norm": 8.197328567504883, + "learning_rate": 0.00014419556339102503, + "loss": 3.3432, + "step": 2405 + }, + { + "epoch": 3.53, + "grad_norm": 3.395263910293579, + "learning_rate": 0.00014415348441107806, + "loss": 3.2308, + "step": 2406 + }, + { + "epoch": 3.53, + "grad_norm": 6.569970607757568, + "learning_rate": 0.00014411139571766042, + "loss": 3.2445, + "step": 2407 + }, + { + "epoch": 3.53, + "grad_norm": 9.0120849609375, + "learning_rate": 0.0001440692973200314, + "loss": 3.2689, + "step": 2408 + }, + { + "epoch": 3.53, + "grad_norm": 5.56130313873291, + "learning_rate": 0.00014402718922745228, + "loss": 3.2685, + "step": 2409 + }, + { + "epoch": 3.53, + "grad_norm": 4.367111682891846, + "learning_rate": 0.00014398507144918656, + "loss": 3.3052, + "step": 2410 + }, + { + "epoch": 3.53, + "grad_norm": 6.1577558517456055, + "learning_rate": 0.00014394294399449996, + "loss": 3.2687, + "step": 2411 + }, + { + "epoch": 3.53, + "grad_norm": 14.60132122039795, + "learning_rate": 0.00014390080687266013, + "loss": 3.2542, + "step": 2412 + }, + { + "epoch": 3.54, + "grad_norm": 19.244531631469727, + "learning_rate": 0.000143858660092937, + "loss": 3.2623, + "step": 2413 + }, + { + "epoch": 3.54, + "grad_norm": 5.000961780548096, + "learning_rate": 0.0001438165036646026, + "loss": 3.2887, + "step": 2414 + }, + { + "epoch": 3.54, + "grad_norm": 1.7813776731491089, + "learning_rate": 0.000143774337596931, + "loss": 3.2522, + "step": 2415 + }, + { + "epoch": 3.54, + "grad_norm": 11.567626953125, + "learning_rate": 0.00014373216189919851, + "loss": 3.3463, + "step": 2416 + }, + { + "epoch": 3.54, + "grad_norm": 17.07219696044922, + "learning_rate": 0.0001436899765806835, + "loss": 3.3653, + "step": 2417 + }, + { + "epoch": 3.54, + "grad_norm": 10.967621803283691, + "learning_rate": 0.00014364778165066642, + "loss": 3.339, + "step": 2418 + }, + { + "epoch": 3.54, + "grad_norm": 5.372880935668945, + "learning_rate": 0.00014360557711842991, + "loss": 3.3025, + "step": 2419 + }, + { + "epoch": 3.55, + "grad_norm": 8.31612777709961, + "learning_rate": 0.00014356336299325864, + "loss": 3.2886, + "step": 2420 + }, + { + "epoch": 3.55, + "grad_norm": 6.364083290100098, + "learning_rate": 0.00014352113928443947, + "loss": 3.3376, + "step": 2421 + }, + { + "epoch": 3.55, + "grad_norm": 39.3469123840332, + "learning_rate": 0.00014347890600126134, + "loss": 3.2798, + "step": 2422 + }, + { + "epoch": 3.55, + "grad_norm": 15.578664779663086, + "learning_rate": 0.00014343666315301526, + "loss": 3.3638, + "step": 2423 + }, + { + "epoch": 3.55, + "grad_norm": 15.445240020751953, + "learning_rate": 0.0001433944107489944, + "loss": 3.3672, + "step": 2424 + }, + { + "epoch": 3.55, + "grad_norm": 13.184659957885742, + "learning_rate": 0.00014335214879849394, + "loss": 3.3172, + "step": 2425 + }, + { + "epoch": 3.56, + "grad_norm": 4.225596904754639, + "learning_rate": 0.0001433098773108113, + "loss": 3.2598, + "step": 2426 + }, + { + "epoch": 3.56, + "grad_norm": 9.756654739379883, + "learning_rate": 0.00014326759629524594, + "loss": 3.3621, + "step": 2427 + }, + { + "epoch": 3.56, + "grad_norm": 14.042159080505371, + "learning_rate": 0.00014322530576109933, + "loss": 3.3233, + "step": 2428 + }, + { + "epoch": 3.56, + "grad_norm": 14.112271308898926, + "learning_rate": 0.00014318300571767513, + "loss": 3.3309, + "step": 2429 + }, + { + "epoch": 3.56, + "grad_norm": 9.754487037658691, + "learning_rate": 0.00014314069617427907, + "loss": 3.3182, + "step": 2430 + }, + { + "epoch": 3.56, + "grad_norm": 2.0633256435394287, + "learning_rate": 0.00014309837714021902, + "loss": 3.2493, + "step": 2431 + }, + { + "epoch": 3.56, + "grad_norm": 8.613134384155273, + "learning_rate": 0.00014305604862480478, + "loss": 3.3103, + "step": 2432 + }, + { + "epoch": 3.57, + "grad_norm": 9.88028335571289, + "learning_rate": 0.0001430137106373484, + "loss": 3.3556, + "step": 2433 + }, + { + "epoch": 3.57, + "grad_norm": 8.193672180175781, + "learning_rate": 0.00014297136318716395, + "loss": 3.3321, + "step": 2434 + }, + { + "epoch": 3.57, + "grad_norm": 4.98108434677124, + "learning_rate": 0.00014292900628356758, + "loss": 3.265, + "step": 2435 + }, + { + "epoch": 3.57, + "grad_norm": 5.972259521484375, + "learning_rate": 0.0001428866399358775, + "loss": 3.2759, + "step": 2436 + }, + { + "epoch": 3.57, + "grad_norm": 4.662635803222656, + "learning_rate": 0.00014284426415341408, + "loss": 3.3012, + "step": 2437 + }, + { + "epoch": 3.57, + "grad_norm": 4.476476192474365, + "learning_rate": 0.00014280187894549968, + "loss": 3.2919, + "step": 2438 + }, + { + "epoch": 3.57, + "grad_norm": 3.099634885787964, + "learning_rate": 0.00014275948432145876, + "loss": 3.3264, + "step": 2439 + }, + { + "epoch": 3.58, + "grad_norm": 6.200466632843018, + "learning_rate": 0.00014271708029061785, + "loss": 3.2982, + "step": 2440 + }, + { + "epoch": 3.58, + "grad_norm": 3.2733237743377686, + "learning_rate": 0.00014267466686230553, + "loss": 3.2932, + "step": 2441 + }, + { + "epoch": 3.58, + "grad_norm": 3.1728217601776123, + "learning_rate": 0.0001426322440458525, + "loss": 3.2532, + "step": 2442 + }, + { + "epoch": 3.58, + "grad_norm": 3.4622151851654053, + "learning_rate": 0.0001425898118505915, + "loss": 3.2794, + "step": 2443 + }, + { + "epoch": 3.58, + "grad_norm": 7.931596755981445, + "learning_rate": 0.0001425473702858573, + "loss": 3.2692, + "step": 2444 + }, + { + "epoch": 3.58, + "grad_norm": 3.586261749267578, + "learning_rate": 0.00014250491936098678, + "loss": 3.2432, + "step": 2445 + }, + { + "epoch": 3.58, + "grad_norm": 3.096931219100952, + "learning_rate": 0.00014246245908531882, + "loss": 3.2589, + "step": 2446 + }, + { + "epoch": 3.59, + "grad_norm": 3.957979202270508, + "learning_rate": 0.0001424199894681945, + "loss": 3.278, + "step": 2447 + }, + { + "epoch": 3.59, + "grad_norm": 0.8693408370018005, + "learning_rate": 0.00014237751051895676, + "loss": 3.2366, + "step": 2448 + }, + { + "epoch": 3.59, + "grad_norm": 2.3588359355926514, + "learning_rate": 0.0001423350222469507, + "loss": 3.2469, + "step": 2449 + }, + { + "epoch": 3.59, + "grad_norm": 1.2426791191101074, + "learning_rate": 0.00014229252466152345, + "loss": 3.2929, + "step": 2450 + }, + { + "epoch": 3.59, + "grad_norm": 2.7103374004364014, + "learning_rate": 0.00014225001777202424, + "loss": 3.248, + "step": 2451 + }, + { + "epoch": 3.59, + "grad_norm": 2.5343198776245117, + "learning_rate": 0.00014220750158780424, + "loss": 3.257, + "step": 2452 + }, + { + "epoch": 3.6, + "grad_norm": 1.3891563415527344, + "learning_rate": 0.00014216497611821676, + "loss": 3.2348, + "step": 2453 + }, + { + "epoch": 3.6, + "grad_norm": 3.7499053478240967, + "learning_rate": 0.0001421224413726171, + "loss": 3.2513, + "step": 2454 + }, + { + "epoch": 3.6, + "grad_norm": 1.5479190349578857, + "learning_rate": 0.00014207989736036267, + "loss": 3.2785, + "step": 2455 + }, + { + "epoch": 3.6, + "grad_norm": 2.783564567565918, + "learning_rate": 0.0001420373440908128, + "loss": 3.292, + "step": 2456 + }, + { + "epoch": 3.6, + "grad_norm": 3.4745991230010986, + "learning_rate": 0.00014199478157332895, + "loss": 3.2399, + "step": 2457 + }, + { + "epoch": 3.6, + "grad_norm": 2.5730254650115967, + "learning_rate": 0.0001419522098172746, + "loss": 3.257, + "step": 2458 + }, + { + "epoch": 3.6, + "grad_norm": 3.455461025238037, + "learning_rate": 0.00014190962883201523, + "loss": 3.225, + "step": 2459 + }, + { + "epoch": 3.61, + "grad_norm": 4.234341621398926, + "learning_rate": 0.0001418670386269184, + "loss": 3.2997, + "step": 2460 + }, + { + "epoch": 3.61, + "grad_norm": 9.0831937789917, + "learning_rate": 0.00014182443921135362, + "loss": 3.221, + "step": 2461 + }, + { + "epoch": 3.61, + "grad_norm": 3.933647871017456, + "learning_rate": 0.00014178183059469253, + "loss": 3.2355, + "step": 2462 + }, + { + "epoch": 3.61, + "grad_norm": 1.695810079574585, + "learning_rate": 0.00014173921278630872, + "loss": 3.2724, + "step": 2463 + }, + { + "epoch": 3.61, + "grad_norm": 2.3507513999938965, + "learning_rate": 0.0001416965857955778, + "loss": 3.273, + "step": 2464 + }, + { + "epoch": 3.61, + "grad_norm": 2.5299062728881836, + "learning_rate": 0.00014165394963187747, + "loss": 3.2366, + "step": 2465 + }, + { + "epoch": 3.61, + "grad_norm": 0.9136064052581787, + "learning_rate": 0.00014161130430458737, + "loss": 3.2318, + "step": 2466 + }, + { + "epoch": 3.62, + "grad_norm": 2.159097671508789, + "learning_rate": 0.00014156864982308917, + "loss": 3.2139, + "step": 2467 + }, + { + "epoch": 3.62, + "grad_norm": 2.3743743896484375, + "learning_rate": 0.0001415259861967666, + "loss": 3.2442, + "step": 2468 + }, + { + "epoch": 3.62, + "grad_norm": 1.7210527658462524, + "learning_rate": 0.00014148331343500538, + "loss": 3.2565, + "step": 2469 + }, + { + "epoch": 3.62, + "grad_norm": 4.262208938598633, + "learning_rate": 0.0001414406315471932, + "loss": 3.2542, + "step": 2470 + }, + { + "epoch": 3.62, + "grad_norm": 3.7595646381378174, + "learning_rate": 0.00014139794054271977, + "loss": 3.2048, + "step": 2471 + }, + { + "epoch": 3.62, + "grad_norm": 3.6839802265167236, + "learning_rate": 0.00014135524043097693, + "loss": 3.2873, + "step": 2472 + }, + { + "epoch": 3.63, + "grad_norm": 3.4619247913360596, + "learning_rate": 0.0001413125312213583, + "loss": 3.2562, + "step": 2473 + }, + { + "epoch": 3.63, + "grad_norm": 3.164567708969116, + "learning_rate": 0.00014126981292325965, + "loss": 3.2521, + "step": 2474 + }, + { + "epoch": 3.63, + "grad_norm": 3.066319227218628, + "learning_rate": 0.00014122708554607875, + "loss": 3.2671, + "step": 2475 + }, + { + "epoch": 3.63, + "grad_norm": 2.757352590560913, + "learning_rate": 0.00014118434909921534, + "loss": 3.2061, + "step": 2476 + }, + { + "epoch": 3.63, + "grad_norm": 1.9947091341018677, + "learning_rate": 0.0001411416035920711, + "loss": 3.2391, + "step": 2477 + }, + { + "epoch": 3.63, + "grad_norm": 4.6580986976623535, + "learning_rate": 0.0001410988490340498, + "loss": 3.252, + "step": 2478 + }, + { + "epoch": 3.63, + "grad_norm": 4.087027072906494, + "learning_rate": 0.00014105608543455716, + "loss": 3.2371, + "step": 2479 + }, + { + "epoch": 3.64, + "grad_norm": 3.322829008102417, + "learning_rate": 0.0001410133128030009, + "loss": 3.2261, + "step": 2480 + }, + { + "epoch": 3.64, + "grad_norm": 3.1107428073883057, + "learning_rate": 0.00014097053114879061, + "loss": 3.2418, + "step": 2481 + }, + { + "epoch": 3.64, + "grad_norm": 2.6654446125030518, + "learning_rate": 0.00014092774048133809, + "loss": 3.232, + "step": 2482 + }, + { + "epoch": 3.64, + "grad_norm": 1.8028963804244995, + "learning_rate": 0.00014088494081005695, + "loss": 3.2122, + "step": 2483 + }, + { + "epoch": 3.64, + "grad_norm": 9.199921607971191, + "learning_rate": 0.00014084213214436277, + "loss": 3.2409, + "step": 2484 + }, + { + "epoch": 3.64, + "grad_norm": 7.354410648345947, + "learning_rate": 0.00014079931449367328, + "loss": 3.2151, + "step": 2485 + }, + { + "epoch": 3.64, + "grad_norm": 2.6317131519317627, + "learning_rate": 0.00014075648786740806, + "loss": 3.2449, + "step": 2486 + }, + { + "epoch": 3.65, + "grad_norm": 6.4712042808532715, + "learning_rate": 0.00014071365227498857, + "loss": 3.2384, + "step": 2487 + }, + { + "epoch": 3.65, + "grad_norm": 7.566592216491699, + "learning_rate": 0.00014067080772583846, + "loss": 3.2208, + "step": 2488 + }, + { + "epoch": 3.65, + "grad_norm": 3.7115042209625244, + "learning_rate": 0.00014062795422938325, + "loss": 3.2403, + "step": 2489 + }, + { + "epoch": 3.65, + "grad_norm": 7.227043151855469, + "learning_rate": 0.00014058509179505036, + "loss": 3.2533, + "step": 2490 + }, + { + "epoch": 3.65, + "grad_norm": 10.836502075195312, + "learning_rate": 0.00014054222043226926, + "loss": 3.3034, + "step": 2491 + }, + { + "epoch": 3.65, + "grad_norm": 8.85554313659668, + "learning_rate": 0.00014049934015047136, + "loss": 3.2838, + "step": 2492 + }, + { + "epoch": 3.65, + "grad_norm": 7.335647106170654, + "learning_rate": 0.00014045645095909004, + "loss": 3.2506, + "step": 2493 + }, + { + "epoch": 3.66, + "grad_norm": 6.849621295928955, + "learning_rate": 0.00014041355286756066, + "loss": 3.266, + "step": 2494 + }, + { + "epoch": 3.66, + "grad_norm": 5.7214837074279785, + "learning_rate": 0.00014037064588532047, + "loss": 3.229, + "step": 2495 + }, + { + "epoch": 3.66, + "grad_norm": 1.1197165250778198, + "learning_rate": 0.00014032773002180875, + "loss": 3.189, + "step": 2496 + }, + { + "epoch": 3.66, + "grad_norm": 2.7683522701263428, + "learning_rate": 0.00014028480528646667, + "loss": 3.2357, + "step": 2497 + }, + { + "epoch": 3.66, + "grad_norm": 10.467253684997559, + "learning_rate": 0.0001402418716887374, + "loss": 3.2702, + "step": 2498 + }, + { + "epoch": 3.66, + "grad_norm": 5.283757209777832, + "learning_rate": 0.0001401989292380661, + "loss": 3.2404, + "step": 2499 + }, + { + "epoch": 3.67, + "grad_norm": 3.8525490760803223, + "learning_rate": 0.00014015597794389973, + "loss": 3.2352, + "step": 2500 + }, + { + "epoch": 3.67, + "grad_norm": 4.830957412719727, + "learning_rate": 0.0001401130178156873, + "loss": 3.2432, + "step": 2501 + }, + { + "epoch": 3.67, + "grad_norm": 28.55122184753418, + "learning_rate": 0.0001400700488628798, + "loss": 3.2176, + "step": 2502 + }, + { + "epoch": 3.67, + "grad_norm": 6.477649211883545, + "learning_rate": 0.0001400270710949301, + "loss": 3.1952, + "step": 2503 + }, + { + "epoch": 3.67, + "grad_norm": 4.75806188583374, + "learning_rate": 0.00013998408452129302, + "loss": 3.2384, + "step": 2504 + }, + { + "epoch": 3.67, + "grad_norm": 5.451016426086426, + "learning_rate": 0.00013994108915142526, + "loss": 3.2404, + "step": 2505 + }, + { + "epoch": 3.67, + "grad_norm": 2.510890245437622, + "learning_rate": 0.0001398980849947856, + "loss": 3.221, + "step": 2506 + }, + { + "epoch": 3.68, + "grad_norm": 6.08433198928833, + "learning_rate": 0.0001398550720608346, + "loss": 3.2386, + "step": 2507 + }, + { + "epoch": 3.68, + "grad_norm": 7.88702392578125, + "learning_rate": 0.00013981205035903485, + "loss": 3.2582, + "step": 2508 + }, + { + "epoch": 3.68, + "grad_norm": 3.3581769466400146, + "learning_rate": 0.00013976901989885086, + "loss": 3.2437, + "step": 2509 + }, + { + "epoch": 3.68, + "grad_norm": 5.539381980895996, + "learning_rate": 0.000139725980689749, + "loss": 3.2349, + "step": 2510 + }, + { + "epoch": 3.68, + "grad_norm": 6.059652328491211, + "learning_rate": 0.0001396829327411976, + "loss": 3.3016, + "step": 2511 + }, + { + "epoch": 3.68, + "grad_norm": 1.3593266010284424, + "learning_rate": 0.00013963987606266693, + "loss": 3.2587, + "step": 2512 + }, + { + "epoch": 3.68, + "grad_norm": 9.557854652404785, + "learning_rate": 0.0001395968106636292, + "loss": 3.2695, + "step": 2513 + }, + { + "epoch": 3.69, + "grad_norm": 26.98328971862793, + "learning_rate": 0.0001395537365535585, + "loss": 3.3668, + "step": 2514 + }, + { + "epoch": 3.69, + "grad_norm": 14.079690933227539, + "learning_rate": 0.0001395106537419308, + "loss": 3.2736, + "step": 2515 + }, + { + "epoch": 3.69, + "grad_norm": 5.151083946228027, + "learning_rate": 0.00013946756223822413, + "loss": 3.2461, + "step": 2516 + }, + { + "epoch": 3.69, + "grad_norm": 7.186163902282715, + "learning_rate": 0.00013942446205191824, + "loss": 3.2817, + "step": 2517 + }, + { + "epoch": 3.69, + "grad_norm": 9.277379035949707, + "learning_rate": 0.0001393813531924949, + "loss": 3.2008, + "step": 2518 + }, + { + "epoch": 3.69, + "grad_norm": 19.572731018066406, + "learning_rate": 0.00013933823566943777, + "loss": 3.2131, + "step": 2519 + }, + { + "epoch": 3.69, + "grad_norm": 4.6605329513549805, + "learning_rate": 0.00013929510949223245, + "loss": 3.2512, + "step": 2520 + }, + { + "epoch": 3.7, + "grad_norm": 151.2240753173828, + "learning_rate": 0.00013925197467036635, + "loss": 3.2621, + "step": 2521 + }, + { + "epoch": 3.7, + "grad_norm": 13.225665092468262, + "learning_rate": 0.0001392088312133289, + "loss": 3.2814, + "step": 2522 + }, + { + "epoch": 3.7, + "grad_norm": 14.404492378234863, + "learning_rate": 0.00013916567913061133, + "loss": 3.2903, + "step": 2523 + }, + { + "epoch": 3.7, + "grad_norm": 10.506006240844727, + "learning_rate": 0.00013912251843170687, + "loss": 3.226, + "step": 2524 + }, + { + "epoch": 3.7, + "grad_norm": 3.039416790008545, + "learning_rate": 0.0001390793491261105, + "loss": 3.26, + "step": 2525 + }, + { + "epoch": 3.7, + "grad_norm": 6.364565372467041, + "learning_rate": 0.00013903617122331923, + "loss": 3.1935, + "step": 2526 + }, + { + "epoch": 3.71, + "grad_norm": 3.801405668258667, + "learning_rate": 0.0001389929847328319, + "loss": 3.2204, + "step": 2527 + }, + { + "epoch": 3.71, + "grad_norm": 13.04414176940918, + "learning_rate": 0.00013894978966414926, + "loss": 3.2408, + "step": 2528 + }, + { + "epoch": 3.71, + "grad_norm": 14.216852188110352, + "learning_rate": 0.0001389065860267739, + "loss": 3.2739, + "step": 2529 + }, + { + "epoch": 3.71, + "grad_norm": 10.664838790893555, + "learning_rate": 0.0001388633738302104, + "loss": 3.2684, + "step": 2530 + }, + { + "epoch": 3.71, + "grad_norm": 6.671474456787109, + "learning_rate": 0.00013882015308396508, + "loss": 3.3327, + "step": 2531 + }, + { + "epoch": 3.71, + "grad_norm": 15.43130874633789, + "learning_rate": 0.00013877692379754626, + "loss": 3.2355, + "step": 2532 + }, + { + "epoch": 3.71, + "grad_norm": 44.657169342041016, + "learning_rate": 0.00013873368598046404, + "loss": 3.2309, + "step": 2533 + }, + { + "epoch": 3.72, + "grad_norm": 18.107580184936523, + "learning_rate": 0.00013869043964223051, + "loss": 3.2906, + "step": 2534 + }, + { + "epoch": 3.72, + "grad_norm": 11.789257049560547, + "learning_rate": 0.00013864718479235954, + "loss": 3.3045, + "step": 2535 + }, + { + "epoch": 3.72, + "eval_loss": 3.0709898471832275, + "eval_runtime": 1163.6634, + "eval_samples_per_second": 14.731, + "eval_steps_per_second": 0.921, + "step": 2535 + }, + { + "epoch": 3.72, + "grad_norm": 7.7072319984436035, + "learning_rate": 0.00013860392144036693, + "loss": 3.2131, + "step": 2536 + }, + { + "epoch": 3.72, + "grad_norm": 4.083733081817627, + "learning_rate": 0.0001385606495957703, + "loss": 3.2641, + "step": 2537 + }, + { + "epoch": 3.72, + "grad_norm": 11.785395622253418, + "learning_rate": 0.00013851736926808917, + "loss": 3.2106, + "step": 2538 + }, + { + "epoch": 3.72, + "grad_norm": 13.176704406738281, + "learning_rate": 0.00013847408046684497, + "loss": 3.2857, + "step": 2539 + }, + { + "epoch": 3.72, + "grad_norm": 5.579313278198242, + "learning_rate": 0.00013843078320156088, + "loss": 3.2861, + "step": 2540 + }, + { + "epoch": 3.73, + "grad_norm": 4.398528099060059, + "learning_rate": 0.0001383874774817621, + "loss": 3.225, + "step": 2541 + }, + { + "epoch": 3.73, + "grad_norm": 6.542925834655762, + "learning_rate": 0.00013834416331697544, + "loss": 3.204, + "step": 2542 + }, + { + "epoch": 3.73, + "grad_norm": 9.607292175292969, + "learning_rate": 0.00013830084071672985, + "loss": 3.2284, + "step": 2543 + }, + { + "epoch": 3.73, + "grad_norm": 2.1741347312927246, + "learning_rate": 0.00013825750969055604, + "loss": 3.2318, + "step": 2544 + }, + { + "epoch": 3.73, + "grad_norm": 2.4551427364349365, + "learning_rate": 0.00013821417024798642, + "loss": 3.2454, + "step": 2545 + }, + { + "epoch": 3.73, + "grad_norm": 1.8988637924194336, + "learning_rate": 0.0001381708223985555, + "loss": 3.2688, + "step": 2546 + }, + { + "epoch": 3.73, + "grad_norm": 1.2011829614639282, + "learning_rate": 0.00013812746615179944, + "loss": 3.2418, + "step": 2547 + }, + { + "epoch": 3.74, + "grad_norm": 2.4243390560150146, + "learning_rate": 0.0001380841015172563, + "loss": 3.2759, + "step": 2548 + }, + { + "epoch": 3.74, + "grad_norm": 0.9519268274307251, + "learning_rate": 0.00013804072850446612, + "loss": 3.2498, + "step": 2549 + }, + { + "epoch": 3.74, + "grad_norm": 5.196446418762207, + "learning_rate": 0.00013799734712297059, + "loss": 3.2358, + "step": 2550 + }, + { + "epoch": 3.74, + "grad_norm": 3.189202308654785, + "learning_rate": 0.00013795395738231333, + "loss": 3.2582, + "step": 2551 + }, + { + "epoch": 3.74, + "grad_norm": 5.616455078125, + "learning_rate": 0.0001379105592920398, + "loss": 3.2066, + "step": 2552 + }, + { + "epoch": 3.74, + "grad_norm": 9.704096794128418, + "learning_rate": 0.0001378671528616973, + "loss": 3.2245, + "step": 2553 + }, + { + "epoch": 3.75, + "grad_norm": 3.5445096492767334, + "learning_rate": 0.00013782373810083493, + "loss": 3.2097, + "step": 2554 + }, + { + "epoch": 3.75, + "grad_norm": 3.731013774871826, + "learning_rate": 0.00013778031501900367, + "loss": 3.2075, + "step": 2555 + }, + { + "epoch": 3.75, + "grad_norm": 8.907066345214844, + "learning_rate": 0.0001377368836257563, + "loss": 3.2249, + "step": 2556 + }, + { + "epoch": 3.75, + "grad_norm": 4.645308017730713, + "learning_rate": 0.00013769344393064743, + "loss": 3.2139, + "step": 2557 + }, + { + "epoch": 3.75, + "grad_norm": 4.715915679931641, + "learning_rate": 0.00013764999594323353, + "loss": 3.2613, + "step": 2558 + }, + { + "epoch": 3.75, + "grad_norm": 5.21357536315918, + "learning_rate": 0.0001376065396730728, + "loss": 3.2247, + "step": 2559 + }, + { + "epoch": 3.75, + "grad_norm": 1.811521053314209, + "learning_rate": 0.00013756307512972538, + "loss": 3.2714, + "step": 2560 + }, + { + "epoch": 3.76, + "grad_norm": 5.335707664489746, + "learning_rate": 0.0001375196023227532, + "loss": 3.2079, + "step": 2561 + }, + { + "epoch": 3.76, + "grad_norm": 3.844973087310791, + "learning_rate": 0.0001374761212617199, + "loss": 3.2411, + "step": 2562 + }, + { + "epoch": 3.76, + "grad_norm": 4.505429267883301, + "learning_rate": 0.00013743263195619115, + "loss": 3.2508, + "step": 2563 + }, + { + "epoch": 3.76, + "grad_norm": 5.144504547119141, + "learning_rate": 0.00013738913441573418, + "loss": 3.2112, + "step": 2564 + }, + { + "epoch": 3.76, + "grad_norm": 23.4456787109375, + "learning_rate": 0.00013734562864991823, + "loss": 3.2106, + "step": 2565 + }, + { + "epoch": 3.76, + "grad_norm": 4.499874591827393, + "learning_rate": 0.0001373021146683142, + "loss": 3.2324, + "step": 2566 + }, + { + "epoch": 3.76, + "grad_norm": 2.9865691661834717, + "learning_rate": 0.000137258592480495, + "loss": 3.2498, + "step": 2567 + }, + { + "epoch": 3.77, + "grad_norm": 3.308457374572754, + "learning_rate": 0.00013721506209603512, + "loss": 3.2132, + "step": 2568 + }, + { + "epoch": 3.77, + "grad_norm": 2.9479405879974365, + "learning_rate": 0.000137171523524511, + "loss": 3.2817, + "step": 2569 + }, + { + "epoch": 3.77, + "grad_norm": 2.7257778644561768, + "learning_rate": 0.0001371279767755008, + "loss": 3.185, + "step": 2570 + }, + { + "epoch": 3.77, + "grad_norm": 1.7737255096435547, + "learning_rate": 0.00013708442185858454, + "loss": 3.222, + "step": 2571 + }, + { + "epoch": 3.77, + "grad_norm": 3.827313184738159, + "learning_rate": 0.000137040858783344, + "loss": 3.1943, + "step": 2572 + }, + { + "epoch": 3.77, + "grad_norm": 2.6353366374969482, + "learning_rate": 0.00013699728755936277, + "loss": 3.2699, + "step": 2573 + }, + { + "epoch": 3.78, + "grad_norm": 4.652822017669678, + "learning_rate": 0.0001369537081962262, + "loss": 3.2313, + "step": 2574 + }, + { + "epoch": 3.78, + "grad_norm": 8.656730651855469, + "learning_rate": 0.0001369101207035215, + "loss": 3.1737, + "step": 2575 + }, + { + "epoch": 3.78, + "grad_norm": 1.588462471961975, + "learning_rate": 0.0001368665250908376, + "loss": 3.2057, + "step": 2576 + }, + { + "epoch": 3.78, + "grad_norm": 5.477829456329346, + "learning_rate": 0.00013682292136776524, + "loss": 3.2045, + "step": 2577 + }, + { + "epoch": 3.78, + "grad_norm": 6.158089637756348, + "learning_rate": 0.00013677930954389698, + "loss": 3.2183, + "step": 2578 + }, + { + "epoch": 3.78, + "grad_norm": 2.4769415855407715, + "learning_rate": 0.00013673568962882705, + "loss": 3.1601, + "step": 2579 + }, + { + "epoch": 3.78, + "grad_norm": 7.979514122009277, + "learning_rate": 0.00013669206163215164, + "loss": 3.2011, + "step": 2580 + }, + { + "epoch": 3.79, + "grad_norm": 11.770708084106445, + "learning_rate": 0.00013664842556346856, + "loss": 3.2215, + "step": 2581 + }, + { + "epoch": 3.79, + "grad_norm": 7.018968105316162, + "learning_rate": 0.00013660478143237746, + "loss": 3.2163, + "step": 2582 + }, + { + "epoch": 3.79, + "grad_norm": 4.448172569274902, + "learning_rate": 0.00013656112924847978, + "loss": 3.2027, + "step": 2583 + }, + { + "epoch": 3.79, + "grad_norm": 6.86191987991333, + "learning_rate": 0.00013651746902137868, + "loss": 3.1812, + "step": 2584 + }, + { + "epoch": 3.79, + "grad_norm": 4.056680202484131, + "learning_rate": 0.0001364738007606791, + "loss": 3.2857, + "step": 2585 + }, + { + "epoch": 3.79, + "grad_norm": 5.344080924987793, + "learning_rate": 0.0001364301244759878, + "loss": 3.255, + "step": 2586 + }, + { + "epoch": 3.79, + "grad_norm": 7.4425554275512695, + "learning_rate": 0.00013638644017691326, + "loss": 3.2032, + "step": 2587 + }, + { + "epoch": 3.8, + "grad_norm": 4.709530830383301, + "learning_rate": 0.00013634274787306577, + "loss": 3.25, + "step": 2588 + }, + { + "epoch": 3.8, + "grad_norm": 7.912161827087402, + "learning_rate": 0.00013629904757405723, + "loss": 3.2362, + "step": 2589 + }, + { + "epoch": 3.8, + "grad_norm": 24.812524795532227, + "learning_rate": 0.00013625533928950153, + "loss": 3.231, + "step": 2590 + }, + { + "epoch": 3.8, + "grad_norm": 9.660811424255371, + "learning_rate": 0.0001362116230290142, + "loss": 3.2175, + "step": 2591 + }, + { + "epoch": 3.8, + "grad_norm": 6.260350704193115, + "learning_rate": 0.00013616789880221242, + "loss": 3.2169, + "step": 2592 + }, + { + "epoch": 3.8, + "grad_norm": 2.6841564178466797, + "learning_rate": 0.00013612416661871533, + "loss": 3.2084, + "step": 2593 + }, + { + "epoch": 3.8, + "grad_norm": 6.153440475463867, + "learning_rate": 0.00013608042648814364, + "loss": 3.2029, + "step": 2594 + }, + { + "epoch": 3.81, + "grad_norm": 2.9339137077331543, + "learning_rate": 0.00013603667842011994, + "loss": 3.1789, + "step": 2595 + }, + { + "epoch": 3.81, + "grad_norm": 5.685370922088623, + "learning_rate": 0.00013599292242426851, + "loss": 3.2023, + "step": 2596 + }, + { + "epoch": 3.81, + "grad_norm": 12.19677734375, + "learning_rate": 0.00013594915851021532, + "loss": 3.2433, + "step": 2597 + }, + { + "epoch": 3.81, + "grad_norm": 5.506028652191162, + "learning_rate": 0.00013590538668758822, + "loss": 3.2088, + "step": 2598 + }, + { + "epoch": 3.81, + "grad_norm": 1.9152709245681763, + "learning_rate": 0.00013586160696601665, + "loss": 3.2067, + "step": 2599 + }, + { + "epoch": 3.81, + "grad_norm": 3.531026601791382, + "learning_rate": 0.00013581781935513188, + "loss": 3.2368, + "step": 2600 + }, + { + "epoch": 3.82, + "grad_norm": 1.68514883518219, + "learning_rate": 0.00013577402386456692, + "loss": 3.2567, + "step": 2601 + }, + { + "epoch": 3.82, + "grad_norm": 2.6391441822052, + "learning_rate": 0.00013573022050395643, + "loss": 3.1804, + "step": 2602 + }, + { + "epoch": 3.82, + "grad_norm": 2.4722888469696045, + "learning_rate": 0.0001356864092829369, + "loss": 3.2318, + "step": 2603 + }, + { + "epoch": 3.82, + "grad_norm": 1.5304458141326904, + "learning_rate": 0.0001356425902111465, + "loss": 3.2145, + "step": 2604 + }, + { + "epoch": 3.82, + "grad_norm": 4.6478705406188965, + "learning_rate": 0.0001355987632982251, + "loss": 3.2201, + "step": 2605 + }, + { + "epoch": 3.82, + "grad_norm": 3.80239200592041, + "learning_rate": 0.00013555492855381432, + "loss": 3.1819, + "step": 2606 + }, + { + "epoch": 3.82, + "grad_norm": 2.0871238708496094, + "learning_rate": 0.0001355110859875576, + "loss": 3.2169, + "step": 2607 + }, + { + "epoch": 3.83, + "grad_norm": 1.9589481353759766, + "learning_rate": 0.00013546723560909992, + "loss": 3.2077, + "step": 2608 + }, + { + "epoch": 3.83, + "grad_norm": 2.8086769580841064, + "learning_rate": 0.00013542337742808807, + "loss": 3.2386, + "step": 2609 + }, + { + "epoch": 3.83, + "grad_norm": 1.3195122480392456, + "learning_rate": 0.0001353795114541706, + "loss": 3.1651, + "step": 2610 + }, + { + "epoch": 3.83, + "grad_norm": 4.268999099731445, + "learning_rate": 0.00013533563769699774, + "loss": 3.2113, + "step": 2611 + }, + { + "epoch": 3.83, + "grad_norm": 2.7233781814575195, + "learning_rate": 0.00013529175616622133, + "loss": 3.2103, + "step": 2612 + }, + { + "epoch": 3.83, + "grad_norm": 3.8539392948150635, + "learning_rate": 0.00013524786687149512, + "loss": 3.2368, + "step": 2613 + }, + { + "epoch": 3.83, + "grad_norm": 3.89041805267334, + "learning_rate": 0.00013520396982247442, + "loss": 3.1865, + "step": 2614 + }, + { + "epoch": 3.84, + "grad_norm": 2.546916961669922, + "learning_rate": 0.0001351600650288163, + "loss": 3.192, + "step": 2615 + }, + { + "epoch": 3.84, + "grad_norm": 1.615480661392212, + "learning_rate": 0.0001351161525001795, + "loss": 3.2953, + "step": 2616 + }, + { + "epoch": 3.84, + "grad_norm": 3.4749913215637207, + "learning_rate": 0.00013507223224622445, + "loss": 3.1994, + "step": 2617 + }, + { + "epoch": 3.84, + "grad_norm": 1.6761186122894287, + "learning_rate": 0.00013502830427661338, + "loss": 3.1836, + "step": 2618 + }, + { + "epoch": 3.84, + "grad_norm": 6.091635704040527, + "learning_rate": 0.00013498436860101011, + "loss": 3.2183, + "step": 2619 + }, + { + "epoch": 3.84, + "grad_norm": 5.926030158996582, + "learning_rate": 0.00013494042522908022, + "loss": 3.2222, + "step": 2620 + }, + { + "epoch": 3.84, + "grad_norm": 12.00096321105957, + "learning_rate": 0.00013489647417049095, + "loss": 3.2073, + "step": 2621 + }, + { + "epoch": 3.85, + "grad_norm": 4.7124762535095215, + "learning_rate": 0.0001348525154349112, + "loss": 3.1859, + "step": 2622 + }, + { + "epoch": 3.85, + "grad_norm": 3.3024585247039795, + "learning_rate": 0.0001348085490320116, + "loss": 3.177, + "step": 2623 + }, + { + "epoch": 3.85, + "grad_norm": 4.006857395172119, + "learning_rate": 0.00013476457497146456, + "loss": 3.2317, + "step": 2624 + }, + { + "epoch": 3.85, + "grad_norm": 4.812809467315674, + "learning_rate": 0.00013472059326294396, + "loss": 3.2637, + "step": 2625 + }, + { + "epoch": 3.85, + "grad_norm": 1.2310017347335815, + "learning_rate": 0.00013467660391612555, + "loss": 3.2366, + "step": 2626 + }, + { + "epoch": 3.85, + "grad_norm": 2.896761894226074, + "learning_rate": 0.00013463260694068666, + "loss": 3.2069, + "step": 2627 + }, + { + "epoch": 3.86, + "grad_norm": 2.2707176208496094, + "learning_rate": 0.00013458860234630632, + "loss": 3.224, + "step": 2628 + }, + { + "epoch": 3.86, + "grad_norm": 2.2991745471954346, + "learning_rate": 0.00013454459014266526, + "loss": 3.2074, + "step": 2629 + }, + { + "epoch": 3.86, + "grad_norm": 2.838233470916748, + "learning_rate": 0.0001345005703394459, + "loss": 3.2019, + "step": 2630 + }, + { + "epoch": 3.86, + "grad_norm": 2.686645030975342, + "learning_rate": 0.0001344565429463323, + "loss": 3.1873, + "step": 2631 + }, + { + "epoch": 3.86, + "grad_norm": 3.610236883163452, + "learning_rate": 0.00013441250797301016, + "loss": 3.1667, + "step": 2632 + }, + { + "epoch": 3.86, + "grad_norm": 3.012873888015747, + "learning_rate": 0.00013436846542916686, + "loss": 3.2006, + "step": 2633 + }, + { + "epoch": 3.86, + "grad_norm": 2.223695755004883, + "learning_rate": 0.00013432441532449152, + "loss": 3.2004, + "step": 2634 + }, + { + "epoch": 3.87, + "grad_norm": 5.295112609863281, + "learning_rate": 0.00013428035766867484, + "loss": 3.1959, + "step": 2635 + }, + { + "epoch": 3.87, + "grad_norm": 1.9531642198562622, + "learning_rate": 0.00013423629247140919, + "loss": 3.1796, + "step": 2636 + }, + { + "epoch": 3.87, + "grad_norm": 11.508848190307617, + "learning_rate": 0.00013419221974238864, + "loss": 3.2325, + "step": 2637 + }, + { + "epoch": 3.87, + "grad_norm": 3.0934455394744873, + "learning_rate": 0.00013414813949130894, + "loss": 3.1904, + "step": 2638 + }, + { + "epoch": 3.87, + "grad_norm": 1.843153476715088, + "learning_rate": 0.00013410405172786734, + "loss": 3.177, + "step": 2639 + }, + { + "epoch": 3.87, + "grad_norm": 3.7216010093688965, + "learning_rate": 0.00013405995646176294, + "loss": 3.2278, + "step": 2640 + }, + { + "epoch": 3.87, + "grad_norm": 4.631261825561523, + "learning_rate": 0.00013401585370269637, + "loss": 3.2772, + "step": 2641 + }, + { + "epoch": 3.88, + "grad_norm": 2.928041696548462, + "learning_rate": 0.00013397174346036996, + "loss": 3.2489, + "step": 2642 + }, + { + "epoch": 3.88, + "grad_norm": 2.3705878257751465, + "learning_rate": 0.00013392762574448766, + "loss": 3.2577, + "step": 2643 + }, + { + "epoch": 3.88, + "grad_norm": 4.091426372528076, + "learning_rate": 0.00013388350056475504, + "loss": 3.1664, + "step": 2644 + }, + { + "epoch": 3.88, + "grad_norm": 1.9827549457550049, + "learning_rate": 0.0001338393679308794, + "loss": 3.1786, + "step": 2645 + }, + { + "epoch": 3.88, + "grad_norm": 4.402117729187012, + "learning_rate": 0.0001337952278525696, + "loss": 3.1966, + "step": 2646 + }, + { + "epoch": 3.88, + "grad_norm": 5.21814489364624, + "learning_rate": 0.00013375108033953614, + "loss": 3.2393, + "step": 2647 + }, + { + "epoch": 3.88, + "grad_norm": 2.4417850971221924, + "learning_rate": 0.0001337069254014912, + "loss": 3.1729, + "step": 2648 + }, + { + "epoch": 3.89, + "grad_norm": 3.521346092224121, + "learning_rate": 0.0001336627630481486, + "loss": 3.1997, + "step": 2649 + }, + { + "epoch": 3.89, + "grad_norm": 4.180556297302246, + "learning_rate": 0.0001336185932892237, + "loss": 3.1878, + "step": 2650 + }, + { + "epoch": 3.89, + "grad_norm": 2.869370460510254, + "learning_rate": 0.00013357441613443354, + "loss": 3.2052, + "step": 2651 + }, + { + "epoch": 3.89, + "grad_norm": 2.097275495529175, + "learning_rate": 0.0001335302315934969, + "loss": 3.1459, + "step": 2652 + }, + { + "epoch": 3.89, + "grad_norm": 3.70109486579895, + "learning_rate": 0.00013348603967613398, + "loss": 3.2187, + "step": 2653 + }, + { + "epoch": 3.89, + "grad_norm": 3.026033878326416, + "learning_rate": 0.00013344184039206677, + "loss": 3.2251, + "step": 2654 + }, + { + "epoch": 3.9, + "grad_norm": 2.825608015060425, + "learning_rate": 0.00013339763375101876, + "loss": 3.1845, + "step": 2655 + }, + { + "epoch": 3.9, + "grad_norm": 2.3940205574035645, + "learning_rate": 0.00013335341976271518, + "loss": 3.1909, + "step": 2656 + }, + { + "epoch": 3.9, + "grad_norm": 3.781881332397461, + "learning_rate": 0.00013330919843688275, + "loss": 3.1861, + "step": 2657 + }, + { + "epoch": 3.9, + "grad_norm": 3.5430831909179688, + "learning_rate": 0.0001332649697832499, + "loss": 3.2879, + "step": 2658 + }, + { + "epoch": 3.9, + "grad_norm": 3.327752113342285, + "learning_rate": 0.00013322073381154663, + "loss": 3.1407, + "step": 2659 + }, + { + "epoch": 3.9, + "grad_norm": 3.319768190383911, + "learning_rate": 0.00013317649053150456, + "loss": 3.2215, + "step": 2660 + }, + { + "epoch": 3.9, + "grad_norm": 2.317915201187134, + "learning_rate": 0.0001331322399528569, + "loss": 3.251, + "step": 2661 + }, + { + "epoch": 3.91, + "grad_norm": 2.999483108520508, + "learning_rate": 0.00013308798208533853, + "loss": 3.2182, + "step": 2662 + }, + { + "epoch": 3.91, + "grad_norm": 4.7933549880981445, + "learning_rate": 0.0001330437169386858, + "loss": 3.2272, + "step": 2663 + }, + { + "epoch": 3.91, + "grad_norm": 4.440158367156982, + "learning_rate": 0.00013299944452263682, + "loss": 3.202, + "step": 2664 + }, + { + "epoch": 3.91, + "grad_norm": 1.8913938999176025, + "learning_rate": 0.0001329551648469312, + "loss": 3.1705, + "step": 2665 + }, + { + "epoch": 3.91, + "grad_norm": 4.039271831512451, + "learning_rate": 0.00013291087792131017, + "loss": 3.213, + "step": 2666 + }, + { + "epoch": 3.91, + "grad_norm": 4.9916276931762695, + "learning_rate": 0.00013286658375551654, + "loss": 3.2316, + "step": 2667 + }, + { + "epoch": 3.91, + "grad_norm": 2.624239206314087, + "learning_rate": 0.00013282228235929474, + "loss": 3.1902, + "step": 2668 + }, + { + "epoch": 3.92, + "grad_norm": 2.3713390827178955, + "learning_rate": 0.0001327779737423908, + "loss": 3.1698, + "step": 2669 + }, + { + "epoch": 3.92, + "grad_norm": 3.9648921489715576, + "learning_rate": 0.0001327336579145523, + "loss": 3.1811, + "step": 2670 + }, + { + "epoch": 3.92, + "grad_norm": 3.6182374954223633, + "learning_rate": 0.00013268933488552845, + "loss": 3.2013, + "step": 2671 + }, + { + "epoch": 3.92, + "grad_norm": 1.5024638175964355, + "learning_rate": 0.00013264500466506998, + "loss": 3.1996, + "step": 2672 + }, + { + "epoch": 3.92, + "grad_norm": 1.957371473312378, + "learning_rate": 0.00013260066726292926, + "loss": 3.188, + "step": 2673 + }, + { + "epoch": 3.92, + "grad_norm": 1.9777625799179077, + "learning_rate": 0.00013255632268886023, + "loss": 3.2656, + "step": 2674 + }, + { + "epoch": 3.93, + "grad_norm": 1.5048279762268066, + "learning_rate": 0.0001325119709526184, + "loss": 3.2279, + "step": 2675 + }, + { + "epoch": 3.93, + "grad_norm": 35.774295806884766, + "learning_rate": 0.0001324676120639608, + "loss": 3.2366, + "step": 2676 + }, + { + "epoch": 3.93, + "grad_norm": 3.9443695545196533, + "learning_rate": 0.00013242324603264616, + "loss": 3.1658, + "step": 2677 + }, + { + "epoch": 3.93, + "grad_norm": 2.9231414794921875, + "learning_rate": 0.0001323788728684347, + "loss": 3.2058, + "step": 2678 + }, + { + "epoch": 3.93, + "grad_norm": 4.9291157722473145, + "learning_rate": 0.0001323344925810882, + "loss": 3.1528, + "step": 2679 + }, + { + "epoch": 3.93, + "grad_norm": 8.999177932739258, + "learning_rate": 0.00013229010518037002, + "loss": 3.2864, + "step": 2680 + }, + { + "epoch": 3.93, + "grad_norm": 4.351472854614258, + "learning_rate": 0.0001322457106760451, + "loss": 3.2048, + "step": 2681 + }, + { + "epoch": 3.94, + "grad_norm": 9.449458122253418, + "learning_rate": 0.00013220130907787994, + "loss": 3.2409, + "step": 2682 + }, + { + "epoch": 3.94, + "grad_norm": 2.7245359420776367, + "learning_rate": 0.00013215690039564257, + "loss": 3.1844, + "step": 2683 + }, + { + "epoch": 3.94, + "grad_norm": 1.7337415218353271, + "learning_rate": 0.00013211248463910262, + "loss": 3.1732, + "step": 2684 + }, + { + "epoch": 3.94, + "grad_norm": 2.845367431640625, + "learning_rate": 0.0001320680618180313, + "loss": 3.1608, + "step": 2685 + }, + { + "epoch": 3.94, + "grad_norm": 2.7872724533081055, + "learning_rate": 0.00013202363194220125, + "loss": 3.2057, + "step": 2686 + }, + { + "epoch": 3.94, + "grad_norm": 1.4545655250549316, + "learning_rate": 0.00013197919502138677, + "loss": 3.205, + "step": 2687 + }, + { + "epoch": 3.94, + "grad_norm": 3.010228157043457, + "learning_rate": 0.00013193475106536374, + "loss": 3.1844, + "step": 2688 + }, + { + "epoch": 3.95, + "grad_norm": 2.508854866027832, + "learning_rate": 0.00013189030008390945, + "loss": 3.2108, + "step": 2689 + }, + { + "epoch": 3.95, + "grad_norm": 3.1477272510528564, + "learning_rate": 0.0001318458420868029, + "loss": 3.2018, + "step": 2690 + }, + { + "epoch": 3.95, + "grad_norm": 2.543921947479248, + "learning_rate": 0.0001318013770838245, + "loss": 3.2235, + "step": 2691 + }, + { + "epoch": 3.95, + "grad_norm": 3.577669382095337, + "learning_rate": 0.00013175690508475626, + "loss": 3.2087, + "step": 2692 + }, + { + "epoch": 3.95, + "grad_norm": 2.2817633152008057, + "learning_rate": 0.00013171242609938175, + "loss": 3.1847, + "step": 2693 + }, + { + "epoch": 3.95, + "grad_norm": 7.939913272857666, + "learning_rate": 0.00013166794013748598, + "loss": 3.2269, + "step": 2694 + }, + { + "epoch": 3.95, + "grad_norm": 6.408430576324463, + "learning_rate": 0.00013162344720885565, + "loss": 3.1853, + "step": 2695 + }, + { + "epoch": 3.96, + "grad_norm": 95.63408660888672, + "learning_rate": 0.00013157894732327886, + "loss": 3.2441, + "step": 2696 + }, + { + "epoch": 3.96, + "grad_norm": 76.839111328125, + "learning_rate": 0.00013153444049054523, + "loss": 3.2624, + "step": 2697 + }, + { + "epoch": 3.96, + "grad_norm": 9.534677505493164, + "learning_rate": 0.00013148992672044608, + "loss": 3.2354, + "step": 2698 + }, + { + "epoch": 3.96, + "grad_norm": 29.769962310791016, + "learning_rate": 0.0001314454060227741, + "loss": 3.1857, + "step": 2699 + }, + { + "epoch": 3.96, + "grad_norm": 26.48052215576172, + "learning_rate": 0.00013140087840732355, + "loss": 3.2512, + "step": 2700 + }, + { + "epoch": 3.96, + "grad_norm": 12.384693145751953, + "learning_rate": 0.00013135634388389015, + "loss": 3.2054, + "step": 2701 + }, + { + "epoch": 3.97, + "grad_norm": 4.66050910949707, + "learning_rate": 0.00013131180246227125, + "loss": 3.2012, + "step": 2702 + }, + { + "epoch": 3.97, + "grad_norm": 28.54119300842285, + "learning_rate": 0.00013126725415226563, + "loss": 3.202, + "step": 2703 + }, + { + "epoch": 3.97, + "grad_norm": 69.08728790283203, + "learning_rate": 0.00013122269896367365, + "loss": 3.2423, + "step": 2704 + }, + { + "epoch": 3.97, + "eval_loss": 3.091963529586792, + "eval_runtime": 1165.9245, + "eval_samples_per_second": 14.702, + "eval_steps_per_second": 0.919, + "step": 2704 + }, + { + "epoch": 3.97, + "grad_norm": 22.657764434814453, + "learning_rate": 0.00013117813690629714, + "loss": 3.2586, + "step": 2705 + }, + { + "epoch": 3.97, + "grad_norm": 74.64742279052734, + "learning_rate": 0.00013113356798993947, + "loss": 3.2606, + "step": 2706 + }, + { + "epoch": 3.97, + "grad_norm": 14.374595642089844, + "learning_rate": 0.0001310889922244055, + "loss": 3.2525, + "step": 2707 + }, + { + "epoch": 3.97, + "grad_norm": 5.337686538696289, + "learning_rate": 0.00013104440961950154, + "loss": 3.2355, + "step": 2708 + }, + { + "epoch": 3.98, + "grad_norm": 3.205595016479492, + "learning_rate": 0.00013099982018503558, + "loss": 3.2323, + "step": 2709 + }, + { + "epoch": 3.98, + "grad_norm": 4.343560218811035, + "learning_rate": 0.0001309552239308169, + "loss": 3.2265, + "step": 2710 + }, + { + "epoch": 3.98, + "grad_norm": 68.85613250732422, + "learning_rate": 0.0001309106208666564, + "loss": 3.2412, + "step": 2711 + }, + { + "epoch": 3.98, + "grad_norm": 4.4929633140563965, + "learning_rate": 0.00013086601100236646, + "loss": 3.2243, + "step": 2712 + }, + { + "epoch": 3.98, + "grad_norm": 5.574833393096924, + "learning_rate": 0.000130821394347761, + "loss": 3.2, + "step": 2713 + }, + { + "epoch": 3.98, + "grad_norm": 2.189110040664673, + "learning_rate": 0.00013077677091265528, + "loss": 3.2177, + "step": 2714 + }, + { + "epoch": 3.98, + "grad_norm": 1.2807992696762085, + "learning_rate": 0.00013073214070686626, + "loss": 3.1971, + "step": 2715 + }, + { + "epoch": 3.99, + "grad_norm": 3.459458351135254, + "learning_rate": 0.0001306875037402122, + "loss": 3.186, + "step": 2716 + }, + { + "epoch": 3.99, + "grad_norm": 9.210981369018555, + "learning_rate": 0.000130642860022513, + "loss": 3.1966, + "step": 2717 + }, + { + "epoch": 3.99, + "grad_norm": 2.803443670272827, + "learning_rate": 0.00013059820956358998, + "loss": 3.1927, + "step": 2718 + }, + { + "epoch": 3.99, + "grad_norm": 2.078461170196533, + "learning_rate": 0.00013055355237326588, + "loss": 3.1776, + "step": 2719 + }, + { + "epoch": 3.99, + "grad_norm": 2.4591166973114014, + "learning_rate": 0.00013050888846136502, + "loss": 3.2126, + "step": 2720 + }, + { + "epoch": 3.99, + "grad_norm": 2.6859610080718994, + "learning_rate": 0.00013046421783771318, + "loss": 3.1956, + "step": 2721 + }, + { + "epoch": 3.99, + "grad_norm": 3.064992904663086, + "learning_rate": 0.00013041954051213757, + "loss": 3.25, + "step": 2722 + }, + { + "epoch": 4.0, + "grad_norm": 1.7159329652786255, + "learning_rate": 0.0001303748564944669, + "loss": 3.1734, + "step": 2723 + }, + { + "epoch": 4.0, + "grad_norm": 2.7244114875793457, + "learning_rate": 0.00013033016579453138, + "loss": 3.1743, + "step": 2724 + }, + { + "epoch": 4.0, + "grad_norm": 2.2309274673461914, + "learning_rate": 0.0001302854684221626, + "loss": 3.1751, + "step": 2725 + }, + { + "epoch": 4.0, + "grad_norm": 1.5704296827316284, + "learning_rate": 0.0001302407643871938, + "loss": 3.2018, + "step": 2726 + }, + { + "epoch": 4.0, + "grad_norm": 2.5336227416992188, + "learning_rate": 0.0001301960536994595, + "loss": 3.2139, + "step": 2727 + }, + { + "epoch": 4.0, + "grad_norm": 1.6835451126098633, + "learning_rate": 0.00013015133636879567, + "loss": 3.2198, + "step": 2728 + }, + { + "epoch": 4.01, + "grad_norm": 1.5716129541397095, + "learning_rate": 0.00013010661240503999, + "loss": 3.2108, + "step": 2729 + }, + { + "epoch": 4.01, + "grad_norm": 1.5545756816864014, + "learning_rate": 0.0001300618818180313, + "loss": 3.1592, + "step": 2730 + }, + { + "epoch": 4.01, + "grad_norm": 1.9438145160675049, + "learning_rate": 0.00013001714461761008, + "loss": 3.2145, + "step": 2731 + }, + { + "epoch": 4.01, + "grad_norm": 1.7333067655563354, + "learning_rate": 0.00012997240081361823, + "loss": 3.1362, + "step": 2732 + }, + { + "epoch": 4.01, + "grad_norm": 1.632812261581421, + "learning_rate": 0.00012992765041589904, + "loss": 3.2064, + "step": 2733 + }, + { + "epoch": 4.01, + "grad_norm": 1.1884607076644897, + "learning_rate": 0.00012988289343429733, + "loss": 3.1789, + "step": 2734 + }, + { + "epoch": 4.01, + "grad_norm": 2.9912478923797607, + "learning_rate": 0.00012983812987865935, + "loss": 3.1561, + "step": 2735 + }, + { + "epoch": 4.02, + "grad_norm": 1.3816310167312622, + "learning_rate": 0.00012979335975883276, + "loss": 3.1848, + "step": 2736 + }, + { + "epoch": 4.0, + "grad_norm": 19.01185417175293, + "learning_rate": 0.00012974858308466666, + "loss": 3.1801, + "step": 2737 + }, + { + "epoch": 4.0, + "grad_norm": 5.716503143310547, + "learning_rate": 0.0001297037998660117, + "loss": 3.2216, + "step": 2738 + }, + { + "epoch": 4.0, + "grad_norm": 2.985485792160034, + "learning_rate": 0.00012965901011271978, + "loss": 3.2136, + "step": 2739 + }, + { + "epoch": 4.01, + "grad_norm": 2.645477056503296, + "learning_rate": 0.00012961421383464445, + "loss": 3.176, + "step": 2740 + }, + { + "epoch": 4.01, + "grad_norm": 2.804504871368408, + "learning_rate": 0.0001295694110416405, + "loss": 3.1502, + "step": 2741 + }, + { + "epoch": 4.01, + "grad_norm": 2.257525682449341, + "learning_rate": 0.00012952460174356432, + "loss": 3.1909, + "step": 2742 + }, + { + "epoch": 4.01, + "grad_norm": 2.525448799133301, + "learning_rate": 0.0001294797859502736, + "loss": 3.2058, + "step": 2743 + }, + { + "epoch": 4.01, + "grad_norm": 2.589095115661621, + "learning_rate": 0.00012943496367162754, + "loss": 3.2137, + "step": 2744 + }, + { + "epoch": 4.01, + "grad_norm": 1.6353000402450562, + "learning_rate": 0.00012939013491748672, + "loss": 3.2001, + "step": 2745 + }, + { + "epoch": 4.01, + "grad_norm": 2.5494680404663086, + "learning_rate": 0.00012934529969771324, + "loss": 3.1529, + "step": 2746 + }, + { + "epoch": 4.02, + "grad_norm": 2.1689844131469727, + "learning_rate": 0.00012930045802217043, + "loss": 3.1931, + "step": 2747 + }, + { + "epoch": 4.02, + "grad_norm": 3.0633785724639893, + "learning_rate": 0.00012925560990072324, + "loss": 3.1941, + "step": 2748 + }, + { + "epoch": 4.02, + "grad_norm": 2.6445438861846924, + "learning_rate": 0.00012921075534323795, + "loss": 3.1673, + "step": 2749 + }, + { + "epoch": 4.02, + "grad_norm": 1.6328538656234741, + "learning_rate": 0.00012916589435958223, + "loss": 3.1533, + "step": 2750 + }, + { + "epoch": 4.02, + "grad_norm": 7.05519962310791, + "learning_rate": 0.0001291210269596252, + "loss": 3.1762, + "step": 2751 + }, + { + "epoch": 4.02, + "grad_norm": 10.450418472290039, + "learning_rate": 0.0001290761531532374, + "loss": 3.2032, + "step": 2752 + }, + { + "epoch": 4.02, + "grad_norm": 3.726336717605591, + "learning_rate": 0.00012903127295029078, + "loss": 3.1835, + "step": 2753 + }, + { + "epoch": 4.03, + "grad_norm": 2.4860239028930664, + "learning_rate": 0.00012898638636065868, + "loss": 3.1818, + "step": 2754 + }, + { + "epoch": 4.03, + "grad_norm": 2.166196584701538, + "learning_rate": 0.00012894149339421582, + "loss": 3.1609, + "step": 2755 + }, + { + "epoch": 4.03, + "grad_norm": 2.179659128189087, + "learning_rate": 0.00012889659406083837, + "loss": 3.2125, + "step": 2756 + }, + { + "epoch": 4.03, + "grad_norm": 2.2733445167541504, + "learning_rate": 0.0001288516883704039, + "loss": 3.2931, + "step": 2757 + }, + { + "epoch": 4.03, + "grad_norm": 1.3774925470352173, + "learning_rate": 0.00012880677633279132, + "loss": 3.1741, + "step": 2758 + }, + { + "epoch": 4.03, + "grad_norm": 4.090391159057617, + "learning_rate": 0.000128761857957881, + "loss": 3.2181, + "step": 2759 + }, + { + "epoch": 4.04, + "grad_norm": 3.762101411819458, + "learning_rate": 0.00012871693325555472, + "loss": 3.1991, + "step": 2760 + }, + { + "epoch": 4.04, + "grad_norm": 2.435084819793701, + "learning_rate": 0.00012867200223569555, + "loss": 3.1734, + "step": 2761 + }, + { + "epoch": 4.04, + "grad_norm": 11.30912971496582, + "learning_rate": 0.00012862706490818804, + "loss": 3.1755, + "step": 2762 + }, + { + "epoch": 4.04, + "grad_norm": 3.355214834213257, + "learning_rate": 0.0001285821212829181, + "loss": 3.1718, + "step": 2763 + }, + { + "epoch": 4.04, + "grad_norm": 2.4179513454437256, + "learning_rate": 0.00012853717136977305, + "loss": 3.1722, + "step": 2764 + }, + { + "epoch": 4.04, + "grad_norm": 2.18635630607605, + "learning_rate": 0.0001284922151786415, + "loss": 3.1982, + "step": 2765 + }, + { + "epoch": 4.04, + "grad_norm": 2.3029189109802246, + "learning_rate": 0.0001284472527194136, + "loss": 3.2427, + "step": 2766 + }, + { + "epoch": 4.05, + "grad_norm": 2.3949828147888184, + "learning_rate": 0.00012840228400198071, + "loss": 3.1944, + "step": 2767 + }, + { + "epoch": 4.05, + "grad_norm": 2.5206806659698486, + "learning_rate": 0.0001283573090362357, + "loss": 3.1899, + "step": 2768 + }, + { + "epoch": 4.05, + "grad_norm": 1.8850032091140747, + "learning_rate": 0.00012831232783207277, + "loss": 3.2212, + "step": 2769 + }, + { + "epoch": 4.05, + "grad_norm": 2.58017897605896, + "learning_rate": 0.0001282673403993874, + "loss": 3.2026, + "step": 2770 + }, + { + "epoch": 4.05, + "grad_norm": 2.990920305252075, + "learning_rate": 0.00012822234674807662, + "loss": 3.1852, + "step": 2771 + }, + { + "epoch": 4.05, + "grad_norm": 0.5992396473884583, + "learning_rate": 0.00012817734688803868, + "loss": 3.1736, + "step": 2772 + }, + { + "epoch": 4.05, + "grad_norm": 2.821469306945801, + "learning_rate": 0.0001281323408291733, + "loss": 3.1999, + "step": 2773 + }, + { + "epoch": 4.06, + "grad_norm": 15.83332633972168, + "learning_rate": 0.00012808732858138146, + "loss": 3.1844, + "step": 2774 + }, + { + "epoch": 4.06, + "grad_norm": 2.910820960998535, + "learning_rate": 0.0001280423101545655, + "loss": 3.1587, + "step": 2775 + }, + { + "epoch": 4.06, + "grad_norm": 17.933792114257812, + "learning_rate": 0.00012799728555862934, + "loss": 3.2201, + "step": 2776 + }, + { + "epoch": 4.06, + "grad_norm": 2.721127986907959, + "learning_rate": 0.0001279522548034779, + "loss": 3.1507, + "step": 2777 + }, + { + "epoch": 4.06, + "grad_norm": 1.5368844270706177, + "learning_rate": 0.00012790721789901775, + "loss": 3.1902, + "step": 2778 + }, + { + "epoch": 4.06, + "grad_norm": 3.0113532543182373, + "learning_rate": 0.00012786217485515674, + "loss": 3.149, + "step": 2779 + }, + { + "epoch": 4.06, + "grad_norm": 19.78055191040039, + "learning_rate": 0.00012781712568180393, + "loss": 3.1755, + "step": 2780 + }, + { + "epoch": 4.07, + "grad_norm": 2.508270025253296, + "learning_rate": 0.0001277720703888699, + "loss": 3.1915, + "step": 2781 + }, + { + "epoch": 4.07, + "grad_norm": 2.167954206466675, + "learning_rate": 0.00012772700898626652, + "loss": 3.2054, + "step": 2782 + }, + { + "epoch": 4.07, + "grad_norm": 4.016536235809326, + "learning_rate": 0.00012768194148390696, + "loss": 3.195, + "step": 2783 + }, + { + "epoch": 4.07, + "grad_norm": 3.441601276397705, + "learning_rate": 0.0001276368678917058, + "loss": 3.15, + "step": 2784 + }, + { + "epoch": 4.07, + "grad_norm": 7.370872974395752, + "learning_rate": 0.0001275917882195789, + "loss": 3.1633, + "step": 2785 + }, + { + "epoch": 4.07, + "grad_norm": 2.296081781387329, + "learning_rate": 0.00012754670247744354, + "loss": 3.2345, + "step": 2786 + }, + { + "epoch": 4.08, + "grad_norm": 2.009742259979248, + "learning_rate": 0.00012750161067521826, + "loss": 3.1938, + "step": 2787 + }, + { + "epoch": 4.08, + "grad_norm": 4.299360752105713, + "learning_rate": 0.00012745651282282288, + "loss": 3.2213, + "step": 2788 + }, + { + "epoch": 4.08, + "grad_norm": 3.636892080307007, + "learning_rate": 0.00012741140893017872, + "loss": 3.1713, + "step": 2789 + }, + { + "epoch": 4.08, + "grad_norm": 1.2036387920379639, + "learning_rate": 0.0001273662990072083, + "loss": 3.2107, + "step": 2790 + }, + { + "epoch": 4.08, + "grad_norm": 3.7308552265167236, + "learning_rate": 0.0001273211830638355, + "loss": 3.1955, + "step": 2791 + }, + { + "epoch": 4.08, + "grad_norm": 9.248310089111328, + "learning_rate": 0.00012727606110998549, + "loss": 3.1917, + "step": 2792 + }, + { + "epoch": 4.08, + "grad_norm": 15.789203643798828, + "learning_rate": 0.00012723093315558488, + "loss": 3.1611, + "step": 2793 + }, + { + "epoch": 4.09, + "grad_norm": 3.4315154552459717, + "learning_rate": 0.00012718579921056143, + "loss": 3.2364, + "step": 2794 + }, + { + "epoch": 4.09, + "grad_norm": 2.061835289001465, + "learning_rate": 0.00012714065928484436, + "loss": 3.2079, + "step": 2795 + }, + { + "epoch": 4.09, + "grad_norm": 2.4200713634490967, + "learning_rate": 0.00012709551338836413, + "loss": 3.1846, + "step": 2796 + }, + { + "epoch": 4.09, + "grad_norm": 3.561788558959961, + "learning_rate": 0.00012705036153105257, + "loss": 3.2151, + "step": 2797 + }, + { + "epoch": 4.09, + "grad_norm": 2.967207908630371, + "learning_rate": 0.00012700520372284273, + "loss": 3.2036, + "step": 2798 + }, + { + "epoch": 4.09, + "grad_norm": 3.490912437438965, + "learning_rate": 0.00012696003997366906, + "loss": 3.1893, + "step": 2799 + }, + { + "epoch": 4.09, + "grad_norm": 2.477809429168701, + "learning_rate": 0.00012691487029346728, + "loss": 3.1701, + "step": 2800 + }, + { + "epoch": 4.1, + "grad_norm": 3.558258295059204, + "learning_rate": 0.0001268696946921744, + "loss": 3.2343, + "step": 2801 + }, + { + "epoch": 4.1, + "grad_norm": 3.1721646785736084, + "learning_rate": 0.00012682451317972874, + "loss": 3.201, + "step": 2802 + }, + { + "epoch": 4.1, + "grad_norm": 2.562236785888672, + "learning_rate": 0.00012677932576606998, + "loss": 3.2291, + "step": 2803 + }, + { + "epoch": 4.1, + "grad_norm": 9.012580871582031, + "learning_rate": 0.00012673413246113905, + "loss": 3.2074, + "step": 2804 + }, + { + "epoch": 4.1, + "grad_norm": 2.1724753379821777, + "learning_rate": 0.00012668893327487805, + "loss": 3.2104, + "step": 2805 + }, + { + "epoch": 4.1, + "grad_norm": 1.8097708225250244, + "learning_rate": 0.00012664372821723064, + "loss": 3.1534, + "step": 2806 + }, + { + "epoch": 4.11, + "grad_norm": 2.409672498703003, + "learning_rate": 0.00012659851729814162, + "loss": 3.1748, + "step": 2807 + }, + { + "epoch": 4.11, + "grad_norm": 2.8095180988311768, + "learning_rate": 0.000126553300527557, + "loss": 3.1996, + "step": 2808 + }, + { + "epoch": 4.11, + "grad_norm": 2.353957176208496, + "learning_rate": 0.00012650807791542422, + "loss": 3.1945, + "step": 2809 + }, + { + "epoch": 4.11, + "grad_norm": 3.013678789138794, + "learning_rate": 0.00012646284947169195, + "loss": 3.1839, + "step": 2810 + }, + { + "epoch": 4.11, + "grad_norm": 2.154859781265259, + "learning_rate": 0.00012641761520631016, + "loss": 3.1541, + "step": 2811 + }, + { + "epoch": 4.11, + "grad_norm": 2.0929133892059326, + "learning_rate": 0.00012637237512923006, + "loss": 3.1655, + "step": 2812 + }, + { + "epoch": 4.11, + "grad_norm": 1.2277342081069946, + "learning_rate": 0.0001263271292504042, + "loss": 3.1933, + "step": 2813 + }, + { + "epoch": 4.12, + "grad_norm": 2.6938316822052, + "learning_rate": 0.0001262818775797863, + "loss": 3.1441, + "step": 2814 + }, + { + "epoch": 4.12, + "grad_norm": 1.9652117490768433, + "learning_rate": 0.00012623662012733146, + "loss": 3.1616, + "step": 2815 + }, + { + "epoch": 4.12, + "grad_norm": 1.6476244926452637, + "learning_rate": 0.00012619135690299603, + "loss": 3.1839, + "step": 2816 + }, + { + "epoch": 4.12, + "grad_norm": 1.534227967262268, + "learning_rate": 0.00012614608791673763, + "loss": 3.1646, + "step": 2817 + }, + { + "epoch": 4.12, + "grad_norm": 2.850703477859497, + "learning_rate": 0.00012610081317851511, + "loss": 3.207, + "step": 2818 + }, + { + "epoch": 4.12, + "grad_norm": 2.1084752082824707, + "learning_rate": 0.00012605553269828858, + "loss": 3.1688, + "step": 2819 + }, + { + "epoch": 4.12, + "grad_norm": 2.303504705429077, + "learning_rate": 0.0001260102464860195, + "loss": 3.229, + "step": 2820 + }, + { + "epoch": 4.13, + "grad_norm": 1.5470285415649414, + "learning_rate": 0.0001259649545516705, + "loss": 3.1814, + "step": 2821 + }, + { + "epoch": 4.13, + "grad_norm": 1.4991244077682495, + "learning_rate": 0.0001259196569052055, + "loss": 3.1927, + "step": 2822 + }, + { + "epoch": 4.13, + "grad_norm": 1.342314600944519, + "learning_rate": 0.0001258743535565897, + "loss": 3.1743, + "step": 2823 + }, + { + "epoch": 4.13, + "grad_norm": 2.662546396255493, + "learning_rate": 0.0001258290445157895, + "loss": 3.1884, + "step": 2824 + }, + { + "epoch": 4.13, + "grad_norm": 1.1698952913284302, + "learning_rate": 0.00012578372979277265, + "loss": 3.1841, + "step": 2825 + }, + { + "epoch": 4.13, + "grad_norm": 2.8144288063049316, + "learning_rate": 0.000125738409397508, + "loss": 3.1634, + "step": 2826 + }, + { + "epoch": 4.13, + "grad_norm": 1.0438013076782227, + "learning_rate": 0.00012569308333996578, + "loss": 3.1828, + "step": 2827 + }, + { + "epoch": 4.14, + "grad_norm": 5.697520732879639, + "learning_rate": 0.0001256477516301174, + "loss": 3.1897, + "step": 2828 + }, + { + "epoch": 4.14, + "grad_norm": 6.039661884307861, + "learning_rate": 0.00012560241427793555, + "loss": 3.2308, + "step": 2829 + }, + { + "epoch": 4.14, + "grad_norm": 51.954402923583984, + "learning_rate": 0.00012555707129339417, + "loss": 3.1693, + "step": 2830 + }, + { + "epoch": 4.14, + "grad_norm": 5.388270378112793, + "learning_rate": 0.00012551172268646834, + "loss": 3.2463, + "step": 2831 + }, + { + "epoch": 4.14, + "grad_norm": 4.189896583557129, + "learning_rate": 0.00012546636846713454, + "loss": 3.209, + "step": 2832 + }, + { + "epoch": 4.14, + "grad_norm": 2.421431064605713, + "learning_rate": 0.0001254210086453703, + "loss": 3.2312, + "step": 2833 + }, + { + "epoch": 4.15, + "grad_norm": 4.315710544586182, + "learning_rate": 0.00012537564323115454, + "loss": 3.2011, + "step": 2834 + }, + { + "epoch": 4.15, + "grad_norm": 1.7199636697769165, + "learning_rate": 0.00012533027223446733, + "loss": 3.1634, + "step": 2835 + }, + { + "epoch": 4.15, + "grad_norm": 4.567407608032227, + "learning_rate": 0.00012528489566528996, + "loss": 3.1999, + "step": 2836 + }, + { + "epoch": 4.15, + "grad_norm": 5.22501277923584, + "learning_rate": 0.00012523951353360504, + "loss": 3.1743, + "step": 2837 + }, + { + "epoch": 4.15, + "grad_norm": 2.197282314300537, + "learning_rate": 0.00012519412584939627, + "loss": 3.2332, + "step": 2838 + }, + { + "epoch": 4.15, + "grad_norm": 5.91029167175293, + "learning_rate": 0.00012514873262264864, + "loss": 3.1713, + "step": 2839 + }, + { + "epoch": 4.15, + "grad_norm": 6.8516716957092285, + "learning_rate": 0.0001251033338633484, + "loss": 3.1908, + "step": 2840 + }, + { + "epoch": 4.16, + "grad_norm": 3.321195125579834, + "learning_rate": 0.00012505792958148292, + "loss": 3.1693, + "step": 2841 + }, + { + "epoch": 4.16, + "grad_norm": 11.536995887756348, + "learning_rate": 0.00012501251978704088, + "loss": 3.1626, + "step": 2842 + }, + { + "epoch": 4.16, + "grad_norm": 9.267814636230469, + "learning_rate": 0.0001249671044900121, + "loss": 3.1807, + "step": 2843 + }, + { + "epoch": 4.16, + "grad_norm": 8.332723617553711, + "learning_rate": 0.00012492168370038765, + "loss": 3.1934, + "step": 2844 + }, + { + "epoch": 4.16, + "grad_norm": 3.4367361068725586, + "learning_rate": 0.00012487625742815985, + "loss": 3.1797, + "step": 2845 + }, + { + "epoch": 4.16, + "grad_norm": 6.636847019195557, + "learning_rate": 0.00012483082568332207, + "loss": 3.1851, + "step": 2846 + }, + { + "epoch": 4.16, + "grad_norm": 8.669164657592773, + "learning_rate": 0.00012478538847586907, + "loss": 3.2061, + "step": 2847 + }, + { + "epoch": 4.17, + "grad_norm": 6.34977912902832, + "learning_rate": 0.00012473994581579673, + "loss": 3.196, + "step": 2848 + }, + { + "epoch": 4.17, + "grad_norm": 4.2629780769348145, + "learning_rate": 0.00012469449771310208, + "loss": 3.2128, + "step": 2849 + }, + { + "epoch": 4.17, + "grad_norm": 5.895789623260498, + "learning_rate": 0.00012464904417778345, + "loss": 3.2769, + "step": 2850 + }, + { + "epoch": 4.17, + "grad_norm": 7.333654403686523, + "learning_rate": 0.0001246035852198403, + "loss": 3.2108, + "step": 2851 + }, + { + "epoch": 4.17, + "grad_norm": 2.965632915496826, + "learning_rate": 0.00012455812084927325, + "loss": 3.2012, + "step": 2852 + }, + { + "epoch": 4.17, + "grad_norm": 8.51198673248291, + "learning_rate": 0.0001245126510760842, + "loss": 3.2149, + "step": 2853 + }, + { + "epoch": 4.17, + "grad_norm": 7.021212100982666, + "learning_rate": 0.00012446717591027624, + "loss": 3.2624, + "step": 2854 + }, + { + "epoch": 4.18, + "grad_norm": 1.8178749084472656, + "learning_rate": 0.0001244216953618535, + "loss": 3.2099, + "step": 2855 + }, + { + "epoch": 4.18, + "grad_norm": 11.300747871398926, + "learning_rate": 0.00012437620944082145, + "loss": 3.1795, + "step": 2856 + }, + { + "epoch": 4.18, + "grad_norm": 9.994318962097168, + "learning_rate": 0.0001243307181571867, + "loss": 3.2285, + "step": 2857 + }, + { + "epoch": 4.18, + "grad_norm": 11.500404357910156, + "learning_rate": 0.00012428522152095703, + "loss": 3.1869, + "step": 2858 + }, + { + "epoch": 4.18, + "grad_norm": 5.326147556304932, + "learning_rate": 0.00012423971954214137, + "loss": 3.2226, + "step": 2859 + }, + { + "epoch": 4.18, + "grad_norm": 5.258158206939697, + "learning_rate": 0.00012419421223074982, + "loss": 3.1848, + "step": 2860 + }, + { + "epoch": 4.19, + "grad_norm": 7.557288646697998, + "learning_rate": 0.00012414869959679375, + "loss": 3.2594, + "step": 2861 + }, + { + "epoch": 4.19, + "grad_norm": 19.032909393310547, + "learning_rate": 0.0001241031816502856, + "loss": 3.2464, + "step": 2862 + }, + { + "epoch": 4.19, + "grad_norm": 9.38792896270752, + "learning_rate": 0.00012405765840123903, + "loss": 3.4169, + "step": 2863 + }, + { + "epoch": 4.19, + "grad_norm": 58.51255798339844, + "learning_rate": 0.0001240121298596688, + "loss": 3.3102, + "step": 2864 + }, + { + "epoch": 4.19, + "grad_norm": 9.871623992919922, + "learning_rate": 0.00012396659603559099, + "loss": 3.3069, + "step": 2865 + }, + { + "epoch": 4.19, + "grad_norm": 18.70452117919922, + "learning_rate": 0.00012392105693902262, + "loss": 3.2645, + "step": 2866 + }, + { + "epoch": 4.19, + "grad_norm": 6.495476245880127, + "learning_rate": 0.00012387551257998206, + "loss": 3.2188, + "step": 2867 + }, + { + "epoch": 4.2, + "grad_norm": 4.516813278198242, + "learning_rate": 0.00012382996296848875, + "loss": 3.2188, + "step": 2868 + }, + { + "epoch": 4.2, + "grad_norm": 9.024648666381836, + "learning_rate": 0.00012378440811456324, + "loss": 3.251, + "step": 2869 + }, + { + "epoch": 4.2, + "grad_norm": 10.055442810058594, + "learning_rate": 0.00012373884802822737, + "loss": 3.2687, + "step": 2870 + }, + { + "epoch": 4.2, + "grad_norm": 5.984691619873047, + "learning_rate": 0.00012369328271950405, + "loss": 3.2402, + "step": 2871 + }, + { + "epoch": 4.2, + "grad_norm": 4.42611026763916, + "learning_rate": 0.00012364771219841727, + "loss": 3.2241, + "step": 2872 + }, + { + "epoch": 4.2, + "grad_norm": 5.240525722503662, + "learning_rate": 0.0001236021364749923, + "loss": 3.2565, + "step": 2873 + }, + { + "epoch": 4.2, + "eval_loss": 3.031118154525757, + "eval_runtime": 1166.2283, + "eval_samples_per_second": 14.699, + "eval_steps_per_second": 0.919, + "step": 2873 + }, + { + "epoch": 4.2, + "grad_norm": 3.474372148513794, + "learning_rate": 0.00012355655555925548, + "loss": 3.1972, + "step": 2874 + }, + { + "epoch": 4.21, + "grad_norm": 3.6792542934417725, + "learning_rate": 0.00012351096946123431, + "loss": 3.1889, + "step": 2875 + }, + { + "epoch": 4.21, + "grad_norm": 3.1051828861236572, + "learning_rate": 0.00012346537819095743, + "loss": 3.206, + "step": 2876 + }, + { + "epoch": 4.21, + "grad_norm": 2.708695411682129, + "learning_rate": 0.00012341978175845462, + "loss": 3.2389, + "step": 2877 + }, + { + "epoch": 4.21, + "grad_norm": 2.4177584648132324, + "learning_rate": 0.00012337418017375676, + "loss": 3.1948, + "step": 2878 + }, + { + "epoch": 4.21, + "grad_norm": 3.8860323429107666, + "learning_rate": 0.00012332857344689587, + "loss": 3.2037, + "step": 2879 + }, + { + "epoch": 4.21, + "grad_norm": 4.779682159423828, + "learning_rate": 0.0001232829615879052, + "loss": 3.1901, + "step": 2880 + }, + { + "epoch": 4.21, + "grad_norm": 2.2922635078430176, + "learning_rate": 0.00012323734460681902, + "loss": 3.1367, + "step": 2881 + }, + { + "epoch": 4.22, + "grad_norm": 2.1851584911346436, + "learning_rate": 0.00012319172251367276, + "loss": 3.1782, + "step": 2882 + }, + { + "epoch": 4.22, + "grad_norm": 3.3458876609802246, + "learning_rate": 0.00012314609531850293, + "loss": 3.2361, + "step": 2883 + }, + { + "epoch": 4.22, + "grad_norm": 3.9705002307891846, + "learning_rate": 0.00012310046303134732, + "loss": 3.2227, + "step": 2884 + }, + { + "epoch": 4.22, + "grad_norm": 2.2800686359405518, + "learning_rate": 0.0001230548256622446, + "loss": 3.2052, + "step": 2885 + }, + { + "epoch": 4.22, + "grad_norm": 4.4462199211120605, + "learning_rate": 0.00012300918322123475, + "loss": 3.2142, + "step": 2886 + }, + { + "epoch": 4.22, + "grad_norm": 3.9958441257476807, + "learning_rate": 0.0001229635357183588, + "loss": 3.1989, + "step": 2887 + }, + { + "epoch": 4.23, + "grad_norm": 3.798068046569824, + "learning_rate": 0.00012291788316365888, + "loss": 3.1956, + "step": 2888 + }, + { + "epoch": 4.23, + "grad_norm": 5.135157108306885, + "learning_rate": 0.00012287222556717827, + "loss": 3.1854, + "step": 2889 + }, + { + "epoch": 4.23, + "grad_norm": 2.207172393798828, + "learning_rate": 0.0001228265629389613, + "loss": 3.1583, + "step": 2890 + }, + { + "epoch": 4.23, + "grad_norm": 8.646849632263184, + "learning_rate": 0.00012278089528905346, + "loss": 3.245, + "step": 2891 + }, + { + "epoch": 4.23, + "grad_norm": 7.601872444152832, + "learning_rate": 0.00012273522262750132, + "loss": 3.1968, + "step": 2892 + }, + { + "epoch": 4.23, + "grad_norm": 6.095008373260498, + "learning_rate": 0.00012268954496435258, + "loss": 3.1753, + "step": 2893 + }, + { + "epoch": 4.23, + "grad_norm": 3.311551094055176, + "learning_rate": 0.000122643862309656, + "loss": 3.2052, + "step": 2894 + }, + { + "epoch": 4.24, + "grad_norm": 6.019943714141846, + "learning_rate": 0.0001225981746734615, + "loss": 3.2064, + "step": 2895 + }, + { + "epoch": 4.24, + "grad_norm": 5.016195774078369, + "learning_rate": 0.00012255248206581996, + "loss": 3.2018, + "step": 2896 + }, + { + "epoch": 4.24, + "grad_norm": 2.2277472019195557, + "learning_rate": 0.00012250678449678354, + "loss": 3.1877, + "step": 2897 + }, + { + "epoch": 4.24, + "grad_norm": 5.388427734375, + "learning_rate": 0.00012246108197640538, + "loss": 3.2087, + "step": 2898 + }, + { + "epoch": 4.24, + "grad_norm": 3.7437267303466797, + "learning_rate": 0.0001224153745147397, + "loss": 3.1554, + "step": 2899 + }, + { + "epoch": 4.24, + "grad_norm": 2.8456854820251465, + "learning_rate": 0.0001223696621218418, + "loss": 3.1912, + "step": 2900 + }, + { + "epoch": 4.24, + "grad_norm": 2.9107625484466553, + "learning_rate": 0.0001223239448077682, + "loss": 3.1832, + "step": 2901 + }, + { + "epoch": 4.25, + "grad_norm": 2.9731972217559814, + "learning_rate": 0.00012227822258257636, + "loss": 3.1781, + "step": 2902 + }, + { + "epoch": 4.25, + "grad_norm": 3.2368147373199463, + "learning_rate": 0.00012223249545632482, + "loss": 3.1863, + "step": 2903 + }, + { + "epoch": 4.25, + "grad_norm": 1.3652539253234863, + "learning_rate": 0.00012218676343907328, + "loss": 3.1765, + "step": 2904 + }, + { + "epoch": 4.25, + "grad_norm": 6.007261753082275, + "learning_rate": 0.00012214102654088248, + "loss": 3.2073, + "step": 2905 + }, + { + "epoch": 4.25, + "grad_norm": 6.385366439819336, + "learning_rate": 0.00012209528477181422, + "loss": 3.1974, + "step": 2906 + }, + { + "epoch": 4.25, + "grad_norm": 1.5009405612945557, + "learning_rate": 0.00012204953814193139, + "loss": 3.1815, + "step": 2907 + }, + { + "epoch": 4.26, + "grad_norm": 4.717186450958252, + "learning_rate": 0.00012200378666129793, + "loss": 3.166, + "step": 2908 + }, + { + "epoch": 4.26, + "grad_norm": 4.571866035461426, + "learning_rate": 0.0001219580303399789, + "loss": 3.2043, + "step": 2909 + }, + { + "epoch": 4.26, + "grad_norm": 6.416868686676025, + "learning_rate": 0.0001219122691880403, + "loss": 3.1969, + "step": 2910 + }, + { + "epoch": 4.26, + "grad_norm": 4.813213348388672, + "learning_rate": 0.00012186650321554935, + "loss": 3.1706, + "step": 2911 + }, + { + "epoch": 4.26, + "grad_norm": 3.790310859680176, + "learning_rate": 0.00012182073243257426, + "loss": 3.178, + "step": 2912 + }, + { + "epoch": 4.26, + "grad_norm": 4.808967590332031, + "learning_rate": 0.00012177495684918423, + "loss": 3.2266, + "step": 2913 + }, + { + "epoch": 4.26, + "grad_norm": 8.275779724121094, + "learning_rate": 0.00012172917647544964, + "loss": 3.1736, + "step": 2914 + }, + { + "epoch": 4.27, + "grad_norm": 4.079070568084717, + "learning_rate": 0.00012168339132144186, + "loss": 3.1918, + "step": 2915 + }, + { + "epoch": 4.27, + "grad_norm": 2.108287811279297, + "learning_rate": 0.0001216376013972333, + "loss": 3.1543, + "step": 2916 + }, + { + "epoch": 4.27, + "grad_norm": 3.337580442428589, + "learning_rate": 0.00012159180671289743, + "loss": 3.1937, + "step": 2917 + }, + { + "epoch": 4.27, + "grad_norm": 2.6380503177642822, + "learning_rate": 0.0001215460072785088, + "loss": 3.1794, + "step": 2918 + }, + { + "epoch": 4.27, + "grad_norm": 2.371419906616211, + "learning_rate": 0.00012150020310414296, + "loss": 3.1682, + "step": 2919 + }, + { + "epoch": 4.27, + "grad_norm": 1.647721767425537, + "learning_rate": 0.00012145439419987652, + "loss": 3.1618, + "step": 2920 + }, + { + "epoch": 4.27, + "grad_norm": 0.846366822719574, + "learning_rate": 0.00012140858057578713, + "loss": 3.1883, + "step": 2921 + }, + { + "epoch": 4.28, + "grad_norm": 1.4289370775222778, + "learning_rate": 0.00012136276224195348, + "loss": 3.2119, + "step": 2922 + }, + { + "epoch": 4.28, + "grad_norm": 2.510512351989746, + "learning_rate": 0.00012131693920845533, + "loss": 3.1849, + "step": 2923 + }, + { + "epoch": 4.28, + "grad_norm": 1.3627164363861084, + "learning_rate": 0.0001212711114853734, + "loss": 3.1851, + "step": 2924 + }, + { + "epoch": 4.28, + "grad_norm": 2.383141279220581, + "learning_rate": 0.00012122527908278949, + "loss": 3.1599, + "step": 2925 + }, + { + "epoch": 4.28, + "grad_norm": 2.2744383811950684, + "learning_rate": 0.00012117944201078647, + "loss": 3.1814, + "step": 2926 + }, + { + "epoch": 4.28, + "grad_norm": 2.6775460243225098, + "learning_rate": 0.00012113360027944806, + "loss": 3.1495, + "step": 2927 + }, + { + "epoch": 4.28, + "grad_norm": 2.249178647994995, + "learning_rate": 0.00012108775389885929, + "loss": 3.1721, + "step": 2928 + }, + { + "epoch": 4.29, + "grad_norm": 1.3241461515426636, + "learning_rate": 0.00012104190287910597, + "loss": 3.1563, + "step": 2929 + }, + { + "epoch": 4.29, + "grad_norm": 3.977029323577881, + "learning_rate": 0.00012099604723027502, + "loss": 3.254, + "step": 2930 + }, + { + "epoch": 4.29, + "grad_norm": 3.9783012866973877, + "learning_rate": 0.00012095018696245441, + "loss": 3.1861, + "step": 2931 + }, + { + "epoch": 4.29, + "grad_norm": 1.687117338180542, + "learning_rate": 0.00012090432208573306, + "loss": 3.1259, + "step": 2932 + }, + { + "epoch": 4.29, + "grad_norm": 10.139452934265137, + "learning_rate": 0.00012085845261020095, + "loss": 3.1579, + "step": 2933 + }, + { + "epoch": 4.29, + "grad_norm": 4.6292500495910645, + "learning_rate": 0.00012081257854594906, + "loss": 3.2254, + "step": 2934 + }, + { + "epoch": 4.3, + "grad_norm": 3.5405800342559814, + "learning_rate": 0.00012076669990306934, + "loss": 3.1327, + "step": 2935 + }, + { + "epoch": 4.3, + "grad_norm": 1.5894381999969482, + "learning_rate": 0.00012072081669165482, + "loss": 3.1545, + "step": 2936 + }, + { + "epoch": 4.3, + "grad_norm": 3.921757221221924, + "learning_rate": 0.0001206749289217995, + "loss": 3.1646, + "step": 2937 + }, + { + "epoch": 4.3, + "grad_norm": 3.1364986896514893, + "learning_rate": 0.0001206290366035984, + "loss": 3.1835, + "step": 2938 + }, + { + "epoch": 4.3, + "grad_norm": 1.0131809711456299, + "learning_rate": 0.00012058313974714746, + "loss": 3.1446, + "step": 2939 + }, + { + "epoch": 4.3, + "grad_norm": 2.2197444438934326, + "learning_rate": 0.00012053723836254373, + "loss": 3.1805, + "step": 2940 + }, + { + "epoch": 4.3, + "grad_norm": 1.5680094957351685, + "learning_rate": 0.00012049133245988519, + "loss": 3.2005, + "step": 2941 + }, + { + "epoch": 4.31, + "grad_norm": 2.2786076068878174, + "learning_rate": 0.00012044542204927087, + "loss": 3.1663, + "step": 2942 + }, + { + "epoch": 4.31, + "grad_norm": 2.1755716800689697, + "learning_rate": 0.00012039950714080068, + "loss": 3.1709, + "step": 2943 + }, + { + "epoch": 4.31, + "grad_norm": 1.485142469406128, + "learning_rate": 0.00012035358774457563, + "loss": 3.2298, + "step": 2944 + }, + { + "epoch": 4.31, + "grad_norm": 1.9527829885482788, + "learning_rate": 0.00012030766387069771, + "loss": 3.1868, + "step": 2945 + }, + { + "epoch": 4.31, + "grad_norm": 1.5011379718780518, + "learning_rate": 0.00012026173552926981, + "loss": 3.1533, + "step": 2946 + }, + { + "epoch": 4.31, + "grad_norm": 1.8543486595153809, + "learning_rate": 0.00012021580273039585, + "loss": 3.1536, + "step": 2947 + }, + { + "epoch": 4.31, + "grad_norm": 2.353090763092041, + "learning_rate": 0.00012016986548418083, + "loss": 3.1675, + "step": 2948 + }, + { + "epoch": 4.32, + "grad_norm": 3.509500741958618, + "learning_rate": 0.00012012392380073056, + "loss": 3.1963, + "step": 2949 + }, + { + "epoch": 4.32, + "grad_norm": 4.018800258636475, + "learning_rate": 0.00012007797769015192, + "loss": 3.165, + "step": 2950 + }, + { + "epoch": 4.32, + "grad_norm": 2.620981454849243, + "learning_rate": 0.00012003202716255273, + "loss": 3.1379, + "step": 2951 + }, + { + "epoch": 4.32, + "grad_norm": 2501.735107421875, + "learning_rate": 0.00011998607222804181, + "loss": 3.2292, + "step": 2952 + }, + { + "epoch": 4.32, + "grad_norm": 7.561061859130859, + "learning_rate": 0.00011994011289672895, + "loss": 3.1922, + "step": 2953 + }, + { + "epoch": 4.32, + "grad_norm": 7.529016971588135, + "learning_rate": 0.00011989414917872488, + "loss": 3.1894, + "step": 2954 + }, + { + "epoch": 4.32, + "grad_norm": 2.2252986431121826, + "learning_rate": 0.00011984818108414133, + "loss": 3.1745, + "step": 2955 + }, + { + "epoch": 4.33, + "grad_norm": 6.323204040527344, + "learning_rate": 0.00011980220862309097, + "loss": 3.1978, + "step": 2956 + }, + { + "epoch": 4.33, + "grad_norm": 137.4775390625, + "learning_rate": 0.00011975623180568738, + "loss": 3.222, + "step": 2957 + }, + { + "epoch": 4.33, + "grad_norm": 9.54941177368164, + "learning_rate": 0.0001197102506420452, + "loss": 3.1733, + "step": 2958 + }, + { + "epoch": 4.33, + "grad_norm": 7.13044548034668, + "learning_rate": 0.00011966426514228, + "loss": 3.2191, + "step": 2959 + }, + { + "epoch": 4.33, + "grad_norm": 1.8224902153015137, + "learning_rate": 0.00011961827531650824, + "loss": 3.2066, + "step": 2960 + }, + { + "epoch": 4.33, + "grad_norm": 7.74959659576416, + "learning_rate": 0.00011957228117484736, + "loss": 3.2138, + "step": 2961 + }, + { + "epoch": 4.34, + "grad_norm": 8.487194061279297, + "learning_rate": 0.00011952628272741584, + "loss": 3.1951, + "step": 2962 + }, + { + "epoch": 4.34, + "grad_norm": 8.898509979248047, + "learning_rate": 0.00011948027998433296, + "loss": 3.1982, + "step": 2963 + }, + { + "epoch": 4.34, + "grad_norm": 11.637873649597168, + "learning_rate": 0.00011943427295571899, + "loss": 3.2156, + "step": 2964 + }, + { + "epoch": 4.34, + "grad_norm": 7.426854133605957, + "learning_rate": 0.00011938826165169525, + "loss": 3.2193, + "step": 2965 + }, + { + "epoch": 4.34, + "grad_norm": 12.111673355102539, + "learning_rate": 0.00011934224608238389, + "loss": 3.1818, + "step": 2966 + }, + { + "epoch": 4.34, + "grad_norm": 3.1284587383270264, + "learning_rate": 0.00011929622625790802, + "loss": 3.1583, + "step": 2967 + }, + { + "epoch": 4.34, + "grad_norm": 4.960999488830566, + "learning_rate": 0.00011925020218839167, + "loss": 3.1765, + "step": 2968 + }, + { + "epoch": 4.35, + "grad_norm": 7.320558071136475, + "learning_rate": 0.00011920417388395986, + "loss": 3.19, + "step": 2969 + }, + { + "epoch": 4.35, + "grad_norm": 4.303677082061768, + "learning_rate": 0.0001191581413547385, + "loss": 3.1875, + "step": 2970 + }, + { + "epoch": 4.35, + "grad_norm": 4.7796549797058105, + "learning_rate": 0.0001191121046108544, + "loss": 3.1843, + "step": 2971 + }, + { + "epoch": 4.35, + "grad_norm": 5.886425971984863, + "learning_rate": 0.0001190660636624354, + "loss": 3.1532, + "step": 2972 + }, + { + "epoch": 4.35, + "grad_norm": 2.2462821006774902, + "learning_rate": 0.00011902001851961017, + "loss": 3.1873, + "step": 2973 + }, + { + "epoch": 4.35, + "grad_norm": 6.365395545959473, + "learning_rate": 0.00011897396919250831, + "loss": 3.1509, + "step": 2974 + }, + { + "epoch": 4.35, + "grad_norm": 8.911288261413574, + "learning_rate": 0.0001189279156912604, + "loss": 3.1714, + "step": 2975 + }, + { + "epoch": 4.36, + "grad_norm": 5.056528091430664, + "learning_rate": 0.00011888185802599791, + "loss": 3.1708, + "step": 2976 + }, + { + "epoch": 4.36, + "grad_norm": 3.0920491218566895, + "learning_rate": 0.00011883579620685318, + "loss": 3.1315, + "step": 2977 + }, + { + "epoch": 4.36, + "grad_norm": 4.874855041503906, + "learning_rate": 0.00011878973024395952, + "loss": 3.1369, + "step": 2978 + }, + { + "epoch": 4.36, + "grad_norm": 2.600829601287842, + "learning_rate": 0.0001187436601474511, + "loss": 3.1673, + "step": 2979 + }, + { + "epoch": 4.36, + "grad_norm": 5.483844757080078, + "learning_rate": 0.00011869758592746309, + "loss": 3.1439, + "step": 2980 + }, + { + "epoch": 4.36, + "grad_norm": 7.73532772064209, + "learning_rate": 0.00011865150759413147, + "loss": 3.2182, + "step": 2981 + }, + { + "epoch": 4.36, + "grad_norm": 3.6422669887542725, + "learning_rate": 0.00011860542515759318, + "loss": 3.1551, + "step": 2982 + }, + { + "epoch": 4.37, + "grad_norm": 4.651947975158691, + "learning_rate": 0.00011855933862798605, + "loss": 3.1674, + "step": 2983 + }, + { + "epoch": 4.37, + "grad_norm": 6.519864559173584, + "learning_rate": 0.00011851324801544879, + "loss": 3.1561, + "step": 2984 + }, + { + "epoch": 4.37, + "grad_norm": 3.8000521659851074, + "learning_rate": 0.00011846715333012104, + "loss": 3.174, + "step": 2985 + }, + { + "epoch": 4.37, + "grad_norm": 4.087390899658203, + "learning_rate": 0.00011842105458214333, + "loss": 3.1715, + "step": 2986 + }, + { + "epoch": 4.37, + "grad_norm": 5.734311580657959, + "learning_rate": 0.00011837495178165706, + "loss": 3.1486, + "step": 2987 + }, + { + "epoch": 4.37, + "grad_norm": 2.4519381523132324, + "learning_rate": 0.00011832884493880452, + "loss": 3.1528, + "step": 2988 + }, + { + "epoch": 4.38, + "grad_norm": 4.925870895385742, + "learning_rate": 0.00011828273406372894, + "loss": 3.1785, + "step": 2989 + }, + { + "epoch": 4.38, + "grad_norm": 6.140622615814209, + "learning_rate": 0.0001182366191665744, + "loss": 3.207, + "step": 2990 + }, + { + "epoch": 4.38, + "grad_norm": 2.630859851837158, + "learning_rate": 0.00011819050025748583, + "loss": 3.1437, + "step": 2991 + }, + { + "epoch": 4.38, + "grad_norm": 6.897947311401367, + "learning_rate": 0.00011814437734660917, + "loss": 3.2066, + "step": 2992 + }, + { + "epoch": 4.38, + "grad_norm": 10.167567253112793, + "learning_rate": 0.00011809825044409106, + "loss": 3.1581, + "step": 2993 + }, + { + "epoch": 4.38, + "grad_norm": 13.395803451538086, + "learning_rate": 0.00011805211956007914, + "loss": 3.1348, + "step": 2994 + }, + { + "epoch": 4.38, + "grad_norm": 1.3696818351745605, + "learning_rate": 0.00011800598470472189, + "loss": 3.1391, + "step": 2995 + }, + { + "epoch": 4.39, + "grad_norm": 6.5317063331604, + "learning_rate": 0.00011795984588816869, + "loss": 3.1951, + "step": 2996 + }, + { + "epoch": 4.39, + "grad_norm": 63.888736724853516, + "learning_rate": 0.00011791370312056972, + "loss": 3.2487, + "step": 2997 + }, + { + "epoch": 4.39, + "grad_norm": 9.84652042388916, + "learning_rate": 0.00011786755641207614, + "loss": 3.1655, + "step": 2998 + }, + { + "epoch": 4.39, + "grad_norm": 7.442485809326172, + "learning_rate": 0.00011782140577283991, + "loss": 3.1921, + "step": 2999 + }, + { + "epoch": 4.39, + "grad_norm": 1.4993599653244019, + "learning_rate": 0.00011777525121301383, + "loss": 3.1366, + "step": 3000 + }, + { + "epoch": 4.39, + "grad_norm": 23.01314926147461, + "learning_rate": 0.00011772909274275159, + "loss": 3.1806, + "step": 3001 + }, + { + "epoch": 4.39, + "grad_norm": 13.442095756530762, + "learning_rate": 0.00011768293037220778, + "loss": 3.2192, + "step": 3002 + }, + { + "epoch": 4.4, + "grad_norm": 11.32374382019043, + "learning_rate": 0.0001176367641115378, + "loss": 3.1979, + "step": 3003 + }, + { + "epoch": 4.4, + "grad_norm": 4.027917385101318, + "learning_rate": 0.00011759059397089793, + "loss": 3.1963, + "step": 3004 + }, + { + "epoch": 4.4, + "grad_norm": 6.907668113708496, + "learning_rate": 0.00011754441996044523, + "loss": 3.168, + "step": 3005 + }, + { + "epoch": 4.4, + "grad_norm": 8.269980430603027, + "learning_rate": 0.00011749824209033775, + "loss": 3.1789, + "step": 3006 + }, + { + "epoch": 4.4, + "grad_norm": 6.633981227874756, + "learning_rate": 0.00011745206037073426, + "loss": 3.225, + "step": 3007 + }, + { + "epoch": 4.4, + "grad_norm": 5.4061994552612305, + "learning_rate": 0.00011740587481179442, + "loss": 3.2288, + "step": 3008 + }, + { + "epoch": 4.41, + "grad_norm": 5.786009311676025, + "learning_rate": 0.0001173596854236788, + "loss": 3.1826, + "step": 3009 + }, + { + "epoch": 4.41, + "grad_norm": 6.876216888427734, + "learning_rate": 0.00011731349221654875, + "loss": 3.1563, + "step": 3010 + }, + { + "epoch": 4.41, + "grad_norm": 6.093964099884033, + "learning_rate": 0.00011726729520056641, + "loss": 3.1879, + "step": 3011 + }, + { + "epoch": 4.41, + "grad_norm": 5.7057576179504395, + "learning_rate": 0.00011722109438589486, + "loss": 3.2176, + "step": 3012 + }, + { + "epoch": 4.41, + "grad_norm": 6.603635787963867, + "learning_rate": 0.00011717488978269795, + "loss": 3.1905, + "step": 3013 + }, + { + "epoch": 4.41, + "grad_norm": 4.3624067306518555, + "learning_rate": 0.00011712868140114037, + "loss": 3.1549, + "step": 3014 + }, + { + "epoch": 4.41, + "grad_norm": 3.682727098464966, + "learning_rate": 0.00011708246925138771, + "loss": 3.1376, + "step": 3015 + }, + { + "epoch": 4.42, + "grad_norm": 6.1669921875, + "learning_rate": 0.00011703625334360626, + "loss": 3.1513, + "step": 3016 + }, + { + "epoch": 4.42, + "grad_norm": 4.20526123046875, + "learning_rate": 0.00011699003368796328, + "loss": 3.1493, + "step": 3017 + }, + { + "epoch": 4.42, + "grad_norm": 2.534583568572998, + "learning_rate": 0.0001169438102946267, + "loss": 3.1625, + "step": 3018 + }, + { + "epoch": 4.42, + "grad_norm": 2.841315269470215, + "learning_rate": 0.00011689758317376547, + "loss": 3.2036, + "step": 3019 + }, + { + "epoch": 4.42, + "grad_norm": 14.288698196411133, + "learning_rate": 0.00011685135233554917, + "loss": 3.1499, + "step": 3020 + }, + { + "epoch": 4.42, + "grad_norm": 5.600762844085693, + "learning_rate": 0.00011680511779014827, + "loss": 3.1529, + "step": 3021 + }, + { + "epoch": 4.42, + "grad_norm": 2.945816993713379, + "learning_rate": 0.00011675887954773405, + "loss": 3.1832, + "step": 3022 + }, + { + "epoch": 4.43, + "grad_norm": 3.300783395767212, + "learning_rate": 0.0001167126376184787, + "loss": 3.16, + "step": 3023 + }, + { + "epoch": 4.43, + "grad_norm": 10.50096607208252, + "learning_rate": 0.00011666639201255506, + "loss": 3.128, + "step": 3024 + }, + { + "epoch": 4.43, + "grad_norm": 2.324991226196289, + "learning_rate": 0.00011662014274013684, + "loss": 3.1621, + "step": 3025 + }, + { + "epoch": 4.43, + "grad_norm": 3.7734105587005615, + "learning_rate": 0.00011657388981139865, + "loss": 3.1594, + "step": 3026 + }, + { + "epoch": 4.43, + "grad_norm": 3.324181079864502, + "learning_rate": 0.00011652763323651575, + "loss": 3.1495, + "step": 3027 + }, + { + "epoch": 4.43, + "grad_norm": 2.9058284759521484, + "learning_rate": 0.00011648137302566431, + "loss": 3.1342, + "step": 3028 + }, + { + "epoch": 4.43, + "grad_norm": 3.1446614265441895, + "learning_rate": 0.00011643510918902123, + "loss": 3.1408, + "step": 3029 + }, + { + "epoch": 4.44, + "grad_norm": 1.5861114263534546, + "learning_rate": 0.0001163888417367643, + "loss": 3.1833, + "step": 3030 + }, + { + "epoch": 4.44, + "grad_norm": 0.9234637022018433, + "learning_rate": 0.00011634257067907201, + "loss": 3.1692, + "step": 3031 + }, + { + "epoch": 4.44, + "grad_norm": 2.46486759185791, + "learning_rate": 0.00011629629602612368, + "loss": 3.1654, + "step": 3032 + }, + { + "epoch": 4.44, + "grad_norm": 1.57401442527771, + "learning_rate": 0.00011625001778809945, + "loss": 3.1469, + "step": 3033 + }, + { + "epoch": 4.44, + "grad_norm": 3.949265241622925, + "learning_rate": 0.00011620373597518024, + "loss": 3.2028, + "step": 3034 + }, + { + "epoch": 4.44, + "grad_norm": 4.033581256866455, + "learning_rate": 0.00011615745059754766, + "loss": 3.1827, + "step": 3035 + }, + { + "epoch": 4.45, + "grad_norm": 3.1215124130249023, + "learning_rate": 0.00011611116166538425, + "loss": 3.1908, + "step": 3036 + }, + { + "epoch": 4.45, + "grad_norm": 2.8856616020202637, + "learning_rate": 0.00011606486918887327, + "loss": 3.1631, + "step": 3037 + }, + { + "epoch": 4.45, + "grad_norm": 3.271087408065796, + "learning_rate": 0.00011601857317819871, + "loss": 3.1653, + "step": 3038 + }, + { + "epoch": 4.45, + "grad_norm": 3.887037515640259, + "learning_rate": 0.0001159722736435454, + "loss": 3.1656, + "step": 3039 + }, + { + "epoch": 4.45, + "grad_norm": 1.2171587944030762, + "learning_rate": 0.00011592597059509893, + "loss": 3.1612, + "step": 3040 + }, + { + "epoch": 4.45, + "grad_norm": 1.7865351438522339, + "learning_rate": 0.00011587966404304564, + "loss": 3.1736, + "step": 3041 + }, + { + "epoch": 4.45, + "grad_norm": 0.9533948302268982, + "learning_rate": 0.00011583335399757271, + "loss": 3.1167, + "step": 3042 + }, + { + "epoch": 4.45, + "eval_loss": 3.0038862228393555, + "eval_runtime": 1165.2502, + "eval_samples_per_second": 14.711, + "eval_steps_per_second": 0.92, + "step": 3042 + }, + { + "epoch": 4.46, + "grad_norm": 1.679763674736023, + "learning_rate": 0.000115787040468868, + "loss": 3.1827, + "step": 3043 + }, + { + "epoch": 4.46, + "grad_norm": 0.7595570683479309, + "learning_rate": 0.0001157407234671202, + "loss": 3.1615, + "step": 3044 + }, + { + "epoch": 4.46, + "grad_norm": 1.3758699893951416, + "learning_rate": 0.0001156944030025187, + "loss": 3.2312, + "step": 3045 + }, + { + "epoch": 4.46, + "grad_norm": 1.6547698974609375, + "learning_rate": 0.00011564807908525374, + "loss": 3.1552, + "step": 3046 + }, + { + "epoch": 4.46, + "grad_norm": 1.4767943620681763, + "learning_rate": 0.00011560175172551623, + "loss": 3.2091, + "step": 3047 + }, + { + "epoch": 4.46, + "grad_norm": 0.9520362615585327, + "learning_rate": 0.00011555542093349791, + "loss": 3.1592, + "step": 3048 + }, + { + "epoch": 4.46, + "grad_norm": 0.4594898223876953, + "learning_rate": 0.00011550908671939125, + "loss": 3.1426, + "step": 3049 + }, + { + "epoch": 4.47, + "grad_norm": 0.8685529232025146, + "learning_rate": 0.00011546274909338945, + "loss": 3.115, + "step": 3050 + }, + { + "epoch": 4.47, + "grad_norm": 0.986310601234436, + "learning_rate": 0.00011541640806568646, + "loss": 3.1239, + "step": 3051 + }, + { + "epoch": 4.47, + "grad_norm": 2.280745029449463, + "learning_rate": 0.000115370063646477, + "loss": 3.1833, + "step": 3052 + }, + { + "epoch": 4.47, + "grad_norm": 0.696453869342804, + "learning_rate": 0.00011532371584595656, + "loss": 3.1107, + "step": 3053 + }, + { + "epoch": 4.47, + "grad_norm": 3.4933245182037354, + "learning_rate": 0.00011527736467432133, + "loss": 3.1748, + "step": 3054 + }, + { + "epoch": 4.47, + "grad_norm": 2.052901268005371, + "learning_rate": 0.00011523101014176825, + "loss": 3.16, + "step": 3055 + }, + { + "epoch": 4.47, + "grad_norm": 3.9722251892089844, + "learning_rate": 0.00011518465225849502, + "loss": 3.1674, + "step": 3056 + }, + { + "epoch": 4.48, + "grad_norm": 3.952956438064575, + "learning_rate": 0.00011513829103470004, + "loss": 3.1506, + "step": 3057 + }, + { + "epoch": 4.48, + "grad_norm": 2.466555118560791, + "learning_rate": 0.00011509192648058249, + "loss": 3.1843, + "step": 3058 + }, + { + "epoch": 4.48, + "grad_norm": 2.9880712032318115, + "learning_rate": 0.00011504555860634225, + "loss": 3.2101, + "step": 3059 + }, + { + "epoch": 4.48, + "grad_norm": 3.0772719383239746, + "learning_rate": 0.00011499918742217996, + "loss": 3.1815, + "step": 3060 + }, + { + "epoch": 4.48, + "grad_norm": 3.1504437923431396, + "learning_rate": 0.00011495281293829697, + "loss": 3.1493, + "step": 3061 + }, + { + "epoch": 4.48, + "grad_norm": 1.549119472503662, + "learning_rate": 0.00011490643516489532, + "loss": 3.154, + "step": 3062 + }, + { + "epoch": 4.49, + "grad_norm": 5.081751346588135, + "learning_rate": 0.00011486005411217784, + "loss": 3.2329, + "step": 3063 + }, + { + "epoch": 4.49, + "grad_norm": 5.428029537200928, + "learning_rate": 0.00011481366979034806, + "loss": 3.1709, + "step": 3064 + }, + { + "epoch": 4.49, + "grad_norm": 1.1052792072296143, + "learning_rate": 0.00011476728220961015, + "loss": 3.1713, + "step": 3065 + }, + { + "epoch": 4.49, + "grad_norm": 4.415512561798096, + "learning_rate": 0.00011472089138016917, + "loss": 3.1753, + "step": 3066 + }, + { + "epoch": 4.49, + "grad_norm": 4.5699143409729, + "learning_rate": 0.00011467449731223075, + "loss": 3.209, + "step": 3067 + }, + { + "epoch": 4.49, + "grad_norm": 3.3557825088500977, + "learning_rate": 0.00011462810001600123, + "loss": 3.1721, + "step": 3068 + }, + { + "epoch": 4.49, + "grad_norm": 3.846703290939331, + "learning_rate": 0.00011458169950168774, + "loss": 3.1592, + "step": 3069 + }, + { + "epoch": 4.5, + "grad_norm": 3.0023891925811768, + "learning_rate": 0.0001145352957794981, + "loss": 3.1636, + "step": 3070 + }, + { + "epoch": 4.5, + "grad_norm": 6.419353485107422, + "learning_rate": 0.0001144888888596408, + "loss": 3.1472, + "step": 3071 + }, + { + "epoch": 4.5, + "grad_norm": 6.247819900512695, + "learning_rate": 0.00011444247875232503, + "loss": 3.1901, + "step": 3072 + }, + { + "epoch": 4.5, + "grad_norm": 4.539994239807129, + "learning_rate": 0.00011439606546776076, + "loss": 3.1452, + "step": 3073 + }, + { + "epoch": 4.5, + "grad_norm": 3.100914716720581, + "learning_rate": 0.00011434964901615852, + "loss": 3.1582, + "step": 3074 + }, + { + "epoch": 4.5, + "grad_norm": 8.092155456542969, + "learning_rate": 0.0001143032294077297, + "loss": 3.2045, + "step": 3075 + }, + { + "epoch": 4.5, + "grad_norm": 2.9788618087768555, + "learning_rate": 0.00011425680665268626, + "loss": 3.1545, + "step": 3076 + }, + { + "epoch": 4.51, + "grad_norm": 2.915862560272217, + "learning_rate": 0.00011421038076124089, + "loss": 3.1752, + "step": 3077 + }, + { + "epoch": 4.51, + "grad_norm": 3.0964643955230713, + "learning_rate": 0.000114163951743607, + "loss": 3.1225, + "step": 3078 + }, + { + "epoch": 4.51, + "grad_norm": 2.3280699253082275, + "learning_rate": 0.00011411751960999862, + "loss": 3.179, + "step": 3079 + }, + { + "epoch": 4.51, + "grad_norm": 1.9851235151290894, + "learning_rate": 0.00011407108437063057, + "loss": 3.1557, + "step": 3080 + }, + { + "epoch": 4.51, + "grad_norm": 2.6399571895599365, + "learning_rate": 0.00011402464603571825, + "loss": 3.131, + "step": 3081 + }, + { + "epoch": 4.51, + "grad_norm": 10.382826805114746, + "learning_rate": 0.00011397820461547777, + "loss": 3.1541, + "step": 3082 + }, + { + "epoch": 4.51, + "grad_norm": 1.8169869184494019, + "learning_rate": 0.00011393176012012596, + "loss": 3.2169, + "step": 3083 + }, + { + "epoch": 4.52, + "grad_norm": 3.3234314918518066, + "learning_rate": 0.00011388531255988033, + "loss": 3.1631, + "step": 3084 + }, + { + "epoch": 4.52, + "grad_norm": 1.7650176286697388, + "learning_rate": 0.00011383886194495896, + "loss": 3.1561, + "step": 3085 + }, + { + "epoch": 4.52, + "grad_norm": 5.759214878082275, + "learning_rate": 0.00011379240828558072, + "loss": 3.1424, + "step": 3086 + }, + { + "epoch": 4.52, + "grad_norm": 5.468089580535889, + "learning_rate": 0.00011374595159196506, + "loss": 3.1697, + "step": 3087 + }, + { + "epoch": 4.52, + "grad_norm": 25.312021255493164, + "learning_rate": 0.00011369949187433218, + "loss": 3.1632, + "step": 3088 + }, + { + "epoch": 4.52, + "grad_norm": 3.2354679107666016, + "learning_rate": 0.0001136530291429029, + "loss": 3.166, + "step": 3089 + }, + { + "epoch": 4.53, + "grad_norm": 11.292868614196777, + "learning_rate": 0.00011360656340789873, + "loss": 3.1495, + "step": 3090 + }, + { + "epoch": 4.53, + "grad_norm": 9.209776878356934, + "learning_rate": 0.00011356009467954176, + "loss": 3.189, + "step": 3091 + }, + { + "epoch": 4.53, + "grad_norm": 40.07894515991211, + "learning_rate": 0.00011351362296805485, + "loss": 3.1709, + "step": 3092 + }, + { + "epoch": 4.53, + "grad_norm": 4.195234298706055, + "learning_rate": 0.00011346714828366148, + "loss": 3.2041, + "step": 3093 + }, + { + "epoch": 4.53, + "grad_norm": 154.33792114257812, + "learning_rate": 0.0001134206706365857, + "loss": 3.1673, + "step": 3094 + }, + { + "epoch": 4.53, + "grad_norm": 3.392430305480957, + "learning_rate": 0.00011337419003705237, + "loss": 3.1747, + "step": 3095 + }, + { + "epoch": 4.53, + "grad_norm": 156.2117462158203, + "learning_rate": 0.00011332770649528681, + "loss": 3.1426, + "step": 3096 + }, + { + "epoch": 4.54, + "grad_norm": 3.3672966957092285, + "learning_rate": 0.00011328122002151516, + "loss": 3.1428, + "step": 3097 + }, + { + "epoch": 4.54, + "grad_norm": 1.9393489360809326, + "learning_rate": 0.00011323473062596413, + "loss": 3.1491, + "step": 3098 + }, + { + "epoch": 4.54, + "grad_norm": 4.762568473815918, + "learning_rate": 0.00011318823831886103, + "loss": 3.1408, + "step": 3099 + }, + { + "epoch": 4.54, + "grad_norm": 4.439011573791504, + "learning_rate": 0.00011314174311043391, + "loss": 3.149, + "step": 3100 + }, + { + "epoch": 4.54, + "grad_norm": 1099.63525390625, + "learning_rate": 0.00011309524501091135, + "loss": 3.1883, + "step": 3101 + }, + { + "epoch": 4.54, + "grad_norm": 14.831306457519531, + "learning_rate": 0.00011304874403052266, + "loss": 3.2264, + "step": 3102 + }, + { + "epoch": 4.54, + "grad_norm": 1.7882112264633179, + "learning_rate": 0.00011300224017949772, + "loss": 3.1552, + "step": 3103 + }, + { + "epoch": 4.55, + "grad_norm": 5.061419486999512, + "learning_rate": 0.00011295573346806709, + "loss": 3.1629, + "step": 3104 + }, + { + "epoch": 4.55, + "grad_norm": 2.1199474334716797, + "learning_rate": 0.00011290922390646192, + "loss": 3.1672, + "step": 3105 + }, + { + "epoch": 4.55, + "grad_norm": 2.093383312225342, + "learning_rate": 0.000112862711504914, + "loss": 3.1872, + "step": 3106 + }, + { + "epoch": 4.55, + "grad_norm": 1.5643643140792847, + "learning_rate": 0.00011281619627365576, + "loss": 3.2203, + "step": 3107 + }, + { + "epoch": 4.55, + "grad_norm": 3.1214048862457275, + "learning_rate": 0.00011276967822292024, + "loss": 3.1508, + "step": 3108 + }, + { + "epoch": 4.55, + "grad_norm": 17.56413459777832, + "learning_rate": 0.00011272315736294108, + "loss": 3.1855, + "step": 3109 + }, + { + "epoch": 4.56, + "grad_norm": 3.3824844360351562, + "learning_rate": 0.00011267663370395256, + "loss": 3.1449, + "step": 3110 + }, + { + "epoch": 4.56, + "grad_norm": 2.0639045238494873, + "learning_rate": 0.00011263010725618959, + "loss": 3.1227, + "step": 3111 + }, + { + "epoch": 4.56, + "grad_norm": 2.0494930744171143, + "learning_rate": 0.00011258357802988766, + "loss": 3.1386, + "step": 3112 + }, + { + "epoch": 4.56, + "grad_norm": 2.3310718536376953, + "learning_rate": 0.00011253704603528288, + "loss": 3.1649, + "step": 3113 + }, + { + "epoch": 4.56, + "grad_norm": 1.5938396453857422, + "learning_rate": 0.00011249051128261203, + "loss": 3.1524, + "step": 3114 + }, + { + "epoch": 4.56, + "grad_norm": 1.9962538480758667, + "learning_rate": 0.00011244397378211238, + "loss": 3.1949, + "step": 3115 + }, + { + "epoch": 4.56, + "grad_norm": 2.338400363922119, + "learning_rate": 0.00011239743354402188, + "loss": 3.1258, + "step": 3116 + }, + { + "epoch": 4.57, + "grad_norm": 1.9446300268173218, + "learning_rate": 0.00011235089057857911, + "loss": 3.1829, + "step": 3117 + }, + { + "epoch": 4.57, + "grad_norm": 2.5406713485717773, + "learning_rate": 0.00011230434489602317, + "loss": 3.1782, + "step": 3118 + }, + { + "epoch": 4.57, + "grad_norm": 1.6750293970108032, + "learning_rate": 0.00011225779650659383, + "loss": 3.139, + "step": 3119 + }, + { + "epoch": 4.57, + "grad_norm": 10.995355606079102, + "learning_rate": 0.00011221124542053139, + "loss": 3.152, + "step": 3120 + }, + { + "epoch": 4.57, + "grad_norm": 3.1163713932037354, + "learning_rate": 0.00011216469164807679, + "loss": 3.1542, + "step": 3121 + }, + { + "epoch": 4.57, + "grad_norm": 1.6949200630187988, + "learning_rate": 0.00011211813519947159, + "loss": 3.1438, + "step": 3122 + }, + { + "epoch": 4.57, + "grad_norm": 3.045480251312256, + "learning_rate": 0.00011207157608495779, + "loss": 3.1611, + "step": 3123 + }, + { + "epoch": 4.58, + "grad_norm": 2.5979998111724854, + "learning_rate": 0.0001120250143147782, + "loss": 3.1628, + "step": 3124 + }, + { + "epoch": 4.58, + "grad_norm": 2.419813632965088, + "learning_rate": 0.00011197844989917605, + "loss": 3.1876, + "step": 3125 + }, + { + "epoch": 4.58, + "grad_norm": 2.4058735370635986, + "learning_rate": 0.00011193188284839517, + "loss": 3.1617, + "step": 3126 + }, + { + "epoch": 4.58, + "grad_norm": 1.7974401712417603, + "learning_rate": 0.00011188531317268005, + "loss": 3.1276, + "step": 3127 + }, + { + "epoch": 4.58, + "grad_norm": 2.3527932167053223, + "learning_rate": 0.0001118387408822757, + "loss": 3.1843, + "step": 3128 + }, + { + "epoch": 4.58, + "grad_norm": 2.5821757316589355, + "learning_rate": 0.00011179216598742768, + "loss": 3.1492, + "step": 3129 + }, + { + "epoch": 4.58, + "grad_norm": 0.7473565936088562, + "learning_rate": 0.00011174558849838213, + "loss": 3.1384, + "step": 3130 + }, + { + "epoch": 4.59, + "grad_norm": 2.998114585876465, + "learning_rate": 0.00011169900842538588, + "loss": 3.173, + "step": 3131 + }, + { + "epoch": 4.59, + "grad_norm": 8.924659729003906, + "learning_rate": 0.00011165242577868616, + "loss": 3.1929, + "step": 3132 + }, + { + "epoch": 4.59, + "grad_norm": 13.762849807739258, + "learning_rate": 0.00011160584056853082, + "loss": 3.165, + "step": 3133 + }, + { + "epoch": 4.59, + "grad_norm": 2.115877151489258, + "learning_rate": 0.00011155925280516841, + "loss": 3.119, + "step": 3134 + }, + { + "epoch": 4.59, + "grad_norm": 3.3011577129364014, + "learning_rate": 0.00011151266249884783, + "loss": 3.1905, + "step": 3135 + }, + { + "epoch": 4.59, + "grad_norm": 26.584270477294922, + "learning_rate": 0.00011146606965981862, + "loss": 3.2361, + "step": 3136 + }, + { + "epoch": 4.6, + "grad_norm": 1.8690835237503052, + "learning_rate": 0.00011141947429833096, + "loss": 3.1527, + "step": 3137 + }, + { + "epoch": 4.6, + "grad_norm": 3.516594886779785, + "learning_rate": 0.00011137287642463545, + "loss": 3.1821, + "step": 3138 + }, + { + "epoch": 4.6, + "grad_norm": 2.7472705841064453, + "learning_rate": 0.00011132627604898338, + "loss": 3.1594, + "step": 3139 + }, + { + "epoch": 4.6, + "grad_norm": 1.927894115447998, + "learning_rate": 0.00011127967318162644, + "loss": 3.1446, + "step": 3140 + }, + { + "epoch": 4.6, + "grad_norm": 1.823498249053955, + "learning_rate": 0.00011123306783281701, + "loss": 3.1621, + "step": 3141 + }, + { + "epoch": 4.6, + "grad_norm": 2.379281759262085, + "learning_rate": 0.00011118646001280799, + "loss": 3.1441, + "step": 3142 + }, + { + "epoch": 4.6, + "grad_norm": 1.6748161315917969, + "learning_rate": 0.00011113984973185268, + "loss": 3.1687, + "step": 3143 + }, + { + "epoch": 4.61, + "grad_norm": 4.564174652099609, + "learning_rate": 0.00011109323700020508, + "loss": 3.1976, + "step": 3144 + }, + { + "epoch": 4.61, + "grad_norm": 4.5107550621032715, + "learning_rate": 0.00011104662182811974, + "loss": 3.1711, + "step": 3145 + }, + { + "epoch": 4.61, + "grad_norm": 1.1205596923828125, + "learning_rate": 0.00011100000422585158, + "loss": 3.1573, + "step": 3146 + }, + { + "epoch": 4.61, + "grad_norm": 2.339171886444092, + "learning_rate": 0.00011095338420365621, + "loss": 3.1498, + "step": 3147 + }, + { + "epoch": 4.61, + "grad_norm": 2.139634609222412, + "learning_rate": 0.00011090676177178973, + "loss": 3.1493, + "step": 3148 + }, + { + "epoch": 4.61, + "grad_norm": 0.9772034287452698, + "learning_rate": 0.00011086013694050876, + "loss": 3.1551, + "step": 3149 + }, + { + "epoch": 4.61, + "grad_norm": 3.0146749019622803, + "learning_rate": 0.00011081350972007043, + "loss": 3.1186, + "step": 3150 + }, + { + "epoch": 4.62, + "grad_norm": 3.702209711074829, + "learning_rate": 0.00011076688012073244, + "loss": 3.2039, + "step": 3151 + }, + { + "epoch": 4.62, + "grad_norm": 6.443127155303955, + "learning_rate": 0.000110720248152753, + "loss": 3.1794, + "step": 3152 + }, + { + "epoch": 4.62, + "grad_norm": 3.7430996894836426, + "learning_rate": 0.00011067361382639079, + "loss": 3.172, + "step": 3153 + }, + { + "epoch": 4.62, + "grad_norm": 0.7498925924301147, + "learning_rate": 0.00011062697715190507, + "loss": 3.1905, + "step": 3154 + }, + { + "epoch": 4.62, + "grad_norm": 5.21573543548584, + "learning_rate": 0.00011058033813955557, + "loss": 3.1483, + "step": 3155 + }, + { + "epoch": 4.62, + "grad_norm": 5.235692977905273, + "learning_rate": 0.00011053369679960264, + "loss": 3.1197, + "step": 3156 + }, + { + "epoch": 4.62, + "grad_norm": 2.265049934387207, + "learning_rate": 0.00011048705314230696, + "loss": 3.1359, + "step": 3157 + }, + { + "epoch": 4.63, + "grad_norm": 6.700308799743652, + "learning_rate": 0.00011044040717792989, + "loss": 3.141, + "step": 3158 + }, + { + "epoch": 4.63, + "grad_norm": 7.944544792175293, + "learning_rate": 0.00011039375891673322, + "loss": 3.2029, + "step": 3159 + }, + { + "epoch": 4.63, + "grad_norm": 3.51708984375, + "learning_rate": 0.00011034710836897921, + "loss": 3.1747, + "step": 3160 + }, + { + "epoch": 4.63, + "grad_norm": 4.683743000030518, + "learning_rate": 0.00011030045554493075, + "loss": 3.181, + "step": 3161 + }, + { + "epoch": 4.63, + "grad_norm": 7.184266090393066, + "learning_rate": 0.00011025380045485107, + "loss": 3.1523, + "step": 3162 + }, + { + "epoch": 4.63, + "grad_norm": 5.593520164489746, + "learning_rate": 0.00011020714310900401, + "loss": 3.1564, + "step": 3163 + }, + { + "epoch": 4.64, + "grad_norm": 1.7525691986083984, + "learning_rate": 0.0001101604835176539, + "loss": 3.1461, + "step": 3164 + }, + { + "epoch": 4.64, + "grad_norm": 6.076540470123291, + "learning_rate": 0.00011011382169106546, + "loss": 3.1734, + "step": 3165 + }, + { + "epoch": 4.64, + "grad_norm": 7.846095561981201, + "learning_rate": 0.00011006715763950406, + "loss": 3.1622, + "step": 3166 + }, + { + "epoch": 4.64, + "grad_norm": 4.10769510269165, + "learning_rate": 0.00011002049137323543, + "loss": 3.1823, + "step": 3167 + }, + { + "epoch": 4.64, + "grad_norm": 4.750815391540527, + "learning_rate": 0.00010997382290252585, + "loss": 3.1425, + "step": 3168 + }, + { + "epoch": 4.64, + "grad_norm": 4.68991231918335, + "learning_rate": 0.0001099271522376421, + "loss": 3.1231, + "step": 3169 + }, + { + "epoch": 4.64, + "grad_norm": 3.5092368125915527, + "learning_rate": 0.00010988047938885135, + "loss": 3.2153, + "step": 3170 + }, + { + "epoch": 4.65, + "grad_norm": 4.325351715087891, + "learning_rate": 0.00010983380436642136, + "loss": 3.1567, + "step": 3171 + }, + { + "epoch": 4.65, + "grad_norm": 3.288820505142212, + "learning_rate": 0.00010978712718062036, + "loss": 3.1142, + "step": 3172 + }, + { + "epoch": 4.65, + "grad_norm": 3.504354238510132, + "learning_rate": 0.00010974044784171693, + "loss": 3.143, + "step": 3173 + }, + { + "epoch": 4.65, + "grad_norm": 4.500937461853027, + "learning_rate": 0.00010969376635998023, + "loss": 3.1615, + "step": 3174 + }, + { + "epoch": 4.65, + "grad_norm": 39.75580596923828, + "learning_rate": 0.00010964708274567994, + "loss": 3.2307, + "step": 3175 + }, + { + "epoch": 4.65, + "grad_norm": 3.0470621585845947, + "learning_rate": 0.00010960039700908608, + "loss": 3.1374, + "step": 3176 + }, + { + "epoch": 4.65, + "grad_norm": 2.1173510551452637, + "learning_rate": 0.00010955370916046918, + "loss": 3.1752, + "step": 3177 + }, + { + "epoch": 4.66, + "grad_norm": 2.3193938732147217, + "learning_rate": 0.00010950701921010036, + "loss": 3.1474, + "step": 3178 + }, + { + "epoch": 4.66, + "grad_norm": 2.5212008953094482, + "learning_rate": 0.00010946032716825099, + "loss": 3.2541, + "step": 3179 + }, + { + "epoch": 4.66, + "grad_norm": 1.1363469362258911, + "learning_rate": 0.00010941363304519305, + "loss": 3.1424, + "step": 3180 + }, + { + "epoch": 4.66, + "grad_norm": 2.563216209411621, + "learning_rate": 0.00010936693685119892, + "loss": 3.1572, + "step": 3181 + }, + { + "epoch": 4.66, + "grad_norm": 1.520682454109192, + "learning_rate": 0.00010932023859654145, + "loss": 3.1507, + "step": 3182 + }, + { + "epoch": 4.66, + "grad_norm": 3.5463900566101074, + "learning_rate": 0.00010927353829149398, + "loss": 3.1672, + "step": 3183 + }, + { + "epoch": 4.66, + "grad_norm": 4.035703659057617, + "learning_rate": 0.00010922683594633021, + "loss": 3.2053, + "step": 3184 + }, + { + "epoch": 4.67, + "grad_norm": 3.0087568759918213, + "learning_rate": 0.00010918013157132436, + "loss": 3.1566, + "step": 3185 + }, + { + "epoch": 4.67, + "grad_norm": 2.657742500305176, + "learning_rate": 0.00010913342517675113, + "loss": 3.1382, + "step": 3186 + }, + { + "epoch": 4.67, + "grad_norm": 4.097416877746582, + "learning_rate": 0.00010908671677288552, + "loss": 3.1695, + "step": 3187 + }, + { + "epoch": 4.67, + "grad_norm": 3.2829132080078125, + "learning_rate": 0.00010904000637000313, + "loss": 3.2016, + "step": 3188 + }, + { + "epoch": 4.67, + "grad_norm": 0.8120042681694031, + "learning_rate": 0.00010899329397837996, + "loss": 3.1287, + "step": 3189 + }, + { + "epoch": 4.67, + "grad_norm": 1.4076851606369019, + "learning_rate": 0.00010894657960829235, + "loss": 3.1302, + "step": 3190 + }, + { + "epoch": 4.68, + "grad_norm": 2.486541986465454, + "learning_rate": 0.00010889986327001716, + "loss": 3.1751, + "step": 3191 + }, + { + "epoch": 4.68, + "grad_norm": 1.665196418762207, + "learning_rate": 0.00010885314497383175, + "loss": 3.1353, + "step": 3192 + }, + { + "epoch": 4.68, + "grad_norm": 3.312460422515869, + "learning_rate": 0.00010880642473001375, + "loss": 3.1851, + "step": 3193 + }, + { + "epoch": 4.68, + "grad_norm": 6.796298027038574, + "learning_rate": 0.0001087597025488413, + "loss": 3.1513, + "step": 3194 + }, + { + "epoch": 4.68, + "grad_norm": 1.5955018997192383, + "learning_rate": 0.00010871297844059304, + "loss": 3.1609, + "step": 3195 + }, + { + "epoch": 4.68, + "grad_norm": 4.358046531677246, + "learning_rate": 0.00010866625241554789, + "loss": 3.1224, + "step": 3196 + }, + { + "epoch": 4.68, + "grad_norm": 4.621119976043701, + "learning_rate": 0.00010861952448398529, + "loss": 3.1221, + "step": 3197 + }, + { + "epoch": 4.69, + "grad_norm": 1.6193621158599854, + "learning_rate": 0.00010857279465618508, + "loss": 3.1495, + "step": 3198 + }, + { + "epoch": 4.69, + "grad_norm": 3.5418941974639893, + "learning_rate": 0.00010852606294242749, + "loss": 3.182, + "step": 3199 + }, + { + "epoch": 4.69, + "grad_norm": 3.8759021759033203, + "learning_rate": 0.00010847932935299319, + "loss": 3.1323, + "step": 3200 + }, + { + "epoch": 4.69, + "grad_norm": 1.662946343421936, + "learning_rate": 0.00010843259389816325, + "loss": 3.1605, + "step": 3201 + }, + { + "epoch": 4.69, + "grad_norm": 3.667335271835327, + "learning_rate": 0.00010838585658821918, + "loss": 3.1713, + "step": 3202 + }, + { + "epoch": 4.69, + "grad_norm": 4.988526344299316, + "learning_rate": 0.00010833911743344287, + "loss": 3.1812, + "step": 3203 + }, + { + "epoch": 4.69, + "grad_norm": 3.444049596786499, + "learning_rate": 0.00010829237644411659, + "loss": 3.1978, + "step": 3204 + }, + { + "epoch": 4.7, + "grad_norm": 2.074106216430664, + "learning_rate": 0.00010824563363052307, + "loss": 3.1567, + "step": 3205 + }, + { + "epoch": 4.7, + "grad_norm": 5.682570934295654, + "learning_rate": 0.00010819888900294542, + "loss": 3.1777, + "step": 3206 + }, + { + "epoch": 4.7, + "grad_norm": 7.264477729797363, + "learning_rate": 0.00010815214257166712, + "loss": 3.2065, + "step": 3207 + }, + { + "epoch": 4.7, + "grad_norm": 4.073165416717529, + "learning_rate": 0.0001081053943469721, + "loss": 3.1564, + "step": 3208 + }, + { + "epoch": 4.7, + "grad_norm": 3.7732391357421875, + "learning_rate": 0.00010805864433914465, + "loss": 3.1654, + "step": 3209 + }, + { + "epoch": 4.7, + "grad_norm": 4.149001598358154, + "learning_rate": 0.00010801189255846944, + "loss": 3.1563, + "step": 3210 + }, + { + "epoch": 4.71, + "grad_norm": 4.86202335357666, + "learning_rate": 0.00010796513901523155, + "loss": 3.1624, + "step": 3211 + }, + { + "epoch": 4.71, + "eval_loss": 3.0107593536376953, + "eval_runtime": 1169.2042, + "eval_samples_per_second": 14.661, + "eval_steps_per_second": 0.917, + "step": 3211 + }, + { + "epoch": 4.71, + "grad_norm": 4.124433994293213, + "learning_rate": 0.00010791838371971648, + "loss": 3.1329, + "step": 3212 + }, + { + "epoch": 4.71, + "grad_norm": 1.8730932474136353, + "learning_rate": 0.00010787162668221004, + "loss": 3.1344, + "step": 3213 + }, + { + "epoch": 4.71, + "grad_norm": 3.890324831008911, + "learning_rate": 0.00010782486791299849, + "loss": 3.1527, + "step": 3214 + }, + { + "epoch": 4.71, + "grad_norm": 3.587141513824463, + "learning_rate": 0.00010777810742236844, + "loss": 3.1435, + "step": 3215 + }, + { + "epoch": 4.71, + "grad_norm": 1.505163311958313, + "learning_rate": 0.00010773134522060688, + "loss": 3.1474, + "step": 3216 + }, + { + "epoch": 4.71, + "grad_norm": 3.0347487926483154, + "learning_rate": 0.00010768458131800119, + "loss": 3.1363, + "step": 3217 + }, + { + "epoch": 4.72, + "grad_norm": 3.127628803253174, + "learning_rate": 0.0001076378157248391, + "loss": 3.118, + "step": 3218 + }, + { + "epoch": 4.72, + "grad_norm": 2.165717363357544, + "learning_rate": 0.00010759104845140877, + "loss": 3.1966, + "step": 3219 + }, + { + "epoch": 4.72, + "grad_norm": 2.3362936973571777, + "learning_rate": 0.00010754427950799861, + "loss": 3.1494, + "step": 3220 + }, + { + "epoch": 4.72, + "grad_norm": 2.43580961227417, + "learning_rate": 0.00010749750890489754, + "loss": 3.1281, + "step": 3221 + }, + { + "epoch": 4.72, + "grad_norm": 23.91058349609375, + "learning_rate": 0.00010745073665239476, + "loss": 3.1482, + "step": 3222 + }, + { + "epoch": 4.72, + "grad_norm": 3.7118659019470215, + "learning_rate": 0.00010740396276077985, + "loss": 3.1884, + "step": 3223 + }, + { + "epoch": 4.72, + "grad_norm": 3.4350130558013916, + "learning_rate": 0.00010735718724034273, + "loss": 3.1709, + "step": 3224 + }, + { + "epoch": 4.73, + "grad_norm": 0.7842398285865784, + "learning_rate": 0.00010731041010137375, + "loss": 3.162, + "step": 3225 + }, + { + "epoch": 4.73, + "grad_norm": 4.245161533355713, + "learning_rate": 0.00010726363135416351, + "loss": 3.1882, + "step": 3226 + }, + { + "epoch": 4.73, + "grad_norm": 6.0483622550964355, + "learning_rate": 0.00010721685100900308, + "loss": 3.1815, + "step": 3227 + }, + { + "epoch": 4.73, + "grad_norm": 4.542155742645264, + "learning_rate": 0.00010717006907618377, + "loss": 3.1413, + "step": 3228 + }, + { + "epoch": 4.73, + "grad_norm": 1.648128867149353, + "learning_rate": 0.00010712328556599733, + "loss": 3.1296, + "step": 3229 + }, + { + "epoch": 4.73, + "grad_norm": 5.657898902893066, + "learning_rate": 0.00010707650048873579, + "loss": 3.178, + "step": 3230 + }, + { + "epoch": 4.73, + "grad_norm": 5.276024341583252, + "learning_rate": 0.00010702971385469158, + "loss": 3.192, + "step": 3231 + }, + { + "epoch": 4.74, + "grad_norm": 4.145871639251709, + "learning_rate": 0.00010698292567415742, + "loss": 3.134, + "step": 3232 + }, + { + "epoch": 4.74, + "grad_norm": 12.922083854675293, + "learning_rate": 0.00010693613595742644, + "loss": 3.1523, + "step": 3233 + }, + { + "epoch": 4.74, + "grad_norm": 3.349857807159424, + "learning_rate": 0.00010688934471479199, + "loss": 3.1872, + "step": 3234 + }, + { + "epoch": 4.74, + "grad_norm": 2.6229217052459717, + "learning_rate": 0.00010684255195654788, + "loss": 3.1266, + "step": 3235 + }, + { + "epoch": 4.74, + "grad_norm": 3.148838520050049, + "learning_rate": 0.00010679575769298824, + "loss": 3.2267, + "step": 3236 + }, + { + "epoch": 4.74, + "grad_norm": 3.843184471130371, + "learning_rate": 0.00010674896193440742, + "loss": 3.1531, + "step": 3237 + }, + { + "epoch": 4.75, + "grad_norm": 2.107234239578247, + "learning_rate": 0.00010670216469110019, + "loss": 3.1549, + "step": 3238 + }, + { + "epoch": 4.75, + "grad_norm": 1.9575111865997314, + "learning_rate": 0.00010665536597336167, + "loss": 3.1334, + "step": 3239 + }, + { + "epoch": 4.75, + "grad_norm": 2.4358248710632324, + "learning_rate": 0.00010660856579148724, + "loss": 3.1869, + "step": 3240 + }, + { + "epoch": 4.75, + "grad_norm": 1.9233509302139282, + "learning_rate": 0.00010656176415577262, + "loss": 3.1206, + "step": 3241 + }, + { + "epoch": 4.75, + "grad_norm": 2.0048117637634277, + "learning_rate": 0.00010651496107651384, + "loss": 3.1099, + "step": 3242 + }, + { + "epoch": 4.75, + "grad_norm": 4.973776817321777, + "learning_rate": 0.00010646815656400731, + "loss": 3.1621, + "step": 3243 + }, + { + "epoch": 4.75, + "grad_norm": 3.1254444122314453, + "learning_rate": 0.00010642135062854965, + "loss": 3.1564, + "step": 3244 + }, + { + "epoch": 4.76, + "grad_norm": 2.404599666595459, + "learning_rate": 0.00010637454328043793, + "loss": 3.1772, + "step": 3245 + }, + { + "epoch": 4.76, + "grad_norm": 2.372962474822998, + "learning_rate": 0.00010632773452996936, + "loss": 3.1053, + "step": 3246 + }, + { + "epoch": 4.76, + "grad_norm": 1.7531036138534546, + "learning_rate": 0.00010628092438744167, + "loss": 3.1678, + "step": 3247 + }, + { + "epoch": 4.76, + "grad_norm": 15.43344783782959, + "learning_rate": 0.00010623411286315261, + "loss": 3.1616, + "step": 3248 + }, + { + "epoch": 4.76, + "grad_norm": 3.320650577545166, + "learning_rate": 0.00010618729996740056, + "loss": 3.2037, + "step": 3249 + }, + { + "epoch": 4.76, + "grad_norm": 14.557812690734863, + "learning_rate": 0.00010614048571048401, + "loss": 3.221, + "step": 3250 + }, + { + "epoch": 4.76, + "grad_norm": 1.5140577554702759, + "learning_rate": 0.00010609367010270168, + "loss": 3.1666, + "step": 3251 + }, + { + "epoch": 4.77, + "grad_norm": 2.14962100982666, + "learning_rate": 0.00010604685315435281, + "loss": 3.1691, + "step": 3252 + }, + { + "epoch": 4.77, + "grad_norm": 0.9101262092590332, + "learning_rate": 0.0001060000348757368, + "loss": 3.1835, + "step": 3253 + }, + { + "epoch": 4.77, + "grad_norm": 1.9691673517227173, + "learning_rate": 0.00010595321527715327, + "loss": 3.1425, + "step": 3254 + }, + { + "epoch": 4.77, + "grad_norm": 6.10009241104126, + "learning_rate": 0.0001059063943689023, + "loss": 3.163, + "step": 3255 + }, + { + "epoch": 4.77, + "grad_norm": 1.81471848487854, + "learning_rate": 0.00010585957216128415, + "loss": 3.1742, + "step": 3256 + }, + { + "epoch": 4.77, + "grad_norm": 1.8879374265670776, + "learning_rate": 0.0001058127486645994, + "loss": 3.1732, + "step": 3257 + }, + { + "epoch": 4.77, + "grad_norm": 0.9328217506408691, + "learning_rate": 0.0001057659238891489, + "loss": 3.1223, + "step": 3258 + }, + { + "epoch": 4.78, + "grad_norm": 1.236288070678711, + "learning_rate": 0.00010571909784523377, + "loss": 3.114, + "step": 3259 + }, + { + "epoch": 4.78, + "grad_norm": 2.6463773250579834, + "learning_rate": 0.00010567227054315545, + "loss": 3.1444, + "step": 3260 + }, + { + "epoch": 4.78, + "grad_norm": 1.5447609424591064, + "learning_rate": 0.00010562544199321562, + "loss": 3.1348, + "step": 3261 + }, + { + "epoch": 4.78, + "grad_norm": 3.4742603302001953, + "learning_rate": 0.00010557861220571625, + "loss": 3.1773, + "step": 3262 + }, + { + "epoch": 4.78, + "grad_norm": 2.8108527660369873, + "learning_rate": 0.0001055317811909596, + "loss": 3.1871, + "step": 3263 + }, + { + "epoch": 4.78, + "grad_norm": 2.6389920711517334, + "learning_rate": 0.00010548494895924816, + "loss": 3.1609, + "step": 3264 + }, + { + "epoch": 4.79, + "grad_norm": 2.764688491821289, + "learning_rate": 0.00010543811552088468, + "loss": 3.1937, + "step": 3265 + }, + { + "epoch": 4.79, + "grad_norm": 3.0502400398254395, + "learning_rate": 0.00010539128088617225, + "loss": 3.1669, + "step": 3266 + }, + { + "epoch": 4.79, + "grad_norm": 6.166628837585449, + "learning_rate": 0.00010534444506541417, + "loss": 3.142, + "step": 3267 + }, + { + "epoch": 4.79, + "grad_norm": 47.169883728027344, + "learning_rate": 0.00010529760806891395, + "loss": 3.1328, + "step": 3268 + }, + { + "epoch": 4.79, + "grad_norm": 3.0846803188323975, + "learning_rate": 0.0001052507699069755, + "loss": 3.1455, + "step": 3269 + }, + { + "epoch": 4.79, + "grad_norm": 2.2564151287078857, + "learning_rate": 0.00010520393058990285, + "loss": 3.1306, + "step": 3270 + }, + { + "epoch": 4.79, + "grad_norm": 1.4548720121383667, + "learning_rate": 0.00010515709012800033, + "loss": 3.1312, + "step": 3271 + }, + { + "epoch": 4.8, + "grad_norm": 1.8430453538894653, + "learning_rate": 0.00010511024853157255, + "loss": 3.1276, + "step": 3272 + }, + { + "epoch": 4.8, + "grad_norm": 1.2610652446746826, + "learning_rate": 0.00010506340581092433, + "loss": 3.1682, + "step": 3273 + }, + { + "epoch": 4.8, + "grad_norm": 0.8417081832885742, + "learning_rate": 0.00010501656197636078, + "loss": 3.1505, + "step": 3274 + }, + { + "epoch": 4.8, + "grad_norm": 1.3054739236831665, + "learning_rate": 0.00010496971703818721, + "loss": 3.1465, + "step": 3275 + }, + { + "epoch": 4.8, + "grad_norm": 2.4543936252593994, + "learning_rate": 0.00010492287100670919, + "loss": 3.2098, + "step": 3276 + }, + { + "epoch": 4.8, + "grad_norm": 1.5280994176864624, + "learning_rate": 0.00010487602389223257, + "loss": 3.1443, + "step": 3277 + }, + { + "epoch": 4.8, + "grad_norm": 1.9430744647979736, + "learning_rate": 0.00010482917570506334, + "loss": 3.1531, + "step": 3278 + }, + { + "epoch": 4.81, + "grad_norm": 1.8071426153182983, + "learning_rate": 0.00010478232645550782, + "loss": 3.171, + "step": 3279 + }, + { + "epoch": 4.81, + "grad_norm": 3.768815517425537, + "learning_rate": 0.00010473547615387256, + "loss": 3.1586, + "step": 3280 + }, + { + "epoch": 4.81, + "grad_norm": 2.8686270713806152, + "learning_rate": 0.00010468862481046427, + "loss": 3.112, + "step": 3281 + }, + { + "epoch": 4.81, + "grad_norm": 2.201878070831299, + "learning_rate": 0.0001046417724355899, + "loss": 3.1815, + "step": 3282 + }, + { + "epoch": 4.81, + "grad_norm": 1.5182592868804932, + "learning_rate": 0.00010459491903955676, + "loss": 3.1391, + "step": 3283 + }, + { + "epoch": 4.81, + "grad_norm": 23.839717864990234, + "learning_rate": 0.0001045480646326722, + "loss": 3.1331, + "step": 3284 + }, + { + "epoch": 4.81, + "grad_norm": 1.8520891666412354, + "learning_rate": 0.00010450120922524387, + "loss": 3.1622, + "step": 3285 + }, + { + "epoch": 4.82, + "grad_norm": 0.8244103789329529, + "learning_rate": 0.00010445435282757972, + "loss": 3.1308, + "step": 3286 + }, + { + "epoch": 4.82, + "grad_norm": 2.278554677963257, + "learning_rate": 0.00010440749544998778, + "loss": 3.1283, + "step": 3287 + }, + { + "epoch": 4.82, + "grad_norm": 3.664557933807373, + "learning_rate": 0.00010436063710277639, + "loss": 3.1424, + "step": 3288 + }, + { + "epoch": 4.82, + "grad_norm": 1.274968147277832, + "learning_rate": 0.00010431377779625407, + "loss": 3.1358, + "step": 3289 + }, + { + "epoch": 4.82, + "grad_norm": 0.9037652611732483, + "learning_rate": 0.00010426691754072952, + "loss": 3.1329, + "step": 3290 + }, + { + "epoch": 4.82, + "grad_norm": 2.134868860244751, + "learning_rate": 0.00010422005634651173, + "loss": 3.0978, + "step": 3291 + }, + { + "epoch": 4.83, + "grad_norm": 4.81277322769165, + "learning_rate": 0.00010417319422390979, + "loss": 3.118, + "step": 3292 + }, + { + "epoch": 4.83, + "grad_norm": 1.5502432584762573, + "learning_rate": 0.00010412633118323311, + "loss": 3.1258, + "step": 3293 + }, + { + "epoch": 4.83, + "grad_norm": 1.0471981763839722, + "learning_rate": 0.00010407946723479125, + "loss": 3.1163, + "step": 3294 + }, + { + "epoch": 4.83, + "grad_norm": 2.2549943923950195, + "learning_rate": 0.00010403260238889389, + "loss": 3.1441, + "step": 3295 + }, + { + "epoch": 4.83, + "grad_norm": 1.7782256603240967, + "learning_rate": 0.00010398573665585105, + "loss": 3.1251, + "step": 3296 + }, + { + "epoch": 4.83, + "grad_norm": 3.3199939727783203, + "learning_rate": 0.00010393887004597287, + "loss": 3.1731, + "step": 3297 + }, + { + "epoch": 4.83, + "grad_norm": 3.324904441833496, + "learning_rate": 0.00010389200256956968, + "loss": 3.1439, + "step": 3298 + }, + { + "epoch": 4.84, + "grad_norm": 2.5700957775115967, + "learning_rate": 0.00010384513423695196, + "loss": 3.138, + "step": 3299 + }, + { + "epoch": 4.84, + "grad_norm": 3.218172550201416, + "learning_rate": 0.00010379826505843054, + "loss": 3.1335, + "step": 3300 + }, + { + "epoch": 4.84, + "grad_norm": 1.4542341232299805, + "learning_rate": 0.00010375139504431622, + "loss": 3.1655, + "step": 3301 + }, + { + "epoch": 4.84, + "grad_norm": 21.637861251831055, + "learning_rate": 0.00010370452420492012, + "loss": 3.1785, + "step": 3302 + }, + { + "epoch": 4.84, + "grad_norm": 3.2379684448242188, + "learning_rate": 0.00010365765255055358, + "loss": 3.133, + "step": 3303 + }, + { + "epoch": 4.84, + "grad_norm": 1.356064796447754, + "learning_rate": 0.00010361078009152793, + "loss": 3.1157, + "step": 3304 + }, + { + "epoch": 4.84, + "grad_norm": 1.2905539274215698, + "learning_rate": 0.00010356390683815489, + "loss": 3.1529, + "step": 3305 + }, + { + "epoch": 4.85, + "grad_norm": 2.5484352111816406, + "learning_rate": 0.00010351703280074623, + "loss": 3.1459, + "step": 3306 + }, + { + "epoch": 4.85, + "grad_norm": 1.1913775205612183, + "learning_rate": 0.0001034701579896139, + "loss": 3.2115, + "step": 3307 + }, + { + "epoch": 4.85, + "grad_norm": 1.7160303592681885, + "learning_rate": 0.0001034232824150701, + "loss": 3.1572, + "step": 3308 + }, + { + "epoch": 4.85, + "grad_norm": 2.059537887573242, + "learning_rate": 0.0001033764060874271, + "loss": 3.1361, + "step": 3309 + }, + { + "epoch": 4.85, + "grad_norm": 1.258538007736206, + "learning_rate": 0.00010332952901699738, + "loss": 3.1483, + "step": 3310 + }, + { + "epoch": 4.85, + "grad_norm": 2.301053285598755, + "learning_rate": 0.00010328265121409361, + "loss": 3.214, + "step": 3311 + }, + { + "epoch": 4.86, + "grad_norm": 1.366959810256958, + "learning_rate": 0.00010323577268902853, + "loss": 3.1272, + "step": 3312 + }, + { + "epoch": 4.86, + "grad_norm": 1.7083238363265991, + "learning_rate": 0.00010318889345211518, + "loss": 3.1432, + "step": 3313 + }, + { + "epoch": 4.86, + "grad_norm": 1.0292158126831055, + "learning_rate": 0.00010314201351366666, + "loss": 3.16, + "step": 3314 + }, + { + "epoch": 4.86, + "grad_norm": 1.8825881481170654, + "learning_rate": 0.0001030951328839962, + "loss": 3.1021, + "step": 3315 + }, + { + "epoch": 4.86, + "grad_norm": 1.2715975046157837, + "learning_rate": 0.00010304825157341722, + "loss": 3.1305, + "step": 3316 + }, + { + "epoch": 4.86, + "grad_norm": 1.972051978111267, + "learning_rate": 0.00010300136959224335, + "loss": 3.1384, + "step": 3317 + }, + { + "epoch": 4.86, + "grad_norm": 2.031904697418213, + "learning_rate": 0.00010295448695078829, + "loss": 3.1444, + "step": 3318 + }, + { + "epoch": 4.87, + "grad_norm": 1.3081032037734985, + "learning_rate": 0.00010290760365936587, + "loss": 3.1418, + "step": 3319 + }, + { + "epoch": 4.87, + "grad_norm": 1.2461774349212646, + "learning_rate": 0.00010286071972829016, + "loss": 3.1776, + "step": 3320 + }, + { + "epoch": 4.87, + "grad_norm": 614.052978515625, + "learning_rate": 0.00010281383516787528, + "loss": 3.162, + "step": 3321 + }, + { + "epoch": 4.87, + "grad_norm": 16.35662269592285, + "learning_rate": 0.0001027669499884355, + "loss": 3.1498, + "step": 3322 + }, + { + "epoch": 4.87, + "grad_norm": 5.687806606292725, + "learning_rate": 0.00010272006420028528, + "loss": 3.1927, + "step": 3323 + }, + { + "epoch": 4.87, + "grad_norm": 4.659135341644287, + "learning_rate": 0.00010267317781373918, + "loss": 3.1678, + "step": 3324 + }, + { + "epoch": 4.87, + "grad_norm": 6.275466442108154, + "learning_rate": 0.00010262629083911186, + "loss": 3.1716, + "step": 3325 + }, + { + "epoch": 4.88, + "grad_norm": 3.4971659183502197, + "learning_rate": 0.00010257940328671813, + "loss": 3.1567, + "step": 3326 + }, + { + "epoch": 4.88, + "grad_norm": 4.331486225128174, + "learning_rate": 0.00010253251516687303, + "loss": 3.164, + "step": 3327 + }, + { + "epoch": 4.88, + "grad_norm": 2.695127010345459, + "learning_rate": 0.00010248562648989153, + "loss": 3.1222, + "step": 3328 + }, + { + "epoch": 4.88, + "grad_norm": 2.059220314025879, + "learning_rate": 0.00010243873726608881, + "loss": 3.1783, + "step": 3329 + }, + { + "epoch": 4.88, + "grad_norm": 2.7665822505950928, + "learning_rate": 0.0001023918475057803, + "loss": 3.1525, + "step": 3330 + }, + { + "epoch": 4.88, + "grad_norm": 1.4520095586776733, + "learning_rate": 0.00010234495721928137, + "loss": 3.1559, + "step": 3331 + }, + { + "epoch": 4.88, + "grad_norm": 3.7544808387756348, + "learning_rate": 0.00010229806641690754, + "loss": 3.196, + "step": 3332 + }, + { + "epoch": 4.89, + "grad_norm": 4.363980770111084, + "learning_rate": 0.0001022511751089745, + "loss": 3.1835, + "step": 3333 + }, + { + "epoch": 4.89, + "grad_norm": 2.252725124359131, + "learning_rate": 0.00010220428330579804, + "loss": 3.1515, + "step": 3334 + }, + { + "epoch": 4.89, + "grad_norm": 2.367949962615967, + "learning_rate": 0.00010215739101769399, + "loss": 3.1599, + "step": 3335 + }, + { + "epoch": 4.89, + "grad_norm": 2.9788756370544434, + "learning_rate": 0.00010211049825497841, + "loss": 3.147, + "step": 3336 + }, + { + "epoch": 4.89, + "grad_norm": 1.9200280904769897, + "learning_rate": 0.00010206360502796734, + "loss": 3.1723, + "step": 3337 + }, + { + "epoch": 4.89, + "grad_norm": 2.547455072402954, + "learning_rate": 0.00010201671134697699, + "loss": 3.1607, + "step": 3338 + }, + { + "epoch": 4.9, + "grad_norm": 3.103799343109131, + "learning_rate": 0.00010196981722232368, + "loss": 3.1501, + "step": 3339 + }, + { + "epoch": 4.9, + "grad_norm": 0.7216553092002869, + "learning_rate": 0.00010192292266432377, + "loss": 3.1397, + "step": 3340 + }, + { + "epoch": 4.9, + "grad_norm": 3.535027265548706, + "learning_rate": 0.00010187602768329378, + "loss": 3.1029, + "step": 3341 + }, + { + "epoch": 4.9, + "grad_norm": 3.598963975906372, + "learning_rate": 0.00010182913228955025, + "loss": 3.1537, + "step": 3342 + }, + { + "epoch": 4.9, + "grad_norm": 1.5128989219665527, + "learning_rate": 0.00010178223649340985, + "loss": 3.1342, + "step": 3343 + }, + { + "epoch": 4.9, + "grad_norm": 1.6262668371200562, + "learning_rate": 0.00010173534030518941, + "loss": 3.1654, + "step": 3344 + }, + { + "epoch": 4.9, + "grad_norm": 1.0745606422424316, + "learning_rate": 0.00010168844373520572, + "loss": 3.1105, + "step": 3345 + }, + { + "epoch": 4.91, + "grad_norm": 1.7129157781600952, + "learning_rate": 0.00010164154679377569, + "loss": 3.1317, + "step": 3346 + }, + { + "epoch": 4.91, + "grad_norm": 1.3464902639389038, + "learning_rate": 0.00010159464949121641, + "loss": 3.1503, + "step": 3347 + }, + { + "epoch": 4.91, + "grad_norm": 8.157245635986328, + "learning_rate": 0.00010154775183784493, + "loss": 3.1043, + "step": 3348 + }, + { + "epoch": 4.91, + "grad_norm": 4.697030544281006, + "learning_rate": 0.0001015008538439784, + "loss": 3.194, + "step": 3349 + }, + { + "epoch": 4.91, + "grad_norm": 34.21104049682617, + "learning_rate": 0.00010145395551993408, + "loss": 3.171, + "step": 3350 + }, + { + "epoch": 4.91, + "grad_norm": 3.0200319290161133, + "learning_rate": 0.00010140705687602929, + "loss": 3.1502, + "step": 3351 + }, + { + "epoch": 4.91, + "grad_norm": 0.9504959583282471, + "learning_rate": 0.00010136015792258142, + "loss": 3.1142, + "step": 3352 + }, + { + "epoch": 4.92, + "grad_norm": 2.7994866371154785, + "learning_rate": 0.00010131325866990793, + "loss": 3.1242, + "step": 3353 + }, + { + "epoch": 4.92, + "grad_norm": 2.5278077125549316, + "learning_rate": 0.00010126635912832631, + "loss": 3.0951, + "step": 3354 + }, + { + "epoch": 4.92, + "grad_norm": 3.453552007675171, + "learning_rate": 0.00010121945930815422, + "loss": 3.1458, + "step": 3355 + }, + { + "epoch": 4.92, + "grad_norm": 3.2152633666992188, + "learning_rate": 0.00010117255921970921, + "loss": 3.1589, + "step": 3356 + }, + { + "epoch": 4.92, + "grad_norm": 1.388268232345581, + "learning_rate": 0.00010112565887330903, + "loss": 3.1496, + "step": 3357 + }, + { + "epoch": 4.92, + "grad_norm": 3.23463773727417, + "learning_rate": 0.00010107875827927147, + "loss": 3.1184, + "step": 3358 + }, + { + "epoch": 4.92, + "grad_norm": 2.970151662826538, + "learning_rate": 0.00010103185744791431, + "loss": 3.0931, + "step": 3359 + }, + { + "epoch": 4.93, + "grad_norm": 7.270925998687744, + "learning_rate": 0.00010098495638955538, + "loss": 3.1356, + "step": 3360 + }, + { + "epoch": 4.93, + "grad_norm": 18.72118377685547, + "learning_rate": 0.00010093805511451271, + "loss": 3.1538, + "step": 3361 + }, + { + "epoch": 4.93, + "grad_norm": 52.44887161254883, + "learning_rate": 0.00010089115363310417, + "loss": 3.1924, + "step": 3362 + }, + { + "epoch": 4.93, + "grad_norm": 7.027931213378906, + "learning_rate": 0.00010084425195564782, + "loss": 3.1726, + "step": 3363 + }, + { + "epoch": 4.93, + "grad_norm": 17.143966674804688, + "learning_rate": 0.00010079735009246167, + "loss": 3.1921, + "step": 3364 + }, + { + "epoch": 4.93, + "grad_norm": 3.896998167037964, + "learning_rate": 0.00010075044805386386, + "loss": 3.1341, + "step": 3365 + }, + { + "epoch": 4.94, + "grad_norm": 4.068105697631836, + "learning_rate": 0.0001007035458501725, + "loss": 3.1548, + "step": 3366 + }, + { + "epoch": 4.94, + "grad_norm": 3.3968491554260254, + "learning_rate": 0.00010065664349170578, + "loss": 3.146, + "step": 3367 + }, + { + "epoch": 4.94, + "grad_norm": 3.594057083129883, + "learning_rate": 0.00010060974098878189, + "loss": 3.1631, + "step": 3368 + }, + { + "epoch": 4.94, + "grad_norm": 0.987125813961029, + "learning_rate": 0.00010056283835171906, + "loss": 3.1319, + "step": 3369 + }, + { + "epoch": 4.94, + "grad_norm": 4.898683547973633, + "learning_rate": 0.00010051593559083557, + "loss": 3.1316, + "step": 3370 + }, + { + "epoch": 4.94, + "grad_norm": 6.864829063415527, + "learning_rate": 0.0001004690327164497, + "loss": 3.2037, + "step": 3371 + }, + { + "epoch": 4.94, + "grad_norm": 5.136176109313965, + "learning_rate": 0.00010042212973887982, + "loss": 3.1921, + "step": 3372 + }, + { + "epoch": 4.95, + "grad_norm": 2.621640205383301, + "learning_rate": 0.00010037522666844415, + "loss": 3.1716, + "step": 3373 + }, + { + "epoch": 4.95, + "grad_norm": 4.9672746658325195, + "learning_rate": 0.00010032832351546117, + "loss": 3.1374, + "step": 3374 + }, + { + "epoch": 4.95, + "grad_norm": 6.860297203063965, + "learning_rate": 0.00010028142029024923, + "loss": 3.1863, + "step": 3375 + }, + { + "epoch": 4.95, + "grad_norm": 131.51553344726562, + "learning_rate": 0.00010023451700312669, + "loss": 3.1629, + "step": 3376 + }, + { + "epoch": 4.95, + "grad_norm": 4.443822860717773, + "learning_rate": 0.00010018761366441198, + "loss": 3.1513, + "step": 3377 + }, + { + "epoch": 4.95, + "grad_norm": 1.4408992528915405, + "learning_rate": 0.00010014071028442354, + "loss": 3.1317, + "step": 3378 + }, + { + "epoch": 4.95, + "grad_norm": 8.419514656066895, + "learning_rate": 0.00010009380687347975, + "loss": 3.1649, + "step": 3379 + }, + { + "epoch": 4.96, + "grad_norm": 19.037832260131836, + "learning_rate": 0.00010004690344189912, + "loss": 3.1697, + "step": 3380 + }, + { + "epoch": 4.96, + "eval_loss": 3.100813865661621, + "eval_runtime": 1176.1523, + "eval_samples_per_second": 14.575, + "eval_steps_per_second": 0.911, + "step": 3380 + }, + { + "epoch": 4.96, + "grad_norm": 11.52230167388916, + "learning_rate": 0.0001, + "loss": 3.2739, + "step": 3381 + }, + { + "epoch": 4.96, + "grad_norm": 8.755014419555664, + "learning_rate": 9.995309655810093e-05, + "loss": 3.2172, + "step": 3382 + }, + { + "epoch": 4.96, + "grad_norm": 5.496139049530029, + "learning_rate": 9.990619312652026e-05, + "loss": 3.1656, + "step": 3383 + }, + { + "epoch": 4.96, + "grad_norm": 3.737701416015625, + "learning_rate": 9.985928971557649e-05, + "loss": 3.2014, + "step": 3384 + }, + { + "epoch": 4.96, + "grad_norm": 6.020711898803711, + "learning_rate": 9.981238633558805e-05, + "loss": 3.2036, + "step": 3385 + }, + { + "epoch": 4.96, + "grad_norm": 7.874726295471191, + "learning_rate": 9.976548299687332e-05, + "loss": 3.1735, + "step": 3386 + }, + { + "epoch": 4.97, + "grad_norm": 7.075169563293457, + "learning_rate": 9.97185797097508e-05, + "loss": 3.1674, + "step": 3387 + }, + { + "epoch": 4.97, + "grad_norm": 3.235295057296753, + "learning_rate": 9.967167648453885e-05, + "loss": 3.1999, + "step": 3388 + }, + { + "epoch": 4.97, + "grad_norm": 3.599179744720459, + "learning_rate": 9.962477333155585e-05, + "loss": 3.1829, + "step": 3389 + }, + { + "epoch": 4.97, + "grad_norm": 6.6362528800964355, + "learning_rate": 9.95778702611202e-05, + "loss": 3.2165, + "step": 3390 + }, + { + "epoch": 4.97, + "grad_norm": 7.1437153816223145, + "learning_rate": 9.953096728355031e-05, + "loss": 3.1884, + "step": 3391 + }, + { + "epoch": 4.97, + "grad_norm": 4.677809715270996, + "learning_rate": 9.948406440916443e-05, + "loss": 3.1899, + "step": 3392 + }, + { + "epoch": 4.98, + "grad_norm": 1.904357671737671, + "learning_rate": 9.943716164828096e-05, + "loss": 3.1357, + "step": 3393 + }, + { + "epoch": 4.98, + "grad_norm": 5.046501636505127, + "learning_rate": 9.939025901121814e-05, + "loss": 3.1666, + "step": 3394 + }, + { + "epoch": 4.98, + "grad_norm": 6.0771074295043945, + "learning_rate": 9.934335650829425e-05, + "loss": 3.1961, + "step": 3395 + }, + { + "epoch": 4.98, + "grad_norm": 5.269542694091797, + "learning_rate": 9.92964541498275e-05, + "loss": 3.1937, + "step": 3396 + }, + { + "epoch": 4.98, + "grad_norm": 2.9679410457611084, + "learning_rate": 9.924955194613616e-05, + "loss": 3.1376, + "step": 3397 + }, + { + "epoch": 4.98, + "grad_norm": 3.366711139678955, + "learning_rate": 9.920264990753837e-05, + "loss": 3.1116, + "step": 3398 + }, + { + "epoch": 4.98, + "grad_norm": 6.38279390335083, + "learning_rate": 9.91557480443522e-05, + "loss": 3.1385, + "step": 3399 + }, + { + "epoch": 4.99, + "grad_norm": 5.471691608428955, + "learning_rate": 9.910884636689585e-05, + "loss": 3.1874, + "step": 3400 + }, + { + "epoch": 4.99, + "grad_norm": 2.8460099697113037, + "learning_rate": 9.906194488548734e-05, + "loss": 3.1898, + "step": 3401 + }, + { + "epoch": 4.99, + "grad_norm": 3.193862199783325, + "learning_rate": 9.901504361044461e-05, + "loss": 3.1727, + "step": 3402 + }, + { + "epoch": 4.99, + "grad_norm": 4.644516944885254, + "learning_rate": 9.896814255208571e-05, + "loss": 3.1828, + "step": 3403 + }, + { + "epoch": 4.99, + "grad_norm": 4.342918872833252, + "learning_rate": 9.892124172072857e-05, + "loss": 3.1714, + "step": 3404 + }, + { + "epoch": 4.99, + "grad_norm": 2.4274344444274902, + "learning_rate": 9.887434112669097e-05, + "loss": 3.1509, + "step": 3405 + }, + { + "epoch": 4.99, + "grad_norm": 3.326143264770508, + "learning_rate": 9.882744078029081e-05, + "loss": 3.1433, + "step": 3406 + }, + { + "epoch": 5.0, + "grad_norm": 5.951056003570557, + "learning_rate": 9.878054069184583e-05, + "loss": 3.1543, + "step": 3407 + }, + { + "epoch": 5.0, + "grad_norm": 5.0941972732543945, + "learning_rate": 9.873364087167367e-05, + "loss": 3.1558, + "step": 3408 + }, + { + "epoch": 5.0, + "grad_norm": 2.075986623764038, + "learning_rate": 9.868674133009209e-05, + "loss": 3.144, + "step": 3409 + }, + { + "epoch": 5.0, + "grad_norm": 3.3112316131591797, + "learning_rate": 9.863984207741859e-05, + "loss": 3.1444, + "step": 3410 + }, + { + "epoch": 5.0, + "grad_norm": 4.910267353057861, + "learning_rate": 9.85929431239707e-05, + "loss": 3.1823, + "step": 3411 + }, + { + "epoch": 5.0, + "grad_norm": 4.1945624351501465, + "learning_rate": 9.854604448006594e-05, + "loss": 3.1945, + "step": 3412 + }, + { + "epoch": 5.01, + "grad_norm": 1.877642273902893, + "learning_rate": 9.849914615602163e-05, + "loss": 3.1357, + "step": 3413 + }, + { + "epoch": 5.01, + "grad_norm": 3.460303783416748, + "learning_rate": 9.845224816215508e-05, + "loss": 3.1039, + "step": 3414 + }, + { + "epoch": 5.01, + "grad_norm": 5.127685070037842, + "learning_rate": 9.84053505087836e-05, + "loss": 3.1041, + "step": 3415 + }, + { + "epoch": 5.01, + "grad_norm": 3.685288906097412, + "learning_rate": 9.835845320622434e-05, + "loss": 3.1874, + "step": 3416 + }, + { + "epoch": 5.01, + "grad_norm": 1.9846138954162598, + "learning_rate": 9.83115562647943e-05, + "loss": 3.1365, + "step": 3417 + }, + { + "epoch": 5.01, + "grad_norm": 5.992250442504883, + "learning_rate": 9.826465969481061e-05, + "loss": 3.0993, + "step": 3418 + }, + { + "epoch": 5.01, + "grad_norm": 4.008816242218018, + "learning_rate": 9.821776350659017e-05, + "loss": 3.1471, + "step": 3419 + }, + { + "epoch": 5.0, + "grad_norm": 2.797140598297119, + "learning_rate": 9.81708677104498e-05, + "loss": 3.1822, + "step": 3420 + }, + { + "epoch": 5.0, + "grad_norm": 3.2682175636291504, + "learning_rate": 9.812397231670624e-05, + "loss": 3.1501, + "step": 3421 + }, + { + "epoch": 5.0, + "grad_norm": 4.154157638549805, + "learning_rate": 9.807707733567626e-05, + "loss": 3.1572, + "step": 3422 + }, + { + "epoch": 5.01, + "grad_norm": 2.315260648727417, + "learning_rate": 9.803018277767634e-05, + "loss": 3.1305, + "step": 3423 + }, + { + "epoch": 5.01, + "grad_norm": 4.300922870635986, + "learning_rate": 9.798328865302299e-05, + "loss": 3.2001, + "step": 3424 + }, + { + "epoch": 5.01, + "grad_norm": 4.946724891662598, + "learning_rate": 9.793639497203268e-05, + "loss": 3.1431, + "step": 3425 + }, + { + "epoch": 5.01, + "grad_norm": 1.1593897342681885, + "learning_rate": 9.788950174502164e-05, + "loss": 3.1631, + "step": 3426 + }, + { + "epoch": 5.01, + "grad_norm": 88.28909301757812, + "learning_rate": 9.784260898230603e-05, + "loss": 3.1888, + "step": 3427 + }, + { + "epoch": 5.01, + "grad_norm": 7.40468168258667, + "learning_rate": 9.7795716694202e-05, + "loss": 3.1605, + "step": 3428 + }, + { + "epoch": 5.01, + "grad_norm": 8.019392967224121, + "learning_rate": 9.774882489102555e-05, + "loss": 3.2278, + "step": 3429 + }, + { + "epoch": 5.02, + "grad_norm": 5.9179205894470215, + "learning_rate": 9.770193358309248e-05, + "loss": 3.1715, + "step": 3430 + }, + { + "epoch": 5.02, + "grad_norm": 2.9003238677978516, + "learning_rate": 9.765504278071866e-05, + "loss": 3.1651, + "step": 3431 + }, + { + "epoch": 5.02, + "grad_norm": 4.174776554107666, + "learning_rate": 9.760815249421973e-05, + "loss": 3.1573, + "step": 3432 + }, + { + "epoch": 5.02, + "grad_norm": 5.618264198303223, + "learning_rate": 9.756126273391118e-05, + "loss": 3.1799, + "step": 3433 + }, + { + "epoch": 5.02, + "grad_norm": 5.217837333679199, + "learning_rate": 9.75143735101085e-05, + "loss": 3.1674, + "step": 3434 + }, + { + "epoch": 5.02, + "grad_norm": 6.881947994232178, + "learning_rate": 9.746748483312702e-05, + "loss": 3.2233, + "step": 3435 + }, + { + "epoch": 5.02, + "grad_norm": 4.155002117156982, + "learning_rate": 9.742059671328185e-05, + "loss": 3.1715, + "step": 3436 + }, + { + "epoch": 5.03, + "grad_norm": 3.3155972957611084, + "learning_rate": 9.737370916088815e-05, + "loss": 3.1688, + "step": 3437 + }, + { + "epoch": 5.03, + "grad_norm": 4.904897689819336, + "learning_rate": 9.732682218626086e-05, + "loss": 3.1702, + "step": 3438 + }, + { + "epoch": 5.03, + "grad_norm": 27.04771614074707, + "learning_rate": 9.727993579971471e-05, + "loss": 3.1816, + "step": 3439 + }, + { + "epoch": 5.03, + "grad_norm": 3.3083443641662598, + "learning_rate": 9.723305001156452e-05, + "loss": 3.1635, + "step": 3440 + }, + { + "epoch": 5.03, + "grad_norm": 3.0965356826782227, + "learning_rate": 9.718616483212474e-05, + "loss": 3.1631, + "step": 3441 + }, + { + "epoch": 5.03, + "grad_norm": 2.470313310623169, + "learning_rate": 9.713928027170989e-05, + "loss": 3.1788, + "step": 3442 + }, + { + "epoch": 5.04, + "grad_norm": 1.455441951751709, + "learning_rate": 9.709239634063414e-05, + "loss": 3.1543, + "step": 3443 + }, + { + "epoch": 5.04, + "grad_norm": 3.4748833179473877, + "learning_rate": 9.704551304921175e-05, + "loss": 3.1228, + "step": 3444 + }, + { + "epoch": 5.04, + "grad_norm": 3.481116771697998, + "learning_rate": 9.69986304077567e-05, + "loss": 3.1595, + "step": 3445 + }, + { + "epoch": 5.04, + "grad_norm": 4.942490577697754, + "learning_rate": 9.69517484265828e-05, + "loss": 3.1443, + "step": 3446 + }, + { + "epoch": 5.04, + "grad_norm": 3.2956652641296387, + "learning_rate": 9.690486711600384e-05, + "loss": 3.1699, + "step": 3447 + }, + { + "epoch": 5.04, + "grad_norm": 2.4343531131744385, + "learning_rate": 9.68579864863334e-05, + "loss": 3.1488, + "step": 3448 + }, + { + "epoch": 5.04, + "grad_norm": 2.82985782623291, + "learning_rate": 9.681110654788482e-05, + "loss": 3.1032, + "step": 3449 + }, + { + "epoch": 5.05, + "grad_norm": 1.59348464012146, + "learning_rate": 9.676422731097149e-05, + "loss": 3.1256, + "step": 3450 + }, + { + "epoch": 5.05, + "grad_norm": 4.060868740081787, + "learning_rate": 9.671734878590644e-05, + "loss": 3.1592, + "step": 3451 + }, + { + "epoch": 5.05, + "grad_norm": 4.718034744262695, + "learning_rate": 9.667047098300262e-05, + "loss": 3.1666, + "step": 3452 + }, + { + "epoch": 5.05, + "grad_norm": 1.5934293270111084, + "learning_rate": 9.662359391257293e-05, + "loss": 3.1592, + "step": 3453 + }, + { + "epoch": 5.05, + "grad_norm": 4.770664215087891, + "learning_rate": 9.657671758492992e-05, + "loss": 3.1303, + "step": 3454 + }, + { + "epoch": 5.05, + "grad_norm": 5.743010520935059, + "learning_rate": 9.652984201038608e-05, + "loss": 3.1736, + "step": 3455 + }, + { + "epoch": 5.05, + "grad_norm": 22.09770393371582, + "learning_rate": 9.64829671992538e-05, + "loss": 3.1555, + "step": 3456 + }, + { + "epoch": 5.06, + "grad_norm": 4.085333347320557, + "learning_rate": 9.643609316184514e-05, + "loss": 3.1203, + "step": 3457 + }, + { + "epoch": 5.06, + "grad_norm": 3.675048589706421, + "learning_rate": 9.638921990847206e-05, + "loss": 3.128, + "step": 3458 + }, + { + "epoch": 5.06, + "grad_norm": 1.0433051586151123, + "learning_rate": 9.634234744944646e-05, + "loss": 3.1206, + "step": 3459 + }, + { + "epoch": 5.06, + "grad_norm": 4.096065521240234, + "learning_rate": 9.62954757950799e-05, + "loss": 3.1567, + "step": 3460 + }, + { + "epoch": 5.06, + "grad_norm": 5.4872870445251465, + "learning_rate": 9.624860495568379e-05, + "loss": 3.1482, + "step": 3461 + }, + { + "epoch": 5.06, + "grad_norm": 3.236421585083008, + "learning_rate": 9.620173494156949e-05, + "loss": 3.1172, + "step": 3462 + }, + { + "epoch": 5.06, + "grad_norm": 3.7397096157073975, + "learning_rate": 9.615486576304806e-05, + "loss": 3.1485, + "step": 3463 + }, + { + "epoch": 5.07, + "grad_norm": 2.640143632888794, + "learning_rate": 9.610799743043036e-05, + "loss": 3.1444, + "step": 3464 + }, + { + "epoch": 5.07, + "grad_norm": 2.277892589569092, + "learning_rate": 9.606112995402714e-05, + "loss": 3.117, + "step": 3465 + }, + { + "epoch": 5.07, + "grad_norm": 2.7552220821380615, + "learning_rate": 9.601426334414898e-05, + "loss": 3.1557, + "step": 3466 + }, + { + "epoch": 5.07, + "grad_norm": 3.1442368030548096, + "learning_rate": 9.596739761110613e-05, + "loss": 3.1161, + "step": 3467 + }, + { + "epoch": 5.07, + "grad_norm": 1.997355341911316, + "learning_rate": 9.592053276520877e-05, + "loss": 3.1332, + "step": 3468 + }, + { + "epoch": 5.07, + "grad_norm": 1.959526777267456, + "learning_rate": 9.58736688167669e-05, + "loss": 3.141, + "step": 3469 + }, + { + "epoch": 5.08, + "grad_norm": 2.0642502307891846, + "learning_rate": 9.582680577609022e-05, + "loss": 3.1532, + "step": 3470 + }, + { + "epoch": 5.08, + "grad_norm": 1.1112334728240967, + "learning_rate": 9.57799436534883e-05, + "loss": 3.1402, + "step": 3471 + }, + { + "epoch": 5.08, + "grad_norm": 2.348271131515503, + "learning_rate": 9.57330824592705e-05, + "loss": 3.162, + "step": 3472 + }, + { + "epoch": 5.08, + "grad_norm": 1.7725739479064941, + "learning_rate": 9.5686222203746e-05, + "loss": 3.159, + "step": 3473 + }, + { + "epoch": 5.08, + "grad_norm": 1.3065656423568726, + "learning_rate": 9.563936289722363e-05, + "loss": 3.155, + "step": 3474 + }, + { + "epoch": 5.08, + "grad_norm": 2.1432132720947266, + "learning_rate": 9.559250455001223e-05, + "loss": 3.0939, + "step": 3475 + }, + { + "epoch": 5.08, + "grad_norm": 1.2344911098480225, + "learning_rate": 9.554564717242032e-05, + "loss": 3.1094, + "step": 3476 + }, + { + "epoch": 5.09, + "grad_norm": 1.591896653175354, + "learning_rate": 9.549879077475613e-05, + "loss": 3.1632, + "step": 3477 + }, + { + "epoch": 5.09, + "grad_norm": 1.2728568315505981, + "learning_rate": 9.545193536732783e-05, + "loss": 3.1537, + "step": 3478 + }, + { + "epoch": 5.09, + "grad_norm": 0.9794215559959412, + "learning_rate": 9.540508096044329e-05, + "loss": 3.1012, + "step": 3479 + }, + { + "epoch": 5.09, + "grad_norm": 2.092595100402832, + "learning_rate": 9.535822756441011e-05, + "loss": 3.1352, + "step": 3480 + }, + { + "epoch": 5.09, + "grad_norm": 1.6915316581726074, + "learning_rate": 9.531137518953577e-05, + "loss": 3.1332, + "step": 3481 + }, + { + "epoch": 5.09, + "grad_norm": 6.075295925140381, + "learning_rate": 9.526452384612749e-05, + "loss": 3.1654, + "step": 3482 + }, + { + "epoch": 5.09, + "grad_norm": 2.9814789295196533, + "learning_rate": 9.521767354449219e-05, + "loss": 3.1196, + "step": 3483 + }, + { + "epoch": 5.1, + "grad_norm": 1.2538803815841675, + "learning_rate": 9.51708242949367e-05, + "loss": 3.1304, + "step": 3484 + }, + { + "epoch": 5.1, + "grad_norm": 1.274577260017395, + "learning_rate": 9.512397610776747e-05, + "loss": 3.1293, + "step": 3485 + }, + { + "epoch": 5.1, + "grad_norm": 10.467679023742676, + "learning_rate": 9.507712899329079e-05, + "loss": 3.1396, + "step": 3486 + }, + { + "epoch": 5.1, + "grad_norm": 3.3941073417663574, + "learning_rate": 9.503028296181281e-05, + "loss": 3.1455, + "step": 3487 + }, + { + "epoch": 5.1, + "grad_norm": 1.5310273170471191, + "learning_rate": 9.498343802363923e-05, + "loss": 3.141, + "step": 3488 + }, + { + "epoch": 5.1, + "grad_norm": 4.921885967254639, + "learning_rate": 9.493659418907566e-05, + "loss": 3.1257, + "step": 3489 + }, + { + "epoch": 5.1, + "grad_norm": 6.093456268310547, + "learning_rate": 9.488975146842746e-05, + "loss": 3.1542, + "step": 3490 + }, + { + "epoch": 5.11, + "grad_norm": 2.32529616355896, + "learning_rate": 9.484290987199968e-05, + "loss": 3.1527, + "step": 3491 + }, + { + "epoch": 5.11, + "grad_norm": 4.923595905303955, + "learning_rate": 9.47960694100972e-05, + "loss": 3.1667, + "step": 3492 + }, + { + "epoch": 5.11, + "grad_norm": 5.732090950012207, + "learning_rate": 9.474923009302452e-05, + "loss": 3.1218, + "step": 3493 + }, + { + "epoch": 5.11, + "grad_norm": 3.330801486968994, + "learning_rate": 9.470239193108607e-05, + "loss": 3.1252, + "step": 3494 + }, + { + "epoch": 5.11, + "grad_norm": 3.7033092975616455, + "learning_rate": 9.465555493458588e-05, + "loss": 3.1358, + "step": 3495 + }, + { + "epoch": 5.11, + "grad_norm": 6.471573352813721, + "learning_rate": 9.460871911382774e-05, + "loss": 3.1961, + "step": 3496 + }, + { + "epoch": 5.12, + "grad_norm": 6.723712921142578, + "learning_rate": 9.456188447911533e-05, + "loss": 3.0915, + "step": 3497 + }, + { + "epoch": 5.12, + "grad_norm": 1.7002429962158203, + "learning_rate": 9.451505104075188e-05, + "loss": 3.1452, + "step": 3498 + }, + { + "epoch": 5.12, + "grad_norm": 2.4126813411712646, + "learning_rate": 9.44682188090404e-05, + "loss": 3.1457, + "step": 3499 + }, + { + "epoch": 5.12, + "grad_norm": 1.8296704292297363, + "learning_rate": 9.442138779428376e-05, + "loss": 3.1215, + "step": 3500 + }, + { + "epoch": 5.12, + "grad_norm": 1.7919952869415283, + "learning_rate": 9.43745580067844e-05, + "loss": 3.179, + "step": 3501 + }, + { + "epoch": 5.12, + "grad_norm": 1.8230031728744507, + "learning_rate": 9.432772945684454e-05, + "loss": 3.1339, + "step": 3502 + }, + { + "epoch": 5.12, + "grad_norm": 1.5876286029815674, + "learning_rate": 9.428090215476625e-05, + "loss": 3.1248, + "step": 3503 + }, + { + "epoch": 5.13, + "grad_norm": 2.1340315341949463, + "learning_rate": 9.423407611085113e-05, + "loss": 3.1396, + "step": 3504 + }, + { + "epoch": 5.13, + "grad_norm": 1.6990450620651245, + "learning_rate": 9.41872513354006e-05, + "loss": 3.1119, + "step": 3505 + }, + { + "epoch": 5.13, + "grad_norm": 1.792682409286499, + "learning_rate": 9.414042783871586e-05, + "loss": 3.1476, + "step": 3506 + }, + { + "epoch": 5.13, + "grad_norm": 2.685699701309204, + "learning_rate": 9.409360563109773e-05, + "loss": 3.1468, + "step": 3507 + }, + { + "epoch": 5.13, + "grad_norm": 4.750230312347412, + "learning_rate": 9.404678472284673e-05, + "loss": 3.1361, + "step": 3508 + }, + { + "epoch": 5.13, + "grad_norm": 1.0667452812194824, + "learning_rate": 9.399996512426323e-05, + "loss": 3.1625, + "step": 3509 + }, + { + "epoch": 5.13, + "grad_norm": 2.092228889465332, + "learning_rate": 9.395314684564721e-05, + "loss": 3.1709, + "step": 3510 + }, + { + "epoch": 5.14, + "grad_norm": 1.991498589515686, + "learning_rate": 9.390632989729831e-05, + "loss": 3.1122, + "step": 3511 + }, + { + "epoch": 5.14, + "grad_norm": 10.466729164123535, + "learning_rate": 9.385951428951603e-05, + "loss": 3.1214, + "step": 3512 + }, + { + "epoch": 5.14, + "grad_norm": 3.6561026573181152, + "learning_rate": 9.381270003259945e-05, + "loss": 3.1363, + "step": 3513 + }, + { + "epoch": 5.14, + "grad_norm": 1.4475880861282349, + "learning_rate": 9.37658871368474e-05, + "loss": 3.1801, + "step": 3514 + }, + { + "epoch": 5.14, + "grad_norm": 5.58400821685791, + "learning_rate": 9.371907561255837e-05, + "loss": 3.1223, + "step": 3515 + }, + { + "epoch": 5.14, + "grad_norm": 9.835494041442871, + "learning_rate": 9.367226547003065e-05, + "loss": 3.2195, + "step": 3516 + }, + { + "epoch": 5.14, + "grad_norm": 6.080451011657715, + "learning_rate": 9.362545671956214e-05, + "loss": 3.1594, + "step": 3517 + }, + { + "epoch": 5.15, + "grad_norm": 1.3072457313537598, + "learning_rate": 9.357864937145036e-05, + "loss": 3.147, + "step": 3518 + }, + { + "epoch": 5.15, + "grad_norm": 5.248061180114746, + "learning_rate": 9.353184343599273e-05, + "loss": 3.1497, + "step": 3519 + }, + { + "epoch": 5.15, + "grad_norm": 5.6261420249938965, + "learning_rate": 9.34850389234862e-05, + "loss": 3.1824, + "step": 3520 + }, + { + "epoch": 5.15, + "grad_norm": 3.202202320098877, + "learning_rate": 9.343823584422742e-05, + "loss": 3.1609, + "step": 3521 + }, + { + "epoch": 5.15, + "grad_norm": 4.45296573638916, + "learning_rate": 9.339143420851279e-05, + "loss": 3.1792, + "step": 3522 + }, + { + "epoch": 5.15, + "grad_norm": 4.815643310546875, + "learning_rate": 9.334463402663837e-05, + "loss": 3.2226, + "step": 3523 + }, + { + "epoch": 5.16, + "grad_norm": 12.902685165405273, + "learning_rate": 9.32978353088998e-05, + "loss": 3.1462, + "step": 3524 + }, + { + "epoch": 5.16, + "grad_norm": 3.254309892654419, + "learning_rate": 9.32510380655926e-05, + "loss": 3.1306, + "step": 3525 + }, + { + "epoch": 5.16, + "grad_norm": 8.259904861450195, + "learning_rate": 9.320424230701179e-05, + "loss": 3.2094, + "step": 3526 + }, + { + "epoch": 5.16, + "grad_norm": 2.651756525039673, + "learning_rate": 9.31574480434521e-05, + "loss": 3.1366, + "step": 3527 + }, + { + "epoch": 5.16, + "grad_norm": 4.296948432922363, + "learning_rate": 9.311065528520802e-05, + "loss": 3.1398, + "step": 3528 + }, + { + "epoch": 5.16, + "grad_norm": 4.362511157989502, + "learning_rate": 9.30638640425736e-05, + "loss": 3.163, + "step": 3529 + }, + { + "epoch": 5.16, + "grad_norm": 1.9980674982070923, + "learning_rate": 9.301707432584257e-05, + "loss": 3.1884, + "step": 3530 + }, + { + "epoch": 5.17, + "grad_norm": 4.281768321990967, + "learning_rate": 9.297028614530844e-05, + "loss": 3.1488, + "step": 3531 + }, + { + "epoch": 5.17, + "grad_norm": 5.395291328430176, + "learning_rate": 9.292349951126424e-05, + "loss": 3.1816, + "step": 3532 + }, + { + "epoch": 5.17, + "grad_norm": 1.9070907831192017, + "learning_rate": 9.287671443400268e-05, + "loss": 3.1082, + "step": 3533 + }, + { + "epoch": 5.17, + "grad_norm": 5.979832649230957, + "learning_rate": 9.282993092381625e-05, + "loss": 3.1332, + "step": 3534 + }, + { + "epoch": 5.17, + "grad_norm": 7.1100382804870605, + "learning_rate": 9.278314899099694e-05, + "loss": 3.2056, + "step": 3535 + }, + { + "epoch": 5.17, + "grad_norm": 22.313732147216797, + "learning_rate": 9.273636864583647e-05, + "loss": 3.1634, + "step": 3536 + }, + { + "epoch": 5.17, + "grad_norm": 11.737483978271484, + "learning_rate": 9.268958989862628e-05, + "loss": 3.1481, + "step": 3537 + }, + { + "epoch": 5.18, + "grad_norm": 2.638352394104004, + "learning_rate": 9.264281275965728e-05, + "loss": 3.1341, + "step": 3538 + }, + { + "epoch": 5.18, + "grad_norm": 2.807657241821289, + "learning_rate": 9.25960372392202e-05, + "loss": 3.1505, + "step": 3539 + }, + { + "epoch": 5.18, + "grad_norm": 20.156600952148438, + "learning_rate": 9.254926334760526e-05, + "loss": 3.1956, + "step": 3540 + }, + { + "epoch": 5.18, + "grad_norm": 2.164813280105591, + "learning_rate": 9.25024910951025e-05, + "loss": 3.1844, + "step": 3541 + }, + { + "epoch": 5.18, + "grad_norm": 1.7954508066177368, + "learning_rate": 9.245572049200142e-05, + "loss": 3.222, + "step": 3542 + }, + { + "epoch": 5.18, + "grad_norm": 4.100772380828857, + "learning_rate": 9.240895154859125e-05, + "loss": 3.1239, + "step": 3543 + }, + { + "epoch": 5.19, + "grad_norm": 3.8500773906707764, + "learning_rate": 9.236218427516092e-05, + "loss": 3.1278, + "step": 3544 + }, + { + "epoch": 5.19, + "grad_norm": 2.076373815536499, + "learning_rate": 9.231541868199883e-05, + "loss": 3.164, + "step": 3545 + }, + { + "epoch": 5.19, + "grad_norm": 1.8136000633239746, + "learning_rate": 9.226865477939313e-05, + "loss": 3.1479, + "step": 3546 + }, + { + "epoch": 5.19, + "grad_norm": 128.6366424560547, + "learning_rate": 9.222189257763158e-05, + "loss": 3.1239, + "step": 3547 + }, + { + "epoch": 5.19, + "grad_norm": 3.7925548553466797, + "learning_rate": 9.217513208700154e-05, + "loss": 3.1402, + "step": 3548 + }, + { + "epoch": 5.19, + "grad_norm": 11.528712272644043, + "learning_rate": 9.212837331778997e-05, + "loss": 3.1434, + "step": 3549 + }, + { + "epoch": 5.19, + "eval_loss": 2.9914677143096924, + "eval_runtime": 1173.7001, + "eval_samples_per_second": 14.605, + "eval_steps_per_second": 0.913, + "step": 3549 + }, + { + "epoch": 5.19, + "grad_norm": 3.5686726570129395, + "learning_rate": 9.208161628028355e-05, + "loss": 3.1025, + "step": 3550 + }, + { + "epoch": 5.2, + "grad_norm": 19.285165786743164, + "learning_rate": 9.203486098476849e-05, + "loss": 3.1957, + "step": 3551 + }, + { + "epoch": 5.2, + "grad_norm": 7.824908256530762, + "learning_rate": 9.19881074415306e-05, + "loss": 3.1752, + "step": 3552 + }, + { + "epoch": 5.2, + "grad_norm": 3.9910645484924316, + "learning_rate": 9.194135566085537e-05, + "loss": 3.1751, + "step": 3553 + }, + { + "epoch": 5.2, + "grad_norm": 32.02030944824219, + "learning_rate": 9.189460565302794e-05, + "loss": 3.1539, + "step": 3554 + }, + { + "epoch": 5.2, + "grad_norm": 10.703150749206543, + "learning_rate": 9.18478574283329e-05, + "loss": 3.1623, + "step": 3555 + }, + { + "epoch": 5.2, + "grad_norm": 37.447715759277344, + "learning_rate": 9.180111099705459e-05, + "loss": 3.1512, + "step": 3556 + }, + { + "epoch": 5.2, + "grad_norm": 4.359159469604492, + "learning_rate": 9.175436636947697e-05, + "loss": 3.1754, + "step": 3557 + }, + { + "epoch": 5.21, + "grad_norm": 11.65587329864502, + "learning_rate": 9.170762355588342e-05, + "loss": 3.1365, + "step": 3558 + }, + { + "epoch": 5.21, + "grad_norm": 6.701794147491455, + "learning_rate": 9.166088256655715e-05, + "loss": 3.1705, + "step": 3559 + }, + { + "epoch": 5.21, + "grad_norm": 12.4017915725708, + "learning_rate": 9.161414341178084e-05, + "loss": 3.1724, + "step": 3560 + }, + { + "epoch": 5.21, + "grad_norm": 2.782519578933716, + "learning_rate": 9.156740610183674e-05, + "loss": 3.1809, + "step": 3561 + }, + { + "epoch": 5.21, + "grad_norm": 3.470179319381714, + "learning_rate": 9.152067064700683e-05, + "loss": 3.1531, + "step": 3562 + }, + { + "epoch": 5.21, + "grad_norm": 12.016058921813965, + "learning_rate": 9.147393705757254e-05, + "loss": 3.1714, + "step": 3563 + }, + { + "epoch": 5.21, + "grad_norm": 8.318673133850098, + "learning_rate": 9.142720534381497e-05, + "loss": 3.178, + "step": 3564 + }, + { + "epoch": 5.22, + "grad_norm": 26.07087516784668, + "learning_rate": 9.138047551601473e-05, + "loss": 3.1636, + "step": 3565 + }, + { + "epoch": 5.22, + "grad_norm": 55.82182312011719, + "learning_rate": 9.133374758445212e-05, + "loss": 3.1598, + "step": 3566 + }, + { + "epoch": 5.22, + "grad_norm": 3.5638935565948486, + "learning_rate": 9.1287021559407e-05, + "loss": 3.1711, + "step": 3567 + }, + { + "epoch": 5.22, + "grad_norm": 4.856345176696777, + "learning_rate": 9.12402974511587e-05, + "loss": 3.1686, + "step": 3568 + }, + { + "epoch": 5.22, + "grad_norm": 6.552631855010986, + "learning_rate": 9.119357526998628e-05, + "loss": 3.1437, + "step": 3569 + }, + { + "epoch": 5.22, + "grad_norm": 62.54035186767578, + "learning_rate": 9.11468550261683e-05, + "loss": 3.1711, + "step": 3570 + }, + { + "epoch": 5.23, + "grad_norm": 19.124649047851562, + "learning_rate": 9.110013672998284e-05, + "loss": 3.2002, + "step": 3571 + }, + { + "epoch": 5.23, + "grad_norm": 14.90050983428955, + "learning_rate": 9.105342039170768e-05, + "loss": 3.2217, + "step": 3572 + }, + { + "epoch": 5.23, + "grad_norm": 4.7844109535217285, + "learning_rate": 9.100670602162009e-05, + "loss": 3.1506, + "step": 3573 + }, + { + "epoch": 5.23, + "grad_norm": 1.9373784065246582, + "learning_rate": 9.095999362999687e-05, + "loss": 3.1288, + "step": 3574 + }, + { + "epoch": 5.23, + "grad_norm": 60.557308197021484, + "learning_rate": 9.091328322711449e-05, + "loss": 3.1368, + "step": 3575 + }, + { + "epoch": 5.23, + "grad_norm": 6.0116658210754395, + "learning_rate": 9.086657482324892e-05, + "loss": 3.1685, + "step": 3576 + }, + { + "epoch": 5.23, + "grad_norm": 7.286896228790283, + "learning_rate": 9.081986842867562e-05, + "loss": 3.1756, + "step": 3577 + }, + { + "epoch": 5.24, + "grad_norm": 2.469630479812622, + "learning_rate": 9.077316405366981e-05, + "loss": 3.1425, + "step": 3578 + }, + { + "epoch": 5.24, + "grad_norm": 3.226400136947632, + "learning_rate": 9.072646170850605e-05, + "loss": 3.1015, + "step": 3579 + }, + { + "epoch": 5.24, + "grad_norm": 19.016260147094727, + "learning_rate": 9.067976140345854e-05, + "loss": 3.1746, + "step": 3580 + }, + { + "epoch": 5.24, + "grad_norm": 5.787333011627197, + "learning_rate": 9.06330631488011e-05, + "loss": 3.1975, + "step": 3581 + }, + { + "epoch": 5.24, + "grad_norm": 5.138857841491699, + "learning_rate": 9.058636695480698e-05, + "loss": 3.181, + "step": 3582 + }, + { + "epoch": 5.24, + "grad_norm": 11.81260871887207, + "learning_rate": 9.053967283174901e-05, + "loss": 3.1447, + "step": 3583 + }, + { + "epoch": 5.24, + "grad_norm": 6.516455173492432, + "learning_rate": 9.049298078989967e-05, + "loss": 3.1348, + "step": 3584 + }, + { + "epoch": 5.25, + "grad_norm": 6.460428237915039, + "learning_rate": 9.044629083953082e-05, + "loss": 3.1878, + "step": 3585 + }, + { + "epoch": 5.25, + "grad_norm": 5.189371585845947, + "learning_rate": 9.039960299091397e-05, + "loss": 3.1585, + "step": 3586 + }, + { + "epoch": 5.25, + "grad_norm": 4.265483379364014, + "learning_rate": 9.035291725432008e-05, + "loss": 3.1688, + "step": 3587 + }, + { + "epoch": 5.25, + "grad_norm": 2.625492572784424, + "learning_rate": 9.030623364001979e-05, + "loss": 3.1438, + "step": 3588 + }, + { + "epoch": 5.25, + "grad_norm": 3.3770155906677246, + "learning_rate": 9.025955215828312e-05, + "loss": 3.1574, + "step": 3589 + }, + { + "epoch": 5.25, + "grad_norm": 21.14669418334961, + "learning_rate": 9.021287281937967e-05, + "loss": 3.1473, + "step": 3590 + }, + { + "epoch": 5.25, + "grad_norm": 30.993927001953125, + "learning_rate": 9.016619563357865e-05, + "loss": 3.2186, + "step": 3591 + }, + { + "epoch": 5.26, + "grad_norm": 5.043474197387695, + "learning_rate": 9.011952061114867e-05, + "loss": 3.1602, + "step": 3592 + }, + { + "epoch": 5.26, + "grad_norm": 17.649402618408203, + "learning_rate": 9.007284776235791e-05, + "loss": 3.139, + "step": 3593 + }, + { + "epoch": 5.26, + "grad_norm": 50.776214599609375, + "learning_rate": 9.002617709747416e-05, + "loss": 3.1279, + "step": 3594 + }, + { + "epoch": 5.26, + "grad_norm": 12.535754203796387, + "learning_rate": 8.997950862676462e-05, + "loss": 3.174, + "step": 3595 + }, + { + "epoch": 5.26, + "grad_norm": 4.332036972045898, + "learning_rate": 8.993284236049598e-05, + "loss": 3.106, + "step": 3596 + }, + { + "epoch": 5.26, + "grad_norm": 1.4636400938034058, + "learning_rate": 8.988617830893455e-05, + "loss": 3.1456, + "step": 3597 + }, + { + "epoch": 5.27, + "grad_norm": 4.1321024894714355, + "learning_rate": 8.983951648234617e-05, + "loss": 3.102, + "step": 3598 + }, + { + "epoch": 5.27, + "grad_norm": 4.079101085662842, + "learning_rate": 8.979285689099601e-05, + "loss": 3.1632, + "step": 3599 + }, + { + "epoch": 5.27, + "grad_norm": 7.1509690284729, + "learning_rate": 8.974619954514895e-05, + "loss": 3.1862, + "step": 3600 + }, + { + "epoch": 5.27, + "grad_norm": 3.1132075786590576, + "learning_rate": 8.96995444550693e-05, + "loss": 3.1587, + "step": 3601 + }, + { + "epoch": 5.27, + "grad_norm": 15.199933052062988, + "learning_rate": 8.965289163102078e-05, + "loss": 3.1385, + "step": 3602 + }, + { + "epoch": 5.27, + "grad_norm": 3.226987600326538, + "learning_rate": 8.960624108326679e-05, + "loss": 3.1793, + "step": 3603 + }, + { + "epoch": 5.27, + "grad_norm": 3.0855495929718018, + "learning_rate": 8.955959282207013e-05, + "loss": 3.121, + "step": 3604 + }, + { + "epoch": 5.28, + "grad_norm": 1.8328437805175781, + "learning_rate": 8.951294685769303e-05, + "loss": 3.1515, + "step": 3605 + }, + { + "epoch": 5.28, + "grad_norm": 2.6738593578338623, + "learning_rate": 8.946630320039737e-05, + "loss": 3.1412, + "step": 3606 + }, + { + "epoch": 5.28, + "grad_norm": 2.790065288543701, + "learning_rate": 8.941966186044444e-05, + "loss": 3.1307, + "step": 3607 + }, + { + "epoch": 5.28, + "grad_norm": 3.6638455390930176, + "learning_rate": 8.937302284809494e-05, + "loss": 3.1155, + "step": 3608 + }, + { + "epoch": 5.28, + "grad_norm": 3.6409106254577637, + "learning_rate": 8.932638617360925e-05, + "loss": 3.1242, + "step": 3609 + }, + { + "epoch": 5.28, + "grad_norm": 2.8414742946624756, + "learning_rate": 8.927975184724704e-05, + "loss": 3.1433, + "step": 3610 + }, + { + "epoch": 5.28, + "grad_norm": 4.518361568450928, + "learning_rate": 8.92331198792676e-05, + "loss": 3.1883, + "step": 3611 + }, + { + "epoch": 5.29, + "grad_norm": 7.363691806793213, + "learning_rate": 8.918649027992958e-05, + "loss": 3.1231, + "step": 3612 + }, + { + "epoch": 5.29, + "grad_norm": 2.302523136138916, + "learning_rate": 8.913986305949125e-05, + "loss": 3.0972, + "step": 3613 + }, + { + "epoch": 5.29, + "grad_norm": 1.6068569421768188, + "learning_rate": 8.909323822821031e-05, + "loss": 3.1498, + "step": 3614 + }, + { + "epoch": 5.29, + "grad_norm": 1.9950542449951172, + "learning_rate": 8.90466157963438e-05, + "loss": 3.1339, + "step": 3615 + }, + { + "epoch": 5.29, + "grad_norm": 1.7338231801986694, + "learning_rate": 8.899999577414845e-05, + "loss": 3.1382, + "step": 3616 + }, + { + "epoch": 5.29, + "grad_norm": 2.231691598892212, + "learning_rate": 8.895337817188031e-05, + "loss": 3.1417, + "step": 3617 + }, + { + "epoch": 5.29, + "grad_norm": 1.9115151166915894, + "learning_rate": 8.890676299979492e-05, + "loss": 3.1338, + "step": 3618 + }, + { + "epoch": 5.3, + "grad_norm": 2.1142683029174805, + "learning_rate": 8.886015026814736e-05, + "loss": 3.1768, + "step": 3619 + }, + { + "epoch": 5.3, + "grad_norm": 2.7301149368286133, + "learning_rate": 8.881353998719206e-05, + "loss": 3.1843, + "step": 3620 + }, + { + "epoch": 5.3, + "grad_norm": 3.0939254760742188, + "learning_rate": 8.876693216718297e-05, + "loss": 3.1582, + "step": 3621 + }, + { + "epoch": 5.3, + "grad_norm": 1.2294076681137085, + "learning_rate": 8.872032681837358e-05, + "loss": 3.1484, + "step": 3622 + }, + { + "epoch": 5.3, + "grad_norm": 2.869823694229126, + "learning_rate": 8.867372395101666e-05, + "loss": 3.1455, + "step": 3623 + }, + { + "epoch": 5.3, + "grad_norm": 3.04744291305542, + "learning_rate": 8.862712357536454e-05, + "loss": 3.151, + "step": 3624 + }, + { + "epoch": 5.31, + "grad_norm": 1.6261351108551025, + "learning_rate": 8.858052570166908e-05, + "loss": 3.1434, + "step": 3625 + }, + { + "epoch": 5.31, + "grad_norm": 2.8542559146881104, + "learning_rate": 8.85339303401814e-05, + "loss": 3.1384, + "step": 3626 + }, + { + "epoch": 5.31, + "grad_norm": 0.9066136479377747, + "learning_rate": 8.848733750115218e-05, + "loss": 3.1576, + "step": 3627 + }, + { + "epoch": 5.31, + "grad_norm": 2.466120719909668, + "learning_rate": 8.84407471948316e-05, + "loss": 3.1478, + "step": 3628 + }, + { + "epoch": 5.31, + "grad_norm": 1.9956951141357422, + "learning_rate": 8.839415943146919e-05, + "loss": 3.1492, + "step": 3629 + }, + { + "epoch": 5.31, + "grad_norm": 10.860211372375488, + "learning_rate": 8.834757422131385e-05, + "loss": 3.198, + "step": 3630 + }, + { + "epoch": 5.31, + "grad_norm": 4.2302327156066895, + "learning_rate": 8.830099157461413e-05, + "loss": 3.146, + "step": 3631 + }, + { + "epoch": 5.32, + "grad_norm": 13.344452857971191, + "learning_rate": 8.825441150161788e-05, + "loss": 3.1234, + "step": 3632 + }, + { + "epoch": 5.32, + "grad_norm": 3.983452558517456, + "learning_rate": 8.820783401257236e-05, + "loss": 3.1775, + "step": 3633 + }, + { + "epoch": 5.32, + "grad_norm": 2.76181960105896, + "learning_rate": 8.816125911772433e-05, + "loss": 3.1776, + "step": 3634 + }, + { + "epoch": 5.32, + "grad_norm": 5.169229030609131, + "learning_rate": 8.811468682731997e-05, + "loss": 3.1174, + "step": 3635 + }, + { + "epoch": 5.32, + "grad_norm": 2.876774549484253, + "learning_rate": 8.806811715160485e-05, + "loss": 3.1506, + "step": 3636 + }, + { + "epoch": 5.32, + "grad_norm": 1.740065336227417, + "learning_rate": 8.802155010082396e-05, + "loss": 3.1375, + "step": 3637 + }, + { + "epoch": 5.32, + "grad_norm": 2.2916555404663086, + "learning_rate": 8.797498568522182e-05, + "loss": 3.1736, + "step": 3638 + }, + { + "epoch": 5.33, + "grad_norm": 13.75604248046875, + "learning_rate": 8.792842391504222e-05, + "loss": 3.1382, + "step": 3639 + }, + { + "epoch": 5.33, + "grad_norm": 8.454766273498535, + "learning_rate": 8.788186480052844e-05, + "loss": 3.1663, + "step": 3640 + }, + { + "epoch": 5.33, + "grad_norm": 3.627807140350342, + "learning_rate": 8.783530835192322e-05, + "loss": 3.1633, + "step": 3641 + }, + { + "epoch": 5.33, + "grad_norm": 1.9736120700836182, + "learning_rate": 8.778875457946866e-05, + "loss": 3.2027, + "step": 3642 + }, + { + "epoch": 5.33, + "grad_norm": 4.044384479522705, + "learning_rate": 8.77422034934062e-05, + "loss": 3.1484, + "step": 3643 + }, + { + "epoch": 5.33, + "grad_norm": 4.4502763748168945, + "learning_rate": 8.769565510397684e-05, + "loss": 3.1589, + "step": 3644 + }, + { + "epoch": 5.34, + "grad_norm": 2.187070369720459, + "learning_rate": 8.764910942142093e-05, + "loss": 3.1593, + "step": 3645 + }, + { + "epoch": 5.34, + "grad_norm": 2.6429078578948975, + "learning_rate": 8.760256645597813e-05, + "loss": 3.1466, + "step": 3646 + }, + { + "epoch": 5.34, + "grad_norm": 1.6300479173660278, + "learning_rate": 8.755602621788764e-05, + "loss": 3.1133, + "step": 3647 + }, + { + "epoch": 5.34, + "grad_norm": 2.9914662837982178, + "learning_rate": 8.750948871738802e-05, + "loss": 3.114, + "step": 3648 + }, + { + "epoch": 5.34, + "grad_norm": 2.315000534057617, + "learning_rate": 8.746295396471711e-05, + "loss": 3.1427, + "step": 3649 + }, + { + "epoch": 5.34, + "grad_norm": 3.0107791423797607, + "learning_rate": 8.741642197011235e-05, + "loss": 3.1085, + "step": 3650 + }, + { + "epoch": 5.34, + "grad_norm": 11.682272911071777, + "learning_rate": 8.736989274381045e-05, + "loss": 3.1428, + "step": 3651 + }, + { + "epoch": 5.35, + "grad_norm": 2.8496906757354736, + "learning_rate": 8.732336629604745e-05, + "loss": 3.1269, + "step": 3652 + }, + { + "epoch": 5.35, + "grad_norm": 2.9771509170532227, + "learning_rate": 8.727684263705895e-05, + "loss": 3.147, + "step": 3653 + }, + { + "epoch": 5.35, + "grad_norm": 2.937957525253296, + "learning_rate": 8.723032177707978e-05, + "loss": 3.1731, + "step": 3654 + }, + { + "epoch": 5.35, + "grad_norm": 3.5484066009521484, + "learning_rate": 8.718380372634422e-05, + "loss": 3.1602, + "step": 3655 + }, + { + "epoch": 5.35, + "grad_norm": 9.137898445129395, + "learning_rate": 8.7137288495086e-05, + "loss": 3.1692, + "step": 3656 + }, + { + "epoch": 5.35, + "grad_norm": 10.832776069641113, + "learning_rate": 8.709077609353809e-05, + "loss": 3.1829, + "step": 3657 + }, + { + "epoch": 5.35, + "grad_norm": 3.5576894283294678, + "learning_rate": 8.704426653193289e-05, + "loss": 3.1439, + "step": 3658 + }, + { + "epoch": 5.36, + "grad_norm": 2.771862745285034, + "learning_rate": 8.699775982050229e-05, + "loss": 3.1435, + "step": 3659 + }, + { + "epoch": 5.36, + "grad_norm": 7.5768046379089355, + "learning_rate": 8.695125596947736e-05, + "loss": 3.1898, + "step": 3660 + }, + { + "epoch": 5.36, + "grad_norm": 3.332085132598877, + "learning_rate": 8.690475498908868e-05, + "loss": 3.1365, + "step": 3661 + }, + { + "epoch": 5.36, + "grad_norm": 8.380770683288574, + "learning_rate": 8.685825688956613e-05, + "loss": 3.1733, + "step": 3662 + }, + { + "epoch": 5.36, + "grad_norm": 9.0435791015625, + "learning_rate": 8.6811761681139e-05, + "loss": 3.1752, + "step": 3663 + }, + { + "epoch": 5.36, + "grad_norm": 12.012619018554688, + "learning_rate": 8.676526937403591e-05, + "loss": 3.1615, + "step": 3664 + }, + { + "epoch": 5.36, + "grad_norm": 6.051881313323975, + "learning_rate": 8.671877997848484e-05, + "loss": 3.1719, + "step": 3665 + }, + { + "epoch": 5.37, + "grad_norm": 5.059371471405029, + "learning_rate": 8.667229350471322e-05, + "loss": 3.1775, + "step": 3666 + }, + { + "epoch": 5.37, + "grad_norm": 9.862431526184082, + "learning_rate": 8.662580996294767e-05, + "loss": 3.1795, + "step": 3667 + }, + { + "epoch": 5.37, + "grad_norm": 15.149982452392578, + "learning_rate": 8.657932936341428e-05, + "loss": 3.2141, + "step": 3668 + }, + { + "epoch": 5.37, + "grad_norm": 9.565960884094238, + "learning_rate": 8.653285171633855e-05, + "loss": 3.1831, + "step": 3669 + }, + { + "epoch": 5.37, + "grad_norm": 19.23344612121582, + "learning_rate": 8.648637703194516e-05, + "loss": 3.1925, + "step": 3670 + }, + { + "epoch": 5.37, + "grad_norm": 6.268587589263916, + "learning_rate": 8.643990532045822e-05, + "loss": 3.1621, + "step": 3671 + }, + { + "epoch": 5.38, + "grad_norm": 56.983970642089844, + "learning_rate": 8.639343659210131e-05, + "loss": 3.1715, + "step": 3672 + }, + { + "epoch": 5.38, + "grad_norm": 50.29884719848633, + "learning_rate": 8.634697085709711e-05, + "loss": 3.1474, + "step": 3673 + }, + { + "epoch": 5.38, + "grad_norm": 13.747417449951172, + "learning_rate": 8.630050812566781e-05, + "loss": 3.1805, + "step": 3674 + }, + { + "epoch": 5.38, + "grad_norm": 20.72515106201172, + "learning_rate": 8.625404840803498e-05, + "loss": 3.1528, + "step": 3675 + }, + { + "epoch": 5.38, + "grad_norm": 12.049095153808594, + "learning_rate": 8.620759171441935e-05, + "loss": 3.181, + "step": 3676 + }, + { + "epoch": 5.38, + "grad_norm": 11.251856803894043, + "learning_rate": 8.616113805504106e-05, + "loss": 3.1962, + "step": 3677 + }, + { + "epoch": 5.38, + "grad_norm": 6.218480587005615, + "learning_rate": 8.61146874401197e-05, + "loss": 3.2132, + "step": 3678 + }, + { + "epoch": 5.39, + "grad_norm": 10.480876922607422, + "learning_rate": 8.606823987987405e-05, + "loss": 3.1913, + "step": 3679 + }, + { + "epoch": 5.39, + "grad_norm": 12.099661827087402, + "learning_rate": 8.602179538452222e-05, + "loss": 3.178, + "step": 3680 + }, + { + "epoch": 5.39, + "grad_norm": 23.720500946044922, + "learning_rate": 8.597535396428177e-05, + "loss": 3.1558, + "step": 3681 + }, + { + "epoch": 5.39, + "grad_norm": 16.246944427490234, + "learning_rate": 8.592891562936947e-05, + "loss": 3.1834, + "step": 3682 + }, + { + "epoch": 5.39, + "grad_norm": 37.614784240722656, + "learning_rate": 8.58824803900014e-05, + "loss": 3.174, + "step": 3683 + }, + { + "epoch": 5.39, + "grad_norm": 9.572596549987793, + "learning_rate": 8.583604825639302e-05, + "loss": 3.1871, + "step": 3684 + }, + { + "epoch": 5.39, + "grad_norm": 6.099634170532227, + "learning_rate": 8.578961923875915e-05, + "loss": 3.1663, + "step": 3685 + }, + { + "epoch": 5.4, + "grad_norm": 18.694616317749023, + "learning_rate": 8.57431933473138e-05, + "loss": 3.1668, + "step": 3686 + }, + { + "epoch": 5.4, + "grad_norm": 133.21104431152344, + "learning_rate": 8.569677059227032e-05, + "loss": 3.1881, + "step": 3687 + }, + { + "epoch": 5.4, + "grad_norm": 33.713504791259766, + "learning_rate": 8.565035098384149e-05, + "loss": 3.1648, + "step": 3688 + }, + { + "epoch": 5.4, + "grad_norm": 9.273075103759766, + "learning_rate": 8.56039345322393e-05, + "loss": 3.2306, + "step": 3689 + }, + { + "epoch": 5.4, + "grad_norm": 18.504985809326172, + "learning_rate": 8.555752124767498e-05, + "loss": 3.2082, + "step": 3690 + }, + { + "epoch": 5.4, + "grad_norm": 13.903569221496582, + "learning_rate": 8.551111114035922e-05, + "loss": 3.1519, + "step": 3691 + }, + { + "epoch": 5.4, + "grad_norm": 48.394412994384766, + "learning_rate": 8.546470422050193e-05, + "loss": 3.1922, + "step": 3692 + }, + { + "epoch": 5.41, + "grad_norm": 9.13818073272705, + "learning_rate": 8.541830049831227e-05, + "loss": 3.2432, + "step": 3693 + }, + { + "epoch": 5.41, + "grad_norm": 7.568233013153076, + "learning_rate": 8.537189998399879e-05, + "loss": 3.2007, + "step": 3694 + }, + { + "epoch": 5.41, + "grad_norm": 110.37165832519531, + "learning_rate": 8.53255026877693e-05, + "loss": 3.1837, + "step": 3695 + }, + { + "epoch": 5.41, + "grad_norm": 8.272992134094238, + "learning_rate": 8.527910861983084e-05, + "loss": 3.169, + "step": 3696 + }, + { + "epoch": 5.41, + "grad_norm": 29.426700592041016, + "learning_rate": 8.523271779038986e-05, + "loss": 3.1793, + "step": 3697 + }, + { + "epoch": 5.41, + "grad_norm": 9.58652400970459, + "learning_rate": 8.518633020965199e-05, + "loss": 3.1662, + "step": 3698 + }, + { + "epoch": 5.42, + "grad_norm": 5.326894283294678, + "learning_rate": 8.513994588782217e-05, + "loss": 3.1817, + "step": 3699 + }, + { + "epoch": 5.42, + "grad_norm": 8.72939395904541, + "learning_rate": 8.509356483510471e-05, + "loss": 3.1888, + "step": 3700 + }, + { + "epoch": 5.42, + "grad_norm": 9.870657920837402, + "learning_rate": 8.504718706170306e-05, + "loss": 3.1772, + "step": 3701 + }, + { + "epoch": 5.42, + "grad_norm": 39.32550811767578, + "learning_rate": 8.500081257782003e-05, + "loss": 3.203, + "step": 3702 + }, + { + "epoch": 5.42, + "grad_norm": 20.613840103149414, + "learning_rate": 8.495444139365776e-05, + "loss": 3.1721, + "step": 3703 + }, + { + "epoch": 5.42, + "grad_norm": 10.104787826538086, + "learning_rate": 8.490807351941753e-05, + "loss": 3.1791, + "step": 3704 + }, + { + "epoch": 5.42, + "grad_norm": 8.861245155334473, + "learning_rate": 8.486170896529996e-05, + "loss": 3.2222, + "step": 3705 + }, + { + "epoch": 5.43, + "grad_norm": 40.4407844543457, + "learning_rate": 8.481534774150502e-05, + "loss": 3.3014, + "step": 3706 + }, + { + "epoch": 5.43, + "grad_norm": 4.760523319244385, + "learning_rate": 8.476898985823177e-05, + "loss": 3.1825, + "step": 3707 + }, + { + "epoch": 5.43, + "grad_norm": 1924.5662841796875, + "learning_rate": 8.472263532567871e-05, + "loss": 3.1802, + "step": 3708 + }, + { + "epoch": 5.43, + "grad_norm": 5.515054702758789, + "learning_rate": 8.467628415404345e-05, + "loss": 3.1833, + "step": 3709 + }, + { + "epoch": 5.43, + "grad_norm": 2.0650458335876465, + "learning_rate": 8.462993635352303e-05, + "loss": 3.184, + "step": 3710 + }, + { + "epoch": 5.43, + "grad_norm": 4.477024078369141, + "learning_rate": 8.458359193431358e-05, + "loss": 3.1489, + "step": 3711 + }, + { + "epoch": 5.43, + "grad_norm": 8.2962646484375, + "learning_rate": 8.453725090661057e-05, + "loss": 3.1318, + "step": 3712 + }, + { + "epoch": 5.44, + "grad_norm": 4.55861234664917, + "learning_rate": 8.449091328060877e-05, + "loss": 3.185, + "step": 3713 + }, + { + "epoch": 5.44, + "grad_norm": 6.914697647094727, + "learning_rate": 8.44445790665021e-05, + "loss": 3.1659, + "step": 3714 + }, + { + "epoch": 5.44, + "grad_norm": 5.971372127532959, + "learning_rate": 8.439824827448376e-05, + "loss": 3.1691, + "step": 3715 + }, + { + "epoch": 5.44, + "grad_norm": 3.358804702758789, + "learning_rate": 8.435192091474628e-05, + "loss": 3.187, + "step": 3716 + }, + { + "epoch": 5.44, + "grad_norm": 50.52218246459961, + "learning_rate": 8.430559699748132e-05, + "loss": 3.1595, + "step": 3717 + }, + { + "epoch": 5.44, + "grad_norm": 4.3211774826049805, + "learning_rate": 8.425927653287981e-05, + "loss": 3.2301, + "step": 3718 + }, + { + "epoch": 5.44, + "eval_loss": 3.003319025039673, + "eval_runtime": 1168.0997, + "eval_samples_per_second": 14.675, + "eval_steps_per_second": 0.918, + "step": 3718 + }, + { + "epoch": 5.44, + "grad_norm": 3.5544116497039795, + "learning_rate": 8.421295953113201e-05, + "loss": 3.1628, + "step": 3719 + }, + { + "epoch": 5.45, + "grad_norm": 4.997739791870117, + "learning_rate": 8.416664600242733e-05, + "loss": 3.1741, + "step": 3720 + }, + { + "epoch": 5.45, + "grad_norm": 2.1352174282073975, + "learning_rate": 8.412033595695437e-05, + "loss": 3.1518, + "step": 3721 + }, + { + "epoch": 5.45, + "grad_norm": 7.084630489349365, + "learning_rate": 8.40740294049011e-05, + "loss": 3.1283, + "step": 3722 + }, + { + "epoch": 5.45, + "grad_norm": 2.8684113025665283, + "learning_rate": 8.402772635645465e-05, + "loss": 3.1823, + "step": 3723 + }, + { + "epoch": 5.45, + "grad_norm": 14.201617240905762, + "learning_rate": 8.398142682180131e-05, + "loss": 3.14, + "step": 3724 + }, + { + "epoch": 5.45, + "grad_norm": 4.209918975830078, + "learning_rate": 8.393513081112674e-05, + "loss": 3.1867, + "step": 3725 + }, + { + "epoch": 5.46, + "grad_norm": 2.907388687133789, + "learning_rate": 8.388883833461578e-05, + "loss": 3.2095, + "step": 3726 + }, + { + "epoch": 5.46, + "grad_norm": 11.771166801452637, + "learning_rate": 8.384254940245233e-05, + "loss": 3.1938, + "step": 3727 + }, + { + "epoch": 5.46, + "grad_norm": 15.321097373962402, + "learning_rate": 8.379626402481977e-05, + "loss": 3.1755, + "step": 3728 + }, + { + "epoch": 5.46, + "grad_norm": 3.204758882522583, + "learning_rate": 8.374998221190056e-05, + "loss": 3.1617, + "step": 3729 + }, + { + "epoch": 5.46, + "grad_norm": 7.3266754150390625, + "learning_rate": 8.37037039738763e-05, + "loss": 3.1647, + "step": 3730 + }, + { + "epoch": 5.46, + "grad_norm": 108.37910461425781, + "learning_rate": 8.365742932092801e-05, + "loss": 3.1807, + "step": 3731 + }, + { + "epoch": 5.46, + "grad_norm": 4.3307623863220215, + "learning_rate": 8.361115826323572e-05, + "loss": 3.2109, + "step": 3732 + }, + { + "epoch": 5.47, + "grad_norm": 1.9860690832138062, + "learning_rate": 8.35648908109788e-05, + "loss": 3.1484, + "step": 3733 + }, + { + "epoch": 5.47, + "grad_norm": 3.039022207260132, + "learning_rate": 8.351862697433573e-05, + "loss": 3.175, + "step": 3734 + }, + { + "epoch": 5.47, + "grad_norm": 7.283559799194336, + "learning_rate": 8.347236676348426e-05, + "loss": 3.219, + "step": 3735 + }, + { + "epoch": 5.47, + "grad_norm": 3.774641513824463, + "learning_rate": 8.342611018860139e-05, + "loss": 3.1932, + "step": 3736 + }, + { + "epoch": 5.47, + "grad_norm": 4.264026641845703, + "learning_rate": 8.337985725986314e-05, + "loss": 3.193, + "step": 3737 + }, + { + "epoch": 5.47, + "grad_norm": 1.858871340751648, + "learning_rate": 8.333360798744496e-05, + "loss": 3.122, + "step": 3738 + }, + { + "epoch": 5.47, + "grad_norm": 4.920361518859863, + "learning_rate": 8.328736238152133e-05, + "loss": 3.1425, + "step": 3739 + }, + { + "epoch": 5.48, + "grad_norm": 4.10701847076416, + "learning_rate": 8.324112045226593e-05, + "loss": 3.1595, + "step": 3740 + }, + { + "epoch": 5.48, + "grad_norm": 5.043720245361328, + "learning_rate": 8.319488220985174e-05, + "loss": 3.1407, + "step": 3741 + }, + { + "epoch": 5.48, + "grad_norm": 3.795624256134033, + "learning_rate": 8.314864766445088e-05, + "loss": 3.2385, + "step": 3742 + }, + { + "epoch": 5.48, + "grad_norm": 3.801236867904663, + "learning_rate": 8.310241682623454e-05, + "loss": 3.152, + "step": 3743 + }, + { + "epoch": 5.48, + "grad_norm": 2.491832971572876, + "learning_rate": 8.30561897053733e-05, + "loss": 3.1737, + "step": 3744 + }, + { + "epoch": 5.48, + "grad_norm": 2.1997487545013428, + "learning_rate": 8.300996631203675e-05, + "loss": 3.1273, + "step": 3745 + }, + { + "epoch": 5.49, + "grad_norm": 2.1604340076446533, + "learning_rate": 8.296374665639373e-05, + "loss": 3.2015, + "step": 3746 + }, + { + "epoch": 5.49, + "grad_norm": 14.758793830871582, + "learning_rate": 8.291753074861233e-05, + "loss": 3.1594, + "step": 3747 + }, + { + "epoch": 5.49, + "grad_norm": 1.7705718278884888, + "learning_rate": 8.287131859885964e-05, + "loss": 3.1615, + "step": 3748 + }, + { + "epoch": 5.49, + "grad_norm": 3.673743724822998, + "learning_rate": 8.282511021730206e-05, + "loss": 3.1665, + "step": 3749 + }, + { + "epoch": 5.49, + "grad_norm": 2.580552101135254, + "learning_rate": 8.277890561410518e-05, + "loss": 3.1972, + "step": 3750 + }, + { + "epoch": 5.49, + "grad_norm": 1.8429394960403442, + "learning_rate": 8.273270479943361e-05, + "loss": 3.1316, + "step": 3751 + }, + { + "epoch": 5.49, + "grad_norm": 1.4963501691818237, + "learning_rate": 8.268650778345126e-05, + "loss": 3.1498, + "step": 3752 + }, + { + "epoch": 5.5, + "grad_norm": 3.3234293460845947, + "learning_rate": 8.26403145763212e-05, + "loss": 3.1475, + "step": 3753 + }, + { + "epoch": 5.5, + "grad_norm": 3.2355213165283203, + "learning_rate": 8.25941251882056e-05, + "loss": 3.1404, + "step": 3754 + }, + { + "epoch": 5.5, + "grad_norm": 1.3712440729141235, + "learning_rate": 8.254793962926579e-05, + "loss": 3.1316, + "step": 3755 + }, + { + "epoch": 5.5, + "grad_norm": 1.9082093238830566, + "learning_rate": 8.250175790966228e-05, + "loss": 3.1388, + "step": 3756 + }, + { + "epoch": 5.5, + "grad_norm": 2.096512794494629, + "learning_rate": 8.245558003955481e-05, + "loss": 3.1753, + "step": 3757 + }, + { + "epoch": 5.5, + "grad_norm": 1.1812965869903564, + "learning_rate": 8.240940602910212e-05, + "loss": 3.1647, + "step": 3758 + }, + { + "epoch": 5.5, + "grad_norm": 2.799182415008545, + "learning_rate": 8.23632358884622e-05, + "loss": 3.1474, + "step": 3759 + }, + { + "epoch": 5.51, + "grad_norm": 2.4367105960845947, + "learning_rate": 8.231706962779222e-05, + "loss": 3.1312, + "step": 3760 + }, + { + "epoch": 5.51, + "grad_norm": 2.1872997283935547, + "learning_rate": 8.227090725724842e-05, + "loss": 3.0935, + "step": 3761 + }, + { + "epoch": 5.51, + "grad_norm": 3.3611397743225098, + "learning_rate": 8.222474878698617e-05, + "loss": 3.0987, + "step": 3762 + }, + { + "epoch": 5.51, + "grad_norm": 1.0651507377624512, + "learning_rate": 8.217859422716012e-05, + "loss": 3.183, + "step": 3763 + }, + { + "epoch": 5.51, + "grad_norm": 0.8456958532333374, + "learning_rate": 8.213244358792389e-05, + "loss": 3.123, + "step": 3764 + }, + { + "epoch": 5.51, + "grad_norm": 0.9663138389587402, + "learning_rate": 8.208629687943029e-05, + "loss": 3.16, + "step": 3765 + }, + { + "epoch": 5.51, + "grad_norm": 135.12335205078125, + "learning_rate": 8.204015411183135e-05, + "loss": 3.1066, + "step": 3766 + }, + { + "epoch": 5.52, + "grad_norm": 1.7855501174926758, + "learning_rate": 8.199401529527816e-05, + "loss": 3.0972, + "step": 3767 + }, + { + "epoch": 5.52, + "grad_norm": 1.187641978263855, + "learning_rate": 8.194788043992089e-05, + "loss": 3.1476, + "step": 3768 + }, + { + "epoch": 5.52, + "grad_norm": 6.5035834312438965, + "learning_rate": 8.190174955590897e-05, + "loss": 3.134, + "step": 3769 + }, + { + "epoch": 5.52, + "grad_norm": 3.347416400909424, + "learning_rate": 8.185562265339088e-05, + "loss": 3.1592, + "step": 3770 + }, + { + "epoch": 5.52, + "grad_norm": 2.2179319858551025, + "learning_rate": 8.180949974251416e-05, + "loss": 3.1656, + "step": 3771 + }, + { + "epoch": 5.52, + "grad_norm": 3.2867071628570557, + "learning_rate": 8.176338083342561e-05, + "loss": 3.1231, + "step": 3772 + }, + { + "epoch": 5.53, + "grad_norm": 4.15543794631958, + "learning_rate": 8.17172659362711e-05, + "loss": 3.1501, + "step": 3773 + }, + { + "epoch": 5.53, + "grad_norm": 3.2263450622558594, + "learning_rate": 8.167115506119549e-05, + "loss": 3.1269, + "step": 3774 + }, + { + "epoch": 5.53, + "grad_norm": 1.2961052656173706, + "learning_rate": 8.162504821834295e-05, + "loss": 3.1204, + "step": 3775 + }, + { + "epoch": 5.53, + "grad_norm": 2.9037585258483887, + "learning_rate": 8.157894541785671e-05, + "loss": 3.1801, + "step": 3776 + }, + { + "epoch": 5.53, + "grad_norm": 2.3933663368225098, + "learning_rate": 8.153284666987897e-05, + "loss": 3.1537, + "step": 3777 + }, + { + "epoch": 5.53, + "grad_norm": 1.6720260381698608, + "learning_rate": 8.148675198455123e-05, + "loss": 3.1501, + "step": 3778 + }, + { + "epoch": 5.53, + "grad_norm": 11.296658515930176, + "learning_rate": 8.144066137201398e-05, + "loss": 3.1364, + "step": 3779 + }, + { + "epoch": 5.54, + "grad_norm": 14.168936729431152, + "learning_rate": 8.139457484240687e-05, + "loss": 3.1629, + "step": 3780 + }, + { + "epoch": 5.54, + "grad_norm": 5.073040008544922, + "learning_rate": 8.134849240586854e-05, + "loss": 3.1765, + "step": 3781 + }, + { + "epoch": 5.54, + "grad_norm": 2.859703779220581, + "learning_rate": 8.130241407253693e-05, + "loss": 3.1252, + "step": 3782 + }, + { + "epoch": 5.54, + "grad_norm": 6.422140121459961, + "learning_rate": 8.125633985254893e-05, + "loss": 3.126, + "step": 3783 + }, + { + "epoch": 5.54, + "grad_norm": 16.3258056640625, + "learning_rate": 8.121026975604052e-05, + "loss": 3.1953, + "step": 3784 + }, + { + "epoch": 5.54, + "grad_norm": 7.476664066314697, + "learning_rate": 8.116420379314684e-05, + "loss": 3.1981, + "step": 3785 + }, + { + "epoch": 5.54, + "grad_norm": 57.822113037109375, + "learning_rate": 8.111814197400213e-05, + "loss": 3.1517, + "step": 3786 + }, + { + "epoch": 5.55, + "grad_norm": 6.158851146697998, + "learning_rate": 8.10720843087396e-05, + "loss": 3.1655, + "step": 3787 + }, + { + "epoch": 5.55, + "grad_norm": 7.291560173034668, + "learning_rate": 8.102603080749171e-05, + "loss": 3.113, + "step": 3788 + }, + { + "epoch": 5.55, + "grad_norm": 6.275900840759277, + "learning_rate": 8.097998148038985e-05, + "loss": 3.1209, + "step": 3789 + }, + { + "epoch": 5.55, + "grad_norm": 4.1805338859558105, + "learning_rate": 8.093393633756459e-05, + "loss": 3.1049, + "step": 3790 + }, + { + "epoch": 5.55, + "grad_norm": 6.27461051940918, + "learning_rate": 8.088789538914561e-05, + "loss": 3.169, + "step": 3791 + }, + { + "epoch": 5.55, + "grad_norm": 63.376373291015625, + "learning_rate": 8.084185864526153e-05, + "loss": 3.2346, + "step": 3792 + }, + { + "epoch": 5.55, + "grad_norm": 17.007471084594727, + "learning_rate": 8.079582611604015e-05, + "loss": 3.1799, + "step": 3793 + }, + { + "epoch": 5.56, + "grad_norm": 3.8963406085968018, + "learning_rate": 8.074979781160836e-05, + "loss": 3.1546, + "step": 3794 + }, + { + "epoch": 5.56, + "grad_norm": 1.5541036128997803, + "learning_rate": 8.070377374209202e-05, + "loss": 3.17, + "step": 3795 + }, + { + "epoch": 5.56, + "grad_norm": 4.511335372924805, + "learning_rate": 8.06577539176161e-05, + "loss": 3.1307, + "step": 3796 + }, + { + "epoch": 5.56, + "grad_norm": 3.2192275524139404, + "learning_rate": 8.061173834830476e-05, + "loss": 3.1262, + "step": 3797 + }, + { + "epoch": 5.56, + "grad_norm": 10.192057609558105, + "learning_rate": 8.056572704428103e-05, + "loss": 3.1341, + "step": 3798 + }, + { + "epoch": 5.56, + "grad_norm": 10.097148895263672, + "learning_rate": 8.051972001566708e-05, + "loss": 3.1627, + "step": 3799 + }, + { + "epoch": 5.57, + "grad_norm": 5.317000865936279, + "learning_rate": 8.04737172725842e-05, + "loss": 3.1695, + "step": 3800 + }, + { + "epoch": 5.57, + "grad_norm": 3.3502695560455322, + "learning_rate": 8.042771882515265e-05, + "loss": 3.1111, + "step": 3801 + }, + { + "epoch": 5.57, + "grad_norm": 3.3459572792053223, + "learning_rate": 8.038172468349178e-05, + "loss": 3.1737, + "step": 3802 + }, + { + "epoch": 5.57, + "grad_norm": 4.4290900230407715, + "learning_rate": 8.033573485772001e-05, + "loss": 3.1335, + "step": 3803 + }, + { + "epoch": 5.57, + "grad_norm": 3.444740056991577, + "learning_rate": 8.028974935795481e-05, + "loss": 3.1575, + "step": 3804 + }, + { + "epoch": 5.57, + "grad_norm": 6.389871120452881, + "learning_rate": 8.024376819431266e-05, + "loss": 3.1654, + "step": 3805 + }, + { + "epoch": 5.57, + "grad_norm": 4.629978179931641, + "learning_rate": 8.019779137690906e-05, + "loss": 3.1554, + "step": 3806 + }, + { + "epoch": 5.58, + "grad_norm": 2.2746245861053467, + "learning_rate": 8.015181891585871e-05, + "loss": 3.1065, + "step": 3807 + }, + { + "epoch": 5.58, + "grad_norm": 4.390033721923828, + "learning_rate": 8.010585082127513e-05, + "loss": 3.142, + "step": 3808 + }, + { + "epoch": 5.58, + "grad_norm": 8.950037956237793, + "learning_rate": 8.005988710327105e-05, + "loss": 3.1759, + "step": 3809 + }, + { + "epoch": 5.58, + "grad_norm": 7.786754608154297, + "learning_rate": 8.001392777195821e-05, + "loss": 3.1518, + "step": 3810 + }, + { + "epoch": 5.58, + "grad_norm": 1.803338646888733, + "learning_rate": 7.996797283744732e-05, + "loss": 3.1452, + "step": 3811 + }, + { + "epoch": 5.58, + "grad_norm": 5.436160087585449, + "learning_rate": 7.992202230984811e-05, + "loss": 3.1284, + "step": 3812 + }, + { + "epoch": 5.58, + "grad_norm": 5.9366912841796875, + "learning_rate": 7.987607619926946e-05, + "loss": 3.1547, + "step": 3813 + }, + { + "epoch": 5.59, + "grad_norm": 5.989798545837402, + "learning_rate": 7.98301345158192e-05, + "loss": 3.1264, + "step": 3814 + }, + { + "epoch": 5.59, + "grad_norm": 2.005276679992676, + "learning_rate": 7.978419726960413e-05, + "loss": 3.1387, + "step": 3815 + }, + { + "epoch": 5.59, + "grad_norm": 4.961897373199463, + "learning_rate": 7.97382644707302e-05, + "loss": 3.1506, + "step": 3816 + }, + { + "epoch": 5.59, + "grad_norm": 5.611112594604492, + "learning_rate": 7.969233612930234e-05, + "loss": 3.1237, + "step": 3817 + }, + { + "epoch": 5.59, + "grad_norm": 2.2314023971557617, + "learning_rate": 7.964641225542438e-05, + "loss": 3.1433, + "step": 3818 + }, + { + "epoch": 5.59, + "grad_norm": 27.2403564453125, + "learning_rate": 7.960049285919933e-05, + "loss": 3.1587, + "step": 3819 + }, + { + "epoch": 5.59, + "grad_norm": 7.609023571014404, + "learning_rate": 7.955457795072918e-05, + "loss": 3.2031, + "step": 3820 + }, + { + "epoch": 5.6, + "grad_norm": 6.942310810089111, + "learning_rate": 7.95086675401148e-05, + "loss": 3.1571, + "step": 3821 + }, + { + "epoch": 5.6, + "grad_norm": 5.55133056640625, + "learning_rate": 7.94627616374563e-05, + "loss": 3.1465, + "step": 3822 + }, + { + "epoch": 5.6, + "grad_norm": 2.6575374603271484, + "learning_rate": 7.941686025285257e-05, + "loss": 3.1384, + "step": 3823 + }, + { + "epoch": 5.6, + "grad_norm": 4.183603763580322, + "learning_rate": 7.937096339640162e-05, + "loss": 3.1551, + "step": 3824 + }, + { + "epoch": 5.6, + "grad_norm": 203.84231567382812, + "learning_rate": 7.932507107820051e-05, + "loss": 3.1756, + "step": 3825 + }, + { + "epoch": 5.6, + "grad_norm": 1.8645808696746826, + "learning_rate": 7.92791833083452e-05, + "loss": 3.1254, + "step": 3826 + }, + { + "epoch": 5.61, + "grad_norm": 3.2966809272766113, + "learning_rate": 7.923330009693071e-05, + "loss": 3.1409, + "step": 3827 + }, + { + "epoch": 5.61, + "grad_norm": 16.701486587524414, + "learning_rate": 7.918742145405098e-05, + "loss": 3.1687, + "step": 3828 + }, + { + "epoch": 5.61, + "grad_norm": 5.320707321166992, + "learning_rate": 7.914154738979908e-05, + "loss": 3.166, + "step": 3829 + }, + { + "epoch": 5.61, + "grad_norm": 3.770569086074829, + "learning_rate": 7.909567791426697e-05, + "loss": 3.129, + "step": 3830 + }, + { + "epoch": 5.61, + "grad_norm": 1.9080885648727417, + "learning_rate": 7.904981303754561e-05, + "loss": 3.1792, + "step": 3831 + }, + { + "epoch": 5.61, + "grad_norm": 2.2089335918426514, + "learning_rate": 7.900395276972502e-05, + "loss": 3.1417, + "step": 3832 + }, + { + "epoch": 5.61, + "grad_norm": 2.3139476776123047, + "learning_rate": 7.895809712089406e-05, + "loss": 3.1278, + "step": 3833 + }, + { + "epoch": 5.62, + "grad_norm": 3.4870378971099854, + "learning_rate": 7.891224610114072e-05, + "loss": 3.1562, + "step": 3834 + }, + { + "epoch": 5.62, + "grad_norm": 4.3670244216918945, + "learning_rate": 7.886639972055195e-05, + "loss": 3.1389, + "step": 3835 + }, + { + "epoch": 5.62, + "grad_norm": 5.9168596267700195, + "learning_rate": 7.882055798921358e-05, + "loss": 3.1104, + "step": 3836 + }, + { + "epoch": 5.62, + "grad_norm": 7.184929370880127, + "learning_rate": 7.87747209172105e-05, + "loss": 3.1555, + "step": 3837 + }, + { + "epoch": 5.62, + "grad_norm": 4.011050701141357, + "learning_rate": 7.872888851462663e-05, + "loss": 3.1192, + "step": 3838 + }, + { + "epoch": 5.62, + "grad_norm": 1.6208895444869995, + "learning_rate": 7.86830607915447e-05, + "loss": 3.1861, + "step": 3839 + }, + { + "epoch": 5.62, + "grad_norm": 1.3240761756896973, + "learning_rate": 7.863723775804651e-05, + "loss": 3.1656, + "step": 3840 + }, + { + "epoch": 5.63, + "grad_norm": 2.556436538696289, + "learning_rate": 7.85914194242129e-05, + "loss": 3.1392, + "step": 3841 + }, + { + "epoch": 5.63, + "grad_norm": 1.8541210889816284, + "learning_rate": 7.854560580012351e-05, + "loss": 3.1391, + "step": 3842 + }, + { + "epoch": 5.63, + "grad_norm": 2.840593099594116, + "learning_rate": 7.849979689585705e-05, + "loss": 3.0979, + "step": 3843 + }, + { + "epoch": 5.63, + "grad_norm": 1.50877845287323, + "learning_rate": 7.845399272149124e-05, + "loss": 3.1231, + "step": 3844 + }, + { + "epoch": 5.63, + "grad_norm": 2.6628737449645996, + "learning_rate": 7.84081932871026e-05, + "loss": 3.1347, + "step": 3845 + }, + { + "epoch": 5.63, + "grad_norm": 33.729000091552734, + "learning_rate": 7.836239860276671e-05, + "loss": 3.1319, + "step": 3846 + }, + { + "epoch": 5.64, + "grad_norm": 13.277280807495117, + "learning_rate": 7.831660867855816e-05, + "loss": 3.1439, + "step": 3847 + }, + { + "epoch": 5.64, + "grad_norm": 1.8134537935256958, + "learning_rate": 7.827082352455039e-05, + "loss": 3.1507, + "step": 3848 + }, + { + "epoch": 5.64, + "grad_norm": 2.101228952407837, + "learning_rate": 7.822504315081578e-05, + "loss": 3.1133, + "step": 3849 + }, + { + "epoch": 5.64, + "grad_norm": 2.533062696456909, + "learning_rate": 7.817926756742576e-05, + "loss": 3.1362, + "step": 3850 + }, + { + "epoch": 5.64, + "grad_norm": 4.869479656219482, + "learning_rate": 7.813349678445067e-05, + "loss": 3.1618, + "step": 3851 + }, + { + "epoch": 5.64, + "grad_norm": 2.6743147373199463, + "learning_rate": 7.808773081195972e-05, + "loss": 3.1278, + "step": 3852 + }, + { + "epoch": 5.64, + "grad_norm": 0.8585074543952942, + "learning_rate": 7.804196966002113e-05, + "loss": 3.1386, + "step": 3853 + }, + { + "epoch": 5.65, + "grad_norm": 7.510640621185303, + "learning_rate": 7.799621333870208e-05, + "loss": 3.1786, + "step": 3854 + }, + { + "epoch": 5.65, + "grad_norm": 2.670872449874878, + "learning_rate": 7.795046185806866e-05, + "loss": 3.1483, + "step": 3855 + }, + { + "epoch": 5.65, + "grad_norm": 22.928964614868164, + "learning_rate": 7.79047152281858e-05, + "loss": 3.1633, + "step": 3856 + }, + { + "epoch": 5.65, + "grad_norm": 7.530433177947998, + "learning_rate": 7.785897345911754e-05, + "loss": 3.1601, + "step": 3857 + }, + { + "epoch": 5.65, + "grad_norm": 7.938250541687012, + "learning_rate": 7.781323656092677e-05, + "loss": 3.1554, + "step": 3858 + }, + { + "epoch": 5.65, + "grad_norm": 2.2199909687042236, + "learning_rate": 7.776750454367521e-05, + "loss": 3.1512, + "step": 3859 + }, + { + "epoch": 5.65, + "grad_norm": 92.11273956298828, + "learning_rate": 7.772177741742369e-05, + "loss": 3.1275, + "step": 3860 + }, + { + "epoch": 5.66, + "grad_norm": 4.906170845031738, + "learning_rate": 7.767605519223183e-05, + "loss": 3.1245, + "step": 3861 + }, + { + "epoch": 5.66, + "grad_norm": 3.870858907699585, + "learning_rate": 7.76303378781582e-05, + "loss": 3.1105, + "step": 3862 + }, + { + "epoch": 5.66, + "grad_norm": 2.2505300045013428, + "learning_rate": 7.758462548526033e-05, + "loss": 3.1065, + "step": 3863 + }, + { + "epoch": 5.66, + "grad_norm": 2.947871208190918, + "learning_rate": 7.753891802359466e-05, + "loss": 3.1869, + "step": 3864 + }, + { + "epoch": 5.66, + "grad_norm": 3.654209613800049, + "learning_rate": 7.749321550321645e-05, + "loss": 3.1427, + "step": 3865 + }, + { + "epoch": 5.66, + "grad_norm": 3.259089469909668, + "learning_rate": 7.744751793418005e-05, + "loss": 3.1221, + "step": 3866 + }, + { + "epoch": 5.66, + "grad_norm": 2.3307595252990723, + "learning_rate": 7.740182532653853e-05, + "loss": 3.1556, + "step": 3867 + }, + { + "epoch": 5.67, + "grad_norm": 6.026763439178467, + "learning_rate": 7.735613769034399e-05, + "loss": 3.1646, + "step": 3868 + }, + { + "epoch": 5.67, + "grad_norm": 9.292561531066895, + "learning_rate": 7.731045503564743e-05, + "loss": 3.1498, + "step": 3869 + }, + { + "epoch": 5.67, + "grad_norm": 4.8578314781188965, + "learning_rate": 7.726477737249869e-05, + "loss": 3.1159, + "step": 3870 + }, + { + "epoch": 5.67, + "grad_norm": 5.385825157165527, + "learning_rate": 7.721910471094653e-05, + "loss": 3.1494, + "step": 3871 + }, + { + "epoch": 5.67, + "grad_norm": 3.9133341312408447, + "learning_rate": 7.717343706103873e-05, + "loss": 3.1402, + "step": 3872 + }, + { + "epoch": 5.67, + "grad_norm": 6.215272903442383, + "learning_rate": 7.712777443282175e-05, + "loss": 3.195, + "step": 3873 + }, + { + "epoch": 5.68, + "grad_norm": 3.6820201873779297, + "learning_rate": 7.708211683634112e-05, + "loss": 3.2287, + "step": 3874 + }, + { + "epoch": 5.68, + "grad_norm": 3.4176886081695557, + "learning_rate": 7.703646428164122e-05, + "loss": 3.1445, + "step": 3875 + }, + { + "epoch": 5.68, + "grad_norm": 8.820064544677734, + "learning_rate": 7.699081677876526e-05, + "loss": 3.2085, + "step": 3876 + }, + { + "epoch": 5.68, + "grad_norm": 17.15256118774414, + "learning_rate": 7.694517433775543e-05, + "loss": 3.1627, + "step": 3877 + }, + { + "epoch": 5.68, + "grad_norm": 1.7407811880111694, + "learning_rate": 7.689953696865271e-05, + "loss": 3.1868, + "step": 3878 + }, + { + "epoch": 5.68, + "grad_norm": 3.167860746383667, + "learning_rate": 7.685390468149707e-05, + "loss": 3.1838, + "step": 3879 + }, + { + "epoch": 5.68, + "grad_norm": 4.515937805175781, + "learning_rate": 7.680827748632728e-05, + "loss": 3.1624, + "step": 3880 + }, + { + "epoch": 5.69, + "grad_norm": 5.7442755699157715, + "learning_rate": 7.676265539318098e-05, + "loss": 3.1957, + "step": 3881 + }, + { + "epoch": 5.69, + "grad_norm": 4.945896625518799, + "learning_rate": 7.671703841209481e-05, + "loss": 3.1714, + "step": 3882 + }, + { + "epoch": 5.69, + "grad_norm": 8.167930603027344, + "learning_rate": 7.667142655310414e-05, + "loss": 3.1944, + "step": 3883 + }, + { + "epoch": 5.69, + "grad_norm": 17.595746994018555, + "learning_rate": 7.662581982624326e-05, + "loss": 3.1853, + "step": 3884 + }, + { + "epoch": 5.69, + "grad_norm": 2.751532554626465, + "learning_rate": 7.658021824154542e-05, + "loss": 3.1348, + "step": 3885 + }, + { + "epoch": 5.69, + "grad_norm": 6.661686897277832, + "learning_rate": 7.653462180904258e-05, + "loss": 3.2022, + "step": 3886 + }, + { + "epoch": 5.69, + "grad_norm": 9.880109786987305, + "learning_rate": 7.648903053876568e-05, + "loss": 3.1686, + "step": 3887 + }, + { + "epoch": 5.69, + "eval_loss": 2.989283800125122, + "eval_runtime": 1170.3187, + "eval_samples_per_second": 14.647, + "eval_steps_per_second": 0.916, + "step": 3887 + }, + { + "epoch": 5.7, + "grad_norm": 3.414788246154785, + "learning_rate": 7.644344444074453e-05, + "loss": 3.1669, + "step": 3888 + }, + { + "epoch": 5.7, + "grad_norm": 3.5142674446105957, + "learning_rate": 7.639786352500773e-05, + "loss": 3.1526, + "step": 3889 + }, + { + "epoch": 5.7, + "grad_norm": 7.385360240936279, + "learning_rate": 7.635228780158274e-05, + "loss": 3.1528, + "step": 3890 + }, + { + "epoch": 5.7, + "grad_norm": 7.439982891082764, + "learning_rate": 7.630671728049598e-05, + "loss": 3.1417, + "step": 3891 + }, + { + "epoch": 5.7, + "grad_norm": 4.842582702636719, + "learning_rate": 7.626115197177266e-05, + "loss": 3.2058, + "step": 3892 + }, + { + "epoch": 5.7, + "grad_norm": 6.158822536468506, + "learning_rate": 7.621559188543677e-05, + "loss": 3.2051, + "step": 3893 + }, + { + "epoch": 5.7, + "grad_norm": 4.262391090393066, + "learning_rate": 7.617003703151129e-05, + "loss": 3.2001, + "step": 3894 + }, + { + "epoch": 5.71, + "grad_norm": 22.497093200683594, + "learning_rate": 7.612448742001797e-05, + "loss": 3.1808, + "step": 3895 + }, + { + "epoch": 5.71, + "grad_norm": 3.1336076259613037, + "learning_rate": 7.607894306097739e-05, + "loss": 3.2113, + "step": 3896 + }, + { + "epoch": 5.71, + "grad_norm": 4.773746967315674, + "learning_rate": 7.603340396440902e-05, + "loss": 3.1906, + "step": 3897 + }, + { + "epoch": 5.71, + "grad_norm": 5.188856601715088, + "learning_rate": 7.59878701403312e-05, + "loss": 3.1663, + "step": 3898 + }, + { + "epoch": 5.71, + "grad_norm": 7.35548734664917, + "learning_rate": 7.594234159876098e-05, + "loss": 3.1467, + "step": 3899 + }, + { + "epoch": 5.71, + "grad_norm": 7.549004077911377, + "learning_rate": 7.589681834971443e-05, + "loss": 3.1928, + "step": 3900 + }, + { + "epoch": 5.72, + "grad_norm": 6.5583086013793945, + "learning_rate": 7.585130040320627e-05, + "loss": 3.2067, + "step": 3901 + }, + { + "epoch": 5.72, + "grad_norm": 14.854689598083496, + "learning_rate": 7.580578776925022e-05, + "loss": 3.2208, + "step": 3902 + }, + { + "epoch": 5.72, + "grad_norm": 11.917444229125977, + "learning_rate": 7.576028045785868e-05, + "loss": 3.1828, + "step": 3903 + }, + { + "epoch": 5.72, + "grad_norm": 6.152881145477295, + "learning_rate": 7.5714778479043e-05, + "loss": 3.204, + "step": 3904 + }, + { + "epoch": 5.72, + "grad_norm": 4.3145294189453125, + "learning_rate": 7.566928184281334e-05, + "loss": 3.2282, + "step": 3905 + }, + { + "epoch": 5.72, + "grad_norm": 3.392911434173584, + "learning_rate": 7.562379055917856e-05, + "loss": 3.1795, + "step": 3906 + }, + { + "epoch": 5.72, + "grad_norm": 5.250206470489502, + "learning_rate": 7.557830463814652e-05, + "loss": 3.1574, + "step": 3907 + }, + { + "epoch": 5.73, + "grad_norm": 11.659444808959961, + "learning_rate": 7.553282408972382e-05, + "loss": 3.1975, + "step": 3908 + }, + { + "epoch": 5.73, + "grad_norm": 6.504120826721191, + "learning_rate": 7.54873489239158e-05, + "loss": 3.2318, + "step": 3909 + }, + { + "epoch": 5.73, + "grad_norm": 3.6852314472198486, + "learning_rate": 7.544187915072676e-05, + "loss": 3.1501, + "step": 3910 + }, + { + "epoch": 5.73, + "grad_norm": 4.429998874664307, + "learning_rate": 7.539641478015976e-05, + "loss": 3.1623, + "step": 3911 + }, + { + "epoch": 5.73, + "grad_norm": 2.792065382003784, + "learning_rate": 7.535095582221656e-05, + "loss": 3.1624, + "step": 3912 + }, + { + "epoch": 5.73, + "grad_norm": 21.118961334228516, + "learning_rate": 7.530550228689794e-05, + "loss": 3.2112, + "step": 3913 + }, + { + "epoch": 5.73, + "grad_norm": 18.30959701538086, + "learning_rate": 7.526005418420332e-05, + "loss": 3.2394, + "step": 3914 + }, + { + "epoch": 5.74, + "grad_norm": 8.257102012634277, + "learning_rate": 7.521461152413092e-05, + "loss": 3.2588, + "step": 3915 + }, + { + "epoch": 5.74, + "grad_norm": 3.5304644107818604, + "learning_rate": 7.516917431667795e-05, + "loss": 3.2331, + "step": 3916 + }, + { + "epoch": 5.74, + "grad_norm": 24.953046798706055, + "learning_rate": 7.512374257184019e-05, + "loss": 3.1763, + "step": 3917 + }, + { + "epoch": 5.74, + "grad_norm": 5.283435821533203, + "learning_rate": 7.507831629961234e-05, + "loss": 3.1631, + "step": 3918 + }, + { + "epoch": 5.74, + "grad_norm": 7.136661052703857, + "learning_rate": 7.503289550998791e-05, + "loss": 3.2551, + "step": 3919 + }, + { + "epoch": 5.74, + "grad_norm": 3.8187716007232666, + "learning_rate": 7.498748021295914e-05, + "loss": 3.1214, + "step": 3920 + }, + { + "epoch": 5.74, + "grad_norm": 10.967880249023438, + "learning_rate": 7.494207041851708e-05, + "loss": 3.144, + "step": 3921 + }, + { + "epoch": 5.75, + "grad_norm": 30.177751541137695, + "learning_rate": 7.489666613665163e-05, + "loss": 3.1466, + "step": 3922 + }, + { + "epoch": 5.75, + "grad_norm": 6.267055511474609, + "learning_rate": 7.48512673773514e-05, + "loss": 3.2192, + "step": 3923 + }, + { + "epoch": 5.75, + "grad_norm": 3.4286611080169678, + "learning_rate": 7.480587415060378e-05, + "loss": 3.1519, + "step": 3924 + }, + { + "epoch": 5.75, + "grad_norm": 9.199522972106934, + "learning_rate": 7.476048646639498e-05, + "loss": 3.1738, + "step": 3925 + }, + { + "epoch": 5.75, + "grad_norm": 9.091236114501953, + "learning_rate": 7.471510433471005e-05, + "loss": 3.1704, + "step": 3926 + }, + { + "epoch": 5.75, + "grad_norm": 7.052969455718994, + "learning_rate": 7.46697277655327e-05, + "loss": 3.1835, + "step": 3927 + }, + { + "epoch": 5.76, + "grad_norm": 30.890663146972656, + "learning_rate": 7.462435676884547e-05, + "loss": 3.1883, + "step": 3928 + }, + { + "epoch": 5.76, + "grad_norm": 3.2805261611938477, + "learning_rate": 7.457899135462974e-05, + "loss": 3.2009, + "step": 3929 + }, + { + "epoch": 5.76, + "grad_norm": 11.584432601928711, + "learning_rate": 7.45336315328655e-05, + "loss": 3.2384, + "step": 3930 + }, + { + "epoch": 5.76, + "grad_norm": 6.337993144989014, + "learning_rate": 7.448827731353166e-05, + "loss": 3.1973, + "step": 3931 + }, + { + "epoch": 5.76, + "grad_norm": 13.461885452270508, + "learning_rate": 7.444292870660586e-05, + "loss": 3.2435, + "step": 3932 + }, + { + "epoch": 5.76, + "grad_norm": 9.56443977355957, + "learning_rate": 7.439758572206448e-05, + "loss": 3.1465, + "step": 3933 + }, + { + "epoch": 5.76, + "grad_norm": 7.9625115394592285, + "learning_rate": 7.43522483698826e-05, + "loss": 3.1635, + "step": 3934 + }, + { + "epoch": 5.77, + "grad_norm": 6.776467800140381, + "learning_rate": 7.430691666003424e-05, + "loss": 3.2405, + "step": 3935 + }, + { + "epoch": 5.77, + "grad_norm": 43.86566162109375, + "learning_rate": 7.426159060249204e-05, + "loss": 3.192, + "step": 3936 + }, + { + "epoch": 5.77, + "grad_norm": 12.768895149230957, + "learning_rate": 7.421627020722738e-05, + "loss": 3.1741, + "step": 3937 + }, + { + "epoch": 5.77, + "grad_norm": 6.572139263153076, + "learning_rate": 7.41709554842105e-05, + "loss": 3.1637, + "step": 3938 + }, + { + "epoch": 5.77, + "grad_norm": 5.105015277862549, + "learning_rate": 7.412564644341033e-05, + "loss": 3.2329, + "step": 3939 + }, + { + "epoch": 5.77, + "grad_norm": 4.994419574737549, + "learning_rate": 7.40803430947945e-05, + "loss": 3.2124, + "step": 3940 + }, + { + "epoch": 5.77, + "grad_norm": 13.615537643432617, + "learning_rate": 7.403504544832951e-05, + "loss": 3.2148, + "step": 3941 + }, + { + "epoch": 5.78, + "grad_norm": 6.730780124664307, + "learning_rate": 7.398975351398053e-05, + "loss": 3.2086, + "step": 3942 + }, + { + "epoch": 5.78, + "grad_norm": 8.399843215942383, + "learning_rate": 7.394446730171141e-05, + "loss": 3.2455, + "step": 3943 + }, + { + "epoch": 5.78, + "grad_norm": 7.892696857452393, + "learning_rate": 7.389918682148491e-05, + "loss": 3.263, + "step": 3944 + }, + { + "epoch": 5.78, + "grad_norm": 3.2410852909088135, + "learning_rate": 7.38539120832624e-05, + "loss": 3.2023, + "step": 3945 + }, + { + "epoch": 5.78, + "grad_norm": 43.567543029785156, + "learning_rate": 7.380864309700396e-05, + "loss": 3.2505, + "step": 3946 + }, + { + "epoch": 5.78, + "grad_norm": 8.303853988647461, + "learning_rate": 7.376337987266856e-05, + "loss": 3.2253, + "step": 3947 + }, + { + "epoch": 5.79, + "grad_norm": 5.137606620788574, + "learning_rate": 7.371812242021374e-05, + "loss": 3.2614, + "step": 3948 + }, + { + "epoch": 5.79, + "grad_norm": 7.571026802062988, + "learning_rate": 7.367287074959586e-05, + "loss": 3.3035, + "step": 3949 + }, + { + "epoch": 5.79, + "grad_norm": 5.098783016204834, + "learning_rate": 7.362762487076995e-05, + "loss": 3.24, + "step": 3950 + }, + { + "epoch": 5.79, + "grad_norm": 5.123687744140625, + "learning_rate": 7.358238479368985e-05, + "loss": 3.2333, + "step": 3951 + }, + { + "epoch": 5.79, + "grad_norm": 9.861023902893066, + "learning_rate": 7.353715052830808e-05, + "loss": 3.2665, + "step": 3952 + }, + { + "epoch": 5.79, + "grad_norm": 30.322845458984375, + "learning_rate": 7.349192208457579e-05, + "loss": 3.2337, + "step": 3953 + }, + { + "epoch": 5.79, + "grad_norm": 10.311440467834473, + "learning_rate": 7.344669947244302e-05, + "loss": 3.2307, + "step": 3954 + }, + { + "epoch": 5.8, + "grad_norm": 13.122282981872559, + "learning_rate": 7.340148270185842e-05, + "loss": 3.2456, + "step": 3955 + }, + { + "epoch": 5.8, + "grad_norm": 16.757612228393555, + "learning_rate": 7.335627178276937e-05, + "loss": 3.3208, + "step": 3956 + }, + { + "epoch": 5.8, + "grad_norm": 36.39468002319336, + "learning_rate": 7.331106672512197e-05, + "loss": 3.2313, + "step": 3957 + }, + { + "epoch": 5.8, + "grad_norm": 39.07672882080078, + "learning_rate": 7.326586753886102e-05, + "loss": 3.3304, + "step": 3958 + }, + { + "epoch": 5.8, + "grad_norm": 67.7435302734375, + "learning_rate": 7.322067423393002e-05, + "loss": 3.3623, + "step": 3959 + }, + { + "epoch": 5.8, + "grad_norm": 102.08399963378906, + "learning_rate": 7.317548682027127e-05, + "loss": 3.3498, + "step": 3960 + }, + { + "epoch": 5.8, + "grad_norm": 68.23622131347656, + "learning_rate": 7.313030530782564e-05, + "loss": 3.4284, + "step": 3961 + }, + { + "epoch": 5.81, + "grad_norm": 54.879310607910156, + "learning_rate": 7.308512970653274e-05, + "loss": 3.4101, + "step": 3962 + }, + { + "epoch": 5.81, + "grad_norm": 154.8038787841797, + "learning_rate": 7.303996002633096e-05, + "loss": 3.5803, + "step": 3963 + }, + { + "epoch": 5.81, + "grad_norm": 228.81173706054688, + "learning_rate": 7.29947962771573e-05, + "loss": 3.489, + "step": 3964 + }, + { + "epoch": 5.81, + "grad_norm": 350.7447509765625, + "learning_rate": 7.294963846894744e-05, + "loss": 3.6323, + "step": 3965 + }, + { + "epoch": 5.81, + "grad_norm": 858.3496704101562, + "learning_rate": 7.290448661163587e-05, + "loss": 4.0812, + "step": 3966 + }, + { + "epoch": 5.81, + "grad_norm": 730.7702026367188, + "learning_rate": 7.285934071515568e-05, + "loss": 4.0206, + "step": 3967 + }, + { + "epoch": 5.81, + "grad_norm": 658.6021118164062, + "learning_rate": 7.281420078943855e-05, + "loss": 4.3312, + "step": 3968 + }, + { + "epoch": 5.82, + "grad_norm": 475.8223876953125, + "learning_rate": 7.276906684441514e-05, + "loss": 4.5484, + "step": 3969 + }, + { + "epoch": 5.82, + "grad_norm": 309.2330322265625, + "learning_rate": 7.272393889001454e-05, + "loss": 4.6918, + "step": 3970 + }, + { + "epoch": 5.82, + "grad_norm": 195.97572326660156, + "learning_rate": 7.267881693616452e-05, + "loss": 4.7661, + "step": 3971 + }, + { + "epoch": 5.82, + "grad_norm": 134.8620147705078, + "learning_rate": 7.263370099279172e-05, + "loss": 4.6478, + "step": 3972 + }, + { + "epoch": 5.82, + "grad_norm": 108.79113006591797, + "learning_rate": 7.25885910698213e-05, + "loss": 4.6325, + "step": 3973 + }, + { + "epoch": 5.82, + "grad_norm": 149.6044921875, + "learning_rate": 7.254348717717714e-05, + "loss": 4.8721, + "step": 3974 + }, + { + "epoch": 5.83, + "grad_norm": 117.6059341430664, + "learning_rate": 7.249838932478177e-05, + "loss": 4.8513, + "step": 3975 + }, + { + "epoch": 5.83, + "grad_norm": 56.039024353027344, + "learning_rate": 7.245329752255647e-05, + "loss": 4.8064, + "step": 3976 + }, + { + "epoch": 5.83, + "grad_norm": 100.5042953491211, + "learning_rate": 7.24082117804211e-05, + "loss": 4.8077, + "step": 3977 + }, + { + "epoch": 5.83, + "grad_norm": 101.55089569091797, + "learning_rate": 7.236313210829419e-05, + "loss": 4.9091, + "step": 3978 + }, + { + "epoch": 5.83, + "grad_norm": 54.18492889404297, + "learning_rate": 7.231805851609306e-05, + "loss": 4.9624, + "step": 3979 + }, + { + "epoch": 5.83, + "grad_norm": 90.85334014892578, + "learning_rate": 7.227299101373352e-05, + "loss": 4.8436, + "step": 3980 + }, + { + "epoch": 5.83, + "grad_norm": 108.68850708007812, + "learning_rate": 7.222792961113012e-05, + "loss": 4.7235, + "step": 3981 + }, + { + "epoch": 5.84, + "grad_norm": 46.6330451965332, + "learning_rate": 7.21828743181961e-05, + "loss": 4.7477, + "step": 3982 + }, + { + "epoch": 5.84, + "grad_norm": 116.49347686767578, + "learning_rate": 7.213782514484333e-05, + "loss": 4.6439, + "step": 3983 + }, + { + "epoch": 5.84, + "grad_norm": 46.60487747192383, + "learning_rate": 7.209278210098224e-05, + "loss": 4.6174, + "step": 3984 + }, + { + "epoch": 5.84, + "grad_norm": 74.72257995605469, + "learning_rate": 7.204774519652211e-05, + "loss": 4.6026, + "step": 3985 + }, + { + "epoch": 5.84, + "grad_norm": 117.54887390136719, + "learning_rate": 7.200271444137072e-05, + "loss": 4.3403, + "step": 3986 + }, + { + "epoch": 5.84, + "grad_norm": 66.1172866821289, + "learning_rate": 7.195768984543448e-05, + "loss": 4.2326, + "step": 3987 + }, + { + "epoch": 5.84, + "grad_norm": 23.685178756713867, + "learning_rate": 7.191267141861858e-05, + "loss": 4.2885, + "step": 3988 + }, + { + "epoch": 5.85, + "grad_norm": 33.490779876708984, + "learning_rate": 7.186765917082674e-05, + "loss": 4.2881, + "step": 3989 + }, + { + "epoch": 5.85, + "grad_norm": 49.65608215332031, + "learning_rate": 7.18226531119613e-05, + "loss": 4.0779, + "step": 3990 + }, + { + "epoch": 5.85, + "grad_norm": 21.95450782775879, + "learning_rate": 7.177765325192338e-05, + "loss": 4.0829, + "step": 3991 + }, + { + "epoch": 5.85, + "grad_norm": 36.09683609008789, + "learning_rate": 7.173265960061261e-05, + "loss": 4.1138, + "step": 3992 + }, + { + "epoch": 5.85, + "grad_norm": 135.8228302001953, + "learning_rate": 7.168767216792724e-05, + "loss": 4.0337, + "step": 3993 + }, + { + "epoch": 5.85, + "grad_norm": 42.24320602416992, + "learning_rate": 7.164269096376432e-05, + "loss": 4.0038, + "step": 3994 + }, + { + "epoch": 5.85, + "grad_norm": 90.61548614501953, + "learning_rate": 7.15977159980193e-05, + "loss": 3.9393, + "step": 3995 + }, + { + "epoch": 5.86, + "grad_norm": 83.08573150634766, + "learning_rate": 7.155274728058645e-05, + "loss": 4.085, + "step": 3996 + }, + { + "epoch": 5.86, + "grad_norm": 51.404388427734375, + "learning_rate": 7.150778482135851e-05, + "loss": 4.025, + "step": 3997 + }, + { + "epoch": 5.86, + "grad_norm": 133.36476135253906, + "learning_rate": 7.146282863022699e-05, + "loss": 4.0341, + "step": 3998 + }, + { + "epoch": 5.86, + "grad_norm": 68.87841033935547, + "learning_rate": 7.141787871708194e-05, + "loss": 4.1013, + "step": 3999 + }, + { + "epoch": 5.86, + "grad_norm": 83.18514251708984, + "learning_rate": 7.137293509181197e-05, + "loss": 4.0575, + "step": 4000 + }, + { + "epoch": 5.86, + "grad_norm": 63.65642166137695, + "learning_rate": 7.132799776430446e-05, + "loss": 4.0338, + "step": 4001 + }, + { + "epoch": 5.87, + "grad_norm": 84.51449584960938, + "learning_rate": 7.128306674444531e-05, + "loss": 4.0289, + "step": 4002 + }, + { + "epoch": 5.87, + "grad_norm": 1112.17431640625, + "learning_rate": 7.123814204211899e-05, + "loss": 3.9574, + "step": 4003 + }, + { + "epoch": 5.87, + "grad_norm": 113.361572265625, + "learning_rate": 7.11932236672087e-05, + "loss": 3.9683, + "step": 4004 + }, + { + "epoch": 5.87, + "grad_norm": 94.9935073852539, + "learning_rate": 7.114831162959614e-05, + "loss": 3.867, + "step": 4005 + }, + { + "epoch": 5.87, + "grad_norm": 99.41561126708984, + "learning_rate": 7.110340593916164e-05, + "loss": 4.0003, + "step": 4006 + }, + { + "epoch": 5.87, + "grad_norm": 68.38880157470703, + "learning_rate": 7.105850660578422e-05, + "loss": 3.8946, + "step": 4007 + }, + { + "epoch": 5.87, + "grad_norm": 48.9935188293457, + "learning_rate": 7.101361363934136e-05, + "loss": 3.9, + "step": 4008 + }, + { + "epoch": 5.88, + "grad_norm": 75.62947082519531, + "learning_rate": 7.096872704970923e-05, + "loss": 3.9334, + "step": 4009 + }, + { + "epoch": 5.88, + "grad_norm": 259.18511962890625, + "learning_rate": 7.092384684676262e-05, + "loss": 3.9539, + "step": 4010 + }, + { + "epoch": 5.88, + "grad_norm": 60.03557586669922, + "learning_rate": 7.087897304037483e-05, + "loss": 3.9378, + "step": 4011 + }, + { + "epoch": 5.88, + "grad_norm": 60.427616119384766, + "learning_rate": 7.083410564041779e-05, + "loss": 3.922, + "step": 4012 + }, + { + "epoch": 5.88, + "grad_norm": 86.16722869873047, + "learning_rate": 7.078924465676208e-05, + "loss": 3.9123, + "step": 4013 + }, + { + "epoch": 5.88, + "grad_norm": 92.25879669189453, + "learning_rate": 7.07443900992768e-05, + "loss": 3.9653, + "step": 4014 + }, + { + "epoch": 5.88, + "grad_norm": 63.97455596923828, + "learning_rate": 7.069954197782958e-05, + "loss": 3.9709, + "step": 4015 + }, + { + "epoch": 5.89, + "grad_norm": 85.49566650390625, + "learning_rate": 7.065470030228679e-05, + "loss": 3.9614, + "step": 4016 + }, + { + "epoch": 5.89, + "grad_norm": 200.93997192382812, + "learning_rate": 7.060986508251329e-05, + "loss": 3.9715, + "step": 4017 + }, + { + "epoch": 5.89, + "grad_norm": 47.623600006103516, + "learning_rate": 7.056503632837247e-05, + "loss": 3.9969, + "step": 4018 + }, + { + "epoch": 5.89, + "grad_norm": 57.71437454223633, + "learning_rate": 7.052021404972641e-05, + "loss": 4.0892, + "step": 4019 + }, + { + "epoch": 5.89, + "grad_norm": 101.89254760742188, + "learning_rate": 7.047539825643572e-05, + "loss": 4.1731, + "step": 4020 + }, + { + "epoch": 5.89, + "grad_norm": 56.065486907958984, + "learning_rate": 7.043058895835953e-05, + "loss": 4.1648, + "step": 4021 + }, + { + "epoch": 5.89, + "grad_norm": 98.78327178955078, + "learning_rate": 7.038578616535556e-05, + "loss": 4.3178, + "step": 4022 + }, + { + "epoch": 5.9, + "grad_norm": 383.0876770019531, + "learning_rate": 7.034098988728024e-05, + "loss": 4.3457, + "step": 4023 + }, + { + "epoch": 5.9, + "grad_norm": 162.98101806640625, + "learning_rate": 7.029620013398837e-05, + "loss": 4.3242, + "step": 4024 + }, + { + "epoch": 5.9, + "grad_norm": 149.046142578125, + "learning_rate": 7.025141691533334e-05, + "loss": 4.4494, + "step": 4025 + }, + { + "epoch": 5.9, + "grad_norm": 135.7252655029297, + "learning_rate": 7.020664024116726e-05, + "loss": 4.487, + "step": 4026 + }, + { + "epoch": 5.9, + "grad_norm": 103.65221405029297, + "learning_rate": 7.016187012134068e-05, + "loss": 4.544, + "step": 4027 + }, + { + "epoch": 5.9, + "grad_norm": 95.76521301269531, + "learning_rate": 7.011710656570267e-05, + "loss": 4.5937, + "step": 4028 + }, + { + "epoch": 5.91, + "grad_norm": 287.5489196777344, + "learning_rate": 7.007234958410098e-05, + "loss": 4.6397, + "step": 4029 + }, + { + "epoch": 5.91, + "grad_norm": 55.86167526245117, + "learning_rate": 7.002759918638183e-05, + "loss": 4.6566, + "step": 4030 + }, + { + "epoch": 5.91, + "grad_norm": 50.3218879699707, + "learning_rate": 6.998285538238993e-05, + "loss": 4.6248, + "step": 4031 + }, + { + "epoch": 5.91, + "grad_norm": 414.169921875, + "learning_rate": 6.993811818196872e-05, + "loss": 4.6006, + "step": 4032 + }, + { + "epoch": 5.91, + "grad_norm": 112.764404296875, + "learning_rate": 6.989338759496006e-05, + "loss": 4.6415, + "step": 4033 + }, + { + "epoch": 5.91, + "grad_norm": 53.906734466552734, + "learning_rate": 6.984866363120432e-05, + "loss": 4.6018, + "step": 4034 + }, + { + "epoch": 5.91, + "grad_norm": 45.21807098388672, + "learning_rate": 6.980394630054054e-05, + "loss": 4.5761, + "step": 4035 + }, + { + "epoch": 5.92, + "grad_norm": 123.97179412841797, + "learning_rate": 6.975923561280624e-05, + "loss": 4.4143, + "step": 4036 + }, + { + "epoch": 5.92, + "grad_norm": 39.991905212402344, + "learning_rate": 6.971453157783739e-05, + "loss": 4.3971, + "step": 4037 + }, + { + "epoch": 5.92, + "grad_norm": 245.3745574951172, + "learning_rate": 6.966983420546866e-05, + "loss": 4.3387, + "step": 4038 + }, + { + "epoch": 5.92, + "grad_norm": 535.9085083007812, + "learning_rate": 6.962514350553313e-05, + "loss": 4.2919, + "step": 4039 + }, + { + "epoch": 5.92, + "grad_norm": 31.66384506225586, + "learning_rate": 6.958045948786245e-05, + "loss": 4.2878, + "step": 4040 + }, + { + "epoch": 5.92, + "grad_norm": 104.09449768066406, + "learning_rate": 6.953578216228685e-05, + "loss": 4.2645, + "step": 4041 + }, + { + "epoch": 5.92, + "grad_norm": 73.07312774658203, + "learning_rate": 6.9491111538635e-05, + "loss": 4.2232, + "step": 4042 + }, + { + "epoch": 5.93, + "grad_norm": 55.163421630859375, + "learning_rate": 6.944644762673413e-05, + "loss": 4.2499, + "step": 4043 + }, + { + "epoch": 5.93, + "grad_norm": 97.98957061767578, + "learning_rate": 6.940179043641005e-05, + "loss": 4.2195, + "step": 4044 + }, + { + "epoch": 5.93, + "grad_norm": 35.54196548461914, + "learning_rate": 6.9357139977487e-05, + "loss": 4.2314, + "step": 4045 + }, + { + "epoch": 5.93, + "grad_norm": 36.159202575683594, + "learning_rate": 6.931249625978782e-05, + "loss": 4.1943, + "step": 4046 + }, + { + "epoch": 5.93, + "grad_norm": 119.87337493896484, + "learning_rate": 6.926785929313376e-05, + "loss": 4.1162, + "step": 4047 + }, + { + "epoch": 5.93, + "grad_norm": 7.6323442459106445, + "learning_rate": 6.922322908734474e-05, + "loss": 4.1434, + "step": 4048 + }, + { + "epoch": 5.94, + "grad_norm": 13.205349922180176, + "learning_rate": 6.917860565223905e-05, + "loss": 4.1026, + "step": 4049 + }, + { + "epoch": 5.94, + "grad_norm": 29.02250099182129, + "learning_rate": 6.913398899763352e-05, + "loss": 4.0863, + "step": 4050 + }, + { + "epoch": 5.94, + "grad_norm": 18.842418670654297, + "learning_rate": 6.908937913334362e-05, + "loss": 4.1056, + "step": 4051 + }, + { + "epoch": 5.94, + "grad_norm": 26.442873001098633, + "learning_rate": 6.904477606918313e-05, + "loss": 4.1195, + "step": 4052 + }, + { + "epoch": 5.94, + "grad_norm": 146.73545837402344, + "learning_rate": 6.900017981496443e-05, + "loss": 4.0448, + "step": 4053 + }, + { + "epoch": 5.94, + "grad_norm": 18.860145568847656, + "learning_rate": 6.895559038049846e-05, + "loss": 4.0507, + "step": 4054 + }, + { + "epoch": 5.94, + "grad_norm": 7.4111504554748535, + "learning_rate": 6.891100777559454e-05, + "loss": 3.9955, + "step": 4055 + }, + { + "epoch": 5.95, + "grad_norm": 2.2253613471984863, + "learning_rate": 6.886643201006052e-05, + "loss": 3.9959, + "step": 4056 + }, + { + "epoch": 5.95, + "eval_loss": 3.7561120986938477, + "eval_runtime": 1174.8961, + "eval_samples_per_second": 14.59, + "eval_steps_per_second": 0.912, + "step": 4056 + }, + { + "epoch": 5.95, + "grad_norm": 2.182352066040039, + "learning_rate": 6.882186309370287e-05, + "loss": 3.9472, + "step": 4057 + }, + { + "epoch": 5.95, + "grad_norm": 1.8885817527770996, + "learning_rate": 6.877730103632638e-05, + "loss": 3.9685, + "step": 4058 + }, + { + "epoch": 5.95, + "grad_norm": 1.6018338203430176, + "learning_rate": 6.87327458477344e-05, + "loss": 3.8923, + "step": 4059 + }, + { + "epoch": 5.95, + "grad_norm": 1.4974128007888794, + "learning_rate": 6.868819753772879e-05, + "loss": 3.8199, + "step": 4060 + }, + { + "epoch": 5.95, + "grad_norm": 1.9076157808303833, + "learning_rate": 6.86436561161099e-05, + "loss": 3.8143, + "step": 4061 + }, + { + "epoch": 5.95, + "grad_norm": 1.2935985326766968, + "learning_rate": 6.859912159267648e-05, + "loss": 3.8232, + "step": 4062 + }, + { + "epoch": 5.96, + "grad_norm": 1.5463148355484009, + "learning_rate": 6.85545939772259e-05, + "loss": 3.7682, + "step": 4063 + }, + { + "epoch": 5.96, + "grad_norm": 1.2979236841201782, + "learning_rate": 6.851007327955393e-05, + "loss": 3.812, + "step": 4064 + }, + { + "epoch": 5.96, + "grad_norm": 1.235023856163025, + "learning_rate": 6.846555950945475e-05, + "loss": 3.6794, + "step": 4065 + }, + { + "epoch": 5.96, + "grad_norm": 1.587446689605713, + "learning_rate": 6.842105267672117e-05, + "loss": 3.6736, + "step": 4066 + }, + { + "epoch": 5.96, + "grad_norm": 1.6601265668869019, + "learning_rate": 6.837655279114439e-05, + "loss": 3.7298, + "step": 4067 + }, + { + "epoch": 5.96, + "grad_norm": 1.1815412044525146, + "learning_rate": 6.833205986251404e-05, + "loss": 3.6491, + "step": 4068 + }, + { + "epoch": 5.96, + "grad_norm": 2.2124860286712646, + "learning_rate": 6.828757390061828e-05, + "loss": 3.6432, + "step": 4069 + }, + { + "epoch": 5.97, + "grad_norm": 1.2953498363494873, + "learning_rate": 6.824309491524377e-05, + "loss": 3.6124, + "step": 4070 + }, + { + "epoch": 5.97, + "grad_norm": 1.5017386674880981, + "learning_rate": 6.819862291617555e-05, + "loss": 3.6377, + "step": 4071 + }, + { + "epoch": 5.97, + "grad_norm": 2.5354573726654053, + "learning_rate": 6.815415791319712e-05, + "loss": 3.6211, + "step": 4072 + }, + { + "epoch": 5.97, + "grad_norm": 2.2560806274414062, + "learning_rate": 6.810969991609055e-05, + "loss": 3.5855, + "step": 4073 + }, + { + "epoch": 5.97, + "grad_norm": 2.3098514080047607, + "learning_rate": 6.806524893463631e-05, + "loss": 3.5944, + "step": 4074 + }, + { + "epoch": 5.97, + "grad_norm": 3.807551860809326, + "learning_rate": 6.802080497861326e-05, + "loss": 3.582, + "step": 4075 + }, + { + "epoch": 5.98, + "grad_norm": 0.8520615100860596, + "learning_rate": 6.797636805779878e-05, + "loss": 3.5581, + "step": 4076 + }, + { + "epoch": 5.98, + "grad_norm": 2.77345871925354, + "learning_rate": 6.793193818196876e-05, + "loss": 3.5948, + "step": 4077 + }, + { + "epoch": 5.98, + "grad_norm": 1.2732380628585815, + "learning_rate": 6.788751536089739e-05, + "loss": 3.5607, + "step": 4078 + }, + { + "epoch": 5.98, + "grad_norm": 4.257369518280029, + "learning_rate": 6.784309960435744e-05, + "loss": 3.588, + "step": 4079 + }, + { + "epoch": 5.98, + "grad_norm": 4.027552604675293, + "learning_rate": 6.77986909221201e-05, + "loss": 3.5961, + "step": 4080 + }, + { + "epoch": 5.98, + "grad_norm": 1.3005938529968262, + "learning_rate": 6.77542893239549e-05, + "loss": 3.52, + "step": 4081 + }, + { + "epoch": 5.98, + "grad_norm": 3.389254570007324, + "learning_rate": 6.770989481963001e-05, + "loss": 3.5313, + "step": 4082 + }, + { + "epoch": 5.99, + "grad_norm": 1.3640910387039185, + "learning_rate": 6.766550741891182e-05, + "loss": 3.5291, + "step": 4083 + }, + { + "epoch": 5.99, + "grad_norm": 5.111477375030518, + "learning_rate": 6.76211271315653e-05, + "loss": 3.5717, + "step": 4084 + }, + { + "epoch": 5.99, + "grad_norm": 5.650928974151611, + "learning_rate": 6.757675396735386e-05, + "loss": 3.5892, + "step": 4085 + }, + { + "epoch": 5.99, + "grad_norm": 2.029646158218384, + "learning_rate": 6.753238793603923e-05, + "loss": 3.5179, + "step": 4086 + }, + { + "epoch": 5.99, + "grad_norm": 5.816488742828369, + "learning_rate": 6.748802904738163e-05, + "loss": 3.5582, + "step": 4087 + }, + { + "epoch": 5.99, + "grad_norm": 7.348140716552734, + "learning_rate": 6.74436773111398e-05, + "loss": 3.5371, + "step": 4088 + }, + { + "epoch": 5.99, + "grad_norm": 6.118741512298584, + "learning_rate": 6.739933273707078e-05, + "loss": 3.5298, + "step": 4089 + }, + { + "epoch": 6.0, + "grad_norm": 0.9806273579597473, + "learning_rate": 6.735499533493003e-05, + "loss": 3.5438, + "step": 4090 + }, + { + "epoch": 6.0, + "grad_norm": 9.882466316223145, + "learning_rate": 6.731066511447159e-05, + "loss": 3.5713, + "step": 4091 + }, + { + "epoch": 6.0, + "grad_norm": 14.419783592224121, + "learning_rate": 6.726634208544773e-05, + "loss": 3.569, + "step": 4092 + }, + { + "epoch": 6.0, + "grad_norm": 11.746726036071777, + "learning_rate": 6.722202625760923e-05, + "loss": 3.6069, + "step": 4093 + }, + { + "epoch": 6.0, + "grad_norm": 5.577601432800293, + "learning_rate": 6.717771764070527e-05, + "loss": 3.5287, + "step": 4094 + }, + { + "epoch": 6.0, + "grad_norm": 5.144962310791016, + "learning_rate": 6.713341624448351e-05, + "loss": 3.5013, + "step": 4095 + }, + { + "epoch": 6.0, + "grad_norm": 8.980729103088379, + "learning_rate": 6.708912207868987e-05, + "loss": 3.5225, + "step": 4096 + }, + { + "epoch": 6.01, + "grad_norm": 9.171212196350098, + "learning_rate": 6.70448351530688e-05, + "loss": 3.537, + "step": 4097 + }, + { + "epoch": 6.01, + "grad_norm": 6.954055309295654, + "learning_rate": 6.70005554773632e-05, + "loss": 3.5512, + "step": 4098 + }, + { + "epoch": 6.01, + "grad_norm": 25.35405731201172, + "learning_rate": 6.695628306131422e-05, + "loss": 3.525, + "step": 4099 + }, + { + "epoch": 6.01, + "grad_norm": 7.4213948249816895, + "learning_rate": 6.691201791466148e-05, + "loss": 3.5411, + "step": 4100 + }, + { + "epoch": 6.01, + "grad_norm": 10.75389289855957, + "learning_rate": 6.68677600471431e-05, + "loss": 3.5158, + "step": 4101 + }, + { + "epoch": 6.01, + "grad_norm": 9.7072114944458, + "learning_rate": 6.682350946849546e-05, + "loss": 3.5728, + "step": 4102 + }, + { + "epoch": 6.02, + "grad_norm": 4.010187149047852, + "learning_rate": 6.677926618845337e-05, + "loss": 3.4962, + "step": 4103 + }, + { + "epoch": 6.0, + "grad_norm": 5.2085371017456055, + "learning_rate": 6.673503021675011e-05, + "loss": 3.5387, + "step": 4104 + }, + { + "epoch": 6.0, + "grad_norm": 7.690248012542725, + "learning_rate": 6.669080156311729e-05, + "loss": 3.5012, + "step": 4105 + }, + { + "epoch": 6.0, + "grad_norm": 7.818577766418457, + "learning_rate": 6.664658023728485e-05, + "loss": 3.5594, + "step": 4106 + }, + { + "epoch": 6.01, + "grad_norm": 4.1020355224609375, + "learning_rate": 6.660236624898124e-05, + "loss": 3.4641, + "step": 4107 + }, + { + "epoch": 6.01, + "grad_norm": 3.3743233680725098, + "learning_rate": 6.655815960793328e-05, + "loss": 3.5123, + "step": 4108 + }, + { + "epoch": 6.01, + "grad_norm": 5.35082483291626, + "learning_rate": 6.651396032386603e-05, + "loss": 3.5047, + "step": 4109 + }, + { + "epoch": 6.01, + "grad_norm": 2.8813364505767822, + "learning_rate": 6.646976840650312e-05, + "loss": 3.4699, + "step": 4110 + }, + { + "epoch": 6.01, + "grad_norm": 3.898869037628174, + "learning_rate": 6.642558386556648e-05, + "loss": 3.4723, + "step": 4111 + }, + { + "epoch": 6.01, + "grad_norm": 3.897127389907837, + "learning_rate": 6.638140671077633e-05, + "loss": 3.4887, + "step": 4112 + }, + { + "epoch": 6.01, + "grad_norm": 0.8849645256996155, + "learning_rate": 6.633723695185141e-05, + "loss": 3.4564, + "step": 4113 + }, + { + "epoch": 6.02, + "grad_norm": 4.971652507781982, + "learning_rate": 6.62930745985088e-05, + "loss": 3.5218, + "step": 4114 + }, + { + "epoch": 6.02, + "grad_norm": 4.998004913330078, + "learning_rate": 6.624891966046385e-05, + "loss": 3.5211, + "step": 4115 + }, + { + "epoch": 6.02, + "grad_norm": 1.827019214630127, + "learning_rate": 6.620477214743042e-05, + "loss": 3.4959, + "step": 4116 + }, + { + "epoch": 6.02, + "grad_norm": 4.872849464416504, + "learning_rate": 6.616063206912063e-05, + "loss": 3.4849, + "step": 4117 + }, + { + "epoch": 6.02, + "grad_norm": 5.961864948272705, + "learning_rate": 6.611649943524498e-05, + "loss": 3.4685, + "step": 4118 + }, + { + "epoch": 6.02, + "grad_norm": 3.4859068393707275, + "learning_rate": 6.607237425551236e-05, + "loss": 3.4841, + "step": 4119 + }, + { + "epoch": 6.03, + "grad_norm": 4.598318099975586, + "learning_rate": 6.602825653963006e-05, + "loss": 3.5138, + "step": 4120 + }, + { + "epoch": 6.03, + "grad_norm": 8.418424606323242, + "learning_rate": 6.598414629730367e-05, + "loss": 3.4861, + "step": 4121 + }, + { + "epoch": 6.03, + "grad_norm": 6.949596881866455, + "learning_rate": 6.594004353823708e-05, + "loss": 3.4545, + "step": 4122 + }, + { + "epoch": 6.03, + "grad_norm": 1.4345308542251587, + "learning_rate": 6.589594827213268e-05, + "loss": 3.4615, + "step": 4123 + }, + { + "epoch": 6.03, + "grad_norm": 5.833983898162842, + "learning_rate": 6.585186050869112e-05, + "loss": 3.4469, + "step": 4124 + }, + { + "epoch": 6.03, + "grad_norm": 8.324283599853516, + "learning_rate": 6.580778025761136e-05, + "loss": 3.5428, + "step": 4125 + }, + { + "epoch": 6.03, + "grad_norm": 7.164588451385498, + "learning_rate": 6.576370752859085e-05, + "loss": 3.44, + "step": 4126 + }, + { + "epoch": 6.04, + "grad_norm": 4.524144172668457, + "learning_rate": 6.57196423313252e-05, + "loss": 3.4798, + "step": 4127 + }, + { + "epoch": 6.04, + "grad_norm": 3.6542019844055176, + "learning_rate": 6.56755846755085e-05, + "loss": 3.4431, + "step": 4128 + }, + { + "epoch": 6.04, + "grad_norm": 6.415909290313721, + "learning_rate": 6.563153457083315e-05, + "loss": 3.4509, + "step": 4129 + }, + { + "epoch": 6.04, + "grad_norm": 4.0182623863220215, + "learning_rate": 6.558749202698988e-05, + "loss": 3.4182, + "step": 4130 + }, + { + "epoch": 6.04, + "grad_norm": 22.318021774291992, + "learning_rate": 6.55434570536677e-05, + "loss": 3.5704, + "step": 4131 + }, + { + "epoch": 6.04, + "grad_norm": 6.184370040893555, + "learning_rate": 6.54994296605541e-05, + "loss": 3.3746, + "step": 4132 + }, + { + "epoch": 6.04, + "grad_norm": 7.076589584350586, + "learning_rate": 6.545540985733473e-05, + "loss": 3.4426, + "step": 4133 + }, + { + "epoch": 6.05, + "grad_norm": 3.5989532470703125, + "learning_rate": 6.541139765369368e-05, + "loss": 3.4139, + "step": 4134 + }, + { + "epoch": 6.05, + "grad_norm": 4.95648717880249, + "learning_rate": 6.536739305931338e-05, + "loss": 3.392, + "step": 4135 + }, + { + "epoch": 6.05, + "grad_norm": 8.156216621398926, + "learning_rate": 6.53233960838745e-05, + "loss": 3.4219, + "step": 4136 + }, + { + "epoch": 6.05, + "grad_norm": 6.330258846282959, + "learning_rate": 6.527940673705605e-05, + "loss": 3.4077, + "step": 4137 + }, + { + "epoch": 6.05, + "grad_norm": 1.3029074668884277, + "learning_rate": 6.523542502853547e-05, + "loss": 3.422, + "step": 4138 + }, + { + "epoch": 6.05, + "grad_norm": 5.435974597930908, + "learning_rate": 6.519145096798842e-05, + "loss": 3.4704, + "step": 4139 + }, + { + "epoch": 6.05, + "grad_norm": 5.678780555725098, + "learning_rate": 6.514748456508885e-05, + "loss": 3.419, + "step": 4140 + }, + { + "epoch": 6.06, + "grad_norm": 3.2500386238098145, + "learning_rate": 6.510352582950909e-05, + "loss": 3.4207, + "step": 4141 + }, + { + "epoch": 6.06, + "grad_norm": 4.012285232543945, + "learning_rate": 6.505957477091981e-05, + "loss": 3.3755, + "step": 4142 + }, + { + "epoch": 6.06, + "grad_norm": 6.388498783111572, + "learning_rate": 6.501563139898991e-05, + "loss": 3.4086, + "step": 4143 + }, + { + "epoch": 6.06, + "grad_norm": 3.951239824295044, + "learning_rate": 6.497169572338661e-05, + "loss": 3.3432, + "step": 4144 + }, + { + "epoch": 6.06, + "grad_norm": 4.527097702026367, + "learning_rate": 6.492776775377556e-05, + "loss": 3.3961, + "step": 4145 + }, + { + "epoch": 6.06, + "grad_norm": 5.017838954925537, + "learning_rate": 6.488384749982053e-05, + "loss": 3.3483, + "step": 4146 + }, + { + "epoch": 6.07, + "grad_norm": 4.438302516937256, + "learning_rate": 6.483993497118371e-05, + "loss": 3.456, + "step": 4147 + }, + { + "epoch": 6.07, + "grad_norm": 0.9947657585144043, + "learning_rate": 6.479603017752559e-05, + "loss": 3.3646, + "step": 4148 + }, + { + "epoch": 6.07, + "grad_norm": 2.749354600906372, + "learning_rate": 6.47521331285049e-05, + "loss": 3.4993, + "step": 4149 + }, + { + "epoch": 6.07, + "grad_norm": 2.3780393600463867, + "learning_rate": 6.470824383377867e-05, + "loss": 3.3783, + "step": 4150 + }, + { + "epoch": 6.07, + "grad_norm": 2.840714454650879, + "learning_rate": 6.46643623030023e-05, + "loss": 3.3901, + "step": 4151 + }, + { + "epoch": 6.07, + "grad_norm": 2.8188705444335938, + "learning_rate": 6.462048854582944e-05, + "loss": 3.3313, + "step": 4152 + }, + { + "epoch": 6.07, + "grad_norm": 0.6622484922409058, + "learning_rate": 6.457662257191195e-05, + "loss": 3.3428, + "step": 4153 + }, + { + "epoch": 6.08, + "grad_norm": 2.970710515975952, + "learning_rate": 6.453276439090012e-05, + "loss": 3.3343, + "step": 4154 + }, + { + "epoch": 6.08, + "grad_norm": 2.88561749458313, + "learning_rate": 6.448891401244244e-05, + "loss": 3.3591, + "step": 4155 + }, + { + "epoch": 6.08, + "grad_norm": 2.3622708320617676, + "learning_rate": 6.444507144618566e-05, + "loss": 3.39, + "step": 4156 + }, + { + "epoch": 6.08, + "grad_norm": 1.4716769456863403, + "learning_rate": 6.440123670177492e-05, + "loss": 3.4068, + "step": 4157 + }, + { + "epoch": 6.08, + "grad_norm": 10.4777193069458, + "learning_rate": 6.435740978885353e-05, + "loss": 3.352, + "step": 4158 + }, + { + "epoch": 6.08, + "grad_norm": 6.331012725830078, + "learning_rate": 6.43135907170631e-05, + "loss": 3.3506, + "step": 4159 + }, + { + "epoch": 6.08, + "grad_norm": 4.355317115783691, + "learning_rate": 6.426977949604358e-05, + "loss": 3.3904, + "step": 4160 + }, + { + "epoch": 6.09, + "grad_norm": 1.3198426961898804, + "learning_rate": 6.422597613543309e-05, + "loss": 3.3279, + "step": 4161 + }, + { + "epoch": 6.09, + "grad_norm": 3.2531580924987793, + "learning_rate": 6.41821806448681e-05, + "loss": 3.427, + "step": 4162 + }, + { + "epoch": 6.09, + "grad_norm": 1.780542254447937, + "learning_rate": 6.413839303398337e-05, + "loss": 3.382, + "step": 4163 + }, + { + "epoch": 6.09, + "grad_norm": 4.514572620391846, + "learning_rate": 6.40946133124118e-05, + "loss": 3.3181, + "step": 4164 + }, + { + "epoch": 6.09, + "grad_norm": 4.696477890014648, + "learning_rate": 6.405084148978471e-05, + "loss": 3.3693, + "step": 4165 + }, + { + "epoch": 6.09, + "grad_norm": 1.9454889297485352, + "learning_rate": 6.400707757573152e-05, + "loss": 3.3362, + "step": 4166 + }, + { + "epoch": 6.1, + "grad_norm": 4.424253463745117, + "learning_rate": 6.396332157988007e-05, + "loss": 3.3042, + "step": 4167 + }, + { + "epoch": 6.1, + "grad_norm": 6.185297012329102, + "learning_rate": 6.39195735118564e-05, + "loss": 3.3837, + "step": 4168 + }, + { + "epoch": 6.1, + "grad_norm": 4.592702865600586, + "learning_rate": 6.387583338128471e-05, + "loss": 3.35, + "step": 4169 + }, + { + "epoch": 6.1, + "grad_norm": 1.6287086009979248, + "learning_rate": 6.38321011977876e-05, + "loss": 3.3304, + "step": 4170 + }, + { + "epoch": 6.1, + "grad_norm": 4.277772903442383, + "learning_rate": 6.378837697098586e-05, + "loss": 3.321, + "step": 4171 + }, + { + "epoch": 6.1, + "grad_norm": 2.2683122158050537, + "learning_rate": 6.374466071049847e-05, + "loss": 3.2938, + "step": 4172 + }, + { + "epoch": 6.1, + "grad_norm": 2.8517701625823975, + "learning_rate": 6.370095242594279e-05, + "loss": 3.338, + "step": 4173 + }, + { + "epoch": 6.11, + "grad_norm": 3.4475059509277344, + "learning_rate": 6.365725212693428e-05, + "loss": 3.3719, + "step": 4174 + }, + { + "epoch": 6.11, + "grad_norm": 1.0694806575775146, + "learning_rate": 6.361355982308675e-05, + "loss": 3.3108, + "step": 4175 + }, + { + "epoch": 6.11, + "grad_norm": 4.983922958374023, + "learning_rate": 6.356987552401222e-05, + "loss": 3.326, + "step": 4176 + }, + { + "epoch": 6.11, + "grad_norm": 5.746216773986816, + "learning_rate": 6.352619923932093e-05, + "loss": 3.3831, + "step": 4177 + }, + { + "epoch": 6.11, + "grad_norm": 2.469836473464966, + "learning_rate": 6.348253097862134e-05, + "loss": 3.3593, + "step": 4178 + }, + { + "epoch": 6.11, + "grad_norm": 3.8957202434539795, + "learning_rate": 6.343887075152024e-05, + "loss": 3.3594, + "step": 4179 + }, + { + "epoch": 6.11, + "grad_norm": 5.375302791595459, + "learning_rate": 6.339521856762254e-05, + "loss": 3.3357, + "step": 4180 + }, + { + "epoch": 6.12, + "grad_norm": 4.205276966094971, + "learning_rate": 6.335157443653142e-05, + "loss": 3.3987, + "step": 4181 + }, + { + "epoch": 6.12, + "grad_norm": 1.7369492053985596, + "learning_rate": 6.330793836784836e-05, + "loss": 3.3606, + "step": 4182 + }, + { + "epoch": 6.12, + "grad_norm": 2.639575481414795, + "learning_rate": 6.326431037117296e-05, + "loss": 3.4109, + "step": 4183 + }, + { + "epoch": 6.12, + "grad_norm": 0.6227443218231201, + "learning_rate": 6.322069045610305e-05, + "loss": 3.3122, + "step": 4184 + }, + { + "epoch": 6.12, + "grad_norm": 2.4166953563690186, + "learning_rate": 6.317707863223477e-05, + "loss": 3.339, + "step": 4185 + }, + { + "epoch": 6.12, + "grad_norm": 1.3330849409103394, + "learning_rate": 6.313347490916243e-05, + "loss": 3.2775, + "step": 4186 + }, + { + "epoch": 6.12, + "grad_norm": 3.1274631023406982, + "learning_rate": 6.308987929647852e-05, + "loss": 3.2931, + "step": 4187 + }, + { + "epoch": 6.13, + "grad_norm": 2.9381067752838135, + "learning_rate": 6.30462918037738e-05, + "loss": 3.347, + "step": 4188 + }, + { + "epoch": 6.13, + "grad_norm": 0.6513715982437134, + "learning_rate": 6.300271244063726e-05, + "loss": 3.3084, + "step": 4189 + }, + { + "epoch": 6.13, + "grad_norm": 1.9384881258010864, + "learning_rate": 6.295914121665602e-05, + "loss": 3.3508, + "step": 4190 + }, + { + "epoch": 6.13, + "grad_norm": 1.33826744556427, + "learning_rate": 6.291557814141547e-05, + "loss": 3.3385, + "step": 4191 + }, + { + "epoch": 6.13, + "grad_norm": 1.159730076789856, + "learning_rate": 6.287202322449922e-05, + "loss": 3.3025, + "step": 4192 + }, + { + "epoch": 6.13, + "grad_norm": 15.805403709411621, + "learning_rate": 6.282847647548904e-05, + "loss": 3.3027, + "step": 4193 + }, + { + "epoch": 6.14, + "grad_norm": 4.59617280960083, + "learning_rate": 6.278493790396489e-05, + "loss": 3.3538, + "step": 4194 + }, + { + "epoch": 6.14, + "grad_norm": 2.9616920948028564, + "learning_rate": 6.274140751950501e-05, + "loss": 3.3287, + "step": 4195 + }, + { + "epoch": 6.14, + "grad_norm": 2.299812078475952, + "learning_rate": 6.269788533168581e-05, + "loss": 3.3451, + "step": 4196 + }, + { + "epoch": 6.14, + "grad_norm": 2.9177210330963135, + "learning_rate": 6.26543713500818e-05, + "loss": 3.3583, + "step": 4197 + }, + { + "epoch": 6.14, + "grad_norm": 0.9003336429595947, + "learning_rate": 6.261086558426586e-05, + "loss": 3.2672, + "step": 4198 + }, + { + "epoch": 6.14, + "grad_norm": 6.0708184242248535, + "learning_rate": 6.25673680438089e-05, + "loss": 3.2929, + "step": 4199 + }, + { + "epoch": 6.14, + "grad_norm": 6.111046314239502, + "learning_rate": 6.25238787382801e-05, + "loss": 3.3031, + "step": 4200 + }, + { + "epoch": 6.15, + "grad_norm": 4.454471588134766, + "learning_rate": 6.248039767724683e-05, + "loss": 3.3293, + "step": 4201 + }, + { + "epoch": 6.15, + "grad_norm": 0.9689366221427917, + "learning_rate": 6.243692487027464e-05, + "loss": 3.2844, + "step": 4202 + }, + { + "epoch": 6.15, + "grad_norm": 4.885942459106445, + "learning_rate": 6.239346032692721e-05, + "loss": 3.3346, + "step": 4203 + }, + { + "epoch": 6.15, + "grad_norm": 3.6569879055023193, + "learning_rate": 6.23500040567665e-05, + "loss": 3.2936, + "step": 4204 + }, + { + "epoch": 6.15, + "grad_norm": 1.7212425470352173, + "learning_rate": 6.23065560693526e-05, + "loss": 3.3124, + "step": 4205 + }, + { + "epoch": 6.15, + "grad_norm": 2.9631359577178955, + "learning_rate": 6.226311637424371e-05, + "loss": 3.3716, + "step": 4206 + }, + { + "epoch": 6.15, + "grad_norm": 1.4877088069915771, + "learning_rate": 6.221968498099636e-05, + "loss": 3.2881, + "step": 4207 + }, + { + "epoch": 6.16, + "grad_norm": 5.8026533126831055, + "learning_rate": 6.21762618991651e-05, + "loss": 3.2705, + "step": 4208 + }, + { + "epoch": 6.16, + "grad_norm": 6.365312576293945, + "learning_rate": 6.213284713830271e-05, + "loss": 3.3174, + "step": 4209 + }, + { + "epoch": 6.16, + "grad_norm": 5.353916645050049, + "learning_rate": 6.208944070796022e-05, + "loss": 3.3226, + "step": 4210 + }, + { + "epoch": 6.16, + "grad_norm": 1.4095226526260376, + "learning_rate": 6.204604261768669e-05, + "loss": 3.3037, + "step": 4211 + }, + { + "epoch": 6.16, + "grad_norm": 2.51399827003479, + "learning_rate": 6.200265287702942e-05, + "loss": 3.3118, + "step": 4212 + }, + { + "epoch": 6.16, + "grad_norm": 1.4615494012832642, + "learning_rate": 6.19592714955339e-05, + "loss": 3.2963, + "step": 4213 + }, + { + "epoch": 6.16, + "grad_norm": 2.4555823802948, + "learning_rate": 6.191589848274368e-05, + "loss": 3.275, + "step": 4214 + }, + { + "epoch": 6.17, + "grad_norm": 3.085233211517334, + "learning_rate": 6.187253384820061e-05, + "loss": 3.2726, + "step": 4215 + }, + { + "epoch": 6.17, + "grad_norm": 0.8324146270751953, + "learning_rate": 6.182917760144453e-05, + "loss": 3.3158, + "step": 4216 + }, + { + "epoch": 6.17, + "grad_norm": 1.7136584520339966, + "learning_rate": 6.178582975201359e-05, + "loss": 3.2768, + "step": 4217 + }, + { + "epoch": 6.17, + "grad_norm": 1.937159538269043, + "learning_rate": 6.1742490309444e-05, + "loss": 3.3338, + "step": 4218 + }, + { + "epoch": 6.17, + "grad_norm": 1.2840466499328613, + "learning_rate": 6.169915928327012e-05, + "loss": 3.2964, + "step": 4219 + }, + { + "epoch": 6.17, + "grad_norm": 2.5718352794647217, + "learning_rate": 6.165583668302457e-05, + "loss": 3.287, + "step": 4220 + }, + { + "epoch": 6.18, + "grad_norm": 4.60259485244751, + "learning_rate": 6.161252251823795e-05, + "loss": 3.2494, + "step": 4221 + }, + { + "epoch": 6.18, + "grad_norm": 1.160286545753479, + "learning_rate": 6.15692167984391e-05, + "loss": 3.3453, + "step": 4222 + }, + { + "epoch": 6.18, + "grad_norm": 2.802889823913574, + "learning_rate": 6.152591953315505e-05, + "loss": 3.3133, + "step": 4223 + }, + { + "epoch": 6.18, + "grad_norm": 1.775070071220398, + "learning_rate": 6.148263073191082e-05, + "loss": 3.2936, + "step": 4224 + }, + { + "epoch": 6.18, + "grad_norm": 1.9516983032226562, + "learning_rate": 6.14393504042297e-05, + "loss": 3.3066, + "step": 4225 + }, + { + "epoch": 6.18, + "eval_loss": 3.107570171356201, + "eval_runtime": 1177.648, + "eval_samples_per_second": 14.556, + "eval_steps_per_second": 0.91, + "step": 4225 + }, + { + "epoch": 6.18, + "grad_norm": 1.7126461267471313, + "learning_rate": 6.13960785596331e-05, + "loss": 3.2474, + "step": 4226 + }, + { + "epoch": 6.18, + "grad_norm": 1.2491967678070068, + "learning_rate": 6.135281520764049e-05, + "loss": 3.3243, + "step": 4227 + }, + { + "epoch": 6.19, + "grad_norm": 1.4934719800949097, + "learning_rate": 6.130956035776951e-05, + "loss": 3.3423, + "step": 4228 + }, + { + "epoch": 6.19, + "grad_norm": 1.9269001483917236, + "learning_rate": 6.126631401953598e-05, + "loss": 3.2633, + "step": 4229 + }, + { + "epoch": 6.19, + "grad_norm": 1.3831140995025635, + "learning_rate": 6.12230762024538e-05, + "loss": 3.286, + "step": 4230 + }, + { + "epoch": 6.19, + "grad_norm": 2.75412917137146, + "learning_rate": 6.117984691603493e-05, + "loss": 3.2788, + "step": 4231 + }, + { + "epoch": 6.19, + "grad_norm": 3.342359781265259, + "learning_rate": 6.113662616978961e-05, + "loss": 3.2737, + "step": 4232 + }, + { + "epoch": 6.19, + "grad_norm": 0.9600224494934082, + "learning_rate": 6.109341397322611e-05, + "loss": 3.2899, + "step": 4233 + }, + { + "epoch": 6.19, + "grad_norm": 2.350534200668335, + "learning_rate": 6.105021033585074e-05, + "loss": 3.2768, + "step": 4234 + }, + { + "epoch": 6.2, + "grad_norm": 1.1939469575881958, + "learning_rate": 6.10070152671681e-05, + "loss": 3.2706, + "step": 4235 + }, + { + "epoch": 6.2, + "grad_norm": 4.221333980560303, + "learning_rate": 6.096382877668079e-05, + "loss": 3.2853, + "step": 4236 + }, + { + "epoch": 6.2, + "grad_norm": 4.897353649139404, + "learning_rate": 6.092065087388953e-05, + "loss": 3.3084, + "step": 4237 + }, + { + "epoch": 6.2, + "grad_norm": 2.3200554847717285, + "learning_rate": 6.087748156829316e-05, + "loss": 3.2422, + "step": 4238 + }, + { + "epoch": 6.2, + "grad_norm": 4.042237758636475, + "learning_rate": 6.08343208693887e-05, + "loss": 3.2441, + "step": 4239 + }, + { + "epoch": 6.2, + "grad_norm": 4.729053974151611, + "learning_rate": 6.079116878667115e-05, + "loss": 3.2911, + "step": 4240 + }, + { + "epoch": 6.2, + "grad_norm": 3.419166088104248, + "learning_rate": 6.074802532963368e-05, + "loss": 3.2984, + "step": 4241 + }, + { + "epoch": 6.21, + "grad_norm": 2.1573331356048584, + "learning_rate": 6.0704890507767597e-05, + "loss": 3.2494, + "step": 4242 + }, + { + "epoch": 6.21, + "grad_norm": 3.604609489440918, + "learning_rate": 6.0661764330562276e-05, + "loss": 3.2961, + "step": 4243 + }, + { + "epoch": 6.21, + "grad_norm": 1.5325900316238403, + "learning_rate": 6.061864680750513e-05, + "loss": 3.2401, + "step": 4244 + }, + { + "epoch": 6.21, + "grad_norm": 3.4049322605133057, + "learning_rate": 6.0575537948081793e-05, + "loss": 3.292, + "step": 4245 + }, + { + "epoch": 6.21, + "grad_norm": 4.517602920532227, + "learning_rate": 6.053243776177591e-05, + "loss": 3.2811, + "step": 4246 + }, + { + "epoch": 6.21, + "grad_norm": 2.2732179164886475, + "learning_rate": 6.0489346258069177e-05, + "loss": 3.2883, + "step": 4247 + }, + { + "epoch": 6.22, + "grad_norm": 5.137287139892578, + "learning_rate": 6.044626344644151e-05, + "loss": 3.2465, + "step": 4248 + }, + { + "epoch": 6.22, + "grad_norm": 5.510692596435547, + "learning_rate": 6.040318933637081e-05, + "loss": 3.3609, + "step": 4249 + }, + { + "epoch": 6.22, + "grad_norm": 4.384718894958496, + "learning_rate": 6.036012393733306e-05, + "loss": 3.2768, + "step": 4250 + }, + { + "epoch": 6.22, + "grad_norm": 0.5632131695747375, + "learning_rate": 6.031706725880243e-05, + "loss": 3.2637, + "step": 4251 + }, + { + "epoch": 6.22, + "grad_norm": 4.083331108093262, + "learning_rate": 6.027401931025103e-05, + "loss": 3.2579, + "step": 4252 + }, + { + "epoch": 6.22, + "grad_norm": 4.434952735900879, + "learning_rate": 6.0230980101149136e-05, + "loss": 3.2995, + "step": 4253 + }, + { + "epoch": 6.22, + "grad_norm": 1.9891043901443481, + "learning_rate": 6.0187949640965146e-05, + "loss": 3.2652, + "step": 4254 + }, + { + "epoch": 6.23, + "grad_norm": 3.985672950744629, + "learning_rate": 6.014492793916539e-05, + "loss": 3.256, + "step": 4255 + }, + { + "epoch": 6.23, + "grad_norm": 5.051290988922119, + "learning_rate": 6.010191500521439e-05, + "loss": 3.2773, + "step": 4256 + }, + { + "epoch": 6.23, + "grad_norm": 3.2940456867218018, + "learning_rate": 6.005891084857473e-05, + "loss": 3.3033, + "step": 4257 + }, + { + "epoch": 6.23, + "grad_norm": 2.565002202987671, + "learning_rate": 6.001591547870701e-05, + "loss": 3.2838, + "step": 4258 + }, + { + "epoch": 6.23, + "grad_norm": 3.5721094608306885, + "learning_rate": 5.997292890506989e-05, + "loss": 3.2499, + "step": 4259 + }, + { + "epoch": 6.23, + "grad_norm": 1.7328767776489258, + "learning_rate": 5.992995113712019e-05, + "loss": 3.2925, + "step": 4260 + }, + { + "epoch": 6.23, + "grad_norm": 3.9454033374786377, + "learning_rate": 5.9886982184312725e-05, + "loss": 3.2744, + "step": 4261 + }, + { + "epoch": 6.24, + "grad_norm": 4.74951171875, + "learning_rate": 5.984402205610031e-05, + "loss": 3.2632, + "step": 4262 + }, + { + "epoch": 6.24, + "grad_norm": 2.6633002758026123, + "learning_rate": 5.980107076193393e-05, + "loss": 3.2585, + "step": 4263 + }, + { + "epoch": 6.24, + "grad_norm": 2.6646337509155273, + "learning_rate": 5.975812831126261e-05, + "loss": 3.3006, + "step": 4264 + }, + { + "epoch": 6.24, + "grad_norm": 3.7945520877838135, + "learning_rate": 5.971519471353335e-05, + "loss": 3.2882, + "step": 4265 + }, + { + "epoch": 6.24, + "grad_norm": 1.7113827466964722, + "learning_rate": 5.9672269978191266e-05, + "loss": 3.2554, + "step": 4266 + }, + { + "epoch": 6.24, + "grad_norm": 3.288426399230957, + "learning_rate": 5.9629354114679556e-05, + "loss": 3.329, + "step": 4267 + }, + { + "epoch": 6.25, + "grad_norm": 5.059687614440918, + "learning_rate": 5.958644713243937e-05, + "loss": 3.2254, + "step": 4268 + }, + { + "epoch": 6.25, + "grad_norm": 2.444850206375122, + "learning_rate": 5.9543549040909954e-05, + "loss": 3.2804, + "step": 4269 + }, + { + "epoch": 6.25, + "grad_norm": 2.6846628189086914, + "learning_rate": 5.9500659849528664e-05, + "loss": 3.2151, + "step": 4270 + }, + { + "epoch": 6.25, + "grad_norm": 3.8976359367370605, + "learning_rate": 5.945777956773077e-05, + "loss": 3.3193, + "step": 4271 + }, + { + "epoch": 6.25, + "grad_norm": 2.4183270931243896, + "learning_rate": 5.941490820494965e-05, + "loss": 3.261, + "step": 4272 + }, + { + "epoch": 6.25, + "grad_norm": 3.022622585296631, + "learning_rate": 5.937204577061677e-05, + "loss": 3.2569, + "step": 4273 + }, + { + "epoch": 6.25, + "grad_norm": 3.768277168273926, + "learning_rate": 5.932919227416156e-05, + "loss": 3.2582, + "step": 4274 + }, + { + "epoch": 6.26, + "grad_norm": 2.1014842987060547, + "learning_rate": 5.9286347725011424e-05, + "loss": 3.236, + "step": 4275 + }, + { + "epoch": 6.26, + "grad_norm": 2.9817419052124023, + "learning_rate": 5.9243512132591985e-05, + "loss": 3.2667, + "step": 4276 + }, + { + "epoch": 6.26, + "grad_norm": 4.514947891235352, + "learning_rate": 5.9200685506326734e-05, + "loss": 3.3089, + "step": 4277 + }, + { + "epoch": 6.26, + "grad_norm": 2.3111038208007812, + "learning_rate": 5.915786785563722e-05, + "loss": 3.253, + "step": 4278 + }, + { + "epoch": 6.26, + "grad_norm": 2.7890591621398926, + "learning_rate": 5.911505918994309e-05, + "loss": 3.2643, + "step": 4279 + }, + { + "epoch": 6.26, + "grad_norm": 3.4461700916290283, + "learning_rate": 5.907225951866195e-05, + "loss": 3.2625, + "step": 4280 + }, + { + "epoch": 6.26, + "grad_norm": 2.0314648151397705, + "learning_rate": 5.902946885120939e-05, + "loss": 3.2712, + "step": 4281 + }, + { + "epoch": 6.27, + "grad_norm": 2.992061138153076, + "learning_rate": 5.8986687196999135e-05, + "loss": 3.2365, + "step": 4282 + }, + { + "epoch": 6.27, + "grad_norm": 4.312194347381592, + "learning_rate": 5.8943914565442856e-05, + "loss": 3.2867, + "step": 4283 + }, + { + "epoch": 6.27, + "grad_norm": 2.6370973587036133, + "learning_rate": 5.8901150965950194e-05, + "loss": 3.239, + "step": 4284 + }, + { + "epoch": 6.27, + "grad_norm": 3.0366334915161133, + "learning_rate": 5.885839640792892e-05, + "loss": 3.2651, + "step": 4285 + }, + { + "epoch": 6.27, + "grad_norm": 4.975083827972412, + "learning_rate": 5.881565090078469e-05, + "loss": 3.2896, + "step": 4286 + }, + { + "epoch": 6.27, + "grad_norm": 2.4393575191497803, + "learning_rate": 5.8772914453921276e-05, + "loss": 3.2273, + "step": 4287 + }, + { + "epoch": 6.27, + "grad_norm": 2.5122718811035156, + "learning_rate": 5.873018707674036e-05, + "loss": 3.2617, + "step": 4288 + }, + { + "epoch": 6.28, + "grad_norm": 3.912677526473999, + "learning_rate": 5.868746877864173e-05, + "loss": 3.3112, + "step": 4289 + }, + { + "epoch": 6.28, + "grad_norm": 1.8555430173873901, + "learning_rate": 5.864475956902312e-05, + "loss": 3.322, + "step": 4290 + }, + { + "epoch": 6.28, + "grad_norm": 11.267170906066895, + "learning_rate": 5.860205945728021e-05, + "loss": 3.2759, + "step": 4291 + }, + { + "epoch": 6.28, + "grad_norm": 7.718640327453613, + "learning_rate": 5.855936845280682e-05, + "loss": 3.2407, + "step": 4292 + }, + { + "epoch": 6.28, + "grad_norm": 8.495302200317383, + "learning_rate": 5.851668656499465e-05, + "loss": 3.3493, + "step": 4293 + }, + { + "epoch": 6.28, + "grad_norm": 4.235414981842041, + "learning_rate": 5.8474013803233394e-05, + "loss": 3.2929, + "step": 4294 + }, + { + "epoch": 6.29, + "grad_norm": 2.0188095569610596, + "learning_rate": 5.843135017691085e-05, + "loss": 3.2134, + "step": 4295 + }, + { + "epoch": 6.29, + "grad_norm": 4.253573417663574, + "learning_rate": 5.8388695695412666e-05, + "loss": 3.2903, + "step": 4296 + }, + { + "epoch": 6.29, + "grad_norm": 2.978238582611084, + "learning_rate": 5.834605036812255e-05, + "loss": 3.2783, + "step": 4297 + }, + { + "epoch": 6.29, + "grad_norm": 1.3776922225952148, + "learning_rate": 5.830341420442221e-05, + "loss": 3.2773, + "step": 4298 + }, + { + "epoch": 6.29, + "grad_norm": 2.41166353225708, + "learning_rate": 5.82607872136913e-05, + "loss": 3.2214, + "step": 4299 + }, + { + "epoch": 6.29, + "grad_norm": 0.7983728647232056, + "learning_rate": 5.821816940530749e-05, + "loss": 3.3103, + "step": 4300 + }, + { + "epoch": 6.29, + "grad_norm": 1.7378346920013428, + "learning_rate": 5.8175560788646414e-05, + "loss": 3.2548, + "step": 4301 + }, + { + "epoch": 6.3, + "grad_norm": 0.7058059573173523, + "learning_rate": 5.813296137308164e-05, + "loss": 3.2865, + "step": 4302 + }, + { + "epoch": 6.3, + "grad_norm": 2.132636070251465, + "learning_rate": 5.809037116798479e-05, + "loss": 3.2757, + "step": 4303 + }, + { + "epoch": 6.3, + "grad_norm": 1.359208345413208, + "learning_rate": 5.804779018272544e-05, + "loss": 3.2997, + "step": 4304 + }, + { + "epoch": 6.3, + "grad_norm": 2.7147774696350098, + "learning_rate": 5.800521842667107e-05, + "loss": 3.2789, + "step": 4305 + }, + { + "epoch": 6.3, + "grad_norm": 1.9574331045150757, + "learning_rate": 5.7962655909187226e-05, + "loss": 3.2634, + "step": 4306 + }, + { + "epoch": 6.3, + "grad_norm": 1.1457573175430298, + "learning_rate": 5.792010263963735e-05, + "loss": 3.2802, + "step": 4307 + }, + { + "epoch": 6.3, + "grad_norm": 0.9981921315193176, + "learning_rate": 5.7877558627382886e-05, + "loss": 3.3431, + "step": 4308 + }, + { + "epoch": 6.31, + "grad_norm": 1.708836317062378, + "learning_rate": 5.783502388178328e-05, + "loss": 3.2667, + "step": 4309 + }, + { + "epoch": 6.31, + "grad_norm": 1.0355262756347656, + "learning_rate": 5.7792498412195764e-05, + "loss": 3.2917, + "step": 4310 + }, + { + "epoch": 6.31, + "grad_norm": 3.3587546348571777, + "learning_rate": 5.774998222797577e-05, + "loss": 3.2507, + "step": 4311 + }, + { + "epoch": 6.31, + "grad_norm": 3.278871774673462, + "learning_rate": 5.770747533847658e-05, + "loss": 3.2427, + "step": 4312 + }, + { + "epoch": 6.31, + "grad_norm": 0.9090314507484436, + "learning_rate": 5.766497775304931e-05, + "loss": 3.2401, + "step": 4313 + }, + { + "epoch": 6.31, + "grad_norm": 3.0170810222625732, + "learning_rate": 5.7622489481043254e-05, + "loss": 3.2708, + "step": 4314 + }, + { + "epoch": 6.31, + "grad_norm": 3.1690783500671387, + "learning_rate": 5.758001053180555e-05, + "loss": 3.2354, + "step": 4315 + }, + { + "epoch": 6.32, + "grad_norm": 0.4450177550315857, + "learning_rate": 5.753754091468115e-05, + "loss": 3.2435, + "step": 4316 + }, + { + "epoch": 6.32, + "grad_norm": 3.6252622604370117, + "learning_rate": 5.749508063901328e-05, + "loss": 3.2594, + "step": 4317 + }, + { + "epoch": 6.32, + "grad_norm": 5.237750053405762, + "learning_rate": 5.745262971414276e-05, + "loss": 3.2698, + "step": 4318 + }, + { + "epoch": 6.32, + "grad_norm": 1.0414494276046753, + "learning_rate": 5.741018814940853e-05, + "loss": 3.2509, + "step": 4319 + }, + { + "epoch": 6.32, + "grad_norm": 2.6525533199310303, + "learning_rate": 5.736775595414756e-05, + "loss": 3.2609, + "step": 4320 + }, + { + "epoch": 6.32, + "grad_norm": 3.908628225326538, + "learning_rate": 5.7325333137694525e-05, + "loss": 3.2479, + "step": 4321 + }, + { + "epoch": 6.33, + "grad_norm": 1.5807950496673584, + "learning_rate": 5.728291970938218e-05, + "loss": 3.2861, + "step": 4322 + }, + { + "epoch": 6.33, + "grad_norm": 3.6872596740722656, + "learning_rate": 5.72405156785413e-05, + "loss": 3.2316, + "step": 4323 + }, + { + "epoch": 6.33, + "grad_norm": 4.621198654174805, + "learning_rate": 5.719812105450035e-05, + "loss": 3.245, + "step": 4324 + }, + { + "epoch": 6.33, + "grad_norm": 1.5106016397476196, + "learning_rate": 5.715573584658589e-05, + "loss": 3.215, + "step": 4325 + }, + { + "epoch": 6.33, + "grad_norm": 3.5040335655212402, + "learning_rate": 5.711336006412251e-05, + "loss": 3.2626, + "step": 4326 + }, + { + "epoch": 6.33, + "grad_norm": 4.461084842681885, + "learning_rate": 5.707099371643245e-05, + "loss": 3.2804, + "step": 4327 + }, + { + "epoch": 6.33, + "grad_norm": 2.81062912940979, + "learning_rate": 5.7028636812836035e-05, + "loss": 3.2476, + "step": 4328 + }, + { + "epoch": 6.34, + "grad_norm": 2.1609275341033936, + "learning_rate": 5.6986289362651624e-05, + "loss": 3.2922, + "step": 4329 + }, + { + "epoch": 6.34, + "grad_norm": 39.97700881958008, + "learning_rate": 5.694395137519525e-05, + "loss": 3.262, + "step": 4330 + }, + { + "epoch": 6.34, + "grad_norm": 4.451782703399658, + "learning_rate": 5.690162285978102e-05, + "loss": 3.3011, + "step": 4331 + }, + { + "epoch": 6.34, + "grad_norm": 1.8545382022857666, + "learning_rate": 5.6859303825720935e-05, + "loss": 3.2684, + "step": 4332 + }, + { + "epoch": 6.34, + "grad_norm": 3.6993982791900635, + "learning_rate": 5.6816994282324875e-05, + "loss": 3.2664, + "step": 4333 + }, + { + "epoch": 6.34, + "grad_norm": 4.563552379608154, + "learning_rate": 5.67746942389007e-05, + "loss": 3.2004, + "step": 4334 + }, + { + "epoch": 6.34, + "grad_norm": 3.6278369426727295, + "learning_rate": 5.6732403704754086e-05, + "loss": 3.1965, + "step": 4335 + }, + { + "epoch": 6.35, + "grad_norm": 1.3190866708755493, + "learning_rate": 5.66901226891887e-05, + "loss": 3.2505, + "step": 4336 + }, + { + "epoch": 6.35, + "grad_norm": 3.1047582626342773, + "learning_rate": 5.664785120150607e-05, + "loss": 3.3744, + "step": 4337 + }, + { + "epoch": 6.35, + "grad_norm": 1.4010859727859497, + "learning_rate": 5.660558925100565e-05, + "loss": 3.2555, + "step": 4338 + }, + { + "epoch": 6.35, + "grad_norm": 2.454122304916382, + "learning_rate": 5.656333684698476e-05, + "loss": 3.2401, + "step": 4339 + }, + { + "epoch": 6.35, + "grad_norm": 2.7071664333343506, + "learning_rate": 5.652109399873873e-05, + "loss": 3.2436, + "step": 4340 + }, + { + "epoch": 6.35, + "grad_norm": 0.4050593674182892, + "learning_rate": 5.647886071556054e-05, + "loss": 3.2843, + "step": 4341 + }, + { + "epoch": 6.35, + "grad_norm": 3.2467408180236816, + "learning_rate": 5.6436637006741375e-05, + "loss": 3.2526, + "step": 4342 + }, + { + "epoch": 6.36, + "grad_norm": 3.083259344100952, + "learning_rate": 5.639442288157015e-05, + "loss": 3.2539, + "step": 4343 + }, + { + "epoch": 6.36, + "grad_norm": 1.2270174026489258, + "learning_rate": 5.635221834933359e-05, + "loss": 3.2435, + "step": 4344 + }, + { + "epoch": 6.36, + "grad_norm": 1.6782840490341187, + "learning_rate": 5.63100234193165e-05, + "loss": 3.2838, + "step": 4345 + }, + { + "epoch": 6.36, + "grad_norm": 1.8353314399719238, + "learning_rate": 5.6267838100801516e-05, + "loss": 3.2793, + "step": 4346 + }, + { + "epoch": 6.36, + "grad_norm": 1.642630934715271, + "learning_rate": 5.622566240306899e-05, + "loss": 3.261, + "step": 4347 + }, + { + "epoch": 6.36, + "grad_norm": 1.191734790802002, + "learning_rate": 5.618349633539741e-05, + "loss": 3.2822, + "step": 4348 + }, + { + "epoch": 6.37, + "grad_norm": 0.7267574071884155, + "learning_rate": 5.614133990706304e-05, + "loss": 3.2482, + "step": 4349 + }, + { + "epoch": 6.37, + "grad_norm": 0.9115816950798035, + "learning_rate": 5.6099193127339864e-05, + "loss": 3.2347, + "step": 4350 + }, + { + "epoch": 6.37, + "grad_norm": 1.4531807899475098, + "learning_rate": 5.6057056005500085e-05, + "loss": 3.2186, + "step": 4351 + }, + { + "epoch": 6.37, + "grad_norm": 1.0633835792541504, + "learning_rate": 5.601492855081345e-05, + "loss": 3.2526, + "step": 4352 + }, + { + "epoch": 6.37, + "grad_norm": 0.716886579990387, + "learning_rate": 5.5972810772547724e-05, + "loss": 3.2448, + "step": 4353 + }, + { + "epoch": 6.37, + "grad_norm": 0.9413835406303406, + "learning_rate": 5.5930702679968636e-05, + "loss": 3.2125, + "step": 4354 + }, + { + "epoch": 6.37, + "grad_norm": 1.2338863611221313, + "learning_rate": 5.58886042823396e-05, + "loss": 3.2228, + "step": 4355 + }, + { + "epoch": 6.38, + "grad_norm": 2.2373838424682617, + "learning_rate": 5.584651558892193e-05, + "loss": 3.2777, + "step": 4356 + }, + { + "epoch": 6.38, + "grad_norm": 1.3574821949005127, + "learning_rate": 5.580443660897503e-05, + "loss": 3.2371, + "step": 4357 + }, + { + "epoch": 6.38, + "grad_norm": 1.0699307918548584, + "learning_rate": 5.576236735175584e-05, + "loss": 3.2889, + "step": 4358 + }, + { + "epoch": 6.38, + "grad_norm": 0.737332284450531, + "learning_rate": 5.572030782651938e-05, + "loss": 3.2533, + "step": 4359 + }, + { + "epoch": 6.38, + "grad_norm": 1.1307657957077026, + "learning_rate": 5.567825804251843e-05, + "loss": 3.2832, + "step": 4360 + }, + { + "epoch": 6.38, + "grad_norm": 0.6265879273414612, + "learning_rate": 5.563621800900369e-05, + "loss": 3.2663, + "step": 4361 + }, + { + "epoch": 6.38, + "grad_norm": 2.2958106994628906, + "learning_rate": 5.559418773522367e-05, + "loss": 3.2426, + "step": 4362 + }, + { + "epoch": 6.39, + "grad_norm": 1.0793524980545044, + "learning_rate": 5.555216723042476e-05, + "loss": 3.2336, + "step": 4363 + }, + { + "epoch": 6.39, + "grad_norm": 1.8671746253967285, + "learning_rate": 5.551015650385119e-05, + "loss": 3.2449, + "step": 4364 + }, + { + "epoch": 6.39, + "grad_norm": 1.5960359573364258, + "learning_rate": 5.5468155564745e-05, + "loss": 3.2135, + "step": 4365 + }, + { + "epoch": 6.39, + "grad_norm": 2.3406100273132324, + "learning_rate": 5.542616442234618e-05, + "loss": 3.2034, + "step": 4366 + }, + { + "epoch": 6.39, + "grad_norm": 1.8202327489852905, + "learning_rate": 5.538418308589246e-05, + "loss": 3.2608, + "step": 4367 + }, + { + "epoch": 6.39, + "grad_norm": 1.2679892778396606, + "learning_rate": 5.534221156461945e-05, + "loss": 3.2336, + "step": 4368 + }, + { + "epoch": 6.4, + "grad_norm": 1.0765639543533325, + "learning_rate": 5.530024986776061e-05, + "loss": 3.2034, + "step": 4369 + }, + { + "epoch": 6.4, + "grad_norm": 2.065032720565796, + "learning_rate": 5.5258298004547246e-05, + "loss": 3.2323, + "step": 4370 + }, + { + "epoch": 6.4, + "grad_norm": 1.7677899599075317, + "learning_rate": 5.521635598420847e-05, + "loss": 3.2416, + "step": 4371 + }, + { + "epoch": 6.4, + "grad_norm": 1.2142525911331177, + "learning_rate": 5.5174423815971254e-05, + "loss": 3.2783, + "step": 4372 + }, + { + "epoch": 6.4, + "grad_norm": 1.2158989906311035, + "learning_rate": 5.513250150906039e-05, + "loss": 3.2203, + "step": 4373 + }, + { + "epoch": 6.4, + "grad_norm": 2.0104503631591797, + "learning_rate": 5.509058907269854e-05, + "loss": 3.2627, + "step": 4374 + }, + { + "epoch": 6.4, + "grad_norm": 1.794409155845642, + "learning_rate": 5.504868651610604e-05, + "loss": 3.2247, + "step": 4375 + }, + { + "epoch": 6.41, + "grad_norm": 1.241929531097412, + "learning_rate": 5.5006793848501314e-05, + "loss": 3.2352, + "step": 4376 + }, + { + "epoch": 6.41, + "grad_norm": 1.3682582378387451, + "learning_rate": 5.4964911079100446e-05, + "loss": 3.2834, + "step": 4377 + }, + { + "epoch": 6.41, + "grad_norm": 1.42366623878479, + "learning_rate": 5.492303821711724e-05, + "loss": 3.2356, + "step": 4378 + }, + { + "epoch": 6.41, + "grad_norm": 0.9967645406723022, + "learning_rate": 5.488117527176357e-05, + "loss": 3.2623, + "step": 4379 + }, + { + "epoch": 6.41, + "grad_norm": 1.6339219808578491, + "learning_rate": 5.4839322252249016e-05, + "loss": 3.2162, + "step": 4380 + }, + { + "epoch": 6.41, + "grad_norm": 0.9269053936004639, + "learning_rate": 5.479747916778088e-05, + "loss": 3.2236, + "step": 4381 + }, + { + "epoch": 6.41, + "grad_norm": 3.181241989135742, + "learning_rate": 5.4755646027564336e-05, + "loss": 3.2464, + "step": 4382 + }, + { + "epoch": 6.42, + "grad_norm": 3.343838930130005, + "learning_rate": 5.4713822840802544e-05, + "loss": 3.227, + "step": 4383 + }, + { + "epoch": 6.42, + "grad_norm": 0.8503923416137695, + "learning_rate": 5.467200961669619e-05, + "loss": 3.2975, + "step": 4384 + }, + { + "epoch": 6.42, + "grad_norm": 2.970435380935669, + "learning_rate": 5.4630206364443936e-05, + "loss": 3.2325, + "step": 4385 + }, + { + "epoch": 6.42, + "grad_norm": 3.7299089431762695, + "learning_rate": 5.4588413093242244e-05, + "loss": 3.1893, + "step": 4386 + }, + { + "epoch": 6.42, + "grad_norm": 1.3558520078659058, + "learning_rate": 5.454662981228532e-05, + "loss": 3.2106, + "step": 4387 + }, + { + "epoch": 6.42, + "grad_norm": 3.116187572479248, + "learning_rate": 5.450485653076524e-05, + "loss": 3.2292, + "step": 4388 + }, + { + "epoch": 6.42, + "grad_norm": 3.9433794021606445, + "learning_rate": 5.446309325787181e-05, + "loss": 3.2971, + "step": 4389 + }, + { + "epoch": 6.43, + "grad_norm": 2.001657247543335, + "learning_rate": 5.44213400027927e-05, + "loss": 3.2795, + "step": 4390 + }, + { + "epoch": 6.43, + "grad_norm": 2.557114601135254, + "learning_rate": 5.4379596774713334e-05, + "loss": 3.2177, + "step": 4391 + }, + { + "epoch": 6.43, + "grad_norm": 3.371882677078247, + "learning_rate": 5.4337863582816936e-05, + "loss": 3.2478, + "step": 4392 + }, + { + "epoch": 6.43, + "grad_norm": 1.5067156553268433, + "learning_rate": 5.4296140436284546e-05, + "loss": 3.2362, + "step": 4393 + }, + { + "epoch": 6.43, + "grad_norm": 2.2881274223327637, + "learning_rate": 5.425442734429495e-05, + "loss": 3.2567, + "step": 4394 + }, + { + "epoch": 6.43, + "eval_loss": 3.067939043045044, + "eval_runtime": 1166.4557, + "eval_samples_per_second": 14.696, + "eval_steps_per_second": 0.919, + "step": 4394 + }, + { + "epoch": 6.43, + "grad_norm": 2.7140753269195557, + "learning_rate": 5.4212724316024776e-05, + "loss": 3.1976, + "step": 4395 + }, + { + "epoch": 6.44, + "grad_norm": 1.5862021446228027, + "learning_rate": 5.4171031360648406e-05, + "loss": 3.2628, + "step": 4396 + }, + { + "epoch": 6.44, + "grad_norm": 1.6570472717285156, + "learning_rate": 5.4129348487337995e-05, + "loss": 3.2399, + "step": 4397 + }, + { + "epoch": 6.44, + "grad_norm": 1.0379539728164673, + "learning_rate": 5.408767570526352e-05, + "loss": 3.2495, + "step": 4398 + }, + { + "epoch": 6.44, + "grad_norm": 0.9085617065429688, + "learning_rate": 5.404601302359269e-05, + "loss": 3.2073, + "step": 4399 + }, + { + "epoch": 6.44, + "grad_norm": 1.3575804233551025, + "learning_rate": 5.400436045149102e-05, + "loss": 3.2191, + "step": 4400 + }, + { + "epoch": 6.44, + "grad_norm": 1.4507752656936646, + "learning_rate": 5.39627179981218e-05, + "loss": 3.2083, + "step": 4401 + }, + { + "epoch": 6.44, + "grad_norm": 1.4040168523788452, + "learning_rate": 5.392108567264609e-05, + "loss": 3.2492, + "step": 4402 + }, + { + "epoch": 6.45, + "grad_norm": 1.684368371963501, + "learning_rate": 5.387946348422274e-05, + "loss": 3.2433, + "step": 4403 + }, + { + "epoch": 6.45, + "grad_norm": 1.438843846321106, + "learning_rate": 5.383785144200832e-05, + "loss": 3.2335, + "step": 4404 + }, + { + "epoch": 6.45, + "grad_norm": 1.311325192451477, + "learning_rate": 5.379624955515723e-05, + "loss": 3.2418, + "step": 4405 + }, + { + "epoch": 6.45, + "grad_norm": 1.0702279806137085, + "learning_rate": 5.375465783282163e-05, + "loss": 3.2534, + "step": 4406 + }, + { + "epoch": 6.45, + "grad_norm": 1.4737073183059692, + "learning_rate": 5.371307628415131e-05, + "loss": 3.2167, + "step": 4407 + }, + { + "epoch": 6.45, + "grad_norm": 1.0920416116714478, + "learning_rate": 5.367150491829408e-05, + "loss": 3.2418, + "step": 4408 + }, + { + "epoch": 6.45, + "grad_norm": 2.5086584091186523, + "learning_rate": 5.362994374439525e-05, + "loss": 3.2499, + "step": 4409 + }, + { + "epoch": 6.46, + "grad_norm": 1.8857651948928833, + "learning_rate": 5.3588392771598e-05, + "loss": 3.2232, + "step": 4410 + }, + { + "epoch": 6.46, + "grad_norm": 1.4724599123001099, + "learning_rate": 5.354685200904337e-05, + "loss": 3.2534, + "step": 4411 + }, + { + "epoch": 6.46, + "grad_norm": 1.5440292358398438, + "learning_rate": 5.350532146586996e-05, + "loss": 3.2061, + "step": 4412 + }, + { + "epoch": 6.46, + "grad_norm": 1.186196208000183, + "learning_rate": 5.346380115121419e-05, + "loss": 3.2526, + "step": 4413 + }, + { + "epoch": 6.46, + "grad_norm": 1.4283154010772705, + "learning_rate": 5.342229107421037e-05, + "loss": 3.2498, + "step": 4414 + }, + { + "epoch": 6.46, + "grad_norm": 1.7490146160125732, + "learning_rate": 5.3380791243990325e-05, + "loss": 3.2403, + "step": 4415 + }, + { + "epoch": 6.46, + "grad_norm": 1.5179111957550049, + "learning_rate": 5.333930166968374e-05, + "loss": 3.2239, + "step": 4416 + }, + { + "epoch": 6.47, + "grad_norm": 1.3260568380355835, + "learning_rate": 5.329782236041816e-05, + "loss": 3.2357, + "step": 4417 + }, + { + "epoch": 6.47, + "grad_norm": 0.9309088587760925, + "learning_rate": 5.325635332531864e-05, + "loss": 3.2212, + "step": 4418 + }, + { + "epoch": 6.47, + "grad_norm": 1.8849228620529175, + "learning_rate": 5.321489457350812e-05, + "loss": 3.2158, + "step": 4419 + }, + { + "epoch": 6.47, + "grad_norm": 2.031489610671997, + "learning_rate": 5.3173446114107265e-05, + "loss": 3.229, + "step": 4420 + }, + { + "epoch": 6.47, + "grad_norm": 1.7095611095428467, + "learning_rate": 5.313200795623443e-05, + "loss": 3.2323, + "step": 4421 + }, + { + "epoch": 6.47, + "grad_norm": 2.060396671295166, + "learning_rate": 5.309058010900576e-05, + "loss": 3.2519, + "step": 4422 + }, + { + "epoch": 6.48, + "grad_norm": 0.750653088092804, + "learning_rate": 5.3049162581535094e-05, + "loss": 3.2344, + "step": 4423 + }, + { + "epoch": 6.48, + "grad_norm": 1.6587752103805542, + "learning_rate": 5.300775538293401e-05, + "loss": 3.2506, + "step": 4424 + }, + { + "epoch": 6.48, + "grad_norm": 0.8414655923843384, + "learning_rate": 5.296635852231182e-05, + "loss": 3.2065, + "step": 4425 + }, + { + "epoch": 6.48, + "grad_norm": 0.4840925335884094, + "learning_rate": 5.292497200877553e-05, + "loss": 3.2054, + "step": 4426 + }, + { + "epoch": 6.48, + "grad_norm": 0.6098539233207703, + "learning_rate": 5.288359585142993e-05, + "loss": 3.2024, + "step": 4427 + }, + { + "epoch": 6.48, + "grad_norm": 0.46651723980903625, + "learning_rate": 5.2842230059377486e-05, + "loss": 3.2161, + "step": 4428 + }, + { + "epoch": 6.48, + "grad_norm": 0.508057177066803, + "learning_rate": 5.280087464171838e-05, + "loss": 3.1871, + "step": 4429 + }, + { + "epoch": 6.49, + "grad_norm": 0.5408399701118469, + "learning_rate": 5.275952960755055e-05, + "loss": 3.2624, + "step": 4430 + }, + { + "epoch": 6.49, + "grad_norm": 1.0045307874679565, + "learning_rate": 5.271819496596964e-05, + "loss": 3.2303, + "step": 4431 + }, + { + "epoch": 6.49, + "grad_norm": 0.6911265850067139, + "learning_rate": 5.2676870726068906e-05, + "loss": 3.2087, + "step": 4432 + }, + { + "epoch": 6.49, + "grad_norm": 0.6002917289733887, + "learning_rate": 5.2635556896939486e-05, + "loss": 3.1978, + "step": 4433 + }, + { + "epoch": 6.49, + "grad_norm": 0.9630332589149475, + "learning_rate": 5.259425348767018e-05, + "loss": 3.2238, + "step": 4434 + }, + { + "epoch": 6.49, + "grad_norm": 0.7534378170967102, + "learning_rate": 5.2552960507347325e-05, + "loss": 3.2308, + "step": 4435 + }, + { + "epoch": 6.49, + "grad_norm": 0.9733850359916687, + "learning_rate": 5.251167796505523e-05, + "loss": 3.2392, + "step": 4436 + }, + { + "epoch": 6.5, + "grad_norm": 1.1052712202072144, + "learning_rate": 5.247040586987575e-05, + "loss": 3.2257, + "step": 4437 + }, + { + "epoch": 6.5, + "grad_norm": 0.8000559210777283, + "learning_rate": 5.2429144230888384e-05, + "loss": 3.2837, + "step": 4438 + }, + { + "epoch": 6.5, + "grad_norm": 0.7555025219917297, + "learning_rate": 5.238789305717053e-05, + "loss": 3.2463, + "step": 4439 + }, + { + "epoch": 6.5, + "grad_norm": 1.0577791929244995, + "learning_rate": 5.2346652357797146e-05, + "loss": 3.2328, + "step": 4440 + }, + { + "epoch": 6.5, + "grad_norm": 0.6421762704849243, + "learning_rate": 5.230542214184081e-05, + "loss": 3.2186, + "step": 4441 + }, + { + "epoch": 6.5, + "grad_norm": 1.1943224668502808, + "learning_rate": 5.226420241837204e-05, + "loss": 3.2487, + "step": 4442 + }, + { + "epoch": 6.5, + "grad_norm": 0.6367546916007996, + "learning_rate": 5.22229931964588e-05, + "loss": 3.2938, + "step": 4443 + }, + { + "epoch": 6.51, + "grad_norm": 1.3476903438568115, + "learning_rate": 5.2181794485166824e-05, + "loss": 3.2098, + "step": 4444 + }, + { + "epoch": 6.51, + "grad_norm": 1.686813235282898, + "learning_rate": 5.214060629355968e-05, + "loss": 3.2456, + "step": 4445 + }, + { + "epoch": 6.51, + "grad_norm": 0.49779945611953735, + "learning_rate": 5.209942863069837e-05, + "loss": 3.2537, + "step": 4446 + }, + { + "epoch": 6.51, + "grad_norm": 1.0947519540786743, + "learning_rate": 5.20582615056417e-05, + "loss": 3.3028, + "step": 4447 + }, + { + "epoch": 6.51, + "grad_norm": 0.6501612663269043, + "learning_rate": 5.201710492744629e-05, + "loss": 3.2317, + "step": 4448 + }, + { + "epoch": 6.51, + "grad_norm": 0.4535963535308838, + "learning_rate": 5.19759589051662e-05, + "loss": 3.23, + "step": 4449 + }, + { + "epoch": 6.52, + "grad_norm": 0.5250782370567322, + "learning_rate": 5.193482344785327e-05, + "loss": 3.2435, + "step": 4450 + }, + { + "epoch": 6.52, + "grad_norm": 0.400921493768692, + "learning_rate": 5.1893698564557124e-05, + "loss": 3.2188, + "step": 4451 + }, + { + "epoch": 6.52, + "grad_norm": 0.902998149394989, + "learning_rate": 5.1852584264324866e-05, + "loss": 3.2495, + "step": 4452 + }, + { + "epoch": 6.52, + "grad_norm": 0.7799638509750366, + "learning_rate": 5.1811480556201375e-05, + "loss": 3.234, + "step": 4453 + }, + { + "epoch": 6.52, + "grad_norm": 1.8399096727371216, + "learning_rate": 5.1770387449229276e-05, + "loss": 3.3015, + "step": 4454 + }, + { + "epoch": 6.52, + "grad_norm": 0.94755619764328, + "learning_rate": 5.172930495244869e-05, + "loss": 3.2394, + "step": 4455 + }, + { + "epoch": 6.52, + "grad_norm": 2.66386079788208, + "learning_rate": 5.168823307489751e-05, + "loss": 3.2519, + "step": 4456 + }, + { + "epoch": 6.53, + "grad_norm": 1.893293857574463, + "learning_rate": 5.1647171825611284e-05, + "loss": 3.1926, + "step": 4457 + }, + { + "epoch": 6.53, + "grad_norm": 0.6333227753639221, + "learning_rate": 5.160612121362322e-05, + "loss": 3.2169, + "step": 4458 + }, + { + "epoch": 6.53, + "grad_norm": 1.2479146718978882, + "learning_rate": 5.156508124796415e-05, + "loss": 3.2308, + "step": 4459 + }, + { + "epoch": 6.53, + "grad_norm": 0.3292587399482727, + "learning_rate": 5.152405193766263e-05, + "loss": 3.2359, + "step": 4460 + }, + { + "epoch": 6.53, + "grad_norm": 1.253524899482727, + "learning_rate": 5.1483033291744797e-05, + "loss": 3.2176, + "step": 4461 + }, + { + "epoch": 6.53, + "grad_norm": 0.4894808232784271, + "learning_rate": 5.14420253192345e-05, + "loss": 3.2433, + "step": 4462 + }, + { + "epoch": 6.53, + "grad_norm": 1.1630624532699585, + "learning_rate": 5.140102802915321e-05, + "loss": 3.213, + "step": 4463 + }, + { + "epoch": 6.54, + "grad_norm": 0.8647369742393494, + "learning_rate": 5.136004143052007e-05, + "loss": 3.2736, + "step": 4464 + }, + { + "epoch": 6.54, + "grad_norm": 0.9852306842803955, + "learning_rate": 5.1319065532351865e-05, + "loss": 3.2649, + "step": 4465 + }, + { + "epoch": 6.54, + "grad_norm": 0.7359192371368408, + "learning_rate": 5.1278100343662916e-05, + "loss": 3.2311, + "step": 4466 + }, + { + "epoch": 6.54, + "grad_norm": 0.38752880692481995, + "learning_rate": 5.123714587346543e-05, + "loss": 3.2308, + "step": 4467 + }, + { + "epoch": 6.54, + "grad_norm": 1.2246233224868774, + "learning_rate": 5.119620213076909e-05, + "loss": 3.1778, + "step": 4468 + }, + { + "epoch": 6.54, + "grad_norm": 0.7400122284889221, + "learning_rate": 5.115526912458113e-05, + "loss": 3.2127, + "step": 4469 + }, + { + "epoch": 6.55, + "grad_norm": 0.9233433604240417, + "learning_rate": 5.111434686390666e-05, + "loss": 3.184, + "step": 4470 + }, + { + "epoch": 6.55, + "grad_norm": 1.7230321168899536, + "learning_rate": 5.107343535774831e-05, + "loss": 3.2116, + "step": 4471 + }, + { + "epoch": 6.55, + "grad_norm": 1.6479278802871704, + "learning_rate": 5.1032534615106195e-05, + "loss": 3.2049, + "step": 4472 + }, + { + "epoch": 6.55, + "grad_norm": 2.626024007797241, + "learning_rate": 5.0991644644978356e-05, + "loss": 3.2434, + "step": 4473 + }, + { + "epoch": 6.55, + "grad_norm": 3.1591696739196777, + "learning_rate": 5.095076545636028e-05, + "loss": 3.2116, + "step": 4474 + }, + { + "epoch": 6.55, + "grad_norm": 0.8667658567428589, + "learning_rate": 5.0909897058245025e-05, + "loss": 3.1994, + "step": 4475 + }, + { + "epoch": 6.55, + "grad_norm": 3.626915454864502, + "learning_rate": 5.08690394596235e-05, + "loss": 3.221, + "step": 4476 + }, + { + "epoch": 6.56, + "grad_norm": 4.862388610839844, + "learning_rate": 5.0828192669484e-05, + "loss": 3.2172, + "step": 4477 + }, + { + "epoch": 6.56, + "grad_norm": 2.2208385467529297, + "learning_rate": 5.078735669681257e-05, + "loss": 3.2153, + "step": 4478 + }, + { + "epoch": 6.56, + "grad_norm": 2.200000286102295, + "learning_rate": 5.074653155059286e-05, + "loss": 3.2971, + "step": 4479 + }, + { + "epoch": 6.56, + "grad_norm": 2.9845163822174072, + "learning_rate": 5.0705717239806125e-05, + "loss": 3.2522, + "step": 4480 + }, + { + "epoch": 6.56, + "grad_norm": 1.4984869956970215, + "learning_rate": 5.066491377343126e-05, + "loss": 3.2009, + "step": 4481 + }, + { + "epoch": 6.56, + "grad_norm": 2.6040592193603516, + "learning_rate": 5.0624121160444716e-05, + "loss": 3.1945, + "step": 4482 + }, + { + "epoch": 6.56, + "grad_norm": 3.3462085723876953, + "learning_rate": 5.0583339409820616e-05, + "loss": 3.1927, + "step": 4483 + }, + { + "epoch": 6.57, + "grad_norm": 1.2761520147323608, + "learning_rate": 5.0542568530530676e-05, + "loss": 3.1981, + "step": 4484 + }, + { + "epoch": 6.57, + "grad_norm": 2.611973285675049, + "learning_rate": 5.05018085315442e-05, + "loss": 3.1832, + "step": 4485 + }, + { + "epoch": 6.57, + "grad_norm": 3.3917155265808105, + "learning_rate": 5.046105942182815e-05, + "loss": 3.2319, + "step": 4486 + }, + { + "epoch": 6.57, + "grad_norm": 2.2328946590423584, + "learning_rate": 5.042032121034701e-05, + "loss": 3.2443, + "step": 4487 + }, + { + "epoch": 6.57, + "grad_norm": 2.1951651573181152, + "learning_rate": 5.037959390606294e-05, + "loss": 3.2073, + "step": 4488 + }, + { + "epoch": 6.57, + "grad_norm": 3.5785024166107178, + "learning_rate": 5.033887751793567e-05, + "loss": 3.264, + "step": 4489 + }, + { + "epoch": 6.57, + "grad_norm": 1.865713119506836, + "learning_rate": 5.0298172054922534e-05, + "loss": 3.2455, + "step": 4490 + }, + { + "epoch": 6.58, + "grad_norm": 2.268017292022705, + "learning_rate": 5.0257477525978446e-05, + "loss": 3.2089, + "step": 4491 + }, + { + "epoch": 6.58, + "grad_norm": 2.626833200454712, + "learning_rate": 5.021679394005596e-05, + "loss": 3.2474, + "step": 4492 + }, + { + "epoch": 6.58, + "grad_norm": 1.4748424291610718, + "learning_rate": 5.0176121306105164e-05, + "loss": 3.2171, + "step": 4493 + }, + { + "epoch": 6.58, + "grad_norm": 1.906065583229065, + "learning_rate": 5.0135459633073776e-05, + "loss": 3.2625, + "step": 4494 + }, + { + "epoch": 6.58, + "grad_norm": 2.797215700149536, + "learning_rate": 5.00948089299071e-05, + "loss": 3.2316, + "step": 4495 + }, + { + "epoch": 6.58, + "grad_norm": 0.824268639087677, + "learning_rate": 5.0054169205548016e-05, + "loss": 3.211, + "step": 4496 + }, + { + "epoch": 6.59, + "grad_norm": 2.206386089324951, + "learning_rate": 5.001354046893697e-05, + "loss": 3.2278, + "step": 4497 + }, + { + "epoch": 6.59, + "grad_norm": 2.08886981010437, + "learning_rate": 4.997292272901203e-05, + "loss": 3.2376, + "step": 4498 + }, + { + "epoch": 6.59, + "grad_norm": 0.7808924913406372, + "learning_rate": 4.993231599470887e-05, + "loss": 3.2411, + "step": 4499 + }, + { + "epoch": 6.59, + "grad_norm": 1.8568205833435059, + "learning_rate": 4.9891720274960566e-05, + "loss": 3.2718, + "step": 4500 + }, + { + "epoch": 6.59, + "grad_norm": 1.4308000802993774, + "learning_rate": 4.985113557869803e-05, + "loss": 3.2299, + "step": 4501 + }, + { + "epoch": 6.59, + "grad_norm": 1.8029470443725586, + "learning_rate": 4.981056191484962e-05, + "loss": 3.2458, + "step": 4502 + }, + { + "epoch": 6.59, + "grad_norm": 1.7437525987625122, + "learning_rate": 4.9769999292341194e-05, + "loss": 3.2255, + "step": 4503 + }, + { + "epoch": 6.6, + "grad_norm": 1.1221601963043213, + "learning_rate": 4.9729447720096265e-05, + "loss": 3.1949, + "step": 4504 + }, + { + "epoch": 6.6, + "grad_norm": 1.0662968158721924, + "learning_rate": 4.968890720703602e-05, + "loss": 3.1963, + "step": 4505 + }, + { + "epoch": 6.6, + "grad_norm": 1.716160535812378, + "learning_rate": 4.9648377762078966e-05, + "loss": 3.2037, + "step": 4506 + }, + { + "epoch": 6.6, + "grad_norm": 1.2884470224380493, + "learning_rate": 4.960785939414134e-05, + "loss": 3.2482, + "step": 4507 + }, + { + "epoch": 6.6, + "grad_norm": 2.104020595550537, + "learning_rate": 4.956735211213701e-05, + "loss": 3.2097, + "step": 4508 + }, + { + "epoch": 6.6, + "grad_norm": 2.3521175384521484, + "learning_rate": 4.9526855924977175e-05, + "loss": 3.2618, + "step": 4509 + }, + { + "epoch": 6.6, + "grad_norm": 1.780421257019043, + "learning_rate": 4.9486370841570803e-05, + "loss": 3.2177, + "step": 4510 + }, + { + "epoch": 6.61, + "grad_norm": 1.345507264137268, + "learning_rate": 4.94458968708243e-05, + "loss": 3.1907, + "step": 4511 + }, + { + "epoch": 6.61, + "grad_norm": 1.3987131118774414, + "learning_rate": 4.9405434021641705e-05, + "loss": 3.2361, + "step": 4512 + }, + { + "epoch": 6.61, + "grad_norm": 1.295686960220337, + "learning_rate": 4.9364982302924545e-05, + "loss": 3.2055, + "step": 4513 + }, + { + "epoch": 6.61, + "grad_norm": 1.0267897844314575, + "learning_rate": 4.9324541723571936e-05, + "loss": 3.2089, + "step": 4514 + }, + { + "epoch": 6.61, + "grad_norm": 1.2543052434921265, + "learning_rate": 4.9284112292480535e-05, + "loss": 3.2271, + "step": 4515 + }, + { + "epoch": 6.61, + "grad_norm": 1.0035556554794312, + "learning_rate": 4.924369401854455e-05, + "loss": 3.2686, + "step": 4516 + }, + { + "epoch": 6.61, + "grad_norm": 1.4459877014160156, + "learning_rate": 4.920328691065573e-05, + "loss": 3.2149, + "step": 4517 + }, + { + "epoch": 6.62, + "grad_norm": 1.3623418807983398, + "learning_rate": 4.916289097770337e-05, + "loss": 3.2456, + "step": 4518 + }, + { + "epoch": 6.62, + "grad_norm": 1.6200556755065918, + "learning_rate": 4.9122506228574286e-05, + "loss": 3.1883, + "step": 4519 + }, + { + "epoch": 6.62, + "grad_norm": 1.0919500589370728, + "learning_rate": 4.908213267215287e-05, + "loss": 3.1899, + "step": 4520 + }, + { + "epoch": 6.62, + "grad_norm": 1.162226676940918, + "learning_rate": 4.904177031732103e-05, + "loss": 3.2402, + "step": 4521 + }, + { + "epoch": 6.62, + "grad_norm": 1.1378471851348877, + "learning_rate": 4.900141917295822e-05, + "loss": 3.3377, + "step": 4522 + }, + { + "epoch": 6.62, + "grad_norm": 1.183754563331604, + "learning_rate": 4.896107924794141e-05, + "loss": 3.2368, + "step": 4523 + }, + { + "epoch": 6.63, + "grad_norm": 1.1688711643218994, + "learning_rate": 4.892075055114511e-05, + "loss": 3.2358, + "step": 4524 + }, + { + "epoch": 6.63, + "grad_norm": 0.9402367472648621, + "learning_rate": 4.888043309144138e-05, + "loss": 3.1947, + "step": 4525 + }, + { + "epoch": 6.63, + "grad_norm": 0.9889912009239197, + "learning_rate": 4.8840126877699787e-05, + "loss": 3.2409, + "step": 4526 + }, + { + "epoch": 6.63, + "grad_norm": 1.3844680786132812, + "learning_rate": 4.879983191878742e-05, + "loss": 3.2658, + "step": 4527 + }, + { + "epoch": 6.63, + "grad_norm": 0.8452149033546448, + "learning_rate": 4.875954822356893e-05, + "loss": 3.1865, + "step": 4528 + }, + { + "epoch": 6.63, + "grad_norm": 0.9102594256401062, + "learning_rate": 4.871927580090636e-05, + "loss": 3.2226, + "step": 4529 + }, + { + "epoch": 6.63, + "grad_norm": 0.7820479869842529, + "learning_rate": 4.867901465965949e-05, + "loss": 3.2125, + "step": 4530 + }, + { + "epoch": 6.64, + "grad_norm": 1.0344125032424927, + "learning_rate": 4.863876480868549e-05, + "loss": 3.1863, + "step": 4531 + }, + { + "epoch": 6.64, + "grad_norm": 1.1749505996704102, + "learning_rate": 4.8598526256838947e-05, + "loss": 3.2641, + "step": 4532 + }, + { + "epoch": 6.64, + "grad_norm": 1.117828130722046, + "learning_rate": 4.855829901297223e-05, + "loss": 3.1992, + "step": 4533 + }, + { + "epoch": 6.64, + "grad_norm": 1.255987286567688, + "learning_rate": 4.8518083085934954e-05, + "loss": 3.1951, + "step": 4534 + }, + { + "epoch": 6.64, + "grad_norm": 0.6159536838531494, + "learning_rate": 4.847787848457433e-05, + "loss": 3.2087, + "step": 4535 + }, + { + "epoch": 6.64, + "grad_norm": 1.0251574516296387, + "learning_rate": 4.843768521773523e-05, + "loss": 3.202, + "step": 4536 + }, + { + "epoch": 6.64, + "grad_norm": 1.4189834594726562, + "learning_rate": 4.8397503294259795e-05, + "loss": 3.2329, + "step": 4537 + }, + { + "epoch": 6.65, + "grad_norm": 1.0889549255371094, + "learning_rate": 4.835733272298777e-05, + "loss": 3.2087, + "step": 4538 + }, + { + "epoch": 6.65, + "grad_norm": 1.1167737245559692, + "learning_rate": 4.831717351275653e-05, + "loss": 3.2098, + "step": 4539 + }, + { + "epoch": 6.65, + "grad_norm": 1.141223430633545, + "learning_rate": 4.827702567240072e-05, + "loss": 3.2151, + "step": 4540 + }, + { + "epoch": 6.65, + "grad_norm": 1.1248409748077393, + "learning_rate": 4.823688921075258e-05, + "loss": 3.2147, + "step": 4541 + }, + { + "epoch": 6.65, + "grad_norm": 1.471689224243164, + "learning_rate": 4.8196764136642e-05, + "loss": 3.1923, + "step": 4542 + }, + { + "epoch": 6.65, + "grad_norm": 1.6416572332382202, + "learning_rate": 4.81566504588961e-05, + "loss": 3.2319, + "step": 4543 + }, + { + "epoch": 6.65, + "grad_norm": 1.1048762798309326, + "learning_rate": 4.811654818633966e-05, + "loss": 3.2002, + "step": 4544 + }, + { + "epoch": 6.66, + "grad_norm": 1.3798350095748901, + "learning_rate": 4.807645732779491e-05, + "loss": 3.1903, + "step": 4545 + }, + { + "epoch": 6.66, + "grad_norm": 1.5549540519714355, + "learning_rate": 4.8036377892081565e-05, + "loss": 3.181, + "step": 4546 + }, + { + "epoch": 6.66, + "grad_norm": 0.42195671796798706, + "learning_rate": 4.7996309888016844e-05, + "loss": 3.1872, + "step": 4547 + }, + { + "epoch": 6.66, + "grad_norm": 1.59599769115448, + "learning_rate": 4.795625332441545e-05, + "loss": 3.2153, + "step": 4548 + }, + { + "epoch": 6.66, + "grad_norm": 1.7859076261520386, + "learning_rate": 4.7916208210089516e-05, + "loss": 3.246, + "step": 4549 + }, + { + "epoch": 6.66, + "grad_norm": 1.5909998416900635, + "learning_rate": 4.787617455384874e-05, + "loss": 3.1916, + "step": 4550 + }, + { + "epoch": 6.67, + "grad_norm": 2.1160545349121094, + "learning_rate": 4.783615236450024e-05, + "loss": 3.1881, + "step": 4551 + }, + { + "epoch": 6.67, + "grad_norm": 0.9319918751716614, + "learning_rate": 4.779614165084864e-05, + "loss": 3.2286, + "step": 4552 + }, + { + "epoch": 6.67, + "grad_norm": 2.0976734161376953, + "learning_rate": 4.775614242169606e-05, + "loss": 3.2103, + "step": 4553 + }, + { + "epoch": 6.67, + "grad_norm": 1.695967197418213, + "learning_rate": 4.7716154685841944e-05, + "loss": 3.1638, + "step": 4554 + }, + { + "epoch": 6.67, + "grad_norm": 2.1299004554748535, + "learning_rate": 4.767617845208344e-05, + "loss": 3.21, + "step": 4555 + }, + { + "epoch": 6.67, + "grad_norm": 1.7198915481567383, + "learning_rate": 4.763621372921506e-05, + "loss": 3.2789, + "step": 4556 + }, + { + "epoch": 6.67, + "grad_norm": 0.6392155885696411, + "learning_rate": 4.759626052602866e-05, + "loss": 3.2074, + "step": 4557 + }, + { + "epoch": 6.68, + "grad_norm": 1.7603440284729004, + "learning_rate": 4.7556318851313784e-05, + "loss": 3.1682, + "step": 4558 + }, + { + "epoch": 6.68, + "grad_norm": 0.6075698733329773, + "learning_rate": 4.7516388713857344e-05, + "loss": 3.2082, + "step": 4559 + }, + { + "epoch": 6.68, + "grad_norm": 1.1561111211776733, + "learning_rate": 4.7476470122443576e-05, + "loss": 3.2178, + "step": 4560 + }, + { + "epoch": 6.68, + "grad_norm": 0.6579719185829163, + "learning_rate": 4.743656308585441e-05, + "loss": 3.2283, + "step": 4561 + }, + { + "epoch": 6.68, + "grad_norm": 1.3691036701202393, + "learning_rate": 4.739666761286915e-05, + "loss": 3.2209, + "step": 4562 + }, + { + "epoch": 6.68, + "grad_norm": 0.4905560314655304, + "learning_rate": 4.735678371226441e-05, + "loss": 3.1764, + "step": 4563 + }, + { + "epoch": 6.68, + "eval_loss": 3.0459163188934326, + "eval_runtime": 1166.1122, + "eval_samples_per_second": 14.7, + "eval_steps_per_second": 0.919, + "step": 4563 + }, + { + "epoch": 6.68, + "grad_norm": 0.34124380350112915, + "learning_rate": 4.731691139281449e-05, + "loss": 3.2197, + "step": 4564 + }, + { + "epoch": 6.69, + "grad_norm": 0.7039169073104858, + "learning_rate": 4.7277050663291035e-05, + "loss": 3.2013, + "step": 4565 + }, + { + "epoch": 6.69, + "grad_norm": 1.3341939449310303, + "learning_rate": 4.723720153246303e-05, + "loss": 3.2084, + "step": 4566 + }, + { + "epoch": 6.69, + "grad_norm": 0.4564155638217926, + "learning_rate": 4.719736400909716e-05, + "loss": 3.1865, + "step": 4567 + }, + { + "epoch": 6.69, + "grad_norm": 0.9328960180282593, + "learning_rate": 4.715753810195729e-05, + "loss": 3.2035, + "step": 4568 + }, + { + "epoch": 6.69, + "grad_norm": 0.6796619892120361, + "learning_rate": 4.7117723819804884e-05, + "loss": 3.2024, + "step": 4569 + }, + { + "epoch": 6.69, + "grad_norm": 0.4047408699989319, + "learning_rate": 4.7077921171398895e-05, + "loss": 3.1935, + "step": 4570 + }, + { + "epoch": 6.7, + "grad_norm": 0.6333791017532349, + "learning_rate": 4.7038130165495555e-05, + "loss": 3.1794, + "step": 4571 + }, + { + "epoch": 6.7, + "grad_norm": 0.6958187818527222, + "learning_rate": 4.6998350810848605e-05, + "loss": 3.2401, + "step": 4572 + }, + { + "epoch": 6.7, + "grad_norm": 1.0862935781478882, + "learning_rate": 4.6958583116209344e-05, + "loss": 3.2027, + "step": 4573 + }, + { + "epoch": 6.7, + "grad_norm": 0.5394841432571411, + "learning_rate": 4.691882709032629e-05, + "loss": 3.2092, + "step": 4574 + }, + { + "epoch": 6.7, + "grad_norm": 1.7290090322494507, + "learning_rate": 4.687908274194555e-05, + "loss": 3.2199, + "step": 4575 + }, + { + "epoch": 6.7, + "grad_norm": 1.2097941637039185, + "learning_rate": 4.683935007981061e-05, + "loss": 3.2319, + "step": 4576 + }, + { + "epoch": 6.7, + "grad_norm": 1.8092429637908936, + "learning_rate": 4.6799629112662366e-05, + "loss": 3.1968, + "step": 4577 + }, + { + "epoch": 6.71, + "grad_norm": 1.7616616487503052, + "learning_rate": 4.675991984923921e-05, + "loss": 3.2206, + "step": 4578 + }, + { + "epoch": 6.71, + "grad_norm": 0.6763386726379395, + "learning_rate": 4.672022229827687e-05, + "loss": 3.1857, + "step": 4579 + }, + { + "epoch": 6.71, + "grad_norm": 1.4232481718063354, + "learning_rate": 4.668053646850856e-05, + "loss": 3.184, + "step": 4580 + }, + { + "epoch": 6.71, + "grad_norm": 1.7722804546356201, + "learning_rate": 4.664086236866491e-05, + "loss": 3.225, + "step": 4581 + }, + { + "epoch": 6.71, + "grad_norm": 1.0724995136260986, + "learning_rate": 4.660120000747393e-05, + "loss": 3.1924, + "step": 4582 + }, + { + "epoch": 6.71, + "grad_norm": 0.3242776393890381, + "learning_rate": 4.656154939366109e-05, + "loss": 3.2068, + "step": 4583 + }, + { + "epoch": 6.71, + "grad_norm": 1.4377954006195068, + "learning_rate": 4.652191053594924e-05, + "loss": 3.193, + "step": 4584 + }, + { + "epoch": 6.72, + "grad_norm": 4.462525844573975, + "learning_rate": 4.64822834430587e-05, + "loss": 3.275, + "step": 4585 + }, + { + "epoch": 6.72, + "grad_norm": 1.4812723398208618, + "learning_rate": 4.644266812370714e-05, + "loss": 3.2099, + "step": 4586 + }, + { + "epoch": 6.72, + "grad_norm": 0.6171543002128601, + "learning_rate": 4.640306458660966e-05, + "loss": 3.178, + "step": 4587 + }, + { + "epoch": 6.72, + "grad_norm": 1.2330536842346191, + "learning_rate": 4.636347284047877e-05, + "loss": 3.2013, + "step": 4588 + }, + { + "epoch": 6.72, + "grad_norm": 0.5005393624305725, + "learning_rate": 4.6323892894024414e-05, + "loss": 3.2114, + "step": 4589 + }, + { + "epoch": 6.72, + "grad_norm": 1.613756775856018, + "learning_rate": 4.628432475595393e-05, + "loss": 3.1852, + "step": 4590 + }, + { + "epoch": 6.72, + "grad_norm": 0.8560778498649597, + "learning_rate": 4.6244768434971944e-05, + "loss": 3.2414, + "step": 4591 + }, + { + "epoch": 6.73, + "grad_norm": 1.844510555267334, + "learning_rate": 4.620522393978068e-05, + "loss": 3.1791, + "step": 4592 + }, + { + "epoch": 6.73, + "grad_norm": 1.6803126335144043, + "learning_rate": 4.616569127907968e-05, + "loss": 3.2298, + "step": 4593 + }, + { + "epoch": 6.73, + "grad_norm": 1.2029790878295898, + "learning_rate": 4.6126170461565746e-05, + "loss": 3.2044, + "step": 4594 + }, + { + "epoch": 6.73, + "grad_norm": 0.8809365630149841, + "learning_rate": 4.608666149593328e-05, + "loss": 3.1935, + "step": 4595 + }, + { + "epoch": 6.73, + "grad_norm": 0.6382874846458435, + "learning_rate": 4.604716439087403e-05, + "loss": 3.1965, + "step": 4596 + }, + { + "epoch": 6.73, + "grad_norm": 1.0096112489700317, + "learning_rate": 4.600767915507695e-05, + "loss": 3.1924, + "step": 4597 + }, + { + "epoch": 6.74, + "grad_norm": 0.8474308252334595, + "learning_rate": 4.596820579722867e-05, + "loss": 3.1994, + "step": 4598 + }, + { + "epoch": 6.74, + "grad_norm": 1.0024635791778564, + "learning_rate": 4.592874432601304e-05, + "loss": 3.1834, + "step": 4599 + }, + { + "epoch": 6.74, + "grad_norm": 0.9615989923477173, + "learning_rate": 4.588929475011125e-05, + "loss": 3.1855, + "step": 4600 + }, + { + "epoch": 6.74, + "grad_norm": 0.4216245114803314, + "learning_rate": 4.584985707820198e-05, + "loss": 3.2381, + "step": 4601 + }, + { + "epoch": 6.74, + "grad_norm": 0.5493209362030029, + "learning_rate": 4.581043131896126e-05, + "loss": 3.1567, + "step": 4602 + }, + { + "epoch": 6.74, + "grad_norm": 0.6222708225250244, + "learning_rate": 4.5771017481062475e-05, + "loss": 3.2149, + "step": 4603 + }, + { + "epoch": 6.74, + "grad_norm": 0.5318418741226196, + "learning_rate": 4.573161557317642e-05, + "loss": 3.2158, + "step": 4604 + }, + { + "epoch": 6.75, + "grad_norm": 0.7692480087280273, + "learning_rate": 4.5692225603971264e-05, + "loss": 3.2096, + "step": 4605 + }, + { + "epoch": 6.75, + "grad_norm": 0.5248206257820129, + "learning_rate": 4.56528475821125e-05, + "loss": 3.1933, + "step": 4606 + }, + { + "epoch": 6.75, + "grad_norm": 1.0218887329101562, + "learning_rate": 4.561348151626306e-05, + "loss": 3.1819, + "step": 4607 + }, + { + "epoch": 6.75, + "grad_norm": 2.8102357387542725, + "learning_rate": 4.55741274150832e-05, + "loss": 3.2286, + "step": 4608 + }, + { + "epoch": 6.75, + "grad_norm": 0.5493364334106445, + "learning_rate": 4.5534785287230575e-05, + "loss": 3.1981, + "step": 4609 + }, + { + "epoch": 6.75, + "grad_norm": 0.438005656003952, + "learning_rate": 4.5495455141360165e-05, + "loss": 3.1831, + "step": 4610 + }, + { + "epoch": 6.75, + "grad_norm": 1.3687691688537598, + "learning_rate": 4.5456136986124355e-05, + "loss": 3.1993, + "step": 4611 + }, + { + "epoch": 6.76, + "grad_norm": 0.7376412749290466, + "learning_rate": 4.5416830830172876e-05, + "loss": 3.2093, + "step": 4612 + }, + { + "epoch": 6.76, + "grad_norm": 0.710468590259552, + "learning_rate": 4.53775366821528e-05, + "loss": 3.1834, + "step": 4613 + }, + { + "epoch": 6.76, + "grad_norm": 0.6461208462715149, + "learning_rate": 4.53382545507086e-05, + "loss": 3.2279, + "step": 4614 + }, + { + "epoch": 6.76, + "grad_norm": 0.4084911644458771, + "learning_rate": 4.5298984444482075e-05, + "loss": 3.1921, + "step": 4615 + }, + { + "epoch": 6.76, + "grad_norm": 0.5108605027198792, + "learning_rate": 4.5259726372112376e-05, + "loss": 3.2015, + "step": 4616 + }, + { + "epoch": 6.76, + "grad_norm": 0.36462754011154175, + "learning_rate": 4.522048034223603e-05, + "loss": 3.2099, + "step": 4617 + }, + { + "epoch": 6.76, + "grad_norm": 0.5964351892471313, + "learning_rate": 4.518124636348688e-05, + "loss": 3.223, + "step": 4618 + }, + { + "epoch": 6.77, + "grad_norm": 0.532785177230835, + "learning_rate": 4.514202444449615e-05, + "loss": 3.1739, + "step": 4619 + }, + { + "epoch": 6.77, + "grad_norm": 0.7253513336181641, + "learning_rate": 4.510281459389242e-05, + "loss": 3.202, + "step": 4620 + }, + { + "epoch": 6.77, + "grad_norm": 1.3560603857040405, + "learning_rate": 4.506361682030156e-05, + "loss": 3.2175, + "step": 4621 + }, + { + "epoch": 6.77, + "grad_norm": 0.6363286375999451, + "learning_rate": 4.502443113234688e-05, + "loss": 3.2485, + "step": 4622 + }, + { + "epoch": 6.77, + "grad_norm": 1.2093783617019653, + "learning_rate": 4.498525753864884e-05, + "loss": 3.2012, + "step": 4623 + }, + { + "epoch": 6.77, + "grad_norm": 1.1001776456832886, + "learning_rate": 4.4946096047825515e-05, + "loss": 3.1938, + "step": 4624 + }, + { + "epoch": 6.78, + "grad_norm": 0.4927830398082733, + "learning_rate": 4.490694666849208e-05, + "loss": 3.234, + "step": 4625 + }, + { + "epoch": 6.78, + "grad_norm": 1.6655173301696777, + "learning_rate": 4.486780940926112e-05, + "loss": 3.1661, + "step": 4626 + }, + { + "epoch": 6.78, + "grad_norm": 1.3249220848083496, + "learning_rate": 4.482868427874268e-05, + "loss": 3.2283, + "step": 4627 + }, + { + "epoch": 6.78, + "grad_norm": 2.2860875129699707, + "learning_rate": 4.478957128554392e-05, + "loss": 3.2115, + "step": 4628 + }, + { + "epoch": 6.78, + "grad_norm": 2.1470510959625244, + "learning_rate": 4.4750470438269446e-05, + "loss": 3.2151, + "step": 4629 + }, + { + "epoch": 6.78, + "grad_norm": 1.3575111627578735, + "learning_rate": 4.4711381745521275e-05, + "loss": 3.1938, + "step": 4630 + }, + { + "epoch": 6.78, + "grad_norm": 1.7453752756118774, + "learning_rate": 4.467230521589856e-05, + "loss": 3.1969, + "step": 4631 + }, + { + "epoch": 6.79, + "grad_norm": 1.7523752450942993, + "learning_rate": 4.463324085799786e-05, + "loss": 3.2266, + "step": 4632 + }, + { + "epoch": 6.79, + "grad_norm": 1.433594822883606, + "learning_rate": 4.4594188680413206e-05, + "loss": 3.1962, + "step": 4633 + }, + { + "epoch": 6.79, + "grad_norm": 1.0330843925476074, + "learning_rate": 4.4555148691735685e-05, + "loss": 3.2325, + "step": 4634 + }, + { + "epoch": 6.79, + "grad_norm": 0.9396421313285828, + "learning_rate": 4.451612090055388e-05, + "loss": 3.2274, + "step": 4635 + }, + { + "epoch": 6.79, + "grad_norm": 1.1527962684631348, + "learning_rate": 4.447710531545364e-05, + "loss": 3.2075, + "step": 4636 + }, + { + "epoch": 6.79, + "grad_norm": 0.8900932669639587, + "learning_rate": 4.443810194501813e-05, + "loss": 3.1952, + "step": 4637 + }, + { + "epoch": 6.79, + "grad_norm": 1.3143221139907837, + "learning_rate": 4.4399110797827835e-05, + "loss": 3.2086, + "step": 4638 + }, + { + "epoch": 6.8, + "grad_norm": 1.1299476623535156, + "learning_rate": 4.4360131882460555e-05, + "loss": 3.2, + "step": 4639 + }, + { + "epoch": 6.8, + "grad_norm": 0.8985263109207153, + "learning_rate": 4.432116520749137e-05, + "loss": 3.15, + "step": 4640 + }, + { + "epoch": 6.8, + "grad_norm": 1.0726135969161987, + "learning_rate": 4.4282210781492684e-05, + "loss": 3.2016, + "step": 4641 + }, + { + "epoch": 6.8, + "grad_norm": 0.556115984916687, + "learning_rate": 4.424326861303424e-05, + "loss": 3.2035, + "step": 4642 + }, + { + "epoch": 6.8, + "grad_norm": 1.1372523307800293, + "learning_rate": 4.420433871068302e-05, + "loss": 3.1998, + "step": 4643 + }, + { + "epoch": 6.8, + "grad_norm": 0.9902729988098145, + "learning_rate": 4.416542108300338e-05, + "loss": 3.1661, + "step": 4644 + }, + { + "epoch": 6.8, + "grad_norm": 0.6972740292549133, + "learning_rate": 4.412651573855689e-05, + "loss": 3.191, + "step": 4645 + }, + { + "epoch": 6.81, + "grad_norm": 0.7196264863014221, + "learning_rate": 4.408762268590251e-05, + "loss": 3.1772, + "step": 4646 + }, + { + "epoch": 6.81, + "grad_norm": 0.6130596995353699, + "learning_rate": 4.404874193359646e-05, + "loss": 3.1859, + "step": 4647 + }, + { + "epoch": 6.81, + "grad_norm": 0.8541386723518372, + "learning_rate": 4.4009873490192135e-05, + "loss": 3.1891, + "step": 4648 + }, + { + "epoch": 6.81, + "grad_norm": 0.4068596065044403, + "learning_rate": 4.397101736424046e-05, + "loss": 3.2386, + "step": 4649 + }, + { + "epoch": 6.81, + "grad_norm": 0.35903048515319824, + "learning_rate": 4.3932173564289506e-05, + "loss": 3.2046, + "step": 4650 + }, + { + "epoch": 6.81, + "grad_norm": 1.0451300144195557, + "learning_rate": 4.389334209888455e-05, + "loss": 3.2677, + "step": 4651 + }, + { + "epoch": 6.82, + "grad_norm": 0.3824341595172882, + "learning_rate": 4.385452297656836e-05, + "loss": 3.1955, + "step": 4652 + }, + { + "epoch": 6.82, + "grad_norm": 0.9694157838821411, + "learning_rate": 4.381571620588088e-05, + "loss": 3.2113, + "step": 4653 + }, + { + "epoch": 6.82, + "grad_norm": 0.6231606602668762, + "learning_rate": 4.3776921795359216e-05, + "loss": 3.1968, + "step": 4654 + }, + { + "epoch": 6.82, + "grad_norm": 0.39795351028442383, + "learning_rate": 4.373813975353802e-05, + "loss": 3.1859, + "step": 4655 + }, + { + "epoch": 6.82, + "grad_norm": 0.3494223654270172, + "learning_rate": 4.3699370088949066e-05, + "loss": 3.2194, + "step": 4656 + }, + { + "epoch": 6.82, + "grad_norm": 0.7372296452522278, + "learning_rate": 4.366061281012129e-05, + "loss": 3.2062, + "step": 4657 + }, + { + "epoch": 6.82, + "grad_norm": 1.144978642463684, + "learning_rate": 4.362186792558122e-05, + "loss": 3.1948, + "step": 4658 + }, + { + "epoch": 6.83, + "grad_norm": 0.5863081812858582, + "learning_rate": 4.358313544385233e-05, + "loss": 3.1734, + "step": 4659 + }, + { + "epoch": 6.83, + "grad_norm": 1.334794282913208, + "learning_rate": 4.354441537345553e-05, + "loss": 3.187, + "step": 4660 + }, + { + "epoch": 6.83, + "grad_norm": 0.4816846251487732, + "learning_rate": 4.350570772290906e-05, + "loss": 3.1916, + "step": 4661 + }, + { + "epoch": 6.83, + "grad_norm": 0.8258687853813171, + "learning_rate": 4.3467012500728245e-05, + "loss": 3.2171, + "step": 4662 + }, + { + "epoch": 6.83, + "grad_norm": 0.9091712236404419, + "learning_rate": 4.342832971542578e-05, + "loss": 3.1695, + "step": 4663 + }, + { + "epoch": 6.83, + "grad_norm": 1.3142671585083008, + "learning_rate": 4.3389659375511725e-05, + "loss": 3.1681, + "step": 4664 + }, + { + "epoch": 6.83, + "grad_norm": 2.6121108531951904, + "learning_rate": 4.335100148949318e-05, + "loss": 3.2161, + "step": 4665 + }, + { + "epoch": 6.84, + "grad_norm": 0.6220512986183167, + "learning_rate": 4.3312356065874635e-05, + "loss": 3.1868, + "step": 4666 + }, + { + "epoch": 6.84, + "grad_norm": 1.4593570232391357, + "learning_rate": 4.327372311315792e-05, + "loss": 3.1968, + "step": 4667 + }, + { + "epoch": 6.84, + "grad_norm": 1.4904062747955322, + "learning_rate": 4.323510263984192e-05, + "loss": 3.2013, + "step": 4668 + }, + { + "epoch": 6.84, + "grad_norm": 0.8087517023086548, + "learning_rate": 4.3196494654422916e-05, + "loss": 3.2106, + "step": 4669 + }, + { + "epoch": 6.84, + "grad_norm": 0.5217124223709106, + "learning_rate": 4.315789916539441e-05, + "loss": 3.2188, + "step": 4670 + }, + { + "epoch": 6.84, + "grad_norm": 0.7983100414276123, + "learning_rate": 4.311931618124715e-05, + "loss": 3.217, + "step": 4671 + }, + { + "epoch": 6.84, + "grad_norm": 0.49825963377952576, + "learning_rate": 4.308074571046913e-05, + "loss": 3.2089, + "step": 4672 + }, + { + "epoch": 6.85, + "grad_norm": 0.9408275485038757, + "learning_rate": 4.30421877615456e-05, + "loss": 3.1924, + "step": 4673 + }, + { + "epoch": 6.85, + "grad_norm": 0.6962942481040955, + "learning_rate": 4.300364234295905e-05, + "loss": 3.2022, + "step": 4674 + }, + { + "epoch": 6.85, + "grad_norm": 0.5625227093696594, + "learning_rate": 4.296510946318921e-05, + "loss": 3.2056, + "step": 4675 + }, + { + "epoch": 6.85, + "grad_norm": 0.6208386421203613, + "learning_rate": 4.292658913071306e-05, + "loss": 3.1973, + "step": 4676 + }, + { + "epoch": 6.85, + "grad_norm": 0.5512813925743103, + "learning_rate": 4.2888081354004805e-05, + "loss": 3.2122, + "step": 4677 + }, + { + "epoch": 6.85, + "grad_norm": 1.9432053565979004, + "learning_rate": 4.2849586141535945e-05, + "loss": 3.2239, + "step": 4678 + }, + { + "epoch": 6.86, + "grad_norm": 4.5366291999816895, + "learning_rate": 4.281110350177506e-05, + "loss": 3.1618, + "step": 4679 + }, + { + "epoch": 6.86, + "grad_norm": 0.9245926737785339, + "learning_rate": 4.277263344318817e-05, + "loss": 3.2309, + "step": 4680 + }, + { + "epoch": 6.86, + "grad_norm": 0.3581990897655487, + "learning_rate": 4.2734175974238444e-05, + "loss": 3.1872, + "step": 4681 + }, + { + "epoch": 6.86, + "grad_norm": 0.9905257821083069, + "learning_rate": 4.269573110338615e-05, + "loss": 3.2077, + "step": 4682 + }, + { + "epoch": 6.86, + "grad_norm": 1.2098913192749023, + "learning_rate": 4.265729883908902e-05, + "loss": 3.2249, + "step": 4683 + }, + { + "epoch": 6.86, + "grad_norm": 0.711704671382904, + "learning_rate": 4.2618879189801885e-05, + "loss": 3.1625, + "step": 4684 + }, + { + "epoch": 6.86, + "grad_norm": 2.870743989944458, + "learning_rate": 4.25804721639767e-05, + "loss": 3.1953, + "step": 4685 + }, + { + "epoch": 6.87, + "grad_norm": 1.7720706462860107, + "learning_rate": 4.254207777006287e-05, + "loss": 3.1671, + "step": 4686 + }, + { + "epoch": 6.87, + "grad_norm": 0.8603296875953674, + "learning_rate": 4.25036960165069e-05, + "loss": 3.166, + "step": 4687 + }, + { + "epoch": 6.87, + "grad_norm": 2.4739768505096436, + "learning_rate": 4.246532691175241e-05, + "loss": 3.1946, + "step": 4688 + }, + { + "epoch": 6.87, + "grad_norm": 2.0245025157928467, + "learning_rate": 4.242697046424046e-05, + "loss": 3.2304, + "step": 4689 + }, + { + "epoch": 6.87, + "grad_norm": 0.7027145028114319, + "learning_rate": 4.2388626682409194e-05, + "loss": 3.2526, + "step": 4690 + }, + { + "epoch": 6.87, + "grad_norm": 2.3779094219207764, + "learning_rate": 4.23502955746939e-05, + "loss": 3.2621, + "step": 4691 + }, + { + "epoch": 6.87, + "grad_norm": 1.9404739141464233, + "learning_rate": 4.23119771495273e-05, + "loss": 3.1891, + "step": 4692 + }, + { + "epoch": 6.88, + "grad_norm": 0.6578683257102966, + "learning_rate": 4.227367141533909e-05, + "loss": 3.2023, + "step": 4693 + }, + { + "epoch": 6.88, + "grad_norm": 5.048638820648193, + "learning_rate": 4.2235378380556255e-05, + "loss": 3.136, + "step": 4694 + }, + { + "epoch": 6.88, + "grad_norm": 2.4565985202789307, + "learning_rate": 4.2197098053603136e-05, + "loss": 3.1868, + "step": 4695 + }, + { + "epoch": 6.88, + "grad_norm": 0.770892858505249, + "learning_rate": 4.215883044290103e-05, + "loss": 3.1455, + "step": 4696 + }, + { + "epoch": 6.88, + "grad_norm": 2.0807790756225586, + "learning_rate": 4.2120575556868594e-05, + "loss": 3.1894, + "step": 4697 + }, + { + "epoch": 6.88, + "grad_norm": 2.7063095569610596, + "learning_rate": 4.208233340392164e-05, + "loss": 3.1982, + "step": 4698 + }, + { + "epoch": 6.89, + "grad_norm": 1.1912204027175903, + "learning_rate": 4.2044103992473195e-05, + "loss": 3.1818, + "step": 4699 + }, + { + "epoch": 6.89, + "grad_norm": 2.9936110973358154, + "learning_rate": 4.2005887330933457e-05, + "loss": 3.2109, + "step": 4700 + }, + { + "epoch": 6.89, + "grad_norm": 3.9401204586029053, + "learning_rate": 4.196768342770986e-05, + "loss": 3.2246, + "step": 4701 + }, + { + "epoch": 6.89, + "grad_norm": 2.341156482696533, + "learning_rate": 4.1929492291206983e-05, + "loss": 3.1819, + "step": 4702 + }, + { + "epoch": 6.89, + "grad_norm": 0.7375439405441284, + "learning_rate": 4.1891313929826635e-05, + "loss": 3.2047, + "step": 4703 + }, + { + "epoch": 6.89, + "grad_norm": 1.9721698760986328, + "learning_rate": 4.185314835196781e-05, + "loss": 3.2131, + "step": 4704 + }, + { + "epoch": 6.89, + "grad_norm": 0.8218457102775574, + "learning_rate": 4.181499556602665e-05, + "loss": 3.1891, + "step": 4705 + }, + { + "epoch": 6.9, + "grad_norm": 1.2695006132125854, + "learning_rate": 4.177685558039652e-05, + "loss": 3.1774, + "step": 4706 + }, + { + "epoch": 6.9, + "grad_norm": 1.4132906198501587, + "learning_rate": 4.1738728403467974e-05, + "loss": 3.1856, + "step": 4707 + }, + { + "epoch": 6.9, + "grad_norm": 0.5927780866622925, + "learning_rate": 4.170061404362874e-05, + "loss": 3.2209, + "step": 4708 + }, + { + "epoch": 6.9, + "grad_norm": 1.482879877090454, + "learning_rate": 4.166251250926371e-05, + "loss": 3.1621, + "step": 4709 + }, + { + "epoch": 6.9, + "grad_norm": 0.4204716384410858, + "learning_rate": 4.1624423808754956e-05, + "loss": 3.171, + "step": 4710 + }, + { + "epoch": 6.9, + "grad_norm": 1.5936245918273926, + "learning_rate": 4.158634795048175e-05, + "loss": 3.1746, + "step": 4711 + } + ], + "logging_steps": 1, + "max_steps": 6730, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "save_steps": 673, + "total_flos": 1.1635188876715439e+20, + "train_batch_size": 2, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-4711/training_args.bin b/checkpoint-4711/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..c9318904b092ceb1505e2e8f2b5300dd991baa7f --- /dev/null +++ b/checkpoint-4711/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a4ab9da40c5b7630b88beccc23069550f831ccb9b69b82772516407dd51150c +size 5688 diff --git a/checkpoint-5384/README.md b/checkpoint-5384/README.md new file mode 100644 index 0000000000000000000000000000000000000000..dbeb1c7761226042e1d72de6378a4fa69fdb3f06 --- /dev/null +++ b/checkpoint-5384/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: unsloth/gemma-7b +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.9.0 \ No newline at end of file diff --git a/checkpoint-5384/adapter_config.json b/checkpoint-5384/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..35971161ab187cbdc8af6c44079edb2d2b677921 --- /dev/null +++ b/checkpoint-5384/adapter_config.json @@ -0,0 +1,33 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "unsloth/gemma-7b", + "bias": "none", + "fan_in_fan_out": null, + "inference_mode": true, + "init_lora_weights": true, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 16, + "lora_dropout": 0.05, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 32, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "v_proj", + "o_proj", + "down_proj", + "k_proj", + "up_proj", + "gate_proj", + "q_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": false +} \ No newline at end of file diff --git a/checkpoint-5384/adapter_model.safetensors b/checkpoint-5384/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c93f00b964f9101973e4f7480215389c461fbf9a --- /dev/null +++ b/checkpoint-5384/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3204dffa78550c8071c068d56f5c63b19f2d40c50467540a4ab2680ff670fe7c +size 400084608 diff --git a/checkpoint-5384/optimizer.pt b/checkpoint-5384/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..b70050447005c1609f0ab1f877bea00c1fdccda3 --- /dev/null +++ b/checkpoint-5384/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b7445d2aa634dd1783e07bfa9931d42884b104d7175a20b125ffe79de4bc291 +size 200866804 diff --git a/checkpoint-5384/rng_state_0.pth b/checkpoint-5384/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..85d2b07df07e2d097787ca7ed8bbb41fc2b7da7d --- /dev/null +++ b/checkpoint-5384/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6fc66830faf9278b829a46fe18692e0bb8f7584e46dffa50f6bcf1bda19a475b +size 15984 diff --git a/checkpoint-5384/rng_state_1.pth b/checkpoint-5384/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..e4af162143f4cdf102a44201d56d4522f478b967 --- /dev/null +++ b/checkpoint-5384/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9ddb1cadf8786ca9ae4f2d96f49b64f0ddeaf59de11b97791a2daca46df3bcdf +size 15984 diff --git a/checkpoint-5384/rng_state_2.pth b/checkpoint-5384/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..0b419a2f8f0d61ac1942be7dbe665d59a54a489f --- /dev/null +++ b/checkpoint-5384/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1b5767617fb1de879fc7a1613ca8e14c44d5552c543e4d2d538b108e1a452147 +size 15984 diff --git a/checkpoint-5384/rng_state_3.pth b/checkpoint-5384/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..f86c41ae5983fe167468bb841885c34c8bcb349d --- /dev/null +++ b/checkpoint-5384/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a5a1e0272c3187fa4d091951995b88b67f067d10dd3904c2b867113d3715fbe6 +size 15984 diff --git a/checkpoint-5384/rng_state_4.pth b/checkpoint-5384/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..846e161d2eff19b6549efa1f17668fc749607aed --- /dev/null +++ b/checkpoint-5384/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85e2bee2b899b5b037e00f9a93d3d3582233d0b9434b07095e198cf060768aba +size 15984 diff --git a/checkpoint-5384/rng_state_5.pth b/checkpoint-5384/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..ce70bc6fbca566b9070078699514fe270098d1f7 --- /dev/null +++ b/checkpoint-5384/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:37b42c0b82e6631ed86e395babe85db1d915d21a72caa64e0f17c844ebd95150 +size 15984 diff --git a/checkpoint-5384/rng_state_6.pth b/checkpoint-5384/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..b886109260f2431b45accb71b33b93807fc41eb2 --- /dev/null +++ b/checkpoint-5384/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:993b621d317b8455663a4348f50330ca1e8c6e921993ab4271c9f65ee75ac59d +size 15984 diff --git a/checkpoint-5384/rng_state_7.pth b/checkpoint-5384/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..5906da87813b248605cced723101205be6e0ec6c --- /dev/null +++ b/checkpoint-5384/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bde2b22a539d54f1d88e5b29de127a4ef3107a0c58afb9a8738a30370141720e +size 15984 diff --git a/checkpoint-5384/scheduler.pt b/checkpoint-5384/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..b01975c96aa3824cb3133836e22ef82321f1e620 --- /dev/null +++ b/checkpoint-5384/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e19453bf2f1f46d065069caad1a3fed9474ed9d8714cf64a6cf02a199c97e5dd +size 1064 diff --git a/checkpoint-5384/special_tokens_map.json b/checkpoint-5384/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80 --- /dev/null +++ b/checkpoint-5384/special_tokens_map.json @@ -0,0 +1,34 @@ +{ + "additional_special_tokens": [ + "", + "" + ], + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-5384/tokenizer.json b/checkpoint-5384/tokenizer.json new file mode 100644 index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044 --- /dev/null +++ b/checkpoint-5384/tokenizer.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee +size 17477929 diff --git a/checkpoint-5384/tokenizer.model b/checkpoint-5384/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8 --- /dev/null +++ b/checkpoint-5384/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2 +size 4241003 diff --git a/checkpoint-5384/tokenizer_config.json b/checkpoint-5384/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..1451ca400a880c5056ab40cb960930cf827de2de --- /dev/null +++ b/checkpoint-5384/tokenizer_config.json @@ -0,0 +1,70 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "3": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "106": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "107": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "additional_special_tokens": [ + "", + "" + ], + "bos_token": "", + "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '' + role + '\n' + message['content'] | trim + '\n' }}{% endfor %}{% if add_generation_prompt %}{{'model\n'}}{% endif %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": null, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "tokenizer_class": "GemmaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-5384/trainer_state.json b/checkpoint-5384/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..b14c34ce00cd8c27e15fa3f26b5f6f58d9dd0665 --- /dev/null +++ b/checkpoint-5384/trainer_state.json @@ -0,0 +1,37965 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 7.886950064971227, + "eval_steps": 169, + "global_step": 5384, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.10426733642816544, + "learning_rate": 6.25e-06, + "loss": 39.5782, + "step": 1 + }, + { + "epoch": 0.0, + "eval_loss": 39.257938385009766, + "eval_runtime": 1143.223, + "eval_samples_per_second": 14.994, + "eval_steps_per_second": 0.938, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 0.10457848757505417, + "learning_rate": 1.25e-05, + "loss": 39.5213, + "step": 2 + }, + { + "epoch": 0.0, + "grad_norm": 0.10474710911512375, + "learning_rate": 1.8750000000000002e-05, + "loss": 39.4605, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.11379425972700119, + "learning_rate": 2.5e-05, + "loss": 39.4968, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.12108173966407776, + "learning_rate": 3.125e-05, + "loss": 39.4831, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.13807319104671478, + "learning_rate": 3.7500000000000003e-05, + "loss": 39.5751, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.1619432121515274, + "learning_rate": 4.375e-05, + "loss": 39.4458, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.1936173141002655, + "learning_rate": 5e-05, + "loss": 39.4696, + "step": 8 + }, + { + "epoch": 0.01, + "grad_norm": 0.23368136584758759, + "learning_rate": 5.6250000000000005e-05, + "loss": 39.5398, + "step": 9 + }, + { + "epoch": 0.01, + "grad_norm": 0.28147873282432556, + "learning_rate": 6.25e-05, + "loss": 39.5004, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.3371683359146118, + "learning_rate": 6.875e-05, + "loss": 39.4975, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.4020785391330719, + "learning_rate": 7.500000000000001e-05, + "loss": 39.5237, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.4757504463195801, + "learning_rate": 8.125000000000001e-05, + "loss": 39.5889, + "step": 13 + }, + { + "epoch": 0.02, + "grad_norm": 0.5553041100502014, + "learning_rate": 8.75e-05, + "loss": 39.5117, + "step": 14 + }, + { + "epoch": 0.02, + "grad_norm": 0.6490126848220825, + "learning_rate": 9.375e-05, + "loss": 39.5925, + "step": 15 + }, + { + "epoch": 0.02, + "grad_norm": 0.7478038668632507, + "learning_rate": 0.0001, + "loss": 39.5638, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.8497499227523804, + "learning_rate": 0.00010625000000000001, + "loss": 39.513, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.9426423907279968, + "learning_rate": 0.00011250000000000001, + "loss": 39.1585, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.9930742979049683, + "learning_rate": 0.00011875, + "loss": 38.4983, + "step": 19 + }, + { + "epoch": 0.03, + "grad_norm": 0.9656581878662109, + "learning_rate": 0.000125, + "loss": 37.7409, + "step": 20 + }, + { + "epoch": 0.03, + "grad_norm": 1.0122301578521729, + "learning_rate": 0.00013125000000000002, + "loss": 37.248, + "step": 21 + }, + { + "epoch": 0.03, + "grad_norm": 1.0719304084777832, + "learning_rate": 0.0001375, + "loss": 37.0875, + "step": 22 + }, + { + "epoch": 0.03, + "grad_norm": 1.1606364250183105, + "learning_rate": 0.00014375, + "loss": 36.8509, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 1.2129034996032715, + "learning_rate": 0.00015000000000000001, + "loss": 36.5097, + "step": 24 + }, + { + "epoch": 0.04, + "grad_norm": 1.3106859922409058, + "learning_rate": 0.00015625, + "loss": 36.101, + "step": 25 + }, + { + "epoch": 0.04, + "grad_norm": 1.395408272743225, + "learning_rate": 0.00016250000000000002, + "loss": 35.5219, + "step": 26 + }, + { + "epoch": 0.04, + "grad_norm": 1.5144782066345215, + "learning_rate": 0.00016875, + "loss": 35.1062, + "step": 27 + }, + { + "epoch": 0.04, + "grad_norm": 1.6357661485671997, + "learning_rate": 0.000175, + "loss": 34.562, + "step": 28 + }, + { + "epoch": 0.04, + "grad_norm": 1.7252875566482544, + "learning_rate": 0.00018125000000000001, + "loss": 33.7745, + "step": 29 + }, + { + "epoch": 0.04, + "grad_norm": 1.8169721364974976, + "learning_rate": 0.0001875, + "loss": 33.1849, + "step": 30 + }, + { + "epoch": 0.05, + "grad_norm": 1.8421262502670288, + "learning_rate": 0.00019375000000000002, + "loss": 32.4654, + "step": 31 + }, + { + "epoch": 0.05, + "grad_norm": 1.9798814058303833, + "learning_rate": 0.0002, + "loss": 31.5573, + "step": 32 + }, + { + "epoch": 0.05, + "grad_norm": 2.1344172954559326, + "learning_rate": 0.0001999999890003351, + "loss": 30.9011, + "step": 33 + }, + { + "epoch": 0.05, + "grad_norm": 2.1522722244262695, + "learning_rate": 0.00019999995600134275, + "loss": 30.0969, + "step": 34 + }, + { + "epoch": 0.05, + "grad_norm": 1.9861739873886108, + "learning_rate": 0.0001999999010030303, + "loss": 29.2185, + "step": 35 + }, + { + "epoch": 0.05, + "grad_norm": 1.967444896697998, + "learning_rate": 0.00019999982400540977, + "loss": 28.5804, + "step": 36 + }, + { + "epoch": 0.05, + "grad_norm": 1.9654927253723145, + "learning_rate": 0.00019999972500849814, + "loss": 27.8538, + "step": 37 + }, + { + "epoch": 0.06, + "grad_norm": 1.9123767614364624, + "learning_rate": 0.00019999960401231715, + "loss": 27.1902, + "step": 38 + }, + { + "epoch": 0.06, + "grad_norm": 1.7705096006393433, + "learning_rate": 0.00019999946101689346, + "loss": 26.4983, + "step": 39 + }, + { + "epoch": 0.06, + "grad_norm": 1.6448909044265747, + "learning_rate": 0.00019999929602225854, + "loss": 25.8354, + "step": 40 + }, + { + "epoch": 0.06, + "grad_norm": 1.5614784955978394, + "learning_rate": 0.00019999910902844862, + "loss": 25.2961, + "step": 41 + }, + { + "epoch": 0.06, + "grad_norm": 1.5540868043899536, + "learning_rate": 0.0001999989000355049, + "loss": 24.8097, + "step": 42 + }, + { + "epoch": 0.06, + "grad_norm": 1.4986692667007446, + "learning_rate": 0.00019999866904347332, + "loss": 24.3735, + "step": 43 + }, + { + "epoch": 0.07, + "grad_norm": 1.4384242296218872, + "learning_rate": 0.00019999841605240475, + "loss": 23.8929, + "step": 44 + }, + { + "epoch": 0.07, + "grad_norm": 1.290089726448059, + "learning_rate": 0.00019999814106235478, + "loss": 23.4627, + "step": 45 + }, + { + "epoch": 0.07, + "grad_norm": 1.3345509767532349, + "learning_rate": 0.0001999978440733839, + "loss": 23.1114, + "step": 46 + }, + { + "epoch": 0.07, + "grad_norm": 1.5627790689468384, + "learning_rate": 0.00019999752508555752, + "loss": 22.8306, + "step": 47 + }, + { + "epoch": 0.07, + "grad_norm": 1.67068612575531, + "learning_rate": 0.00019999718409894577, + "loss": 22.5982, + "step": 48 + }, + { + "epoch": 0.07, + "grad_norm": 1.6429119110107422, + "learning_rate": 0.00019999682111362368, + "loss": 22.2669, + "step": 49 + }, + { + "epoch": 0.07, + "grad_norm": 1.5682146549224854, + "learning_rate": 0.00019999643612967106, + "loss": 21.9562, + "step": 50 + }, + { + "epoch": 0.08, + "grad_norm": 1.3983449935913086, + "learning_rate": 0.00019999602914717266, + "loss": 21.6858, + "step": 51 + }, + { + "epoch": 0.08, + "grad_norm": 1.3296594619750977, + "learning_rate": 0.000199995600166218, + "loss": 21.4453, + "step": 52 + }, + { + "epoch": 0.08, + "grad_norm": 1.2462741136550903, + "learning_rate": 0.00019999514918690142, + "loss": 21.1379, + "step": 53 + }, + { + "epoch": 0.08, + "grad_norm": 1.2661601305007935, + "learning_rate": 0.0001999946762093222, + "loss": 20.8178, + "step": 54 + }, + { + "epoch": 0.08, + "grad_norm": 1.5054320096969604, + "learning_rate": 0.00019999418123358432, + "loss": 20.5357, + "step": 55 + }, + { + "epoch": 0.08, + "grad_norm": 1.8500558137893677, + "learning_rate": 0.0001999936642597967, + "loss": 20.2271, + "step": 56 + }, + { + "epoch": 0.08, + "grad_norm": 2.102200508117676, + "learning_rate": 0.00019999312528807308, + "loss": 19.9845, + "step": 57 + }, + { + "epoch": 0.09, + "grad_norm": 2.3885788917541504, + "learning_rate": 0.000199992564318532, + "loss": 19.6821, + "step": 58 + }, + { + "epoch": 0.09, + "grad_norm": 1.3269203901290894, + "learning_rate": 0.00019999198135129689, + "loss": 19.4197, + "step": 59 + }, + { + "epoch": 0.09, + "grad_norm": 1.0758070945739746, + "learning_rate": 0.000199991376386496, + "loss": 19.0985, + "step": 60 + }, + { + "epoch": 0.09, + "grad_norm": 1.1309177875518799, + "learning_rate": 0.00019999074942426246, + "loss": 18.8138, + "step": 61 + }, + { + "epoch": 0.09, + "grad_norm": 1.230851173400879, + "learning_rate": 0.00019999010046473412, + "loss": 18.5005, + "step": 62 + }, + { + "epoch": 0.09, + "grad_norm": 1.2155430316925049, + "learning_rate": 0.00019998942950805377, + "loss": 18.1152, + "step": 63 + }, + { + "epoch": 0.1, + "grad_norm": 1.2195416688919067, + "learning_rate": 0.00019998873655436905, + "loss": 17.7053, + "step": 64 + }, + { + "epoch": 0.1, + "grad_norm": 1.2259167432785034, + "learning_rate": 0.0001999880216038324, + "loss": 17.3123, + "step": 65 + }, + { + "epoch": 0.1, + "grad_norm": 1.2353639602661133, + "learning_rate": 0.00019998728465660105, + "loss": 16.9048, + "step": 66 + }, + { + "epoch": 0.1, + "grad_norm": 1.3028600215911865, + "learning_rate": 0.00019998652571283715, + "loss": 16.477, + "step": 67 + }, + { + "epoch": 0.1, + "grad_norm": 1.3070034980773926, + "learning_rate": 0.00019998574477270772, + "loss": 16.0892, + "step": 68 + }, + { + "epoch": 0.1, + "grad_norm": 1.231437087059021, + "learning_rate": 0.00019998494183638446, + "loss": 15.6727, + "step": 69 + }, + { + "epoch": 0.1, + "grad_norm": 1.143466591835022, + "learning_rate": 0.00019998411690404412, + "loss": 15.3314, + "step": 70 + }, + { + "epoch": 0.11, + "grad_norm": 1.4420065879821777, + "learning_rate": 0.0001999832699758681, + "loss": 15.1217, + "step": 71 + }, + { + "epoch": 0.11, + "grad_norm": 1.9184247255325317, + "learning_rate": 0.00019998240105204276, + "loss": 15.0641, + "step": 72 + }, + { + "epoch": 0.11, + "grad_norm": 2.11586332321167, + "learning_rate": 0.0001999815101327592, + "loss": 14.9107, + "step": 73 + }, + { + "epoch": 0.11, + "grad_norm": 2.1635212898254395, + "learning_rate": 0.0001999805972182135, + "loss": 14.6949, + "step": 74 + }, + { + "epoch": 0.11, + "grad_norm": 1.9647817611694336, + "learning_rate": 0.00019997966230860643, + "loss": 14.3436, + "step": 75 + }, + { + "epoch": 0.11, + "grad_norm": 1.6679894924163818, + "learning_rate": 0.00019997870540414367, + "loss": 13.917, + "step": 76 + }, + { + "epoch": 0.11, + "grad_norm": 1.407972812652588, + "learning_rate": 0.0001999777265050358, + "loss": 13.6019, + "step": 77 + }, + { + "epoch": 0.12, + "grad_norm": 1.442781686782837, + "learning_rate": 0.00019997672561149806, + "loss": 13.3002, + "step": 78 + }, + { + "epoch": 0.12, + "grad_norm": 1.4343981742858887, + "learning_rate": 0.00019997570272375074, + "loss": 13.111, + "step": 79 + }, + { + "epoch": 0.12, + "grad_norm": 1.4623301029205322, + "learning_rate": 0.00019997465784201882, + "loss": 12.9296, + "step": 80 + }, + { + "epoch": 0.12, + "grad_norm": 1.2028108835220337, + "learning_rate": 0.00019997359096653216, + "loss": 12.7946, + "step": 81 + }, + { + "epoch": 0.12, + "grad_norm": 0.7906871438026428, + "learning_rate": 0.0001999725020975255, + "loss": 12.6467, + "step": 82 + }, + { + "epoch": 0.12, + "grad_norm": 0.8495603799819946, + "learning_rate": 0.0001999713912352384, + "loss": 12.5962, + "step": 83 + }, + { + "epoch": 0.12, + "grad_norm": 1.952924370765686, + "learning_rate": 0.00019997025837991514, + "loss": 12.4475, + "step": 84 + }, + { + "epoch": 0.13, + "grad_norm": 2.176668405532837, + "learning_rate": 0.00019996910353180502, + "loss": 12.3597, + "step": 85 + }, + { + "epoch": 0.13, + "grad_norm": 1.4191577434539795, + "learning_rate": 0.0001999679266911621, + "loss": 12.2764, + "step": 86 + }, + { + "epoch": 0.13, + "grad_norm": 1.0766239166259766, + "learning_rate": 0.00019996672785824526, + "loss": 12.2163, + "step": 87 + }, + { + "epoch": 0.13, + "grad_norm": 0.8922137022018433, + "learning_rate": 0.00019996550703331825, + "loss": 12.004, + "step": 88 + }, + { + "epoch": 0.13, + "grad_norm": 0.8033492565155029, + "learning_rate": 0.00019996426421664963, + "loss": 11.8298, + "step": 89 + }, + { + "epoch": 0.13, + "grad_norm": 0.8409807682037354, + "learning_rate": 0.00019996299940851277, + "loss": 11.7005, + "step": 90 + }, + { + "epoch": 0.14, + "grad_norm": 1.0452065467834473, + "learning_rate": 0.000199961712609186, + "loss": 11.5059, + "step": 91 + }, + { + "epoch": 0.14, + "grad_norm": 1.1938868761062622, + "learning_rate": 0.00019996040381895237, + "loss": 11.3725, + "step": 92 + }, + { + "epoch": 0.14, + "grad_norm": 1.1088273525238037, + "learning_rate": 0.0001999590730380998, + "loss": 11.1496, + "step": 93 + }, + { + "epoch": 0.14, + "grad_norm": 0.9627100229263306, + "learning_rate": 0.00019995772026692107, + "loss": 10.9487, + "step": 94 + }, + { + "epoch": 0.14, + "grad_norm": 0.9412399530410767, + "learning_rate": 0.00019995634550571377, + "loss": 10.7797, + "step": 95 + }, + { + "epoch": 0.14, + "grad_norm": 0.962394654750824, + "learning_rate": 0.00019995494875478032, + "loss": 10.5443, + "step": 96 + }, + { + "epoch": 0.14, + "grad_norm": 0.7746237516403198, + "learning_rate": 0.000199953530014428, + "loss": 10.3263, + "step": 97 + }, + { + "epoch": 0.15, + "grad_norm": 0.821438729763031, + "learning_rate": 0.00019995208928496897, + "loss": 10.1812, + "step": 98 + }, + { + "epoch": 0.15, + "grad_norm": 1.4739282131195068, + "learning_rate": 0.00019995062656672012, + "loss": 9.9212, + "step": 99 + }, + { + "epoch": 0.15, + "grad_norm": 0.9201889038085938, + "learning_rate": 0.00019994914186000328, + "loss": 9.8514, + "step": 100 + }, + { + "epoch": 0.15, + "grad_norm": 0.909538745880127, + "learning_rate": 0.00019994763516514505, + "loss": 9.7503, + "step": 101 + }, + { + "epoch": 0.15, + "grad_norm": 0.8758969902992249, + "learning_rate": 0.0001999461064824769, + "loss": 9.6219, + "step": 102 + }, + { + "epoch": 0.15, + "grad_norm": 0.9529918432235718, + "learning_rate": 0.00019994455581233514, + "loss": 9.5321, + "step": 103 + }, + { + "epoch": 0.15, + "grad_norm": 1.0058927536010742, + "learning_rate": 0.0001999429831550609, + "loss": 9.3729, + "step": 104 + }, + { + "epoch": 0.16, + "grad_norm": 0.9739552140235901, + "learning_rate": 0.00019994138851100015, + "loss": 9.3018, + "step": 105 + }, + { + "epoch": 0.16, + "grad_norm": 0.8659080266952515, + "learning_rate": 0.0001999397718805037, + "loss": 9.1481, + "step": 106 + }, + { + "epoch": 0.16, + "grad_norm": 0.9334455132484436, + "learning_rate": 0.00019993813326392722, + "loss": 9.0994, + "step": 107 + }, + { + "epoch": 0.16, + "grad_norm": 0.9508799910545349, + "learning_rate": 0.00019993647266163117, + "loss": 9.0184, + "step": 108 + }, + { + "epoch": 0.16, + "grad_norm": 0.8480530381202698, + "learning_rate": 0.00019993479007398088, + "loss": 8.8884, + "step": 109 + }, + { + "epoch": 0.16, + "grad_norm": 0.7846106290817261, + "learning_rate": 0.0001999330855013465, + "loss": 8.7288, + "step": 110 + }, + { + "epoch": 0.16, + "grad_norm": 0.7611198425292969, + "learning_rate": 0.00019993135894410304, + "loss": 8.7147, + "step": 111 + }, + { + "epoch": 0.17, + "grad_norm": 0.7758539915084839, + "learning_rate": 0.00019992961040263033, + "loss": 8.7001, + "step": 112 + }, + { + "epoch": 0.17, + "grad_norm": 0.6993403434753418, + "learning_rate": 0.000199927839877313, + "loss": 8.5369, + "step": 113 + }, + { + "epoch": 0.17, + "grad_norm": 0.6209515929222107, + "learning_rate": 0.0001999260473685406, + "loss": 8.4196, + "step": 114 + }, + { + "epoch": 0.17, + "grad_norm": 0.6134201884269714, + "learning_rate": 0.00019992423287670748, + "loss": 8.3379, + "step": 115 + }, + { + "epoch": 0.17, + "grad_norm": 0.6407685875892639, + "learning_rate": 0.00019992239640221276, + "loss": 8.2981, + "step": 116 + }, + { + "epoch": 0.17, + "grad_norm": 0.5949248671531677, + "learning_rate": 0.00019992053794546047, + "loss": 8.266, + "step": 117 + }, + { + "epoch": 0.18, + "grad_norm": 0.5390576720237732, + "learning_rate": 0.0001999186575068595, + "loss": 8.2384, + "step": 118 + }, + { + "epoch": 0.18, + "grad_norm": 0.5707695484161377, + "learning_rate": 0.00019991675508682349, + "loss": 8.1337, + "step": 119 + }, + { + "epoch": 0.18, + "grad_norm": 0.6072248816490173, + "learning_rate": 0.00019991483068577097, + "loss": 8.1228, + "step": 120 + }, + { + "epoch": 0.18, + "grad_norm": 0.5436020493507385, + "learning_rate": 0.00019991288430412532, + "loss": 8.0083, + "step": 121 + }, + { + "epoch": 0.18, + "grad_norm": 0.43990185856819153, + "learning_rate": 0.00019991091594231464, + "loss": 7.943, + "step": 122 + }, + { + "epoch": 0.18, + "grad_norm": 0.7439638376235962, + "learning_rate": 0.0001999089256007721, + "loss": 7.9075, + "step": 123 + }, + { + "epoch": 0.18, + "grad_norm": 0.5559173822402954, + "learning_rate": 0.00019990691327993548, + "loss": 7.8031, + "step": 124 + }, + { + "epoch": 0.19, + "grad_norm": 0.5491620302200317, + "learning_rate": 0.0001999048789802475, + "loss": 7.7261, + "step": 125 + }, + { + "epoch": 0.19, + "grad_norm": 0.7270621657371521, + "learning_rate": 0.00019990282270215563, + "loss": 7.8013, + "step": 126 + }, + { + "epoch": 0.19, + "grad_norm": 0.5974888205528259, + "learning_rate": 0.0001999007444461123, + "loss": 7.6982, + "step": 127 + }, + { + "epoch": 0.19, + "grad_norm": 0.4842529892921448, + "learning_rate": 0.00019989864421257467, + "loss": 7.6482, + "step": 128 + }, + { + "epoch": 0.19, + "grad_norm": 0.6567961573600769, + "learning_rate": 0.00019989652200200485, + "loss": 7.6721, + "step": 129 + }, + { + "epoch": 0.19, + "grad_norm": 0.5840482711791992, + "learning_rate": 0.00019989437781486963, + "loss": 7.6075, + "step": 130 + }, + { + "epoch": 0.19, + "grad_norm": 0.41609910130500793, + "learning_rate": 0.0001998922116516408, + "loss": 7.5145, + "step": 131 + }, + { + "epoch": 0.2, + "grad_norm": 0.4465920329093933, + "learning_rate": 0.00019989002351279483, + "loss": 7.5013, + "step": 132 + }, + { + "epoch": 0.2, + "grad_norm": 0.6557760238647461, + "learning_rate": 0.0001998878133988131, + "loss": 7.5361, + "step": 133 + }, + { + "epoch": 0.2, + "grad_norm": 0.4710114598274231, + "learning_rate": 0.00019988558131018186, + "loss": 7.4989, + "step": 134 + }, + { + "epoch": 0.2, + "grad_norm": 0.67153000831604, + "learning_rate": 0.00019988332724739215, + "loss": 7.6135, + "step": 135 + }, + { + "epoch": 0.2, + "grad_norm": 1.6115596294403076, + "learning_rate": 0.00019988105121093982, + "loss": 7.8275, + "step": 136 + }, + { + "epoch": 0.2, + "grad_norm": 0.7458434700965881, + "learning_rate": 0.0001998787532013256, + "loss": 7.9861, + "step": 137 + }, + { + "epoch": 0.2, + "grad_norm": 0.6269705891609192, + "learning_rate": 0.000199876433219055, + "loss": 8.0249, + "step": 138 + }, + { + "epoch": 0.21, + "grad_norm": 0.7430456876754761, + "learning_rate": 0.00019987409126463846, + "loss": 8.0029, + "step": 139 + }, + { + "epoch": 0.21, + "grad_norm": 1.2338321208953857, + "learning_rate": 0.0001998717273385912, + "loss": 7.9936, + "step": 140 + }, + { + "epoch": 0.21, + "grad_norm": 0.7371585965156555, + "learning_rate": 0.00019986934144143316, + "loss": 7.8833, + "step": 141 + }, + { + "epoch": 0.21, + "grad_norm": 0.6107997894287109, + "learning_rate": 0.00019986693357368934, + "loss": 7.7912, + "step": 142 + }, + { + "epoch": 0.21, + "grad_norm": 1.231048583984375, + "learning_rate": 0.00019986450373588946, + "loss": 7.7342, + "step": 143 + }, + { + "epoch": 0.21, + "grad_norm": 0.6497113704681396, + "learning_rate": 0.00019986205192856795, + "loss": 7.7583, + "step": 144 + }, + { + "epoch": 0.22, + "grad_norm": 1.1496895551681519, + "learning_rate": 0.0001998595781522643, + "loss": 7.7372, + "step": 145 + }, + { + "epoch": 0.22, + "grad_norm": 0.949022114276886, + "learning_rate": 0.00019985708240752268, + "loss": 7.7063, + "step": 146 + }, + { + "epoch": 0.22, + "grad_norm": 3.4438350200653076, + "learning_rate": 0.00019985456469489213, + "loss": 7.6564, + "step": 147 + }, + { + "epoch": 0.22, + "grad_norm": 1.3692785501480103, + "learning_rate": 0.00019985202501492656, + "loss": 7.7306, + "step": 148 + }, + { + "epoch": 0.22, + "grad_norm": 1.341970443725586, + "learning_rate": 0.00019984946336818463, + "loss": 7.6558, + "step": 149 + }, + { + "epoch": 0.22, + "grad_norm": 0.8111562728881836, + "learning_rate": 0.00019984687975522993, + "loss": 7.5931, + "step": 150 + }, + { + "epoch": 0.22, + "grad_norm": 1.028735876083374, + "learning_rate": 0.00019984427417663085, + "loss": 7.5077, + "step": 151 + }, + { + "epoch": 0.23, + "grad_norm": 4.875123023986816, + "learning_rate": 0.00019984164663296058, + "loss": 7.4798, + "step": 152 + }, + { + "epoch": 0.23, + "grad_norm": 1.2443698644638062, + "learning_rate": 0.00019983899712479715, + "loss": 7.5151, + "step": 153 + }, + { + "epoch": 0.23, + "grad_norm": 38.03964614868164, + "learning_rate": 0.00019983632565272346, + "loss": 7.3952, + "step": 154 + }, + { + "epoch": 0.23, + "grad_norm": 2.3863110542297363, + "learning_rate": 0.00019983363221732716, + "loss": 7.37, + "step": 155 + }, + { + "epoch": 0.23, + "grad_norm": 1.0871200561523438, + "learning_rate": 0.00019983091681920084, + "loss": 7.4005, + "step": 156 + }, + { + "epoch": 0.23, + "grad_norm": 1.156063437461853, + "learning_rate": 0.0001998281794589419, + "loss": 7.3377, + "step": 157 + }, + { + "epoch": 0.23, + "grad_norm": 3.350292682647705, + "learning_rate": 0.00019982542013715244, + "loss": 7.0855, + "step": 158 + }, + { + "epoch": 0.24, + "grad_norm": 4.875136375427246, + "learning_rate": 0.00019982263885443956, + "loss": 7.0308, + "step": 159 + }, + { + "epoch": 0.24, + "grad_norm": 4.6525068283081055, + "learning_rate": 0.00019981983561141516, + "loss": 7.0757, + "step": 160 + }, + { + "epoch": 0.24, + "grad_norm": 2.9209208488464355, + "learning_rate": 0.00019981701040869585, + "loss": 7.1687, + "step": 161 + }, + { + "epoch": 0.24, + "grad_norm": 2.691115379333496, + "learning_rate": 0.00019981416324690317, + "loss": 7.2762, + "step": 162 + }, + { + "epoch": 0.24, + "grad_norm": 240.15269470214844, + "learning_rate": 0.00019981129412666358, + "loss": 7.3205, + "step": 163 + }, + { + "epoch": 0.24, + "grad_norm": 5.614213466644287, + "learning_rate": 0.00019980840304860813, + "loss": 7.2591, + "step": 164 + }, + { + "epoch": 0.25, + "grad_norm": 6.848481178283691, + "learning_rate": 0.0001998054900133729, + "loss": 7.3047, + "step": 165 + }, + { + "epoch": 0.25, + "grad_norm": 1.3473995923995972, + "learning_rate": 0.00019980255502159873, + "loss": 7.314, + "step": 166 + }, + { + "epoch": 0.25, + "grad_norm": 1.2133899927139282, + "learning_rate": 0.00019979959807393132, + "loss": 7.2487, + "step": 167 + }, + { + "epoch": 0.25, + "grad_norm": 46.43593215942383, + "learning_rate": 0.00019979661917102115, + "loss": 7.2762, + "step": 168 + }, + { + "epoch": 0.25, + "grad_norm": 1.915299892425537, + "learning_rate": 0.00019979361831352357, + "loss": 7.2511, + "step": 169 + }, + { + "epoch": 0.25, + "eval_loss": 7.077121257781982, + "eval_runtime": 1142.7781, + "eval_samples_per_second": 15.0, + "eval_steps_per_second": 0.938, + "step": 169 + }, + { + "epoch": 0.25, + "grad_norm": 2.1094720363616943, + "learning_rate": 0.00019979059550209876, + "loss": 7.2082, + "step": 170 + }, + { + "epoch": 0.25, + "grad_norm": 1.6243547201156616, + "learning_rate": 0.0001997875507374117, + "loss": 7.2193, + "step": 171 + }, + { + "epoch": 0.26, + "grad_norm": 1.493200659751892, + "learning_rate": 0.0001997844840201322, + "loss": 7.2206, + "step": 172 + }, + { + "epoch": 0.26, + "grad_norm": 5.889941215515137, + "learning_rate": 0.00019978139535093496, + "loss": 7.2216, + "step": 173 + }, + { + "epoch": 0.26, + "grad_norm": 16.500181198120117, + "learning_rate": 0.00019977828473049947, + "loss": 7.2762, + "step": 174 + }, + { + "epoch": 0.26, + "grad_norm": 1.3627996444702148, + "learning_rate": 0.00019977515215951, + "loss": 7.2056, + "step": 175 + }, + { + "epoch": 0.26, + "grad_norm": 3.392564535140991, + "learning_rate": 0.0001997719976386557, + "loss": 7.2832, + "step": 176 + }, + { + "epoch": 0.26, + "grad_norm": 4.66141939163208, + "learning_rate": 0.00019976882116863061, + "loss": 7.2547, + "step": 177 + }, + { + "epoch": 0.26, + "grad_norm": 1.6790283918380737, + "learning_rate": 0.00019976562275013344, + "loss": 7.1852, + "step": 178 + }, + { + "epoch": 0.27, + "grad_norm": 3.375530242919922, + "learning_rate": 0.00019976240238386787, + "loss": 7.2132, + "step": 179 + }, + { + "epoch": 0.27, + "grad_norm": 2.9105639457702637, + "learning_rate": 0.0001997591600705424, + "loss": 7.2096, + "step": 180 + }, + { + "epoch": 0.27, + "grad_norm": 3.507265329360962, + "learning_rate": 0.00019975589581087023, + "loss": 7.1184, + "step": 181 + }, + { + "epoch": 0.27, + "grad_norm": 3.2614657878875732, + "learning_rate": 0.0001997526096055695, + "loss": 7.174, + "step": 182 + }, + { + "epoch": 0.27, + "grad_norm": 20.52471923828125, + "learning_rate": 0.0001997493014553632, + "loss": 7.1458, + "step": 183 + }, + { + "epoch": 0.27, + "grad_norm": 7.462634563446045, + "learning_rate": 0.00019974597136097904, + "loss": 7.2018, + "step": 184 + }, + { + "epoch": 0.27, + "grad_norm": 3.244502544403076, + "learning_rate": 0.00019974261932314972, + "loss": 7.2068, + "step": 185 + }, + { + "epoch": 0.28, + "grad_norm": 3.8728437423706055, + "learning_rate": 0.0001997392453426125, + "loss": 7.2269, + "step": 186 + }, + { + "epoch": 0.28, + "grad_norm": 1.4866065979003906, + "learning_rate": 0.00019973584942010978, + "loss": 7.3578, + "step": 187 + }, + { + "epoch": 0.28, + "grad_norm": 11.506189346313477, + "learning_rate": 0.0001997324315563886, + "loss": 7.3122, + "step": 188 + }, + { + "epoch": 0.28, + "grad_norm": 30.93811798095703, + "learning_rate": 0.0001997289917522008, + "loss": 7.3178, + "step": 189 + }, + { + "epoch": 0.28, + "grad_norm": 27.9945125579834, + "learning_rate": 0.00019972553000830324, + "loss": 7.3128, + "step": 190 + }, + { + "epoch": 0.28, + "grad_norm": 144.04150390625, + "learning_rate": 0.00019972204632545736, + "loss": 7.2427, + "step": 191 + }, + { + "epoch": 0.29, + "grad_norm": 77.6709213256836, + "learning_rate": 0.00019971854070442962, + "loss": 7.1204, + "step": 192 + }, + { + "epoch": 0.29, + "grad_norm": 56.658504486083984, + "learning_rate": 0.00019971501314599118, + "loss": 7.1243, + "step": 193 + }, + { + "epoch": 0.29, + "grad_norm": 12.49243450164795, + "learning_rate": 0.00019971146365091813, + "loss": 7.0943, + "step": 194 + }, + { + "epoch": 0.29, + "grad_norm": 12.344218254089355, + "learning_rate": 0.0001997078922199913, + "loss": 7.0644, + "step": 195 + }, + { + "epoch": 0.29, + "grad_norm": 542.6148071289062, + "learning_rate": 0.0001997042988539964, + "loss": 7.1231, + "step": 196 + }, + { + "epoch": 0.29, + "grad_norm": 13.391535758972168, + "learning_rate": 0.00019970068355372393, + "loss": 7.0059, + "step": 197 + }, + { + "epoch": 0.29, + "grad_norm": 438.4042663574219, + "learning_rate": 0.00019969704631996927, + "loss": 7.0568, + "step": 198 + }, + { + "epoch": 0.3, + "grad_norm": 918.09326171875, + "learning_rate": 0.00019969338715353257, + "loss": 7.0436, + "step": 199 + }, + { + "epoch": 0.3, + "grad_norm": 8.38797664642334, + "learning_rate": 0.00019968970605521876, + "loss": 7.0259, + "step": 200 + }, + { + "epoch": 0.3, + "grad_norm": 99.3614501953125, + "learning_rate": 0.00019968600302583774, + "loss": 7.016, + "step": 201 + }, + { + "epoch": 0.3, + "grad_norm": 9.193567276000977, + "learning_rate": 0.0001996822780662041, + "loss": 7.0366, + "step": 202 + }, + { + "epoch": 0.3, + "grad_norm": 20.37701988220215, + "learning_rate": 0.00019967853117713739, + "loss": 7.0319, + "step": 203 + }, + { + "epoch": 0.3, + "grad_norm": 13.956972122192383, + "learning_rate": 0.00019967476235946175, + "loss": 7.1477, + "step": 204 + }, + { + "epoch": 0.3, + "grad_norm": 18.673934936523438, + "learning_rate": 0.00019967097161400642, + "loss": 7.0995, + "step": 205 + }, + { + "epoch": 0.31, + "grad_norm": 52.66343688964844, + "learning_rate": 0.0001996671589416053, + "loss": 7.1156, + "step": 206 + }, + { + "epoch": 0.31, + "grad_norm": 54.673892974853516, + "learning_rate": 0.00019966332434309716, + "loss": 7.0896, + "step": 207 + }, + { + "epoch": 0.31, + "grad_norm": 120.72140502929688, + "learning_rate": 0.00019965946781932558, + "loss": 7.0867, + "step": 208 + }, + { + "epoch": 0.31, + "grad_norm": 22.450502395629883, + "learning_rate": 0.00019965558937113894, + "loss": 7.1196, + "step": 209 + }, + { + "epoch": 0.31, + "grad_norm": 94.62016296386719, + "learning_rate": 0.00019965168899939052, + "loss": 7.1317, + "step": 210 + }, + { + "epoch": 0.31, + "grad_norm": 47.635772705078125, + "learning_rate": 0.00019964776670493837, + "loss": 7.1411, + "step": 211 + }, + { + "epoch": 0.31, + "grad_norm": 5.324208736419678, + "learning_rate": 0.00019964382248864536, + "loss": 7.1979, + "step": 212 + }, + { + "epoch": 0.32, + "grad_norm": 11.261038780212402, + "learning_rate": 0.00019963985635137915, + "loss": 7.1543, + "step": 213 + }, + { + "epoch": 0.32, + "grad_norm": 463.3558044433594, + "learning_rate": 0.0001996358682940123, + "loss": 7.1743, + "step": 214 + }, + { + "epoch": 0.32, + "grad_norm": 3.5599043369293213, + "learning_rate": 0.00019963185831742218, + "loss": 7.1325, + "step": 215 + }, + { + "epoch": 0.32, + "grad_norm": 3.592991828918457, + "learning_rate": 0.00019962782642249092, + "loss": 7.1094, + "step": 216 + }, + { + "epoch": 0.32, + "grad_norm": 59.75080108642578, + "learning_rate": 0.00019962377261010551, + "loss": 7.1571, + "step": 217 + }, + { + "epoch": 0.32, + "grad_norm": 3.108335494995117, + "learning_rate": 0.00019961969688115782, + "loss": 7.061, + "step": 218 + }, + { + "epoch": 0.33, + "grad_norm": 87.59527587890625, + "learning_rate": 0.0001996155992365444, + "loss": 7.0367, + "step": 219 + }, + { + "epoch": 0.33, + "grad_norm": 234.99960327148438, + "learning_rate": 0.00019961147967716673, + "loss": 7.1225, + "step": 220 + }, + { + "epoch": 0.33, + "grad_norm": 2.509389877319336, + "learning_rate": 0.00019960733820393114, + "loss": 7.0294, + "step": 221 + }, + { + "epoch": 0.33, + "grad_norm": 4.749375343322754, + "learning_rate": 0.00019960317481774865, + "loss": 6.9989, + "step": 222 + }, + { + "epoch": 0.33, + "grad_norm": 6.5035319328308105, + "learning_rate": 0.00019959898951953524, + "loss": 6.9879, + "step": 223 + }, + { + "epoch": 0.33, + "grad_norm": 15.936602592468262, + "learning_rate": 0.0001995947823102116, + "loss": 6.8966, + "step": 224 + }, + { + "epoch": 0.33, + "grad_norm": 27.570510864257812, + "learning_rate": 0.00019959055319070332, + "loss": 6.9171, + "step": 225 + }, + { + "epoch": 0.34, + "grad_norm": 31.393901824951172, + "learning_rate": 0.00019958630216194073, + "loss": 6.993, + "step": 226 + }, + { + "epoch": 0.34, + "grad_norm": 2.980260133743286, + "learning_rate": 0.0001995820292248591, + "loss": 7.0092, + "step": 227 + }, + { + "epoch": 0.34, + "grad_norm": 2.1385061740875244, + "learning_rate": 0.0001995777343803984, + "loss": 6.9541, + "step": 228 + }, + { + "epoch": 0.34, + "grad_norm": 4.7213053703308105, + "learning_rate": 0.00019957341762950344, + "loss": 6.8875, + "step": 229 + }, + { + "epoch": 0.34, + "grad_norm": 2.533954381942749, + "learning_rate": 0.000199569078973124, + "loss": 6.934, + "step": 230 + }, + { + "epoch": 0.34, + "grad_norm": 3.432656764984131, + "learning_rate": 0.00019956471841221438, + "loss": 6.8725, + "step": 231 + }, + { + "epoch": 0.34, + "grad_norm": 9.07863998413086, + "learning_rate": 0.000199560335947734, + "loss": 6.8456, + "step": 232 + }, + { + "epoch": 0.35, + "grad_norm": 2.4700138568878174, + "learning_rate": 0.0001995559315806469, + "loss": 6.8803, + "step": 233 + }, + { + "epoch": 0.35, + "grad_norm": 15.215128898620605, + "learning_rate": 0.0001995515053119221, + "loss": 6.7947, + "step": 234 + }, + { + "epoch": 0.35, + "grad_norm": 10.798171997070312, + "learning_rate": 0.00019954705714253328, + "loss": 6.8543, + "step": 235 + }, + { + "epoch": 0.35, + "grad_norm": 2.1929962635040283, + "learning_rate": 0.000199542587073459, + "loss": 6.7983, + "step": 236 + }, + { + "epoch": 0.35, + "grad_norm": 18.292057037353516, + "learning_rate": 0.00019953809510568271, + "loss": 6.704, + "step": 237 + }, + { + "epoch": 0.35, + "grad_norm": 0.7752164602279663, + "learning_rate": 0.00019953358124019253, + "loss": 6.7268, + "step": 238 + }, + { + "epoch": 0.35, + "grad_norm": 1.0236197710037231, + "learning_rate": 0.00019952904547798154, + "loss": 6.6618, + "step": 239 + }, + { + "epoch": 0.36, + "grad_norm": 1.3812897205352783, + "learning_rate": 0.00019952448782004756, + "loss": 6.6367, + "step": 240 + }, + { + "epoch": 0.36, + "grad_norm": 1.219442367553711, + "learning_rate": 0.00019951990826739323, + "loss": 6.6027, + "step": 241 + }, + { + "epoch": 0.36, + "grad_norm": 1.485183596611023, + "learning_rate": 0.00019951530682102604, + "loss": 6.5468, + "step": 242 + }, + { + "epoch": 0.36, + "grad_norm": 20.273746490478516, + "learning_rate": 0.00019951068348195828, + "loss": 6.5281, + "step": 243 + }, + { + "epoch": 0.36, + "grad_norm": 3.194899082183838, + "learning_rate": 0.00019950603825120706, + "loss": 6.5781, + "step": 244 + }, + { + "epoch": 0.36, + "grad_norm": 0.9288484454154968, + "learning_rate": 0.00019950137112979425, + "loss": 6.5043, + "step": 245 + }, + { + "epoch": 0.37, + "grad_norm": 0.9781804084777832, + "learning_rate": 0.00019949668211874665, + "loss": 6.4531, + "step": 246 + }, + { + "epoch": 0.37, + "grad_norm": 5.633649826049805, + "learning_rate": 0.00019949197121909575, + "loss": 6.4262, + "step": 247 + }, + { + "epoch": 0.37, + "grad_norm": 4.805322170257568, + "learning_rate": 0.00019948723843187796, + "loss": 6.3919, + "step": 248 + }, + { + "epoch": 0.37, + "grad_norm": 1.2492502927780151, + "learning_rate": 0.00019948248375813445, + "loss": 6.3459, + "step": 249 + }, + { + "epoch": 0.37, + "grad_norm": 1.5589463710784912, + "learning_rate": 0.0001994777071989112, + "loss": 6.2473, + "step": 250 + }, + { + "epoch": 0.37, + "grad_norm": 1.4179363250732422, + "learning_rate": 0.00019947290875525908, + "loss": 6.2753, + "step": 251 + }, + { + "epoch": 0.37, + "grad_norm": 1.064170479774475, + "learning_rate": 0.00019946808842823366, + "loss": 6.2785, + "step": 252 + }, + { + "epoch": 0.38, + "grad_norm": 1.806328296661377, + "learning_rate": 0.00019946324621889535, + "loss": 6.2752, + "step": 253 + }, + { + "epoch": 0.38, + "grad_norm": 1.2437814474105835, + "learning_rate": 0.0001994583821283095, + "loss": 6.1444, + "step": 254 + }, + { + "epoch": 0.38, + "grad_norm": 17.912715911865234, + "learning_rate": 0.00019945349615754608, + "loss": 6.1001, + "step": 255 + }, + { + "epoch": 0.38, + "grad_norm": 0.8929802775382996, + "learning_rate": 0.00019944858830768002, + "loss": 6.0709, + "step": 256 + }, + { + "epoch": 0.38, + "grad_norm": 2.8683338165283203, + "learning_rate": 0.000199443658579791, + "loss": 5.9837, + "step": 257 + }, + { + "epoch": 0.38, + "grad_norm": 1.3912129402160645, + "learning_rate": 0.00019943870697496357, + "loss": 5.9728, + "step": 258 + }, + { + "epoch": 0.38, + "grad_norm": 1.9068537950515747, + "learning_rate": 0.00019943373349428698, + "loss": 5.9657, + "step": 259 + }, + { + "epoch": 0.39, + "grad_norm": 2.0884199142456055, + "learning_rate": 0.0001994287381388554, + "loss": 5.963, + "step": 260 + }, + { + "epoch": 0.39, + "grad_norm": 0.8210564851760864, + "learning_rate": 0.00019942372090976779, + "loss": 5.8253, + "step": 261 + }, + { + "epoch": 0.39, + "grad_norm": 3.3612587451934814, + "learning_rate": 0.0001994186818081279, + "loss": 5.7743, + "step": 262 + }, + { + "epoch": 0.39, + "grad_norm": 2.1916985511779785, + "learning_rate": 0.00019941362083504425, + "loss": 5.7697, + "step": 263 + }, + { + "epoch": 0.39, + "grad_norm": 1.3054827451705933, + "learning_rate": 0.00019940853799163026, + "loss": 5.7867, + "step": 264 + }, + { + "epoch": 0.39, + "grad_norm": 12.717972755432129, + "learning_rate": 0.00019940343327900416, + "loss": 5.6708, + "step": 265 + }, + { + "epoch": 0.4, + "grad_norm": 3.4190762042999268, + "learning_rate": 0.00019939830669828888, + "loss": 5.6869, + "step": 266 + }, + { + "epoch": 0.4, + "grad_norm": 2.753885507583618, + "learning_rate": 0.00019939315825061223, + "loss": 5.5825, + "step": 267 + }, + { + "epoch": 0.4, + "grad_norm": 1.1797378063201904, + "learning_rate": 0.0001993879879371069, + "loss": 5.6165, + "step": 268 + }, + { + "epoch": 0.4, + "grad_norm": 2.9885246753692627, + "learning_rate": 0.00019938279575891033, + "loss": 5.5817, + "step": 269 + }, + { + "epoch": 0.4, + "grad_norm": 1.8179508447647095, + "learning_rate": 0.00019937758171716468, + "loss": 5.4549, + "step": 270 + }, + { + "epoch": 0.4, + "grad_norm": 7.133963584899902, + "learning_rate": 0.00019937234581301707, + "loss": 5.5333, + "step": 271 + }, + { + "epoch": 0.4, + "grad_norm": 7.718517780303955, + "learning_rate": 0.00019936708804761932, + "loss": 5.4891, + "step": 272 + }, + { + "epoch": 0.41, + "grad_norm": 1.2075421810150146, + "learning_rate": 0.00019936180842212818, + "loss": 5.3542, + "step": 273 + }, + { + "epoch": 0.41, + "grad_norm": 6.181577205657959, + "learning_rate": 0.000199356506937705, + "loss": 5.4501, + "step": 274 + }, + { + "epoch": 0.41, + "grad_norm": 6.531129837036133, + "learning_rate": 0.00019935118359551624, + "loss": 5.5179, + "step": 275 + }, + { + "epoch": 0.41, + "grad_norm": 4.549463272094727, + "learning_rate": 0.00019934583839673287, + "loss": 5.4167, + "step": 276 + }, + { + "epoch": 0.41, + "grad_norm": 1.8859062194824219, + "learning_rate": 0.00019934047134253081, + "loss": 5.2628, + "step": 277 + }, + { + "epoch": 0.41, + "grad_norm": 7.954891681671143, + "learning_rate": 0.00019933508243409086, + "loss": 5.3723, + "step": 278 + }, + { + "epoch": 0.41, + "grad_norm": 6.0001444816589355, + "learning_rate": 0.00019932967167259847, + "loss": 5.2979, + "step": 279 + }, + { + "epoch": 0.42, + "grad_norm": 6.450177192687988, + "learning_rate": 0.000199324239059244, + "loss": 5.2947, + "step": 280 + }, + { + "epoch": 0.42, + "grad_norm": 5.8015642166137695, + "learning_rate": 0.00019931878459522254, + "loss": 5.2849, + "step": 281 + }, + { + "epoch": 0.42, + "grad_norm": 8.846211433410645, + "learning_rate": 0.00019931330828173408, + "loss": 5.1841, + "step": 282 + }, + { + "epoch": 0.42, + "grad_norm": 4.6721086502075195, + "learning_rate": 0.00019930781011998338, + "loss": 5.2056, + "step": 283 + }, + { + "epoch": 0.42, + "grad_norm": 3.9085495471954346, + "learning_rate": 0.00019930229011118, + "loss": 5.1857, + "step": 284 + }, + { + "epoch": 0.42, + "grad_norm": 1.6466888189315796, + "learning_rate": 0.00019929674825653826, + "loss": 5.0631, + "step": 285 + }, + { + "epoch": 0.42, + "grad_norm": 2.1051926612854004, + "learning_rate": 0.0001992911845572774, + "loss": 5.033, + "step": 286 + }, + { + "epoch": 0.43, + "grad_norm": 3.247199058532715, + "learning_rate": 0.00019928559901462133, + "loss": 5.098, + "step": 287 + }, + { + "epoch": 0.43, + "grad_norm": 4.04020881652832, + "learning_rate": 0.00019927999162979885, + "loss": 5.0759, + "step": 288 + }, + { + "epoch": 0.43, + "grad_norm": 1.5520354509353638, + "learning_rate": 0.00019927436240404359, + "loss": 5.0395, + "step": 289 + }, + { + "epoch": 0.43, + "grad_norm": 1.2532438039779663, + "learning_rate": 0.00019926871133859388, + "loss": 4.9942, + "step": 290 + }, + { + "epoch": 0.43, + "grad_norm": 3.3151402473449707, + "learning_rate": 0.00019926303843469296, + "loss": 4.9396, + "step": 291 + }, + { + "epoch": 0.43, + "grad_norm": 1.311003565788269, + "learning_rate": 0.00019925734369358883, + "loss": 4.9452, + "step": 292 + }, + { + "epoch": 0.44, + "grad_norm": 3.930495500564575, + "learning_rate": 0.00019925162711653425, + "loss": 4.9488, + "step": 293 + }, + { + "epoch": 0.44, + "grad_norm": 1.0052605867385864, + "learning_rate": 0.00019924588870478687, + "loss": 4.8627, + "step": 294 + }, + { + "epoch": 0.44, + "grad_norm": 7.544490814208984, + "learning_rate": 0.00019924012845960908, + "loss": 4.9602, + "step": 295 + }, + { + "epoch": 0.44, + "grad_norm": 6.155104637145996, + "learning_rate": 0.00019923434638226812, + "loss": 4.9091, + "step": 296 + }, + { + "epoch": 0.44, + "grad_norm": 3.3449931144714355, + "learning_rate": 0.00019922854247403599, + "loss": 5.002, + "step": 297 + }, + { + "epoch": 0.44, + "grad_norm": 3.9289393424987793, + "learning_rate": 0.0001992227167361895, + "loss": 4.8124, + "step": 298 + }, + { + "epoch": 0.44, + "grad_norm": 6.716512680053711, + "learning_rate": 0.00019921686917001032, + "loss": 4.8421, + "step": 299 + }, + { + "epoch": 0.45, + "grad_norm": 4.2615132331848145, + "learning_rate": 0.00019921099977678484, + "loss": 4.8303, + "step": 300 + }, + { + "epoch": 0.45, + "grad_norm": 4.952175617218018, + "learning_rate": 0.00019920510855780428, + "loss": 4.8718, + "step": 301 + }, + { + "epoch": 0.45, + "grad_norm": 1.830321192741394, + "learning_rate": 0.00019919919551436468, + "loss": 4.8395, + "step": 302 + }, + { + "epoch": 0.45, + "grad_norm": 2.7058656215667725, + "learning_rate": 0.0001991932606477669, + "loss": 4.721, + "step": 303 + }, + { + "epoch": 0.45, + "grad_norm": 2.4603841304779053, + "learning_rate": 0.00019918730395931649, + "loss": 4.7365, + "step": 304 + }, + { + "epoch": 0.45, + "grad_norm": 1.5384434461593628, + "learning_rate": 0.00019918132545032396, + "loss": 4.6979, + "step": 305 + }, + { + "epoch": 0.45, + "grad_norm": 5.274871349334717, + "learning_rate": 0.0001991753251221045, + "loss": 4.7287, + "step": 306 + }, + { + "epoch": 0.46, + "grad_norm": 3.8402764797210693, + "learning_rate": 0.00019916930297597817, + "loss": 4.744, + "step": 307 + }, + { + "epoch": 0.46, + "grad_norm": 3.417302131652832, + "learning_rate": 0.0001991632590132698, + "loss": 4.6439, + "step": 308 + }, + { + "epoch": 0.46, + "grad_norm": 4.113287925720215, + "learning_rate": 0.00019915719323530898, + "loss": 4.6228, + "step": 309 + }, + { + "epoch": 0.46, + "grad_norm": 3.57785964012146, + "learning_rate": 0.0001991511056434302, + "loss": 4.6077, + "step": 310 + }, + { + "epoch": 0.46, + "grad_norm": 1.568590521812439, + "learning_rate": 0.00019914499623897264, + "loss": 4.6056, + "step": 311 + }, + { + "epoch": 0.46, + "grad_norm": 3.293492317199707, + "learning_rate": 0.0001991388650232804, + "loss": 4.5294, + "step": 312 + }, + { + "epoch": 0.46, + "grad_norm": 3.416661262512207, + "learning_rate": 0.00019913271199770217, + "loss": 4.6239, + "step": 313 + }, + { + "epoch": 0.47, + "grad_norm": 2.6851394176483154, + "learning_rate": 0.0001991265371635917, + "loss": 4.6183, + "step": 314 + }, + { + "epoch": 0.47, + "grad_norm": 3.812194585800171, + "learning_rate": 0.0001991203405223074, + "loss": 4.5595, + "step": 315 + }, + { + "epoch": 0.47, + "grad_norm": 1.8244884014129639, + "learning_rate": 0.0001991141220752124, + "loss": 4.5329, + "step": 316 + }, + { + "epoch": 0.47, + "grad_norm": 2.763340950012207, + "learning_rate": 0.0001991078818236748, + "loss": 4.5336, + "step": 317 + }, + { + "epoch": 0.47, + "grad_norm": 0.6997628808021545, + "learning_rate": 0.0001991016197690674, + "loss": 4.5623, + "step": 318 + }, + { + "epoch": 0.47, + "grad_norm": 6.273808479309082, + "learning_rate": 0.0001990953359127678, + "loss": 4.549, + "step": 319 + }, + { + "epoch": 0.48, + "grad_norm": 4.5749359130859375, + "learning_rate": 0.00019908903025615842, + "loss": 4.4894, + "step": 320 + }, + { + "epoch": 0.48, + "grad_norm": 4.788200378417969, + "learning_rate": 0.00019908270280062643, + "loss": 4.514, + "step": 321 + }, + { + "epoch": 0.48, + "grad_norm": 5.6347551345825195, + "learning_rate": 0.00019907635354756385, + "loss": 4.477, + "step": 322 + }, + { + "epoch": 0.48, + "grad_norm": 5.96523904800415, + "learning_rate": 0.00019906998249836745, + "loss": 4.4759, + "step": 323 + }, + { + "epoch": 0.48, + "grad_norm": 2.5423848628997803, + "learning_rate": 0.00019906358965443888, + "loss": 4.4077, + "step": 324 + }, + { + "epoch": 0.48, + "grad_norm": 1.0939940214157104, + "learning_rate": 0.00019905717501718444, + "loss": 4.372, + "step": 325 + }, + { + "epoch": 0.48, + "grad_norm": 2.4741616249084473, + "learning_rate": 0.00019905073858801536, + "loss": 4.4112, + "step": 326 + }, + { + "epoch": 0.49, + "grad_norm": 2.774264335632324, + "learning_rate": 0.00019904428036834763, + "loss": 4.4412, + "step": 327 + }, + { + "epoch": 0.49, + "grad_norm": 1.9465824365615845, + "learning_rate": 0.00019903780035960194, + "loss": 4.4442, + "step": 328 + }, + { + "epoch": 0.49, + "grad_norm": 1.6578731536865234, + "learning_rate": 0.0001990312985632039, + "loss": 4.4368, + "step": 329 + }, + { + "epoch": 0.49, + "grad_norm": 1.1936014890670776, + "learning_rate": 0.00019902477498058387, + "loss": 4.3445, + "step": 330 + }, + { + "epoch": 0.49, + "grad_norm": 1.1214628219604492, + "learning_rate": 0.00019901822961317697, + "loss": 4.4032, + "step": 331 + }, + { + "epoch": 0.49, + "grad_norm": 1.4763222932815552, + "learning_rate": 0.00019901166246242316, + "loss": 4.3713, + "step": 332 + }, + { + "epoch": 0.49, + "grad_norm": 3.219590425491333, + "learning_rate": 0.00019900507352976714, + "loss": 4.3248, + "step": 333 + }, + { + "epoch": 0.5, + "grad_norm": 1.5973373651504517, + "learning_rate": 0.00019899846281665843, + "loss": 4.3571, + "step": 334 + }, + { + "epoch": 0.5, + "grad_norm": 2.3610401153564453, + "learning_rate": 0.0001989918303245514, + "loss": 4.2991, + "step": 335 + }, + { + "epoch": 0.5, + "grad_norm": 2.8560168743133545, + "learning_rate": 0.0001989851760549051, + "loss": 4.3024, + "step": 336 + }, + { + "epoch": 0.5, + "grad_norm": 0.9701719880104065, + "learning_rate": 0.0001989785000091834, + "loss": 4.2602, + "step": 337 + }, + { + "epoch": 0.5, + "grad_norm": 1.6666287183761597, + "learning_rate": 0.00019897180218885507, + "loss": 4.2519, + "step": 338 + }, + { + "epoch": 0.5, + "eval_loss": 4.065431594848633, + "eval_runtime": 1149.6322, + "eval_samples_per_second": 14.911, + "eval_steps_per_second": 0.932, + "step": 338 + }, + { + "epoch": 0.5, + "grad_norm": 2.7144992351531982, + "learning_rate": 0.00019896508259539353, + "loss": 4.2504, + "step": 339 + }, + { + "epoch": 0.5, + "grad_norm": 1.088392734527588, + "learning_rate": 0.00019895834123027703, + "loss": 4.2467, + "step": 340 + }, + { + "epoch": 0.51, + "grad_norm": 1.7031919956207275, + "learning_rate": 0.00019895157809498868, + "loss": 4.2498, + "step": 341 + }, + { + "epoch": 0.51, + "grad_norm": 3.883507013320923, + "learning_rate": 0.00019894479319101626, + "loss": 4.289, + "step": 342 + }, + { + "epoch": 0.51, + "grad_norm": 1.1266521215438843, + "learning_rate": 0.00019893798651985246, + "loss": 4.2371, + "step": 343 + }, + { + "epoch": 0.51, + "grad_norm": 2.436180591583252, + "learning_rate": 0.00019893115808299468, + "loss": 4.213, + "step": 344 + }, + { + "epoch": 0.51, + "grad_norm": 6.882200241088867, + "learning_rate": 0.00019892430788194513, + "loss": 4.2289, + "step": 345 + }, + { + "epoch": 0.51, + "grad_norm": 4.444315433502197, + "learning_rate": 0.00019891743591821077, + "loss": 4.1858, + "step": 346 + }, + { + "epoch": 0.52, + "grad_norm": 2.7933340072631836, + "learning_rate": 0.00019891054219330344, + "loss": 4.1988, + "step": 347 + }, + { + "epoch": 0.52, + "grad_norm": 1.5705324411392212, + "learning_rate": 0.00019890362670873973, + "loss": 4.1639, + "step": 348 + }, + { + "epoch": 0.52, + "grad_norm": 4.15967321395874, + "learning_rate": 0.00019889668946604093, + "loss": 4.1391, + "step": 349 + }, + { + "epoch": 0.52, + "grad_norm": 3.6706178188323975, + "learning_rate": 0.00019888973046673324, + "loss": 4.1706, + "step": 350 + }, + { + "epoch": 0.52, + "grad_norm": 4.234200477600098, + "learning_rate": 0.00019888274971234757, + "loss": 4.3108, + "step": 351 + }, + { + "epoch": 0.52, + "grad_norm": 2.6971170902252197, + "learning_rate": 0.00019887574720441967, + "loss": 4.1871, + "step": 352 + }, + { + "epoch": 0.52, + "grad_norm": 1.4427449703216553, + "learning_rate": 0.00019886872294448997, + "loss": 4.1327, + "step": 353 + }, + { + "epoch": 0.53, + "grad_norm": 4.426568984985352, + "learning_rate": 0.00019886167693410384, + "loss": 4.195, + "step": 354 + }, + { + "epoch": 0.53, + "grad_norm": 4.152656555175781, + "learning_rate": 0.00019885460917481136, + "loss": 4.1858, + "step": 355 + }, + { + "epoch": 0.53, + "grad_norm": 5.731381893157959, + "learning_rate": 0.0001988475196681673, + "loss": 4.179, + "step": 356 + }, + { + "epoch": 0.53, + "grad_norm": 17.160917282104492, + "learning_rate": 0.0001988404084157314, + "loss": 4.0972, + "step": 357 + }, + { + "epoch": 0.53, + "grad_norm": 1.9398918151855469, + "learning_rate": 0.00019883327541906802, + "loss": 4.1238, + "step": 358 + }, + { + "epoch": 0.53, + "grad_norm": 2.0568935871124268, + "learning_rate": 0.00019882612067974643, + "loss": 4.0953, + "step": 359 + }, + { + "epoch": 0.53, + "grad_norm": 3.45890212059021, + "learning_rate": 0.00019881894419934056, + "loss": 4.1171, + "step": 360 + }, + { + "epoch": 0.54, + "grad_norm": 1.6760789155960083, + "learning_rate": 0.00019881174597942923, + "loss": 4.1128, + "step": 361 + }, + { + "epoch": 0.54, + "grad_norm": 307.13031005859375, + "learning_rate": 0.000198804526021596, + "loss": 4.0658, + "step": 362 + }, + { + "epoch": 0.54, + "grad_norm": 15.57320785522461, + "learning_rate": 0.00019879728432742918, + "loss": 4.2352, + "step": 363 + }, + { + "epoch": 0.54, + "grad_norm": 8.536359786987305, + "learning_rate": 0.00019879002089852197, + "loss": 4.2513, + "step": 364 + }, + { + "epoch": 0.54, + "grad_norm": 9.536478996276855, + "learning_rate": 0.00019878273573647218, + "loss": 4.2273, + "step": 365 + }, + { + "epoch": 0.54, + "grad_norm": 11.019887924194336, + "learning_rate": 0.00019877542884288256, + "loss": 4.2037, + "step": 366 + }, + { + "epoch": 0.55, + "grad_norm": 7.40153169631958, + "learning_rate": 0.00019876810021936053, + "loss": 4.1092, + "step": 367 + }, + { + "epoch": 0.55, + "grad_norm": 5.204662322998047, + "learning_rate": 0.0001987607498675184, + "loss": 4.0828, + "step": 368 + }, + { + "epoch": 0.55, + "grad_norm": 3.2306785583496094, + "learning_rate": 0.00019875337778897315, + "loss": 4.0828, + "step": 369 + }, + { + "epoch": 0.55, + "grad_norm": 9.242484092712402, + "learning_rate": 0.00019874598398534658, + "loss": 4.1478, + "step": 370 + }, + { + "epoch": 0.55, + "grad_norm": 10.423043251037598, + "learning_rate": 0.00019873856845826536, + "loss": 4.1709, + "step": 371 + }, + { + "epoch": 0.55, + "grad_norm": 2.9181222915649414, + "learning_rate": 0.00019873113120936074, + "loss": 4.0819, + "step": 372 + }, + { + "epoch": 0.55, + "grad_norm": 5.6168718338012695, + "learning_rate": 0.00019872367224026895, + "loss": 4.09, + "step": 373 + }, + { + "epoch": 0.56, + "grad_norm": 2.738548994064331, + "learning_rate": 0.00019871619155263087, + "loss": 4.1129, + "step": 374 + }, + { + "epoch": 0.56, + "grad_norm": 8.364383697509766, + "learning_rate": 0.0001987086891480922, + "loss": 4.0793, + "step": 375 + }, + { + "epoch": 0.56, + "grad_norm": 8.516968727111816, + "learning_rate": 0.00019870116502830346, + "loss": 4.1287, + "step": 376 + }, + { + "epoch": 0.56, + "grad_norm": 2.941425323486328, + "learning_rate": 0.00019869361919491987, + "loss": 4.1145, + "step": 377 + }, + { + "epoch": 0.56, + "grad_norm": 4.925361633300781, + "learning_rate": 0.00019868605164960145, + "loss": 4.1046, + "step": 378 + }, + { + "epoch": 0.56, + "grad_norm": 1.9746365547180176, + "learning_rate": 0.0001986784623940131, + "loss": 3.9944, + "step": 379 + }, + { + "epoch": 0.56, + "grad_norm": 8.789060592651367, + "learning_rate": 0.00019867085142982426, + "loss": 4.0482, + "step": 380 + }, + { + "epoch": 0.57, + "grad_norm": 7.4179301261901855, + "learning_rate": 0.0001986632187587094, + "loss": 4.0702, + "step": 381 + }, + { + "epoch": 0.57, + "grad_norm": 3.4922187328338623, + "learning_rate": 0.00019865556438234763, + "loss": 4.0579, + "step": 382 + }, + { + "epoch": 0.57, + "grad_norm": 5.763650417327881, + "learning_rate": 0.00019864788830242284, + "loss": 4.0776, + "step": 383 + }, + { + "epoch": 0.57, + "grad_norm": 3.3361189365386963, + "learning_rate": 0.00019864019052062375, + "loss": 4.0426, + "step": 384 + }, + { + "epoch": 0.57, + "grad_norm": 8.98303508758545, + "learning_rate": 0.0001986324710386438, + "loss": 4.0599, + "step": 385 + }, + { + "epoch": 0.57, + "grad_norm": 8.836548805236816, + "learning_rate": 0.00019862472985818124, + "loss": 4.0534, + "step": 386 + }, + { + "epoch": 0.57, + "grad_norm": 2.0397250652313232, + "learning_rate": 0.00019861696698093902, + "loss": 4.0105, + "step": 387 + }, + { + "epoch": 0.58, + "grad_norm": 3.091047763824463, + "learning_rate": 0.000198609182408625, + "loss": 3.9637, + "step": 388 + }, + { + "epoch": 0.58, + "grad_norm": 3.192955255508423, + "learning_rate": 0.00019860137614295168, + "loss": 3.9963, + "step": 389 + }, + { + "epoch": 0.58, + "grad_norm": 1.5237922668457031, + "learning_rate": 0.00019859354818563643, + "loss": 3.964, + "step": 390 + }, + { + "epoch": 0.58, + "grad_norm": 4.192271709442139, + "learning_rate": 0.00019858569853840131, + "loss": 3.9638, + "step": 391 + }, + { + "epoch": 0.58, + "grad_norm": 2.3870139122009277, + "learning_rate": 0.00019857782720297319, + "loss": 4.0272, + "step": 392 + }, + { + "epoch": 0.58, + "grad_norm": 6.654936790466309, + "learning_rate": 0.00019856993418108376, + "loss": 3.9616, + "step": 393 + }, + { + "epoch": 0.59, + "grad_norm": 4.117982387542725, + "learning_rate": 0.00019856201947446937, + "loss": 3.9983, + "step": 394 + }, + { + "epoch": 0.59, + "grad_norm": 5.394196033477783, + "learning_rate": 0.00019855408308487124, + "loss": 4.0334, + "step": 395 + }, + { + "epoch": 0.59, + "grad_norm": 5.410153865814209, + "learning_rate": 0.0001985461250140353, + "loss": 4.0342, + "step": 396 + }, + { + "epoch": 0.59, + "grad_norm": 1.220145583152771, + "learning_rate": 0.00019853814526371228, + "loss": 3.9964, + "step": 397 + }, + { + "epoch": 0.59, + "grad_norm": 1.3831374645233154, + "learning_rate": 0.00019853014383565772, + "loss": 3.8909, + "step": 398 + }, + { + "epoch": 0.59, + "grad_norm": 3.4708709716796875, + "learning_rate": 0.0001985221207316318, + "loss": 3.9548, + "step": 399 + }, + { + "epoch": 0.59, + "grad_norm": 1.6609090566635132, + "learning_rate": 0.00019851407595339958, + "loss": 3.9481, + "step": 400 + }, + { + "epoch": 0.6, + "grad_norm": 1.5527194738388062, + "learning_rate": 0.00019850600950273086, + "loss": 3.9567, + "step": 401 + }, + { + "epoch": 0.6, + "grad_norm": 6.021573066711426, + "learning_rate": 0.00019849792138140022, + "loss": 4.0133, + "step": 402 + }, + { + "epoch": 0.6, + "grad_norm": 4.150272846221924, + "learning_rate": 0.00019848981159118699, + "loss": 3.9614, + "step": 403 + }, + { + "epoch": 0.6, + "grad_norm": 7.311338901519775, + "learning_rate": 0.00019848168013387522, + "loss": 3.9789, + "step": 404 + }, + { + "epoch": 0.6, + "grad_norm": 7.797035217285156, + "learning_rate": 0.00019847352701125385, + "loss": 4.0036, + "step": 405 + }, + { + "epoch": 0.6, + "grad_norm": 2.067911386489868, + "learning_rate": 0.0001984653522251165, + "loss": 3.9421, + "step": 406 + }, + { + "epoch": 0.6, + "grad_norm": 5.364695072174072, + "learning_rate": 0.00019845715577726148, + "loss": 3.9846, + "step": 407 + }, + { + "epoch": 0.61, + "grad_norm": 3.663616895675659, + "learning_rate": 0.00019844893766949207, + "loss": 3.9255, + "step": 408 + }, + { + "epoch": 0.61, + "grad_norm": 8.939522743225098, + "learning_rate": 0.00019844069790361611, + "loss": 3.9433, + "step": 409 + }, + { + "epoch": 0.61, + "grad_norm": 9.791443824768066, + "learning_rate": 0.00019843243648144635, + "loss": 3.9879, + "step": 410 + }, + { + "epoch": 0.61, + "grad_norm": 1.5382161140441895, + "learning_rate": 0.00019842415340480025, + "loss": 3.9755, + "step": 411 + }, + { + "epoch": 0.61, + "grad_norm": 7.2470879554748535, + "learning_rate": 0.00019841584867549998, + "loss": 3.9603, + "step": 412 + }, + { + "epoch": 0.61, + "grad_norm": 7.930785655975342, + "learning_rate": 0.0001984075222953726, + "loss": 3.9416, + "step": 413 + }, + { + "epoch": 0.61, + "grad_norm": 3.421830654144287, + "learning_rate": 0.00019839917426624977, + "loss": 3.9023, + "step": 414 + }, + { + "epoch": 0.62, + "grad_norm": 13.186482429504395, + "learning_rate": 0.00019839080458996806, + "loss": 4.0143, + "step": 415 + }, + { + "epoch": 0.62, + "grad_norm": 18.97665023803711, + "learning_rate": 0.0001983824132683687, + "loss": 4.0361, + "step": 416 + }, + { + "epoch": 0.62, + "grad_norm": 12.884719848632812, + "learning_rate": 0.0001983740003032978, + "loss": 4.0086, + "step": 417 + }, + { + "epoch": 0.62, + "grad_norm": 2.3282957077026367, + "learning_rate": 0.00019836556569660608, + "loss": 3.8853, + "step": 418 + }, + { + "epoch": 0.62, + "grad_norm": 5.4361186027526855, + "learning_rate": 0.0001983571094501491, + "loss": 3.8995, + "step": 419 + }, + { + "epoch": 0.62, + "grad_norm": 2.480142116546631, + "learning_rate": 0.00019834863156578724, + "loss": 3.9204, + "step": 420 + }, + { + "epoch": 0.63, + "grad_norm": 8.683335304260254, + "learning_rate": 0.0001983401320453855, + "loss": 4.04, + "step": 421 + }, + { + "epoch": 0.63, + "grad_norm": 10.6290864944458, + "learning_rate": 0.00019833161089081381, + "loss": 3.9174, + "step": 422 + }, + { + "epoch": 0.63, + "grad_norm": 1.7693558931350708, + "learning_rate": 0.00019832306810394664, + "loss": 3.937, + "step": 423 + }, + { + "epoch": 0.63, + "grad_norm": 6.862966537475586, + "learning_rate": 0.00019831450368666348, + "loss": 3.9314, + "step": 424 + }, + { + "epoch": 0.63, + "grad_norm": 6.495865345001221, + "learning_rate": 0.0001983059176408484, + "loss": 3.8863, + "step": 425 + }, + { + "epoch": 0.63, + "grad_norm": 2.8406107425689697, + "learning_rate": 0.0001982973099683902, + "loss": 3.9525, + "step": 426 + }, + { + "epoch": 0.63, + "grad_norm": 10.725767135620117, + "learning_rate": 0.00019828868067118255, + "loss": 3.894, + "step": 427 + }, + { + "epoch": 0.64, + "grad_norm": 12.51372241973877, + "learning_rate": 0.0001982800297511239, + "loss": 4.0417, + "step": 428 + }, + { + "epoch": 0.64, + "grad_norm": 5.210300445556641, + "learning_rate": 0.00019827135721011736, + "loss": 3.9486, + "step": 429 + }, + { + "epoch": 0.64, + "grad_norm": 8.358055114746094, + "learning_rate": 0.0001982626630500708, + "loss": 4.0152, + "step": 430 + }, + { + "epoch": 0.64, + "grad_norm": 6.729689598083496, + "learning_rate": 0.0001982539472728969, + "loss": 3.8904, + "step": 431 + }, + { + "epoch": 0.64, + "grad_norm": 4.138473033905029, + "learning_rate": 0.00019824520988051306, + "loss": 3.8004, + "step": 432 + }, + { + "epoch": 0.64, + "grad_norm": 6.071408271789551, + "learning_rate": 0.00019823645087484147, + "loss": 4.0332, + "step": 433 + }, + { + "epoch": 0.64, + "grad_norm": 4.820131301879883, + "learning_rate": 0.000198227670257809, + "loss": 4.0076, + "step": 434 + }, + { + "epoch": 0.65, + "grad_norm": 3.6242270469665527, + "learning_rate": 0.00019821886803134738, + "loss": 3.9052, + "step": 435 + }, + { + "epoch": 0.65, + "grad_norm": 4.531324863433838, + "learning_rate": 0.00019821004419739307, + "loss": 3.8389, + "step": 436 + }, + { + "epoch": 0.65, + "grad_norm": 3.6086678504943848, + "learning_rate": 0.00019820119875788716, + "loss": 3.8785, + "step": 437 + }, + { + "epoch": 0.65, + "grad_norm": 3.600398063659668, + "learning_rate": 0.00019819233171477566, + "loss": 3.8425, + "step": 438 + }, + { + "epoch": 0.65, + "grad_norm": 3.8471834659576416, + "learning_rate": 0.00019818344307000925, + "loss": 3.8667, + "step": 439 + }, + { + "epoch": 0.65, + "grad_norm": 3.2653310298919678, + "learning_rate": 0.00019817453282554333, + "loss": 3.8744, + "step": 440 + }, + { + "epoch": 0.65, + "grad_norm": 4.489375591278076, + "learning_rate": 0.00019816560098333817, + "loss": 3.8705, + "step": 441 + }, + { + "epoch": 0.66, + "grad_norm": 3.816256284713745, + "learning_rate": 0.00019815664754535862, + "loss": 3.7886, + "step": 442 + }, + { + "epoch": 0.66, + "grad_norm": 6.624678134918213, + "learning_rate": 0.00019814767251357448, + "loss": 3.865, + "step": 443 + }, + { + "epoch": 0.66, + "grad_norm": 6.498514652252197, + "learning_rate": 0.00019813867588996012, + "loss": 3.8771, + "step": 444 + }, + { + "epoch": 0.66, + "grad_norm": 3.340604305267334, + "learning_rate": 0.00019812965767649474, + "loss": 3.8211, + "step": 445 + }, + { + "epoch": 0.66, + "grad_norm": 5.6404805183410645, + "learning_rate": 0.0001981206178751623, + "loss": 3.8503, + "step": 446 + }, + { + "epoch": 0.66, + "grad_norm": 2.282567024230957, + "learning_rate": 0.00019811155648795153, + "loss": 3.865, + "step": 447 + }, + { + "epoch": 0.67, + "grad_norm": 4.082645893096924, + "learning_rate": 0.00019810247351685587, + "loss": 3.7833, + "step": 448 + }, + { + "epoch": 0.67, + "grad_norm": 2.7963573932647705, + "learning_rate": 0.00019809336896387344, + "loss": 3.7991, + "step": 449 + }, + { + "epoch": 0.67, + "grad_norm": 8.172618865966797, + "learning_rate": 0.00019808424283100726, + "loss": 3.8369, + "step": 450 + }, + { + "epoch": 0.67, + "grad_norm": 8.013967514038086, + "learning_rate": 0.000198075095120265, + "loss": 3.8526, + "step": 451 + }, + { + "epoch": 0.67, + "grad_norm": 0.9244366884231567, + "learning_rate": 0.00019806592583365907, + "loss": 3.7882, + "step": 452 + }, + { + "epoch": 0.67, + "grad_norm": 4.72100305557251, + "learning_rate": 0.00019805673497320662, + "loss": 3.7314, + "step": 453 + }, + { + "epoch": 0.67, + "grad_norm": 2.556580066680908, + "learning_rate": 0.00019804752254092968, + "loss": 3.8743, + "step": 454 + }, + { + "epoch": 0.68, + "grad_norm": 10.014892578125, + "learning_rate": 0.00019803828853885485, + "loss": 3.8263, + "step": 455 + }, + { + "epoch": 0.68, + "grad_norm": 12.621180534362793, + "learning_rate": 0.00019802903296901357, + "loss": 3.8423, + "step": 456 + }, + { + "epoch": 0.68, + "grad_norm": 6.725605010986328, + "learning_rate": 0.000198019755833442, + "loss": 3.7592, + "step": 457 + }, + { + "epoch": 0.68, + "grad_norm": 5.9739603996276855, + "learning_rate": 0.00019801045713418102, + "loss": 3.8252, + "step": 458 + }, + { + "epoch": 0.68, + "grad_norm": 6.546073913574219, + "learning_rate": 0.00019800113687327633, + "loss": 3.8135, + "step": 459 + }, + { + "epoch": 0.68, + "grad_norm": 1.279313325881958, + "learning_rate": 0.0001979917950527783, + "loss": 3.8242, + "step": 460 + }, + { + "epoch": 0.68, + "grad_norm": 14.773170471191406, + "learning_rate": 0.00019798243167474207, + "loss": 3.9442, + "step": 461 + }, + { + "epoch": 0.69, + "grad_norm": 20.262290954589844, + "learning_rate": 0.00019797304674122753, + "loss": 3.9303, + "step": 462 + }, + { + "epoch": 0.69, + "grad_norm": 13.257652282714844, + "learning_rate": 0.00019796364025429926, + "loss": 3.8547, + "step": 463 + }, + { + "epoch": 0.69, + "grad_norm": 1.5865988731384277, + "learning_rate": 0.00019795421221602668, + "loss": 3.781, + "step": 464 + }, + { + "epoch": 0.69, + "grad_norm": 4.687791347503662, + "learning_rate": 0.00019794476262848388, + "loss": 3.7924, + "step": 465 + }, + { + "epoch": 0.69, + "grad_norm": 1.8210766315460205, + "learning_rate": 0.0001979352914937497, + "loss": 3.8706, + "step": 466 + }, + { + "epoch": 0.69, + "grad_norm": 10.269219398498535, + "learning_rate": 0.00019792579881390768, + "loss": 3.8314, + "step": 467 + }, + { + "epoch": 0.7, + "grad_norm": 11.758221626281738, + "learning_rate": 0.00019791628459104625, + "loss": 3.8405, + "step": 468 + }, + { + "epoch": 0.7, + "grad_norm": 3.8874833583831787, + "learning_rate": 0.0001979067488272584, + "loss": 3.7889, + "step": 469 + }, + { + "epoch": 0.7, + "grad_norm": 8.348686218261719, + "learning_rate": 0.00019789719152464192, + "loss": 3.815, + "step": 470 + }, + { + "epoch": 0.7, + "grad_norm": 11.831920623779297, + "learning_rate": 0.0001978876126852994, + "loss": 3.8381, + "step": 471 + }, + { + "epoch": 0.7, + "grad_norm": 7.59721040725708, + "learning_rate": 0.00019787801231133814, + "loss": 3.8394, + "step": 472 + }, + { + "epoch": 0.7, + "grad_norm": 4.788586616516113, + "learning_rate": 0.00019786839040487005, + "loss": 3.8047, + "step": 473 + }, + { + "epoch": 0.7, + "grad_norm": 9.041956901550293, + "learning_rate": 0.00019785874696801202, + "loss": 3.7394, + "step": 474 + }, + { + "epoch": 0.71, + "grad_norm": 3.841315746307373, + "learning_rate": 0.00019784908200288546, + "loss": 3.7839, + "step": 475 + }, + { + "epoch": 0.71, + "grad_norm": 6.382961750030518, + "learning_rate": 0.0001978393955116166, + "loss": 3.7502, + "step": 476 + }, + { + "epoch": 0.71, + "grad_norm": 8.265791893005371, + "learning_rate": 0.00019782968749633642, + "loss": 3.7888, + "step": 477 + }, + { + "epoch": 0.71, + "grad_norm": 4.210597038269043, + "learning_rate": 0.00019781995795918063, + "loss": 3.7096, + "step": 478 + }, + { + "epoch": 0.71, + "grad_norm": 10.053391456604004, + "learning_rate": 0.00019781020690228963, + "loss": 3.7276, + "step": 479 + }, + { + "epoch": 0.71, + "grad_norm": 14.174814224243164, + "learning_rate": 0.0001978004343278086, + "loss": 3.891, + "step": 480 + }, + { + "epoch": 0.71, + "grad_norm": 7.2850661277771, + "learning_rate": 0.00019779064023788746, + "loss": 3.8149, + "step": 481 + }, + { + "epoch": 0.72, + "grad_norm": 5.421473503112793, + "learning_rate": 0.00019778082463468085, + "loss": 3.8328, + "step": 482 + }, + { + "epoch": 0.72, + "grad_norm": 8.712848663330078, + "learning_rate": 0.00019777098752034807, + "loss": 3.826, + "step": 483 + }, + { + "epoch": 0.72, + "grad_norm": 4.683780670166016, + "learning_rate": 0.00019776112889705334, + "loss": 3.7366, + "step": 484 + }, + { + "epoch": 0.72, + "grad_norm": 5.190871715545654, + "learning_rate": 0.00019775124876696536, + "loss": 3.7527, + "step": 485 + }, + { + "epoch": 0.72, + "grad_norm": 8.225262641906738, + "learning_rate": 0.00019774134713225778, + "loss": 3.8112, + "step": 486 + }, + { + "epoch": 0.72, + "grad_norm": 2.200317859649658, + "learning_rate": 0.00019773142399510883, + "loss": 3.7121, + "step": 487 + }, + { + "epoch": 0.72, + "grad_norm": 9.15087604522705, + "learning_rate": 0.0001977214793577016, + "loss": 3.7827, + "step": 488 + }, + { + "epoch": 0.73, + "grad_norm": 11.307198524475098, + "learning_rate": 0.0001977115132222238, + "loss": 3.7558, + "step": 489 + }, + { + "epoch": 0.73, + "grad_norm": 8.0049409866333, + "learning_rate": 0.00019770152559086794, + "loss": 3.7133, + "step": 490 + }, + { + "epoch": 0.73, + "grad_norm": 4.207780361175537, + "learning_rate": 0.0001976915164658312, + "loss": 3.7517, + "step": 491 + }, + { + "epoch": 0.73, + "grad_norm": 9.044132232666016, + "learning_rate": 0.00019768148584931554, + "loss": 3.7111, + "step": 492 + }, + { + "epoch": 0.73, + "grad_norm": 6.856382846832275, + "learning_rate": 0.00019767143374352762, + "loss": 3.7271, + "step": 493 + }, + { + "epoch": 0.73, + "grad_norm": 3.427558183670044, + "learning_rate": 0.00019766136015067886, + "loss": 3.7357, + "step": 494 + }, + { + "epoch": 0.74, + "grad_norm": 4.354124069213867, + "learning_rate": 0.00019765126507298534, + "loss": 3.7107, + "step": 495 + }, + { + "epoch": 0.74, + "grad_norm": 3.53426194190979, + "learning_rate": 0.00019764114851266796, + "loss": 3.7585, + "step": 496 + }, + { + "epoch": 0.74, + "grad_norm": 2.1305716037750244, + "learning_rate": 0.00019763101047195225, + "loss": 3.688, + "step": 497 + }, + { + "epoch": 0.74, + "grad_norm": 108.82621002197266, + "learning_rate": 0.00019762085095306856, + "loss": 3.6435, + "step": 498 + }, + { + "epoch": 0.74, + "grad_norm": 8.900164604187012, + "learning_rate": 0.00019761066995825183, + "loss": 3.7853, + "step": 499 + }, + { + "epoch": 0.74, + "grad_norm": 8.347526550292969, + "learning_rate": 0.0001976004674897419, + "loss": 3.7845, + "step": 500 + }, + { + "epoch": 0.74, + "grad_norm": 3.418429374694824, + "learning_rate": 0.00019759024354978322, + "loss": 3.7529, + "step": 501 + }, + { + "epoch": 0.75, + "grad_norm": 14.40571403503418, + "learning_rate": 0.00019757999814062495, + "loss": 3.795, + "step": 502 + }, + { + "epoch": 0.75, + "grad_norm": 21.624300003051758, + "learning_rate": 0.00019756973126452108, + "loss": 4.0035, + "step": 503 + }, + { + "epoch": 0.75, + "grad_norm": 14.849161148071289, + "learning_rate": 0.00019755944292373017, + "loss": 3.9246, + "step": 504 + }, + { + "epoch": 0.75, + "grad_norm": 7.422677993774414, + "learning_rate": 0.00019754913312051568, + "loss": 3.9148, + "step": 505 + }, + { + "epoch": 0.75, + "grad_norm": 11.331315994262695, + "learning_rate": 0.00019753880185714563, + "loss": 3.7997, + "step": 506 + }, + { + "epoch": 0.75, + "grad_norm": 34.16909408569336, + "learning_rate": 0.00019752844913589285, + "loss": 3.7348, + "step": 507 + }, + { + "epoch": 0.75, + "eval_loss": 3.593658447265625, + "eval_runtime": 1156.954, + "eval_samples_per_second": 14.816, + "eval_steps_per_second": 0.927, + "step": 507 + }, + { + "epoch": 0.75, + "grad_norm": 1562.1485595703125, + "learning_rate": 0.00019751807495903484, + "loss": 3.8149, + "step": 508 + }, + { + "epoch": 0.76, + "grad_norm": 12.771462440490723, + "learning_rate": 0.0001975076793288539, + "loss": 3.9449, + "step": 509 + }, + { + "epoch": 0.76, + "grad_norm": 7.921811580657959, + "learning_rate": 0.00019749726224763697, + "loss": 3.9061, + "step": 510 + }, + { + "epoch": 0.76, + "grad_norm": 4.624162197113037, + "learning_rate": 0.00019748682371767575, + "loss": 3.8572, + "step": 511 + }, + { + "epoch": 0.76, + "grad_norm": 33.93694305419922, + "learning_rate": 0.00019747636374126663, + "loss": 3.8013, + "step": 512 + }, + { + "epoch": 0.76, + "grad_norm": 67.59894561767578, + "learning_rate": 0.00019746588232071076, + "loss": 3.9281, + "step": 513 + }, + { + "epoch": 0.76, + "grad_norm": 18.626140594482422, + "learning_rate": 0.00019745537945831397, + "loss": 4.0322, + "step": 514 + }, + { + "epoch": 0.76, + "grad_norm": 9.813076972961426, + "learning_rate": 0.00019744485515638677, + "loss": 4.0781, + "step": 515 + }, + { + "epoch": 0.77, + "grad_norm": 12.949821472167969, + "learning_rate": 0.00019743430941724453, + "loss": 4.0615, + "step": 516 + }, + { + "epoch": 0.77, + "grad_norm": 9.89267635345459, + "learning_rate": 0.00019742374224320717, + "loss": 3.9752, + "step": 517 + }, + { + "epoch": 0.77, + "grad_norm": 4.396897792816162, + "learning_rate": 0.00019741315363659942, + "loss": 3.768, + "step": 518 + }, + { + "epoch": 0.77, + "grad_norm": 3.6423070430755615, + "learning_rate": 0.00019740254359975072, + "loss": 3.7554, + "step": 519 + }, + { + "epoch": 0.77, + "grad_norm": 5.352052688598633, + "learning_rate": 0.0001973919121349952, + "loss": 3.8254, + "step": 520 + }, + { + "epoch": 0.77, + "grad_norm": 4.971579074859619, + "learning_rate": 0.00019738125924467168, + "loss": 3.9223, + "step": 521 + }, + { + "epoch": 0.78, + "grad_norm": 3.2794764041900635, + "learning_rate": 0.00019737058493112376, + "loss": 3.7504, + "step": 522 + }, + { + "epoch": 0.78, + "grad_norm": 3.714524269104004, + "learning_rate": 0.00019735988919669968, + "loss": 3.7884, + "step": 523 + }, + { + "epoch": 0.78, + "grad_norm": 5.291292190551758, + "learning_rate": 0.0001973491720437525, + "loss": 3.8398, + "step": 524 + }, + { + "epoch": 0.78, + "grad_norm": 4.59107780456543, + "learning_rate": 0.00019733843347463984, + "loss": 3.9075, + "step": 525 + }, + { + "epoch": 0.78, + "grad_norm": 6.6806793212890625, + "learning_rate": 0.00019732767349172413, + "loss": 3.7432, + "step": 526 + }, + { + "epoch": 0.78, + "grad_norm": 48.79610824584961, + "learning_rate": 0.00019731689209737257, + "loss": 3.7873, + "step": 527 + }, + { + "epoch": 0.78, + "grad_norm": 7.798208236694336, + "learning_rate": 0.0001973060892939569, + "loss": 3.8176, + "step": 528 + }, + { + "epoch": 0.79, + "grad_norm": 7.387636661529541, + "learning_rate": 0.00019729526508385369, + "loss": 3.892, + "step": 529 + }, + { + "epoch": 0.79, + "grad_norm": 3.599566698074341, + "learning_rate": 0.00019728441946944422, + "loss": 3.754, + "step": 530 + }, + { + "epoch": 0.79, + "grad_norm": 6.45220947265625, + "learning_rate": 0.00019727355245311446, + "loss": 3.7462, + "step": 531 + }, + { + "epoch": 0.79, + "grad_norm": 6.4421563148498535, + "learning_rate": 0.00019726266403725503, + "loss": 3.8014, + "step": 532 + }, + { + "epoch": 0.79, + "grad_norm": 9.638814926147461, + "learning_rate": 0.00019725175422426135, + "loss": 3.9211, + "step": 533 + }, + { + "epoch": 0.79, + "grad_norm": 4.907318592071533, + "learning_rate": 0.00019724082301653345, + "loss": 3.8431, + "step": 534 + }, + { + "epoch": 0.79, + "grad_norm": 3.3494908809661865, + "learning_rate": 0.00019722987041647622, + "loss": 3.8407, + "step": 535 + }, + { + "epoch": 0.8, + "grad_norm": 2.774167060852051, + "learning_rate": 0.00019721889642649912, + "loss": 3.7459, + "step": 536 + }, + { + "epoch": 0.8, + "grad_norm": 4.917426586151123, + "learning_rate": 0.0001972079010490163, + "loss": 3.8871, + "step": 537 + }, + { + "epoch": 0.8, + "grad_norm": 2.627586841583252, + "learning_rate": 0.0001971968842864467, + "loss": 3.7812, + "step": 538 + }, + { + "epoch": 0.8, + "grad_norm": 6.549022197723389, + "learning_rate": 0.00019718584614121396, + "loss": 3.7288, + "step": 539 + }, + { + "epoch": 0.8, + "grad_norm": 13.424997329711914, + "learning_rate": 0.00019717478661574638, + "loss": 3.7931, + "step": 540 + }, + { + "epoch": 0.8, + "grad_norm": 10.896014213562012, + "learning_rate": 0.00019716370571247696, + "loss": 3.8668, + "step": 541 + }, + { + "epoch": 0.8, + "grad_norm": 7.297813415527344, + "learning_rate": 0.00019715260343384347, + "loss": 3.8884, + "step": 542 + }, + { + "epoch": 0.81, + "grad_norm": 3.8043060302734375, + "learning_rate": 0.00019714147978228832, + "loss": 3.8604, + "step": 543 + }, + { + "epoch": 0.81, + "grad_norm": 25.301651000976562, + "learning_rate": 0.0001971303347602586, + "loss": 3.8794, + "step": 544 + }, + { + "epoch": 0.81, + "grad_norm": 215.01597595214844, + "learning_rate": 0.0001971191683702062, + "loss": 3.9304, + "step": 545 + }, + { + "epoch": 0.81, + "grad_norm": 253.02139282226562, + "learning_rate": 0.0001971079806145876, + "loss": 4.3177, + "step": 546 + }, + { + "epoch": 0.81, + "grad_norm": 41.9672966003418, + "learning_rate": 0.00019709677149586409, + "loss": 4.7431, + "step": 547 + }, + { + "epoch": 0.81, + "grad_norm": 34.384029388427734, + "learning_rate": 0.0001970855410165015, + "loss": 5.1589, + "step": 548 + }, + { + "epoch": 0.82, + "grad_norm": 18.074525833129883, + "learning_rate": 0.00019707428917897058, + "loss": 5.4518, + "step": 549 + }, + { + "epoch": 0.82, + "grad_norm": 10.624735832214355, + "learning_rate": 0.0001970630159857466, + "loss": 5.0911, + "step": 550 + }, + { + "epoch": 0.82, + "grad_norm": 15.860513687133789, + "learning_rate": 0.00019705172143930953, + "loss": 4.8221, + "step": 551 + }, + { + "epoch": 0.82, + "grad_norm": 9.580286979675293, + "learning_rate": 0.00019704040554214422, + "loss": 4.7721, + "step": 552 + }, + { + "epoch": 0.82, + "grad_norm": 16.33072280883789, + "learning_rate": 0.00019702906829674002, + "loss": 4.7985, + "step": 553 + }, + { + "epoch": 0.82, + "grad_norm": 15.135660171508789, + "learning_rate": 0.00019701770970559105, + "loss": 4.7531, + "step": 554 + }, + { + "epoch": 0.82, + "grad_norm": 15.646889686584473, + "learning_rate": 0.0001970063297711961, + "loss": 4.6453, + "step": 555 + }, + { + "epoch": 0.83, + "grad_norm": 13.652432441711426, + "learning_rate": 0.00019699492849605874, + "loss": 4.671, + "step": 556 + }, + { + "epoch": 0.83, + "grad_norm": 18.923864364624023, + "learning_rate": 0.00019698350588268713, + "loss": 4.7306, + "step": 557 + }, + { + "epoch": 0.83, + "grad_norm": 15.4611177444458, + "learning_rate": 0.0001969720619335942, + "loss": 4.6082, + "step": 558 + }, + { + "epoch": 0.83, + "grad_norm": 6.054503440856934, + "learning_rate": 0.0001969605966512975, + "loss": 4.4091, + "step": 559 + }, + { + "epoch": 0.83, + "grad_norm": 11.314775466918945, + "learning_rate": 0.00019694911003831933, + "loss": 4.1236, + "step": 560 + }, + { + "epoch": 0.83, + "grad_norm": 12.968482971191406, + "learning_rate": 0.0001969376020971867, + "loss": 4.023, + "step": 561 + }, + { + "epoch": 0.83, + "grad_norm": 8.118325233459473, + "learning_rate": 0.00019692607283043124, + "loss": 3.9331, + "step": 562 + }, + { + "epoch": 0.84, + "grad_norm": 4.455306529998779, + "learning_rate": 0.00019691452224058933, + "loss": 3.9195, + "step": 563 + }, + { + "epoch": 0.84, + "grad_norm": 5.133811950683594, + "learning_rate": 0.000196902950330202, + "loss": 3.8438, + "step": 564 + }, + { + "epoch": 0.84, + "grad_norm": 4.062127590179443, + "learning_rate": 0.00019689135710181503, + "loss": 3.7435, + "step": 565 + }, + { + "epoch": 0.84, + "grad_norm": 4.880388259887695, + "learning_rate": 0.00019687974255797885, + "loss": 3.7828, + "step": 566 + }, + { + "epoch": 0.84, + "grad_norm": 5.3774027824401855, + "learning_rate": 0.00019686810670124856, + "loss": 3.7977, + "step": 567 + }, + { + "epoch": 0.84, + "grad_norm": 5.431920528411865, + "learning_rate": 0.00019685644953418397, + "loss": 3.8113, + "step": 568 + }, + { + "epoch": 0.84, + "grad_norm": 3.2028257846832275, + "learning_rate": 0.00019684477105934956, + "loss": 3.7359, + "step": 569 + }, + { + "epoch": 0.85, + "grad_norm": 3.04168701171875, + "learning_rate": 0.00019683307127931458, + "loss": 3.6924, + "step": 570 + }, + { + "epoch": 0.85, + "grad_norm": 3.9302849769592285, + "learning_rate": 0.00019682135019665286, + "loss": 3.7543, + "step": 571 + }, + { + "epoch": 0.85, + "grad_norm": 3.5611188411712646, + "learning_rate": 0.00019680960781394297, + "loss": 3.751, + "step": 572 + }, + { + "epoch": 0.85, + "grad_norm": 2.1138036251068115, + "learning_rate": 0.00019679784413376811, + "loss": 3.7196, + "step": 573 + }, + { + "epoch": 0.85, + "grad_norm": 3.7343742847442627, + "learning_rate": 0.0001967860591587163, + "loss": 3.6888, + "step": 574 + }, + { + "epoch": 0.85, + "grad_norm": 3.8373682498931885, + "learning_rate": 0.00019677425289138007, + "loss": 3.6819, + "step": 575 + }, + { + "epoch": 0.86, + "grad_norm": 3.580359935760498, + "learning_rate": 0.00019676242533435678, + "loss": 3.662, + "step": 576 + }, + { + "epoch": 0.86, + "grad_norm": 3.4737274646759033, + "learning_rate": 0.00019675057649024837, + "loss": 3.6293, + "step": 577 + }, + { + "epoch": 0.86, + "grad_norm": 2.3330490589141846, + "learning_rate": 0.00019673870636166156, + "loss": 3.6554, + "step": 578 + }, + { + "epoch": 0.86, + "grad_norm": 4.6455512046813965, + "learning_rate": 0.00019672681495120763, + "loss": 3.6991, + "step": 579 + }, + { + "epoch": 0.86, + "grad_norm": 2.803642988204956, + "learning_rate": 0.00019671490226150264, + "loss": 3.6473, + "step": 580 + }, + { + "epoch": 0.86, + "grad_norm": 5.52310848236084, + "learning_rate": 0.00019670296829516732, + "loss": 3.6435, + "step": 581 + }, + { + "epoch": 0.86, + "grad_norm": 5.066474914550781, + "learning_rate": 0.00019669101305482705, + "loss": 3.6769, + "step": 582 + }, + { + "epoch": 0.87, + "grad_norm": 2.860041379928589, + "learning_rate": 0.00019667903654311194, + "loss": 3.6205, + "step": 583 + }, + { + "epoch": 0.87, + "grad_norm": 2.413363456726074, + "learning_rate": 0.00019666703876265666, + "loss": 3.6287, + "step": 584 + }, + { + "epoch": 0.87, + "grad_norm": 4.433047771453857, + "learning_rate": 0.0001966550197161007, + "loss": 3.6912, + "step": 585 + }, + { + "epoch": 0.87, + "grad_norm": 117.41354370117188, + "learning_rate": 0.00019664297940608812, + "loss": 3.6996, + "step": 586 + }, + { + "epoch": 0.87, + "grad_norm": 2.949551582336426, + "learning_rate": 0.0001966309178352678, + "loss": 3.6726, + "step": 587 + }, + { + "epoch": 0.87, + "grad_norm": 6.152908802032471, + "learning_rate": 0.0001966188350062931, + "loss": 3.6525, + "step": 588 + }, + { + "epoch": 0.87, + "grad_norm": 6.42512321472168, + "learning_rate": 0.00019660673092182225, + "loss": 3.6739, + "step": 589 + }, + { + "epoch": 0.88, + "grad_norm": 3.4023945331573486, + "learning_rate": 0.000196594605584518, + "loss": 3.6757, + "step": 590 + }, + { + "epoch": 0.88, + "grad_norm": 4.428458213806152, + "learning_rate": 0.00019658245899704789, + "loss": 3.6438, + "step": 591 + }, + { + "epoch": 0.88, + "grad_norm": 4.402431488037109, + "learning_rate": 0.00019657029116208405, + "loss": 3.6902, + "step": 592 + }, + { + "epoch": 0.88, + "grad_norm": 4.742648124694824, + "learning_rate": 0.00019655810208230333, + "loss": 3.5998, + "step": 593 + }, + { + "epoch": 0.88, + "grad_norm": 1.4105639457702637, + "learning_rate": 0.00019654589176038728, + "loss": 3.5836, + "step": 594 + }, + { + "epoch": 0.88, + "grad_norm": 5.09644079208374, + "learning_rate": 0.00019653366019902206, + "loss": 3.616, + "step": 595 + }, + { + "epoch": 0.89, + "grad_norm": 5.025322914123535, + "learning_rate": 0.00019652140740089854, + "loss": 3.6783, + "step": 596 + }, + { + "epoch": 0.89, + "grad_norm": 4.536464691162109, + "learning_rate": 0.00019650913336871222, + "loss": 3.6274, + "step": 597 + }, + { + "epoch": 0.89, + "grad_norm": 3.1868488788604736, + "learning_rate": 0.00019649683810516336, + "loss": 3.6113, + "step": 598 + }, + { + "epoch": 0.89, + "grad_norm": 4.735551834106445, + "learning_rate": 0.00019648452161295682, + "loss": 3.5986, + "step": 599 + }, + { + "epoch": 0.89, + "grad_norm": 4.411576747894287, + "learning_rate": 0.0001964721838948021, + "loss": 3.5865, + "step": 600 + }, + { + "epoch": 0.89, + "grad_norm": 2.665224075317383, + "learning_rate": 0.00019645982495341348, + "loss": 3.5546, + "step": 601 + }, + { + "epoch": 0.89, + "grad_norm": 4.546963691711426, + "learning_rate": 0.00019644744479150981, + "loss": 3.6332, + "step": 602 + }, + { + "epoch": 0.9, + "grad_norm": 1.664005160331726, + "learning_rate": 0.00019643504341181466, + "loss": 3.6299, + "step": 603 + }, + { + "epoch": 0.9, + "grad_norm": 2.6972901821136475, + "learning_rate": 0.0001964226208170562, + "loss": 3.6158, + "step": 604 + }, + { + "epoch": 0.9, + "grad_norm": 3.175882339477539, + "learning_rate": 0.00019641017700996736, + "loss": 3.5895, + "step": 605 + }, + { + "epoch": 0.9, + "grad_norm": 3.3170359134674072, + "learning_rate": 0.00019639771199328574, + "loss": 3.5989, + "step": 606 + }, + { + "epoch": 0.9, + "grad_norm": 3.245979070663452, + "learning_rate": 0.00019638522576975346, + "loss": 3.5763, + "step": 607 + }, + { + "epoch": 0.9, + "grad_norm": 1.0223109722137451, + "learning_rate": 0.00019637271834211748, + "loss": 3.5668, + "step": 608 + }, + { + "epoch": 0.9, + "grad_norm": 1.9072734117507935, + "learning_rate": 0.00019636018971312932, + "loss": 3.5738, + "step": 609 + }, + { + "epoch": 0.91, + "grad_norm": 2.925701141357422, + "learning_rate": 0.00019634763988554522, + "loss": 3.616, + "step": 610 + }, + { + "epoch": 0.91, + "grad_norm": 2.6939620971679688, + "learning_rate": 0.00019633506886212603, + "loss": 3.5764, + "step": 611 + }, + { + "epoch": 0.91, + "grad_norm": 2.020845413208008, + "learning_rate": 0.0001963224766456373, + "loss": 3.5868, + "step": 612 + }, + { + "epoch": 0.91, + "grad_norm": 3.805741548538208, + "learning_rate": 0.0001963098632388492, + "loss": 3.5979, + "step": 613 + }, + { + "epoch": 0.91, + "grad_norm": 1.6603795289993286, + "learning_rate": 0.00019629722864453663, + "loss": 3.5792, + "step": 614 + }, + { + "epoch": 0.91, + "grad_norm": 8.938972473144531, + "learning_rate": 0.00019628457286547914, + "loss": 3.5978, + "step": 615 + }, + { + "epoch": 0.91, + "grad_norm": 9.437355041503906, + "learning_rate": 0.0001962718959044609, + "loss": 3.6115, + "step": 616 + }, + { + "epoch": 0.92, + "grad_norm": 1.7519173622131348, + "learning_rate": 0.0001962591977642707, + "loss": 3.5745, + "step": 617 + }, + { + "epoch": 0.92, + "grad_norm": 8.598402976989746, + "learning_rate": 0.00019624647844770213, + "loss": 3.5939, + "step": 618 + }, + { + "epoch": 0.92, + "grad_norm": 11.709813117980957, + "learning_rate": 0.0001962337379575533, + "loss": 3.7499, + "step": 619 + }, + { + "epoch": 0.92, + "grad_norm": 8.366117477416992, + "learning_rate": 0.00019622097629662708, + "loss": 3.5727, + "step": 620 + }, + { + "epoch": 0.92, + "grad_norm": 2.4304378032684326, + "learning_rate": 0.0001962081934677309, + "loss": 3.6257, + "step": 621 + }, + { + "epoch": 0.92, + "grad_norm": 6.7245025634765625, + "learning_rate": 0.00019619538947367692, + "loss": 3.5817, + "step": 622 + }, + { + "epoch": 0.93, + "grad_norm": 2.964221239089966, + "learning_rate": 0.00019618256431728194, + "loss": 3.5903, + "step": 623 + }, + { + "epoch": 0.93, + "grad_norm": 5.846991539001465, + "learning_rate": 0.00019616971800136734, + "loss": 3.5198, + "step": 624 + }, + { + "epoch": 0.93, + "grad_norm": 6.497872829437256, + "learning_rate": 0.00019615685052875933, + "loss": 3.5556, + "step": 625 + }, + { + "epoch": 0.93, + "grad_norm": 1.711037039756775, + "learning_rate": 0.00019614396190228864, + "loss": 3.5108, + "step": 626 + }, + { + "epoch": 0.93, + "grad_norm": 9.816991806030273, + "learning_rate": 0.0001961310521247906, + "loss": 3.5854, + "step": 627 + }, + { + "epoch": 0.93, + "grad_norm": 12.794452667236328, + "learning_rate": 0.00019611812119910537, + "loss": 3.6549, + "step": 628 + }, + { + "epoch": 0.93, + "grad_norm": 6.7332916259765625, + "learning_rate": 0.0001961051691280776, + "loss": 3.6605, + "step": 629 + }, + { + "epoch": 0.94, + "grad_norm": 6.012147426605225, + "learning_rate": 0.00019609219591455675, + "loss": 3.6014, + "step": 630 + }, + { + "epoch": 0.94, + "grad_norm": 6.8346381187438965, + "learning_rate": 0.00019607920156139671, + "loss": 3.5713, + "step": 631 + }, + { + "epoch": 0.94, + "grad_norm": 3.088141679763794, + "learning_rate": 0.0001960661860714563, + "loss": 3.5767, + "step": 632 + }, + { + "epoch": 0.94, + "grad_norm": 8.472504615783691, + "learning_rate": 0.0001960531494475987, + "loss": 3.6236, + "step": 633 + }, + { + "epoch": 0.94, + "grad_norm": 9.718921661376953, + "learning_rate": 0.00019604009169269193, + "loss": 3.5673, + "step": 634 + }, + { + "epoch": 0.94, + "grad_norm": 2.083310127258301, + "learning_rate": 0.00019602701280960866, + "loss": 3.5786, + "step": 635 + }, + { + "epoch": 0.94, + "grad_norm": 8.658992767333984, + "learning_rate": 0.0001960139128012261, + "loss": 3.6109, + "step": 636 + }, + { + "epoch": 0.95, + "grad_norm": 10.503134727478027, + "learning_rate": 0.0001960007916704262, + "loss": 3.6035, + "step": 637 + }, + { + "epoch": 0.95, + "grad_norm": 7.250131130218506, + "learning_rate": 0.00019598764942009548, + "loss": 3.6211, + "step": 638 + }, + { + "epoch": 0.95, + "grad_norm": 4.671655178070068, + "learning_rate": 0.00019597448605312515, + "loss": 3.5175, + "step": 639 + }, + { + "epoch": 0.95, + "grad_norm": 9.631921768188477, + "learning_rate": 0.00019596130157241112, + "loss": 3.6144, + "step": 640 + }, + { + "epoch": 0.95, + "grad_norm": 5.835899353027344, + "learning_rate": 0.00019594809598085381, + "loss": 3.5662, + "step": 641 + }, + { + "epoch": 0.95, + "grad_norm": 5.0995893478393555, + "learning_rate": 0.00019593486928135841, + "loss": 3.5438, + "step": 642 + }, + { + "epoch": 0.95, + "grad_norm": 6.497475624084473, + "learning_rate": 0.00019592162147683473, + "loss": 3.536, + "step": 643 + }, + { + "epoch": 0.96, + "grad_norm": 2.704535722732544, + "learning_rate": 0.00019590835257019714, + "loss": 3.5268, + "step": 644 + }, + { + "epoch": 0.96, + "grad_norm": 8.847089767456055, + "learning_rate": 0.00019589506256436473, + "loss": 3.6095, + "step": 645 + }, + { + "epoch": 0.96, + "grad_norm": 10.879183769226074, + "learning_rate": 0.00019588175146226122, + "loss": 3.657, + "step": 646 + }, + { + "epoch": 0.96, + "grad_norm": 4.468491077423096, + "learning_rate": 0.00019586841926681495, + "loss": 3.5589, + "step": 647 + }, + { + "epoch": 0.96, + "grad_norm": 6.55201530456543, + "learning_rate": 0.00019585506598095897, + "loss": 3.5587, + "step": 648 + }, + { + "epoch": 0.96, + "grad_norm": 8.522590637207031, + "learning_rate": 0.0001958416916076308, + "loss": 3.5354, + "step": 649 + }, + { + "epoch": 0.97, + "grad_norm": 4.8020806312561035, + "learning_rate": 0.00019582829614977283, + "loss": 3.506, + "step": 650 + }, + { + "epoch": 0.97, + "grad_norm": 7.987406253814697, + "learning_rate": 0.00019581487961033188, + "loss": 3.5615, + "step": 651 + }, + { + "epoch": 0.97, + "grad_norm": 11.70871353149414, + "learning_rate": 0.00019580144199225953, + "loss": 3.5993, + "step": 652 + }, + { + "epoch": 0.97, + "grad_norm": 4.69012975692749, + "learning_rate": 0.00019578798329851203, + "loss": 3.542, + "step": 653 + }, + { + "epoch": 0.97, + "grad_norm": 6.301438331604004, + "learning_rate": 0.0001957745035320501, + "loss": 3.5313, + "step": 654 + }, + { + "epoch": 0.97, + "grad_norm": 8.368910789489746, + "learning_rate": 0.0001957610026958393, + "loss": 3.5855, + "step": 655 + }, + { + "epoch": 0.97, + "grad_norm": 5.793453693389893, + "learning_rate": 0.00019574748079284962, + "loss": 3.5403, + "step": 656 + }, + { + "epoch": 0.98, + "grad_norm": 4.990311622619629, + "learning_rate": 0.00019573393782605582, + "loss": 3.5196, + "step": 657 + }, + { + "epoch": 0.98, + "grad_norm": 9.033957481384277, + "learning_rate": 0.0001957203737984373, + "loss": 3.528, + "step": 658 + }, + { + "epoch": 0.98, + "grad_norm": 3.766148805618286, + "learning_rate": 0.00019570678871297804, + "loss": 3.5015, + "step": 659 + }, + { + "epoch": 0.98, + "grad_norm": 7.3363518714904785, + "learning_rate": 0.00019569318257266666, + "loss": 3.5311, + "step": 660 + }, + { + "epoch": 0.98, + "grad_norm": 9.119929313659668, + "learning_rate": 0.00019567955538049643, + "loss": 3.5767, + "step": 661 + }, + { + "epoch": 0.98, + "grad_norm": 5.856534957885742, + "learning_rate": 0.00019566590713946522, + "loss": 3.5836, + "step": 662 + }, + { + "epoch": 0.98, + "grad_norm": 4.941824436187744, + "learning_rate": 0.00019565223785257557, + "loss": 3.5366, + "step": 663 + }, + { + "epoch": 0.99, + "grad_norm": 8.7649507522583, + "learning_rate": 0.00019563854752283465, + "loss": 3.5562, + "step": 664 + }, + { + "epoch": 0.99, + "grad_norm": 3.0613012313842773, + "learning_rate": 0.00019562483615325418, + "loss": 3.5715, + "step": 665 + }, + { + "epoch": 0.99, + "grad_norm": 6.858389377593994, + "learning_rate": 0.00019561110374685062, + "loss": 3.4948, + "step": 666 + }, + { + "epoch": 0.99, + "grad_norm": 8.862404823303223, + "learning_rate": 0.00019559735030664496, + "loss": 3.5523, + "step": 667 + }, + { + "epoch": 0.99, + "grad_norm": 5.061213970184326, + "learning_rate": 0.00019558357583566297, + "loss": 3.4475, + "step": 668 + }, + { + "epoch": 0.99, + "grad_norm": 7.312166690826416, + "learning_rate": 0.0001955697803369348, + "loss": 3.5274, + "step": 669 + }, + { + "epoch": 0.99, + "grad_norm": 11.606597900390625, + "learning_rate": 0.00019555596381349547, + "loss": 3.5426, + "step": 670 + }, + { + "epoch": 1.0, + "grad_norm": 5.651201248168945, + "learning_rate": 0.0001955421262683845, + "loss": 3.533, + "step": 671 + }, + { + "epoch": 1.0, + "grad_norm": 5.392771244049072, + "learning_rate": 0.00019552826770464598, + "loss": 3.4873, + "step": 672 + }, + { + "epoch": 1.0, + "grad_norm": 7.194543361663818, + "learning_rate": 0.00019551438812532882, + "loss": 3.5313, + "step": 673 + }, + { + "epoch": 1.0, + "grad_norm": 3.8677308559417725, + "learning_rate": 0.00019550048753348638, + "loss": 3.5539, + "step": 674 + }, + { + "epoch": 1.0, + "grad_norm": 8.319040298461914, + "learning_rate": 0.0001954865659321767, + "loss": 3.5204, + "step": 675 + }, + { + "epoch": 1.0, + "grad_norm": 12.673453330993652, + "learning_rate": 0.00019547262332446241, + "loss": 3.4573, + "step": 676 + }, + { + "epoch": 1.0, + "eval_loss": 3.3125858306884766, + "eval_runtime": 1160.3441, + "eval_samples_per_second": 14.773, + "eval_steps_per_second": 0.924, + "step": 676 + }, + { + "epoch": 1.01, + "grad_norm": 6.508894920349121, + "learning_rate": 0.00019545865971341085, + "loss": 3.5108, + "step": 677 + }, + { + "epoch": 1.01, + "grad_norm": 5.700242519378662, + "learning_rate": 0.00019544467510209388, + "loss": 3.4928, + "step": 678 + }, + { + "epoch": 1.01, + "grad_norm": 7.889042854309082, + "learning_rate": 0.00019543066949358803, + "loss": 3.5309, + "step": 679 + }, + { + "epoch": 1.01, + "grad_norm": 4.5509724617004395, + "learning_rate": 0.0001954166428909744, + "loss": 3.5496, + "step": 680 + }, + { + "epoch": 1.01, + "grad_norm": 7.286065578460693, + "learning_rate": 0.00019540259529733883, + "loss": 3.5685, + "step": 681 + }, + { + "epoch": 1.01, + "grad_norm": 10.92095947265625, + "learning_rate": 0.00019538852671577165, + "loss": 3.5438, + "step": 682 + }, + { + "epoch": 1.01, + "grad_norm": 4.9438557624816895, + "learning_rate": 0.00019537443714936787, + "loss": 3.4908, + "step": 683 + }, + { + "epoch": 1.0, + "grad_norm": 6.148499011993408, + "learning_rate": 0.00019536032660122707, + "loss": 3.5438, + "step": 684 + }, + { + "epoch": 1.0, + "grad_norm": 8.24433708190918, + "learning_rate": 0.00019534619507445352, + "loss": 3.6016, + "step": 685 + }, + { + "epoch": 1.0, + "grad_norm": 5.07331657409668, + "learning_rate": 0.000195332042572156, + "loss": 3.6093, + "step": 686 + }, + { + "epoch": 1.01, + "grad_norm": 7.187658309936523, + "learning_rate": 0.00019531786909744803, + "loss": 3.5025, + "step": 687 + }, + { + "epoch": 1.01, + "grad_norm": 10.835110664367676, + "learning_rate": 0.0001953036746534476, + "loss": 3.5561, + "step": 688 + }, + { + "epoch": 1.01, + "grad_norm": 4.704497814178467, + "learning_rate": 0.00019528945924327751, + "loss": 3.5617, + "step": 689 + }, + { + "epoch": 1.01, + "grad_norm": 6.398228168487549, + "learning_rate": 0.00019527522287006495, + "loss": 3.5302, + "step": 690 + }, + { + "epoch": 1.01, + "grad_norm": 8.401050567626953, + "learning_rate": 0.00019526096553694187, + "loss": 3.5882, + "step": 691 + }, + { + "epoch": 1.01, + "grad_norm": 4.914839267730713, + "learning_rate": 0.00019524668724704478, + "loss": 3.4931, + "step": 692 + }, + { + "epoch": 1.01, + "grad_norm": 6.374270915985107, + "learning_rate": 0.00019523238800351483, + "loss": 3.4807, + "step": 693 + }, + { + "epoch": 1.02, + "grad_norm": 9.898938179016113, + "learning_rate": 0.0001952180678094977, + "loss": 3.5138, + "step": 694 + }, + { + "epoch": 1.02, + "grad_norm": 4.583561897277832, + "learning_rate": 0.00019520372666814378, + "loss": 3.4739, + "step": 695 + }, + { + "epoch": 1.02, + "grad_norm": 6.307168483734131, + "learning_rate": 0.00019518936458260807, + "loss": 3.5473, + "step": 696 + }, + { + "epoch": 1.02, + "grad_norm": 8.191679000854492, + "learning_rate": 0.00019517498155605003, + "loss": 3.5002, + "step": 697 + }, + { + "epoch": 1.02, + "grad_norm": 5.078609466552734, + "learning_rate": 0.0001951605775916339, + "loss": 3.4731, + "step": 698 + }, + { + "epoch": 1.02, + "grad_norm": 6.6793060302734375, + "learning_rate": 0.00019514615269252845, + "loss": 3.5543, + "step": 699 + }, + { + "epoch": 1.02, + "grad_norm": 10.510246276855469, + "learning_rate": 0.00019513170686190706, + "loss": 3.5589, + "step": 700 + }, + { + "epoch": 1.03, + "grad_norm": 4.769564151763916, + "learning_rate": 0.00019511724010294766, + "loss": 3.4587, + "step": 701 + }, + { + "epoch": 1.03, + "grad_norm": 6.157220363616943, + "learning_rate": 0.00019510275241883292, + "loss": 3.4658, + "step": 702 + }, + { + "epoch": 1.03, + "grad_norm": 8.317898750305176, + "learning_rate": 0.00019508824381274996, + "loss": 3.5147, + "step": 703 + }, + { + "epoch": 1.03, + "grad_norm": 4.906617164611816, + "learning_rate": 0.00019507371428789064, + "loss": 3.5319, + "step": 704 + }, + { + "epoch": 1.03, + "grad_norm": 6.44058084487915, + "learning_rate": 0.00019505916384745134, + "loss": 3.5268, + "step": 705 + }, + { + "epoch": 1.03, + "grad_norm": 10.473220825195312, + "learning_rate": 0.00019504459249463304, + "loss": 3.535, + "step": 706 + }, + { + "epoch": 1.04, + "grad_norm": 4.5666656494140625, + "learning_rate": 0.00019503000023264136, + "loss": 3.4976, + "step": 707 + }, + { + "epoch": 1.04, + "grad_norm": 6.2709221839904785, + "learning_rate": 0.0001950153870646865, + "loss": 3.5246, + "step": 708 + }, + { + "epoch": 1.04, + "grad_norm": 8.397104263305664, + "learning_rate": 0.00019500075299398324, + "loss": 3.5369, + "step": 709 + }, + { + "epoch": 1.04, + "grad_norm": 4.958198547363281, + "learning_rate": 0.00019498609802375097, + "loss": 3.4644, + "step": 710 + }, + { + "epoch": 1.04, + "grad_norm": 6.521581172943115, + "learning_rate": 0.00019497142215721373, + "loss": 3.4931, + "step": 711 + }, + { + "epoch": 1.04, + "grad_norm": 10.69038200378418, + "learning_rate": 0.00019495672539760007, + "loss": 3.4746, + "step": 712 + }, + { + "epoch": 1.04, + "grad_norm": 5.08820915222168, + "learning_rate": 0.0001949420077481432, + "loss": 3.4963, + "step": 713 + }, + { + "epoch": 1.05, + "grad_norm": 5.867018222808838, + "learning_rate": 0.00019492726921208092, + "loss": 3.4735, + "step": 714 + }, + { + "epoch": 1.05, + "grad_norm": 7.802560329437256, + "learning_rate": 0.00019491250979265554, + "loss": 3.514, + "step": 715 + }, + { + "epoch": 1.05, + "grad_norm": 4.548806190490723, + "learning_rate": 0.0001948977294931141, + "loss": 3.4695, + "step": 716 + }, + { + "epoch": 1.05, + "grad_norm": 7.293185710906982, + "learning_rate": 0.00019488292831670818, + "loss": 3.5301, + "step": 717 + }, + { + "epoch": 1.05, + "grad_norm": 10.711014747619629, + "learning_rate": 0.00019486810626669388, + "loss": 3.5166, + "step": 718 + }, + { + "epoch": 1.05, + "grad_norm": 5.207925796508789, + "learning_rate": 0.00019485326334633198, + "loss": 3.4741, + "step": 719 + }, + { + "epoch": 1.05, + "grad_norm": 5.915352821350098, + "learning_rate": 0.00019483839955888784, + "loss": 3.4837, + "step": 720 + }, + { + "epoch": 1.06, + "grad_norm": 8.018194198608398, + "learning_rate": 0.00019482351490763135, + "loss": 3.4991, + "step": 721 + }, + { + "epoch": 1.06, + "grad_norm": 4.433293342590332, + "learning_rate": 0.00019480860939583705, + "loss": 3.417, + "step": 722 + }, + { + "epoch": 1.06, + "grad_norm": 6.9495038986206055, + "learning_rate": 0.0001947936830267841, + "loss": 3.5061, + "step": 723 + }, + { + "epoch": 1.06, + "grad_norm": 10.787668228149414, + "learning_rate": 0.00019477873580375614, + "loss": 3.5036, + "step": 724 + }, + { + "epoch": 1.06, + "grad_norm": 5.106097221374512, + "learning_rate": 0.0001947637677300415, + "loss": 3.4732, + "step": 725 + }, + { + "epoch": 1.06, + "grad_norm": 5.890613555908203, + "learning_rate": 0.00019474877880893302, + "loss": 3.5222, + "step": 726 + }, + { + "epoch": 1.06, + "grad_norm": 7.86448335647583, + "learning_rate": 0.0001947337690437282, + "loss": 3.4886, + "step": 727 + }, + { + "epoch": 1.07, + "grad_norm": 4.345864772796631, + "learning_rate": 0.00019471873843772904, + "loss": 3.4902, + "step": 728 + }, + { + "epoch": 1.07, + "grad_norm": 7.271783351898193, + "learning_rate": 0.00019470368699424218, + "loss": 3.4832, + "step": 729 + }, + { + "epoch": 1.07, + "grad_norm": 11.227890968322754, + "learning_rate": 0.0001946886147165789, + "loss": 3.4815, + "step": 730 + }, + { + "epoch": 1.07, + "grad_norm": 5.0216264724731445, + "learning_rate": 0.00019467352160805498, + "loss": 3.4912, + "step": 731 + }, + { + "epoch": 1.07, + "grad_norm": 5.998202800750732, + "learning_rate": 0.00019465840767199073, + "loss": 3.4653, + "step": 732 + }, + { + "epoch": 1.07, + "grad_norm": 7.844666004180908, + "learning_rate": 0.00019464327291171117, + "loss": 3.5708, + "step": 733 + }, + { + "epoch": 1.08, + "grad_norm": 4.441153526306152, + "learning_rate": 0.00019462811733054588, + "loss": 3.479, + "step": 734 + }, + { + "epoch": 1.08, + "grad_norm": 7.58824348449707, + "learning_rate": 0.0001946129409318289, + "loss": 3.4787, + "step": 735 + }, + { + "epoch": 1.08, + "grad_norm": 11.293692588806152, + "learning_rate": 0.00019459774371889901, + "loss": 3.5108, + "step": 736 + }, + { + "epoch": 1.08, + "grad_norm": 6.113391399383545, + "learning_rate": 0.00019458252569509946, + "loss": 3.4456, + "step": 737 + }, + { + "epoch": 1.08, + "grad_norm": 4.897597312927246, + "learning_rate": 0.00019456728686377813, + "loss": 3.5001, + "step": 738 + }, + { + "epoch": 1.08, + "grad_norm": 6.808078765869141, + "learning_rate": 0.00019455202722828743, + "loss": 3.5318, + "step": 739 + }, + { + "epoch": 1.08, + "grad_norm": 2.7436025142669678, + "learning_rate": 0.0001945367467919844, + "loss": 3.452, + "step": 740 + }, + { + "epoch": 1.09, + "grad_norm": 10.119528770446777, + "learning_rate": 0.00019452144555823065, + "loss": 3.5012, + "step": 741 + }, + { + "epoch": 1.09, + "grad_norm": 15.063286781311035, + "learning_rate": 0.00019450612353039235, + "loss": 3.5615, + "step": 742 + }, + { + "epoch": 1.09, + "grad_norm": 9.542817115783691, + "learning_rate": 0.0001944907807118402, + "loss": 3.5215, + "step": 743 + }, + { + "epoch": 1.09, + "grad_norm": 3.414539098739624, + "learning_rate": 0.00019447541710594952, + "loss": 3.4854, + "step": 744 + }, + { + "epoch": 1.09, + "grad_norm": 5.853447437286377, + "learning_rate": 0.00019446003271610026, + "loss": 3.4727, + "step": 745 + }, + { + "epoch": 1.09, + "grad_norm": 2.5720930099487305, + "learning_rate": 0.00019444462754567682, + "loss": 3.4853, + "step": 746 + }, + { + "epoch": 1.09, + "grad_norm": 8.554189682006836, + "learning_rate": 0.0001944292015980683, + "loss": 3.4712, + "step": 747 + }, + { + "epoch": 1.1, + "grad_norm": 11.265769958496094, + "learning_rate": 0.00019441375487666822, + "loss": 3.5236, + "step": 748 + }, + { + "epoch": 1.1, + "grad_norm": 4.981095314025879, + "learning_rate": 0.00019439828738487483, + "loss": 3.5258, + "step": 749 + }, + { + "epoch": 1.1, + "grad_norm": 5.837092876434326, + "learning_rate": 0.00019438279912609082, + "loss": 3.5503, + "step": 750 + }, + { + "epoch": 1.1, + "grad_norm": 8.067007064819336, + "learning_rate": 0.00019436729010372355, + "loss": 3.5129, + "step": 751 + }, + { + "epoch": 1.1, + "grad_norm": 4.977865219116211, + "learning_rate": 0.00019435176032118486, + "loss": 3.4686, + "step": 752 + }, + { + "epoch": 1.1, + "grad_norm": 6.810112953186035, + "learning_rate": 0.00019433620978189122, + "loss": 3.4424, + "step": 753 + }, + { + "epoch": 1.1, + "grad_norm": 10.288273811340332, + "learning_rate": 0.00019432063848926365, + "loss": 3.4829, + "step": 754 + }, + { + "epoch": 1.11, + "grad_norm": 4.5541839599609375, + "learning_rate": 0.0001943050464467277, + "loss": 3.4592, + "step": 755 + }, + { + "epoch": 1.11, + "grad_norm": 5.883044719696045, + "learning_rate": 0.0001942894336577136, + "loss": 3.5044, + "step": 756 + }, + { + "epoch": 1.11, + "grad_norm": 7.786341667175293, + "learning_rate": 0.00019427380012565593, + "loss": 3.5129, + "step": 757 + }, + { + "epoch": 1.11, + "grad_norm": 4.628783226013184, + "learning_rate": 0.00019425814585399405, + "loss": 3.43, + "step": 758 + }, + { + "epoch": 1.11, + "grad_norm": 6.876189708709717, + "learning_rate": 0.00019424247084617177, + "loss": 3.4606, + "step": 759 + }, + { + "epoch": 1.11, + "grad_norm": 10.343745231628418, + "learning_rate": 0.00019422677510563748, + "loss": 3.5079, + "step": 760 + }, + { + "epoch": 1.12, + "grad_norm": 4.678600311279297, + "learning_rate": 0.00019421105863584413, + "loss": 3.5135, + "step": 761 + }, + { + "epoch": 1.12, + "grad_norm": 6.044957160949707, + "learning_rate": 0.0001941953214402493, + "loss": 3.5067, + "step": 762 + }, + { + "epoch": 1.12, + "grad_norm": 7.979968070983887, + "learning_rate": 0.00019417956352231502, + "loss": 3.4955, + "step": 763 + }, + { + "epoch": 1.12, + "grad_norm": 4.172381401062012, + "learning_rate": 0.00019416378488550787, + "loss": 3.4781, + "step": 764 + }, + { + "epoch": 1.12, + "grad_norm": 7.783565998077393, + "learning_rate": 0.00019414798553329916, + "loss": 3.4649, + "step": 765 + }, + { + "epoch": 1.12, + "grad_norm": 11.02284049987793, + "learning_rate": 0.00019413216546916457, + "loss": 3.5242, + "step": 766 + }, + { + "epoch": 1.12, + "grad_norm": 5.28484582901001, + "learning_rate": 0.00019411632469658443, + "loss": 3.4645, + "step": 767 + }, + { + "epoch": 1.13, + "grad_norm": 6.049994945526123, + "learning_rate": 0.00019410046321904356, + "loss": 3.4601, + "step": 768 + }, + { + "epoch": 1.13, + "grad_norm": 7.867702960968018, + "learning_rate": 0.00019408458104003143, + "loss": 3.4853, + "step": 769 + }, + { + "epoch": 1.13, + "grad_norm": 5.088557720184326, + "learning_rate": 0.00019406867816304204, + "loss": 3.5329, + "step": 770 + }, + { + "epoch": 1.13, + "grad_norm": 6.444199085235596, + "learning_rate": 0.00019405275459157383, + "loss": 3.4624, + "step": 771 + }, + { + "epoch": 1.13, + "grad_norm": 9.823208808898926, + "learning_rate": 0.00019403681032912994, + "loss": 3.4496, + "step": 772 + }, + { + "epoch": 1.13, + "grad_norm": 4.309892654418945, + "learning_rate": 0.00019402084537921798, + "loss": 3.4607, + "step": 773 + }, + { + "epoch": 1.13, + "grad_norm": 6.139223575592041, + "learning_rate": 0.00019400485974535014, + "loss": 3.4246, + "step": 774 + }, + { + "epoch": 1.14, + "grad_norm": 8.018882751464844, + "learning_rate": 0.00019398885343104315, + "loss": 3.4731, + "step": 775 + }, + { + "epoch": 1.14, + "grad_norm": 4.7042555809021, + "learning_rate": 0.00019397282643981826, + "loss": 3.4863, + "step": 776 + }, + { + "epoch": 1.14, + "grad_norm": 6.583994388580322, + "learning_rate": 0.00019395677877520137, + "loss": 3.4354, + "step": 777 + }, + { + "epoch": 1.14, + "grad_norm": 9.998130798339844, + "learning_rate": 0.00019394071044072283, + "loss": 3.4882, + "step": 778 + }, + { + "epoch": 1.14, + "grad_norm": 4.592165946960449, + "learning_rate": 0.00019392462143991753, + "loss": 3.4696, + "step": 779 + }, + { + "epoch": 1.14, + "grad_norm": 6.2591681480407715, + "learning_rate": 0.00019390851177632497, + "loss": 3.4393, + "step": 780 + }, + { + "epoch": 1.14, + "grad_norm": 8.170223236083984, + "learning_rate": 0.00019389238145348916, + "loss": 3.421, + "step": 781 + }, + { + "epoch": 1.15, + "grad_norm": 5.1918559074401855, + "learning_rate": 0.00019387623047495867, + "loss": 3.452, + "step": 782 + }, + { + "epoch": 1.15, + "grad_norm": 6.183322429656982, + "learning_rate": 0.0001938600588442866, + "loss": 3.5035, + "step": 783 + }, + { + "epoch": 1.15, + "grad_norm": 9.819025039672852, + "learning_rate": 0.00019384386656503064, + "loss": 3.4897, + "step": 784 + }, + { + "epoch": 1.15, + "grad_norm": 4.417552471160889, + "learning_rate": 0.00019382765364075292, + "loss": 3.5126, + "step": 785 + }, + { + "epoch": 1.15, + "grad_norm": 6.277004241943359, + "learning_rate": 0.0001938114200750202, + "loss": 3.4848, + "step": 786 + }, + { + "epoch": 1.15, + "grad_norm": 8.565668106079102, + "learning_rate": 0.0001937951658714038, + "loss": 3.5328, + "step": 787 + }, + { + "epoch": 1.16, + "grad_norm": 5.04703426361084, + "learning_rate": 0.00019377889103347945, + "loss": 3.4752, + "step": 788 + }, + { + "epoch": 1.16, + "grad_norm": 5.934195518493652, + "learning_rate": 0.0001937625955648276, + "loss": 3.5186, + "step": 789 + }, + { + "epoch": 1.16, + "grad_norm": 9.426629066467285, + "learning_rate": 0.00019374627946903305, + "loss": 3.4537, + "step": 790 + }, + { + "epoch": 1.16, + "grad_norm": 3.8387088775634766, + "learning_rate": 0.00019372994274968527, + "loss": 3.4172, + "step": 791 + }, + { + "epoch": 1.16, + "grad_norm": 6.488461971282959, + "learning_rate": 0.00019371358541037825, + "loss": 3.5177, + "step": 792 + }, + { + "epoch": 1.16, + "grad_norm": 8.451127052307129, + "learning_rate": 0.0001936972074547105, + "loss": 3.4701, + "step": 793 + }, + { + "epoch": 1.16, + "grad_norm": 5.298417568206787, + "learning_rate": 0.00019368080888628502, + "loss": 3.4579, + "step": 794 + }, + { + "epoch": 1.17, + "grad_norm": 6.457042694091797, + "learning_rate": 0.00019366438970870942, + "loss": 3.4666, + "step": 795 + }, + { + "epoch": 1.17, + "grad_norm": 9.55510139465332, + "learning_rate": 0.00019364794992559578, + "loss": 3.4449, + "step": 796 + }, + { + "epoch": 1.17, + "grad_norm": 3.862884998321533, + "learning_rate": 0.00019363148954056077, + "loss": 3.4185, + "step": 797 + }, + { + "epoch": 1.17, + "grad_norm": 6.591372966766357, + "learning_rate": 0.00019361500855722555, + "loss": 3.537, + "step": 798 + }, + { + "epoch": 1.17, + "grad_norm": 8.397138595581055, + "learning_rate": 0.0001935985069792158, + "loss": 3.4205, + "step": 799 + }, + { + "epoch": 1.17, + "grad_norm": 5.131923675537109, + "learning_rate": 0.00019358198481016182, + "loss": 3.4916, + "step": 800 + }, + { + "epoch": 1.17, + "grad_norm": 5.620177745819092, + "learning_rate": 0.00019356544205369834, + "loss": 3.4604, + "step": 801 + }, + { + "epoch": 1.18, + "grad_norm": 9.219866752624512, + "learning_rate": 0.00019354887871346463, + "loss": 3.467, + "step": 802 + }, + { + "epoch": 1.18, + "grad_norm": 4.237349033355713, + "learning_rate": 0.00019353229479310456, + "loss": 3.4847, + "step": 803 + }, + { + "epoch": 1.18, + "grad_norm": 6.673110485076904, + "learning_rate": 0.00019351569029626644, + "loss": 3.4412, + "step": 804 + }, + { + "epoch": 1.18, + "grad_norm": 8.31098747253418, + "learning_rate": 0.0001934990652266032, + "loss": 3.4838, + "step": 805 + }, + { + "epoch": 1.18, + "grad_norm": 5.009383201599121, + "learning_rate": 0.0001934824195877722, + "loss": 3.4408, + "step": 806 + }, + { + "epoch": 1.18, + "grad_norm": 5.483675956726074, + "learning_rate": 0.00019346575338343537, + "loss": 3.4663, + "step": 807 + }, + { + "epoch": 1.19, + "grad_norm": 9.371169090270996, + "learning_rate": 0.00019344906661725918, + "loss": 3.5116, + "step": 808 + }, + { + "epoch": 1.19, + "grad_norm": 3.6695902347564697, + "learning_rate": 0.00019343235929291462, + "loss": 3.4776, + "step": 809 + }, + { + "epoch": 1.19, + "grad_norm": 6.492842197418213, + "learning_rate": 0.00019341563141407714, + "loss": 3.4725, + "step": 810 + }, + { + "epoch": 1.19, + "grad_norm": 8.520849227905273, + "learning_rate": 0.00019339888298442682, + "loss": 3.4845, + "step": 811 + }, + { + "epoch": 1.19, + "grad_norm": 4.986138820648193, + "learning_rate": 0.00019338211400764816, + "loss": 3.4313, + "step": 812 + }, + { + "epoch": 1.19, + "grad_norm": 6.272872447967529, + "learning_rate": 0.00019336532448743023, + "loss": 3.4337, + "step": 813 + }, + { + "epoch": 1.19, + "grad_norm": 10.374552726745605, + "learning_rate": 0.00019334851442746664, + "loss": 3.4586, + "step": 814 + }, + { + "epoch": 1.2, + "grad_norm": 4.865425109863281, + "learning_rate": 0.00019333168383145547, + "loss": 3.5045, + "step": 815 + }, + { + "epoch": 1.2, + "grad_norm": 6.018640995025635, + "learning_rate": 0.0001933148327030993, + "loss": 3.4509, + "step": 816 + }, + { + "epoch": 1.2, + "grad_norm": 8.015512466430664, + "learning_rate": 0.00019329796104610537, + "loss": 3.4657, + "step": 817 + }, + { + "epoch": 1.2, + "grad_norm": 4.071451187133789, + "learning_rate": 0.00019328106886418522, + "loss": 3.4079, + "step": 818 + }, + { + "epoch": 1.2, + "grad_norm": 7.4237823486328125, + "learning_rate": 0.00019326415616105507, + "loss": 3.4906, + "step": 819 + }, + { + "epoch": 1.2, + "grad_norm": 10.776564598083496, + "learning_rate": 0.00019324722294043558, + "loss": 3.4639, + "step": 820 + }, + { + "epoch": 1.2, + "grad_norm": 5.143712997436523, + "learning_rate": 0.00019323026920605197, + "loss": 3.4356, + "step": 821 + }, + { + "epoch": 1.21, + "grad_norm": 5.7633891105651855, + "learning_rate": 0.00019321329496163397, + "loss": 3.428, + "step": 822 + }, + { + "epoch": 1.21, + "grad_norm": 7.827281475067139, + "learning_rate": 0.0001931963002109157, + "loss": 3.4913, + "step": 823 + }, + { + "epoch": 1.21, + "grad_norm": 4.470641136169434, + "learning_rate": 0.00019317928495763602, + "loss": 3.4441, + "step": 824 + }, + { + "epoch": 1.21, + "grad_norm": 7.5875091552734375, + "learning_rate": 0.00019316224920553807, + "loss": 3.468, + "step": 825 + }, + { + "epoch": 1.21, + "grad_norm": 10.94041919708252, + "learning_rate": 0.00019314519295836964, + "loss": 3.4932, + "step": 826 + }, + { + "epoch": 1.21, + "grad_norm": 5.624682426452637, + "learning_rate": 0.000193128116219883, + "loss": 3.4411, + "step": 827 + }, + { + "epoch": 1.21, + "grad_norm": 5.68918514251709, + "learning_rate": 0.00019311101899383492, + "loss": 3.4896, + "step": 828 + }, + { + "epoch": 1.22, + "grad_norm": 7.500881671905518, + "learning_rate": 0.00019309390128398665, + "loss": 3.4429, + "step": 829 + }, + { + "epoch": 1.22, + "grad_norm": 4.407763481140137, + "learning_rate": 0.000193076763094104, + "loss": 3.4694, + "step": 830 + }, + { + "epoch": 1.22, + "grad_norm": 7.296525955200195, + "learning_rate": 0.0001930596044279572, + "loss": 3.4065, + "step": 831 + }, + { + "epoch": 1.22, + "grad_norm": 11.3180513381958, + "learning_rate": 0.00019304242528932113, + "loss": 3.5025, + "step": 832 + }, + { + "epoch": 1.22, + "grad_norm": 5.404650688171387, + "learning_rate": 0.00019302522568197502, + "loss": 3.4129, + "step": 833 + }, + { + "epoch": 1.22, + "grad_norm": 5.595582485198975, + "learning_rate": 0.0001930080056097027, + "loss": 3.5212, + "step": 834 + }, + { + "epoch": 1.23, + "grad_norm": 7.934787273406982, + "learning_rate": 0.00019299076507629245, + "loss": 3.5551, + "step": 835 + }, + { + "epoch": 1.23, + "grad_norm": 4.486216068267822, + "learning_rate": 0.00019297350408553708, + "loss": 3.441, + "step": 836 + }, + { + "epoch": 1.23, + "grad_norm": 7.029922962188721, + "learning_rate": 0.0001929562226412339, + "loss": 3.4982, + "step": 837 + }, + { + "epoch": 1.23, + "grad_norm": 10.693680763244629, + "learning_rate": 0.0001929389207471847, + "loss": 3.4494, + "step": 838 + }, + { + "epoch": 1.23, + "grad_norm": 4.879317760467529, + "learning_rate": 0.00019292159840719576, + "loss": 3.4357, + "step": 839 + }, + { + "epoch": 1.23, + "grad_norm": 6.057565212249756, + "learning_rate": 0.00019290425562507792, + "loss": 3.4738, + "step": 840 + }, + { + "epoch": 1.23, + "grad_norm": 7.925189018249512, + "learning_rate": 0.0001928868924046465, + "loss": 3.4606, + "step": 841 + }, + { + "epoch": 1.24, + "grad_norm": 4.621508598327637, + "learning_rate": 0.0001928695087497212, + "loss": 3.4088, + "step": 842 + }, + { + "epoch": 1.24, + "grad_norm": 6.142604351043701, + "learning_rate": 0.00019285210466412635, + "loss": 3.4484, + "step": 843 + }, + { + "epoch": 1.24, + "grad_norm": 10.20028305053711, + "learning_rate": 0.00019283468015169076, + "loss": 3.5059, + "step": 844 + }, + { + "epoch": 1.24, + "grad_norm": 4.734007358551025, + "learning_rate": 0.0001928172352162477, + "loss": 3.4299, + "step": 845 + }, + { + "epoch": 1.24, + "eval_loss": 3.2429304122924805, + "eval_runtime": 1163.5436, + "eval_samples_per_second": 14.733, + "eval_steps_per_second": 0.921, + "step": 845 + }, + { + "epoch": 1.24, + "grad_norm": 6.036166191101074, + "learning_rate": 0.00019279976986163493, + "loss": 3.3913, + "step": 846 + }, + { + "epoch": 1.24, + "grad_norm": 8.014019012451172, + "learning_rate": 0.00019278228409169468, + "loss": 3.4981, + "step": 847 + }, + { + "epoch": 1.24, + "grad_norm": 4.856557369232178, + "learning_rate": 0.00019276477791027374, + "loss": 3.4562, + "step": 848 + }, + { + "epoch": 1.25, + "grad_norm": 6.650398254394531, + "learning_rate": 0.00019274725132122336, + "loss": 3.4407, + "step": 849 + }, + { + "epoch": 1.25, + "grad_norm": 10.177898406982422, + "learning_rate": 0.00019272970432839922, + "loss": 3.4437, + "step": 850 + }, + { + "epoch": 1.25, + "grad_norm": 4.559935569763184, + "learning_rate": 0.0001927121369356616, + "loss": 3.418, + "step": 851 + }, + { + "epoch": 1.25, + "grad_norm": 5.8679423332214355, + "learning_rate": 0.00019269454914687517, + "loss": 3.4338, + "step": 852 + }, + { + "epoch": 1.25, + "grad_norm": 8.044879913330078, + "learning_rate": 0.00019267694096590914, + "loss": 3.4146, + "step": 853 + }, + { + "epoch": 1.25, + "grad_norm": 4.654249668121338, + "learning_rate": 0.0001926593123966372, + "loss": 3.4353, + "step": 854 + }, + { + "epoch": 1.25, + "grad_norm": 6.326594352722168, + "learning_rate": 0.0001926416634429375, + "loss": 3.447, + "step": 855 + }, + { + "epoch": 1.26, + "grad_norm": 10.194587707519531, + "learning_rate": 0.0001926239941086927, + "loss": 3.4594, + "step": 856 + }, + { + "epoch": 1.26, + "grad_norm": 4.672416687011719, + "learning_rate": 0.00019260630439778994, + "loss": 3.4292, + "step": 857 + }, + { + "epoch": 1.26, + "grad_norm": 5.656294822692871, + "learning_rate": 0.00019258859431412083, + "loss": 3.4478, + "step": 858 + }, + { + "epoch": 1.26, + "grad_norm": 8.051904678344727, + "learning_rate": 0.00019257086386158148, + "loss": 3.4809, + "step": 859 + }, + { + "epoch": 1.26, + "grad_norm": 4.461113452911377, + "learning_rate": 0.00019255311304407243, + "loss": 3.4361, + "step": 860 + }, + { + "epoch": 1.26, + "grad_norm": 6.34873104095459, + "learning_rate": 0.00019253534186549883, + "loss": 3.424, + "step": 861 + }, + { + "epoch": 1.27, + "grad_norm": 10.255434036254883, + "learning_rate": 0.00019251755032977013, + "loss": 3.4578, + "step": 862 + }, + { + "epoch": 1.27, + "grad_norm": 4.895047187805176, + "learning_rate": 0.00019249973844080037, + "loss": 3.4606, + "step": 863 + }, + { + "epoch": 1.27, + "grad_norm": 6.042967319488525, + "learning_rate": 0.00019248190620250807, + "loss": 3.4398, + "step": 864 + }, + { + "epoch": 1.27, + "grad_norm": 7.916470050811768, + "learning_rate": 0.00019246405361881622, + "loss": 3.4921, + "step": 865 + }, + { + "epoch": 1.27, + "grad_norm": 4.985673904418945, + "learning_rate": 0.0001924461806936522, + "loss": 3.4425, + "step": 866 + }, + { + "epoch": 1.27, + "grad_norm": 6.128829002380371, + "learning_rate": 0.000192428287430948, + "loss": 3.4719, + "step": 867 + }, + { + "epoch": 1.27, + "grad_norm": 9.511234283447266, + "learning_rate": 0.00019241037383463998, + "loss": 3.4018, + "step": 868 + }, + { + "epoch": 1.28, + "grad_norm": 3.816689968109131, + "learning_rate": 0.000192392439908669, + "loss": 3.375, + "step": 869 + }, + { + "epoch": 1.28, + "grad_norm": 6.579503059387207, + "learning_rate": 0.00019237448565698046, + "loss": 3.4039, + "step": 870 + }, + { + "epoch": 1.28, + "grad_norm": 8.701712608337402, + "learning_rate": 0.00019235651108352416, + "loss": 3.3992, + "step": 871 + }, + { + "epoch": 1.28, + "grad_norm": 5.65900993347168, + "learning_rate": 0.00019233851619225432, + "loss": 3.4225, + "step": 872 + }, + { + "epoch": 1.28, + "grad_norm": 4.935523509979248, + "learning_rate": 0.00019232050098712974, + "loss": 3.4198, + "step": 873 + }, + { + "epoch": 1.28, + "grad_norm": 8.355915069580078, + "learning_rate": 0.00019230246547211367, + "loss": 3.4283, + "step": 874 + }, + { + "epoch": 1.28, + "grad_norm": 3.1533217430114746, + "learning_rate": 0.0001922844096511738, + "loss": 3.4058, + "step": 875 + }, + { + "epoch": 1.29, + "grad_norm": 7.010662078857422, + "learning_rate": 0.00019226633352828224, + "loss": 3.3914, + "step": 876 + }, + { + "epoch": 1.29, + "grad_norm": 8.548685073852539, + "learning_rate": 0.00019224823710741569, + "loss": 3.5008, + "step": 877 + }, + { + "epoch": 1.29, + "grad_norm": 5.439549446105957, + "learning_rate": 0.00019223012039255516, + "loss": 3.4421, + "step": 878 + }, + { + "epoch": 1.29, + "grad_norm": 5.446386814117432, + "learning_rate": 0.00019221198338768622, + "loss": 3.4071, + "step": 879 + }, + { + "epoch": 1.29, + "grad_norm": 9.378525733947754, + "learning_rate": 0.00019219382609679894, + "loss": 3.409, + "step": 880 + }, + { + "epoch": 1.29, + "grad_norm": 4.573349475860596, + "learning_rate": 0.0001921756485238878, + "loss": 3.4156, + "step": 881 + }, + { + "epoch": 1.29, + "grad_norm": 5.857920169830322, + "learning_rate": 0.00019215745067295169, + "loss": 3.433, + "step": 882 + }, + { + "epoch": 1.3, + "grad_norm": 7.654170036315918, + "learning_rate": 0.00019213923254799402, + "loss": 3.4011, + "step": 883 + }, + { + "epoch": 1.3, + "grad_norm": 4.262071132659912, + "learning_rate": 0.0001921209941530227, + "loss": 3.401, + "step": 884 + }, + { + "epoch": 1.3, + "grad_norm": 7.949652194976807, + "learning_rate": 0.00019210273549205003, + "loss": 3.5334, + "step": 885 + }, + { + "epoch": 1.3, + "grad_norm": 11.388256072998047, + "learning_rate": 0.0001920844565690928, + "loss": 3.4325, + "step": 886 + }, + { + "epoch": 1.3, + "grad_norm": 6.005974292755127, + "learning_rate": 0.00019206615738817225, + "loss": 3.4264, + "step": 887 + }, + { + "epoch": 1.3, + "grad_norm": 5.020950794219971, + "learning_rate": 0.0001920478379533141, + "loss": 3.426, + "step": 888 + }, + { + "epoch": 1.31, + "grad_norm": 6.8441386222839355, + "learning_rate": 0.00019202949826854844, + "loss": 3.4234, + "step": 889 + }, + { + "epoch": 1.31, + "grad_norm": 3.4975130558013916, + "learning_rate": 0.0001920111383379099, + "loss": 3.4667, + "step": 890 + }, + { + "epoch": 1.31, + "grad_norm": 7.94019889831543, + "learning_rate": 0.00019199275816543762, + "loss": 3.4208, + "step": 891 + }, + { + "epoch": 1.31, + "grad_norm": 11.041260719299316, + "learning_rate": 0.00019197435775517501, + "loss": 3.4392, + "step": 892 + }, + { + "epoch": 1.31, + "grad_norm": 5.074487209320068, + "learning_rate": 0.0001919559371111701, + "loss": 3.3906, + "step": 893 + }, + { + "epoch": 1.31, + "grad_norm": 5.909255027770996, + "learning_rate": 0.00019193749623747525, + "loss": 3.4508, + "step": 894 + }, + { + "epoch": 1.31, + "grad_norm": 7.794407367706299, + "learning_rate": 0.0001919190351381474, + "loss": 3.3957, + "step": 895 + }, + { + "epoch": 1.32, + "grad_norm": 4.833076000213623, + "learning_rate": 0.00019190055381724781, + "loss": 3.403, + "step": 896 + }, + { + "epoch": 1.32, + "grad_norm": 6.273806095123291, + "learning_rate": 0.0001918820522788423, + "loss": 3.4114, + "step": 897 + }, + { + "epoch": 1.32, + "grad_norm": 9.626479148864746, + "learning_rate": 0.00019186353052700104, + "loss": 3.4555, + "step": 898 + }, + { + "epoch": 1.32, + "grad_norm": 4.462099075317383, + "learning_rate": 0.00019184498856579868, + "loss": 3.4001, + "step": 899 + }, + { + "epoch": 1.32, + "grad_norm": 5.839742660522461, + "learning_rate": 0.00019182642639931438, + "loss": 3.4041, + "step": 900 + }, + { + "epoch": 1.32, + "grad_norm": 7.5265583992004395, + "learning_rate": 0.00019180784403163167, + "loss": 3.3898, + "step": 901 + }, + { + "epoch": 1.32, + "grad_norm": 4.089066028594971, + "learning_rate": 0.0001917892414668385, + "loss": 3.4162, + "step": 902 + }, + { + "epoch": 1.33, + "grad_norm": 7.329177379608154, + "learning_rate": 0.0001917706187090274, + "loss": 3.5439, + "step": 903 + }, + { + "epoch": 1.33, + "grad_norm": 10.022590637207031, + "learning_rate": 0.0001917519757622952, + "loss": 3.4009, + "step": 904 + }, + { + "epoch": 1.33, + "grad_norm": 5.030527591705322, + "learning_rate": 0.00019173331263074324, + "loss": 3.4314, + "step": 905 + }, + { + "epoch": 1.33, + "grad_norm": 5.504607677459717, + "learning_rate": 0.00019171462931847724, + "loss": 3.4463, + "step": 906 + }, + { + "epoch": 1.33, + "grad_norm": 7.219310283660889, + "learning_rate": 0.00019169592582960742, + "loss": 3.4878, + "step": 907 + }, + { + "epoch": 1.33, + "grad_norm": 4.046157360076904, + "learning_rate": 0.00019167720216824847, + "loss": 3.3643, + "step": 908 + }, + { + "epoch": 1.34, + "grad_norm": 7.463158130645752, + "learning_rate": 0.00019165845833851941, + "loss": 3.472, + "step": 909 + }, + { + "epoch": 1.34, + "grad_norm": 9.688533782958984, + "learning_rate": 0.00019163969434454382, + "loss": 3.4541, + "step": 910 + }, + { + "epoch": 1.34, + "grad_norm": 4.718085765838623, + "learning_rate": 0.00019162091019044958, + "loss": 3.4252, + "step": 911 + }, + { + "epoch": 1.34, + "grad_norm": 6.263822555541992, + "learning_rate": 0.00019160210588036911, + "loss": 3.4105, + "step": 912 + }, + { + "epoch": 1.34, + "grad_norm": 7.928195953369141, + "learning_rate": 0.00019158328141843928, + "loss": 3.4249, + "step": 913 + }, + { + "epoch": 1.34, + "grad_norm": 4.594407081604004, + "learning_rate": 0.00019156443680880126, + "loss": 3.453, + "step": 914 + }, + { + "epoch": 1.34, + "grad_norm": 6.126087188720703, + "learning_rate": 0.0001915455720556008, + "loss": 3.3944, + "step": 915 + }, + { + "epoch": 1.35, + "grad_norm": 9.558304786682129, + "learning_rate": 0.000191526687162988, + "loss": 3.3926, + "step": 916 + }, + { + "epoch": 1.35, + "grad_norm": 4.792506217956543, + "learning_rate": 0.0001915077821351174, + "loss": 3.4081, + "step": 917 + }, + { + "epoch": 1.35, + "grad_norm": 5.998870372772217, + "learning_rate": 0.00019148885697614797, + "loss": 3.4326, + "step": 918 + }, + { + "epoch": 1.35, + "grad_norm": 7.541780948638916, + "learning_rate": 0.00019146991169024315, + "loss": 3.4647, + "step": 919 + }, + { + "epoch": 1.35, + "grad_norm": 4.514527320861816, + "learning_rate": 0.00019145094628157078, + "loss": 3.3601, + "step": 920 + }, + { + "epoch": 1.35, + "grad_norm": 6.531849384307861, + "learning_rate": 0.0001914319607543031, + "loss": 3.4428, + "step": 921 + }, + { + "epoch": 1.35, + "grad_norm": 9.34991455078125, + "learning_rate": 0.00019141295511261678, + "loss": 3.393, + "step": 922 + }, + { + "epoch": 1.36, + "grad_norm": 4.035532474517822, + "learning_rate": 0.000191393929360693, + "loss": 3.3938, + "step": 923 + }, + { + "epoch": 1.36, + "grad_norm": 6.159792900085449, + "learning_rate": 0.0001913748835027172, + "loss": 3.4207, + "step": 924 + }, + { + "epoch": 1.36, + "grad_norm": 8.069183349609375, + "learning_rate": 0.00019135581754287944, + "loss": 3.4143, + "step": 925 + }, + { + "epoch": 1.36, + "grad_norm": 4.528735160827637, + "learning_rate": 0.00019133673148537405, + "loss": 3.4282, + "step": 926 + }, + { + "epoch": 1.36, + "grad_norm": 6.5554399490356445, + "learning_rate": 0.00019131762533439983, + "loss": 3.4149, + "step": 927 + }, + { + "epoch": 1.36, + "grad_norm": 9.77618408203125, + "learning_rate": 0.00019129849909416005, + "loss": 3.4509, + "step": 928 + }, + { + "epoch": 1.36, + "grad_norm": 4.406791687011719, + "learning_rate": 0.0001912793527688623, + "loss": 3.3955, + "step": 929 + }, + { + "epoch": 1.37, + "grad_norm": 6.1382622718811035, + "learning_rate": 0.00019126018636271867, + "loss": 3.4066, + "step": 930 + }, + { + "epoch": 1.37, + "grad_norm": 8.133618354797363, + "learning_rate": 0.00019124099987994565, + "loss": 3.4536, + "step": 931 + }, + { + "epoch": 1.37, + "grad_norm": 4.887189865112305, + "learning_rate": 0.0001912217933247641, + "loss": 3.4516, + "step": 932 + }, + { + "epoch": 1.37, + "grad_norm": 5.946772575378418, + "learning_rate": 0.0001912025667013994, + "loss": 3.3842, + "step": 933 + }, + { + "epoch": 1.37, + "grad_norm": 9.542755126953125, + "learning_rate": 0.00019118332001408122, + "loss": 3.3833, + "step": 934 + }, + { + "epoch": 1.37, + "grad_norm": 4.512613296508789, + "learning_rate": 0.00019116405326704373, + "loss": 3.3754, + "step": 935 + }, + { + "epoch": 1.38, + "grad_norm": 6.039595603942871, + "learning_rate": 0.00019114476646452547, + "loss": 3.4341, + "step": 936 + }, + { + "epoch": 1.38, + "grad_norm": 7.820281982421875, + "learning_rate": 0.0001911254596107694, + "loss": 3.3616, + "step": 937 + }, + { + "epoch": 1.38, + "grad_norm": 4.0570759773254395, + "learning_rate": 0.0001911061327100229, + "loss": 3.3999, + "step": 938 + }, + { + "epoch": 1.38, + "grad_norm": 6.659790992736816, + "learning_rate": 0.0001910867857665378, + "loss": 3.361, + "step": 939 + }, + { + "epoch": 1.38, + "grad_norm": 10.188064575195312, + "learning_rate": 0.00019106741878457026, + "loss": 3.3805, + "step": 940 + }, + { + "epoch": 1.38, + "grad_norm": 4.66610860824585, + "learning_rate": 0.0001910480317683809, + "loss": 3.3881, + "step": 941 + }, + { + "epoch": 1.38, + "grad_norm": 5.840310573577881, + "learning_rate": 0.0001910286247222347, + "loss": 3.4162, + "step": 942 + }, + { + "epoch": 1.39, + "grad_norm": 7.710467338562012, + "learning_rate": 0.00019100919765040113, + "loss": 3.396, + "step": 943 + }, + { + "epoch": 1.39, + "grad_norm": 4.53704833984375, + "learning_rate": 0.00019098975055715397, + "loss": 3.3605, + "step": 944 + }, + { + "epoch": 1.39, + "grad_norm": 6.515756607055664, + "learning_rate": 0.00019097028344677153, + "loss": 3.4119, + "step": 945 + }, + { + "epoch": 1.39, + "grad_norm": 10.256503105163574, + "learning_rate": 0.00019095079632353633, + "loss": 3.4173, + "step": 946 + }, + { + "epoch": 1.39, + "grad_norm": 4.8957672119140625, + "learning_rate": 0.0001909312891917355, + "loss": 3.3812, + "step": 947 + }, + { + "epoch": 1.39, + "grad_norm": 5.8118205070495605, + "learning_rate": 0.00019091176205566043, + "loss": 3.3734, + "step": 948 + }, + { + "epoch": 1.39, + "grad_norm": 7.7631025314331055, + "learning_rate": 0.00019089221491960695, + "loss": 3.4171, + "step": 949 + }, + { + "epoch": 1.4, + "grad_norm": 4.2992377281188965, + "learning_rate": 0.00019087264778787534, + "loss": 3.3724, + "step": 950 + }, + { + "epoch": 1.4, + "grad_norm": 6.633007049560547, + "learning_rate": 0.00019085306066477021, + "loss": 3.4198, + "step": 951 + }, + { + "epoch": 1.4, + "grad_norm": 9.867729187011719, + "learning_rate": 0.00019083345355460062, + "loss": 3.4247, + "step": 952 + }, + { + "epoch": 1.4, + "grad_norm": 4.476417541503906, + "learning_rate": 0.00019081382646167998, + "loss": 3.3617, + "step": 953 + }, + { + "epoch": 1.4, + "grad_norm": 5.963509559631348, + "learning_rate": 0.00019079417939032612, + "loss": 3.3945, + "step": 954 + }, + { + "epoch": 1.4, + "grad_norm": 7.921098232269287, + "learning_rate": 0.00019077451234486128, + "loss": 3.4305, + "step": 955 + }, + { + "epoch": 1.4, + "grad_norm": 4.772178649902344, + "learning_rate": 0.00019075482532961204, + "loss": 3.3836, + "step": 956 + }, + { + "epoch": 1.41, + "grad_norm": 6.6379265785217285, + "learning_rate": 0.00019073511834890949, + "loss": 3.3646, + "step": 957 + }, + { + "epoch": 1.41, + "grad_norm": 9.883600234985352, + "learning_rate": 0.00019071539140708894, + "loss": 3.398, + "step": 958 + }, + { + "epoch": 1.41, + "grad_norm": 4.0690178871154785, + "learning_rate": 0.00019069564450849025, + "loss": 3.3394, + "step": 959 + }, + { + "epoch": 1.41, + "grad_norm": 6.045716762542725, + "learning_rate": 0.00019067587765745757, + "loss": 3.4695, + "step": 960 + }, + { + "epoch": 1.41, + "grad_norm": 7.893038272857666, + "learning_rate": 0.00019065609085833947, + "loss": 3.4336, + "step": 961 + }, + { + "epoch": 1.41, + "grad_norm": 4.4083075523376465, + "learning_rate": 0.00019063628411548897, + "loss": 3.3471, + "step": 962 + }, + { + "epoch": 1.42, + "grad_norm": 6.27316951751709, + "learning_rate": 0.00019061645743326337, + "loss": 3.4311, + "step": 963 + }, + { + "epoch": 1.42, + "grad_norm": 9.92757797241211, + "learning_rate": 0.00019059661081602443, + "loss": 3.4114, + "step": 964 + }, + { + "epoch": 1.42, + "grad_norm": 5.099198341369629, + "learning_rate": 0.00019057674426813822, + "loss": 3.4386, + "step": 965 + }, + { + "epoch": 1.42, + "grad_norm": 5.424035549163818, + "learning_rate": 0.00019055685779397532, + "loss": 3.4275, + "step": 966 + }, + { + "epoch": 1.42, + "grad_norm": 7.513932704925537, + "learning_rate": 0.0001905369513979106, + "loss": 3.4135, + "step": 967 + }, + { + "epoch": 1.42, + "grad_norm": 4.388814449310303, + "learning_rate": 0.0001905170250843233, + "loss": 3.3889, + "step": 968 + }, + { + "epoch": 1.42, + "grad_norm": 6.554171562194824, + "learning_rate": 0.00019049707885759713, + "loss": 3.4184, + "step": 969 + }, + { + "epoch": 1.43, + "grad_norm": 9.21511173248291, + "learning_rate": 0.00019047711272212006, + "loss": 3.393, + "step": 970 + }, + { + "epoch": 1.43, + "grad_norm": 4.280778408050537, + "learning_rate": 0.00019045712668228456, + "loss": 3.3808, + "step": 971 + }, + { + "epoch": 1.43, + "grad_norm": 5.782205581665039, + "learning_rate": 0.0001904371207424874, + "loss": 3.3168, + "step": 972 + }, + { + "epoch": 1.43, + "grad_norm": 7.35162878036499, + "learning_rate": 0.00019041709490712978, + "loss": 3.4373, + "step": 973 + }, + { + "epoch": 1.43, + "grad_norm": 3.4769937992095947, + "learning_rate": 0.0001903970491806172, + "loss": 3.4039, + "step": 974 + }, + { + "epoch": 1.43, + "grad_norm": 8.196279525756836, + "learning_rate": 0.00019037698356735963, + "loss": 3.38, + "step": 975 + }, + { + "epoch": 1.43, + "grad_norm": 11.59374713897705, + "learning_rate": 0.00019035689807177136, + "loss": 3.4377, + "step": 976 + }, + { + "epoch": 1.44, + "grad_norm": 6.463580131530762, + "learning_rate": 0.00019033679269827105, + "loss": 3.3706, + "step": 977 + }, + { + "epoch": 1.44, + "grad_norm": 4.297844409942627, + "learning_rate": 0.00019031666745128177, + "loss": 3.4047, + "step": 978 + }, + { + "epoch": 1.44, + "grad_norm": 6.652123928070068, + "learning_rate": 0.00019029652233523093, + "loss": 3.3968, + "step": 979 + }, + { + "epoch": 1.44, + "grad_norm": 3.399991750717163, + "learning_rate": 0.00019027635735455027, + "loss": 3.3552, + "step": 980 + }, + { + "epoch": 1.44, + "grad_norm": 7.520792007446289, + "learning_rate": 0.00019025617251367604, + "loss": 3.3979, + "step": 981 + }, + { + "epoch": 1.44, + "grad_norm": 10.621402740478516, + "learning_rate": 0.00019023596781704872, + "loss": 3.4131, + "step": 982 + }, + { + "epoch": 1.44, + "grad_norm": 5.002892971038818, + "learning_rate": 0.0001902157432691132, + "loss": 3.3583, + "step": 983 + }, + { + "epoch": 1.45, + "grad_norm": 5.647467136383057, + "learning_rate": 0.00019019549887431877, + "loss": 3.384, + "step": 984 + }, + { + "epoch": 1.45, + "grad_norm": 7.81645393371582, + "learning_rate": 0.00019017523463711906, + "loss": 3.4198, + "step": 985 + }, + { + "epoch": 1.45, + "grad_norm": 4.508808135986328, + "learning_rate": 0.00019015495056197207, + "loss": 3.4205, + "step": 986 + }, + { + "epoch": 1.45, + "grad_norm": 6.004973411560059, + "learning_rate": 0.00019013464665334012, + "loss": 3.396, + "step": 987 + }, + { + "epoch": 1.45, + "grad_norm": 9.286575317382812, + "learning_rate": 0.00019011432291568995, + "loss": 3.4053, + "step": 988 + }, + { + "epoch": 1.45, + "grad_norm": 4.02089786529541, + "learning_rate": 0.00019009397935349268, + "loss": 3.3216, + "step": 989 + }, + { + "epoch": 1.46, + "grad_norm": 6.502218246459961, + "learning_rate": 0.00019007361597122373, + "loss": 3.4367, + "step": 990 + }, + { + "epoch": 1.46, + "grad_norm": 8.069449424743652, + "learning_rate": 0.0001900532327733629, + "loss": 3.3666, + "step": 991 + }, + { + "epoch": 1.46, + "grad_norm": 4.8657917976379395, + "learning_rate": 0.0001900328297643944, + "loss": 3.3401, + "step": 992 + }, + { + "epoch": 1.46, + "grad_norm": 5.57826042175293, + "learning_rate": 0.00019001240694880672, + "loss": 3.393, + "step": 993 + }, + { + "epoch": 1.46, + "grad_norm": 8.596120834350586, + "learning_rate": 0.00018999196433109274, + "loss": 3.416, + "step": 994 + }, + { + "epoch": 1.46, + "grad_norm": 4.3653130531311035, + "learning_rate": 0.0001899715019157497, + "loss": 3.3789, + "step": 995 + }, + { + "epoch": 1.46, + "grad_norm": 6.067572593688965, + "learning_rate": 0.0001899510197072792, + "loss": 3.3577, + "step": 996 + }, + { + "epoch": 1.47, + "grad_norm": 7.723760604858398, + "learning_rate": 0.00018993051771018719, + "loss": 3.4, + "step": 997 + }, + { + "epoch": 1.47, + "grad_norm": 4.619333744049072, + "learning_rate": 0.00018990999592898396, + "loss": 3.3979, + "step": 998 + }, + { + "epoch": 1.47, + "grad_norm": 6.598495006561279, + "learning_rate": 0.00018988945436818422, + "loss": 3.4331, + "step": 999 + }, + { + "epoch": 1.47, + "grad_norm": 9.450194358825684, + "learning_rate": 0.0001898688930323069, + "loss": 3.4156, + "step": 1000 + }, + { + "epoch": 1.47, + "grad_norm": 4.10836935043335, + "learning_rate": 0.00018984831192587538, + "loss": 3.3787, + "step": 1001 + }, + { + "epoch": 1.47, + "grad_norm": 6.087796211242676, + "learning_rate": 0.00018982771105341737, + "loss": 3.3969, + "step": 1002 + }, + { + "epoch": 1.47, + "grad_norm": 7.915310859680176, + "learning_rate": 0.00018980709041946497, + "loss": 3.3948, + "step": 1003 + }, + { + "epoch": 1.48, + "grad_norm": 4.227004528045654, + "learning_rate": 0.0001897864500285545, + "loss": 3.3238, + "step": 1004 + }, + { + "epoch": 1.48, + "grad_norm": 6.717720031738281, + "learning_rate": 0.00018976578988522675, + "loss": 3.3781, + "step": 1005 + }, + { + "epoch": 1.48, + "grad_norm": 10.050957679748535, + "learning_rate": 0.0001897451099940268, + "loss": 3.4204, + "step": 1006 + }, + { + "epoch": 1.48, + "grad_norm": 4.475175857543945, + "learning_rate": 0.0001897244103595041, + "loss": 3.4221, + "step": 1007 + }, + { + "epoch": 1.48, + "grad_norm": 5.719995975494385, + "learning_rate": 0.00018970369098621244, + "loss": 3.3555, + "step": 1008 + }, + { + "epoch": 1.48, + "grad_norm": 7.594791889190674, + "learning_rate": 0.00018968295187870996, + "loss": 3.4073, + "step": 1009 + }, + { + "epoch": 1.49, + "grad_norm": 4.333056926727295, + "learning_rate": 0.00018966219304155906, + "loss": 3.3556, + "step": 1010 + }, + { + "epoch": 1.49, + "grad_norm": 6.078528881072998, + "learning_rate": 0.00018964141447932656, + "loss": 3.3484, + "step": 1011 + }, + { + "epoch": 1.49, + "grad_norm": 8.950298309326172, + "learning_rate": 0.00018962061619658366, + "loss": 3.4055, + "step": 1012 + }, + { + "epoch": 1.49, + "grad_norm": 4.340825080871582, + "learning_rate": 0.0001895997981979058, + "loss": 3.334, + "step": 1013 + }, + { + "epoch": 1.49, + "grad_norm": 6.103879451751709, + "learning_rate": 0.00018957896048787278, + "loss": 3.4908, + "step": 1014 + }, + { + "epoch": 1.49, + "eval_loss": 3.206282138824463, + "eval_runtime": 1164.0205, + "eval_samples_per_second": 14.727, + "eval_steps_per_second": 0.921, + "step": 1014 + }, + { + "epoch": 1.49, + "grad_norm": 7.772497653961182, + "learning_rate": 0.00018955810307106881, + "loss": 3.3821, + "step": 1015 + }, + { + "epoch": 1.49, + "grad_norm": 4.639206886291504, + "learning_rate": 0.00018953722595208236, + "loss": 3.3776, + "step": 1016 + }, + { + "epoch": 1.5, + "grad_norm": 5.658569812774658, + "learning_rate": 0.00018951632913550626, + "loss": 3.339, + "step": 1017 + }, + { + "epoch": 1.5, + "grad_norm": 8.45776081085205, + "learning_rate": 0.00018949541262593762, + "loss": 3.4214, + "step": 1018 + }, + { + "epoch": 1.5, + "grad_norm": 3.2400307655334473, + "learning_rate": 0.000189474476427978, + "loss": 3.3666, + "step": 1019 + }, + { + "epoch": 1.5, + "grad_norm": 6.397870063781738, + "learning_rate": 0.0001894535205462332, + "loss": 3.4133, + "step": 1020 + }, + { + "epoch": 1.5, + "grad_norm": 8.270928382873535, + "learning_rate": 0.00018943254498531337, + "loss": 3.4133, + "step": 1021 + }, + { + "epoch": 1.5, + "grad_norm": 5.132183074951172, + "learning_rate": 0.00018941154974983298, + "loss": 3.3984, + "step": 1022 + }, + { + "epoch": 1.5, + "grad_norm": 5.842728137969971, + "learning_rate": 0.0001893905348444109, + "loss": 3.4019, + "step": 1023 + }, + { + "epoch": 1.51, + "grad_norm": 8.618907928466797, + "learning_rate": 0.00018936950027367017, + "loss": 3.4036, + "step": 1024 + }, + { + "epoch": 1.51, + "grad_norm": 3.155228853225708, + "learning_rate": 0.00018934844604223835, + "loss": 3.3512, + "step": 1025 + }, + { + "epoch": 1.51, + "grad_norm": 6.209192752838135, + "learning_rate": 0.00018932737215474714, + "loss": 3.388, + "step": 1026 + }, + { + "epoch": 1.51, + "grad_norm": 7.953395843505859, + "learning_rate": 0.00018930627861583275, + "loss": 3.402, + "step": 1027 + }, + { + "epoch": 1.51, + "grad_norm": 4.540157794952393, + "learning_rate": 0.00018928516543013552, + "loss": 3.3162, + "step": 1028 + }, + { + "epoch": 1.51, + "grad_norm": 6.285725116729736, + "learning_rate": 0.0001892640326023003, + "loss": 3.4329, + "step": 1029 + }, + { + "epoch": 1.51, + "grad_norm": 9.266477584838867, + "learning_rate": 0.0001892428801369761, + "loss": 3.381, + "step": 1030 + }, + { + "epoch": 1.52, + "grad_norm": 4.977963924407959, + "learning_rate": 0.00018922170803881635, + "loss": 3.3758, + "step": 1031 + }, + { + "epoch": 1.52, + "grad_norm": 5.400416374206543, + "learning_rate": 0.0001892005163124788, + "loss": 3.3279, + "step": 1032 + }, + { + "epoch": 1.52, + "grad_norm": 6.933363914489746, + "learning_rate": 0.00018917930496262542, + "loss": 3.3655, + "step": 1033 + }, + { + "epoch": 1.52, + "grad_norm": 3.6254875659942627, + "learning_rate": 0.00018915807399392262, + "loss": 3.3451, + "step": 1034 + }, + { + "epoch": 1.52, + "grad_norm": 7.3330841064453125, + "learning_rate": 0.00018913682341104105, + "loss": 3.4291, + "step": 1035 + }, + { + "epoch": 1.52, + "grad_norm": 10.801877975463867, + "learning_rate": 0.0001891155532186557, + "loss": 3.4253, + "step": 1036 + }, + { + "epoch": 1.53, + "grad_norm": 5.197689533233643, + "learning_rate": 0.00018909426342144587, + "loss": 3.3764, + "step": 1037 + }, + { + "epoch": 1.53, + "grad_norm": 5.4762797355651855, + "learning_rate": 0.00018907295402409514, + "loss": 3.3669, + "step": 1038 + }, + { + "epoch": 1.53, + "grad_norm": 9.155226707458496, + "learning_rate": 0.0001890516250312915, + "loss": 3.3748, + "step": 1039 + }, + { + "epoch": 1.53, + "grad_norm": 4.974143028259277, + "learning_rate": 0.00018903027644772715, + "loss": 3.3812, + "step": 1040 + }, + { + "epoch": 1.53, + "grad_norm": 3.935563802719116, + "learning_rate": 0.00018900890827809862, + "loss": 3.3748, + "step": 1041 + }, + { + "epoch": 1.53, + "grad_norm": 5.06904935836792, + "learning_rate": 0.0001889875205271068, + "loss": 3.4328, + "step": 1042 + }, + { + "epoch": 1.53, + "grad_norm": 1.7735739946365356, + "learning_rate": 0.0001889661131994568, + "loss": 3.3331, + "step": 1043 + }, + { + "epoch": 1.54, + "grad_norm": 3.266242504119873, + "learning_rate": 0.00018894468629985814, + "loss": 3.3797, + "step": 1044 + }, + { + "epoch": 1.54, + "grad_norm": 1.9417365789413452, + "learning_rate": 0.0001889232398330246, + "loss": 3.3556, + "step": 1045 + }, + { + "epoch": 1.54, + "grad_norm": 2.2055392265319824, + "learning_rate": 0.00018890177380367419, + "loss": 3.3695, + "step": 1046 + }, + { + "epoch": 1.54, + "grad_norm": 3.9984004497528076, + "learning_rate": 0.00018888028821652936, + "loss": 3.4158, + "step": 1047 + }, + { + "epoch": 1.54, + "grad_norm": 2.413911819458008, + "learning_rate": 0.0001888587830763168, + "loss": 3.4492, + "step": 1048 + }, + { + "epoch": 1.54, + "grad_norm": 4.597828388214111, + "learning_rate": 0.00018883725838776742, + "loss": 3.3518, + "step": 1049 + }, + { + "epoch": 1.54, + "grad_norm": 4.952722549438477, + "learning_rate": 0.00018881571415561656, + "loss": 3.371, + "step": 1050 + }, + { + "epoch": 1.55, + "grad_norm": 2.318856954574585, + "learning_rate": 0.00018879415038460383, + "loss": 3.3974, + "step": 1051 + }, + { + "epoch": 1.55, + "grad_norm": 2.227694034576416, + "learning_rate": 0.00018877256707947306, + "loss": 3.3784, + "step": 1052 + }, + { + "epoch": 1.55, + "grad_norm": 4.430450916290283, + "learning_rate": 0.00018875096424497248, + "loss": 3.3792, + "step": 1053 + }, + { + "epoch": 1.55, + "grad_norm": 2.9697399139404297, + "learning_rate": 0.00018872934188585454, + "loss": 3.4059, + "step": 1054 + }, + { + "epoch": 1.55, + "grad_norm": 4.346686840057373, + "learning_rate": 0.000188707700006876, + "loss": 3.4155, + "step": 1055 + }, + { + "epoch": 1.55, + "grad_norm": 4.2472381591796875, + "learning_rate": 0.000188686038612798, + "loss": 3.4242, + "step": 1056 + }, + { + "epoch": 1.55, + "grad_norm": 3.157696008682251, + "learning_rate": 0.0001886643577083858, + "loss": 3.4913, + "step": 1057 + }, + { + "epoch": 1.56, + "grad_norm": 2.940396308898926, + "learning_rate": 0.00018864265729840913, + "loss": 3.4734, + "step": 1058 + }, + { + "epoch": 1.56, + "grad_norm": 4.0289154052734375, + "learning_rate": 0.00018862093738764192, + "loss": 3.443, + "step": 1059 + }, + { + "epoch": 1.56, + "grad_norm": 3.4877893924713135, + "learning_rate": 0.00018859919798086238, + "loss": 3.5322, + "step": 1060 + }, + { + "epoch": 1.56, + "grad_norm": 4.22461462020874, + "learning_rate": 0.00018857743908285305, + "loss": 3.4562, + "step": 1061 + }, + { + "epoch": 1.56, + "grad_norm": 4.031467437744141, + "learning_rate": 0.00018855566069840077, + "loss": 3.4653, + "step": 1062 + }, + { + "epoch": 1.56, + "grad_norm": 3.602692127227783, + "learning_rate": 0.00018853386283229658, + "loss": 3.4555, + "step": 1063 + }, + { + "epoch": 1.57, + "grad_norm": 2.7948570251464844, + "learning_rate": 0.00018851204548933593, + "loss": 3.4255, + "step": 1064 + }, + { + "epoch": 1.57, + "grad_norm": 4.970945358276367, + "learning_rate": 0.00018849020867431842, + "loss": 3.4463, + "step": 1065 + }, + { + "epoch": 1.57, + "grad_norm": 4.409501552581787, + "learning_rate": 0.00018846835239204804, + "loss": 3.3992, + "step": 1066 + }, + { + "epoch": 1.57, + "grad_norm": 2.8853704929351807, + "learning_rate": 0.000188446476647333, + "loss": 3.4709, + "step": 1067 + }, + { + "epoch": 1.57, + "grad_norm": 2.3841402530670166, + "learning_rate": 0.00018842458144498588, + "loss": 3.4342, + "step": 1068 + }, + { + "epoch": 1.57, + "grad_norm": 4.642658710479736, + "learning_rate": 0.00018840266678982342, + "loss": 3.4102, + "step": 1069 + }, + { + "epoch": 1.57, + "grad_norm": 4.265832424163818, + "learning_rate": 0.00018838073268666672, + "loss": 3.4367, + "step": 1070 + }, + { + "epoch": 1.58, + "grad_norm": 2.900568723678589, + "learning_rate": 0.00018835877914034113, + "loss": 3.4052, + "step": 1071 + }, + { + "epoch": 1.58, + "grad_norm": 2.4887197017669678, + "learning_rate": 0.00018833680615567628, + "loss": 3.4175, + "step": 1072 + }, + { + "epoch": 1.58, + "grad_norm": 5.449963092803955, + "learning_rate": 0.0001883148137375061, + "loss": 3.4521, + "step": 1073 + }, + { + "epoch": 1.58, + "grad_norm": 5.6257643699646, + "learning_rate": 0.00018829280189066873, + "loss": 3.4741, + "step": 1074 + }, + { + "epoch": 1.58, + "grad_norm": 2.799276828765869, + "learning_rate": 0.00018827077062000668, + "loss": 3.4431, + "step": 1075 + }, + { + "epoch": 1.58, + "grad_norm": 2.863265037536621, + "learning_rate": 0.00018824871993036664, + "loss": 3.3978, + "step": 1076 + }, + { + "epoch": 1.58, + "grad_norm": 4.204624176025391, + "learning_rate": 0.00018822664982659964, + "loss": 3.4559, + "step": 1077 + }, + { + "epoch": 1.59, + "grad_norm": 3.0599143505096436, + "learning_rate": 0.00018820456031356093, + "loss": 3.3883, + "step": 1078 + }, + { + "epoch": 1.59, + "grad_norm": 4.109722137451172, + "learning_rate": 0.00018818245139611011, + "loss": 3.4014, + "step": 1079 + }, + { + "epoch": 1.59, + "grad_norm": 3.4017956256866455, + "learning_rate": 0.00018816032307911092, + "loss": 3.4348, + "step": 1080 + }, + { + "epoch": 1.59, + "grad_norm": 4.216831207275391, + "learning_rate": 0.00018813817536743148, + "loss": 3.452, + "step": 1081 + }, + { + "epoch": 1.59, + "grad_norm": 3.819659471511841, + "learning_rate": 0.00018811600826594416, + "loss": 3.3937, + "step": 1082 + }, + { + "epoch": 1.59, + "grad_norm": 3.5902295112609863, + "learning_rate": 0.00018809382177952552, + "loss": 3.3992, + "step": 1083 + }, + { + "epoch": 1.59, + "grad_norm": 2.6973037719726562, + "learning_rate": 0.0001880716159130565, + "loss": 3.4278, + "step": 1084 + }, + { + "epoch": 1.6, + "grad_norm": 5.055513858795166, + "learning_rate": 0.0001880493906714222, + "loss": 3.3913, + "step": 1085 + }, + { + "epoch": 1.6, + "grad_norm": 4.533970832824707, + "learning_rate": 0.00018802714605951199, + "loss": 3.4047, + "step": 1086 + }, + { + "epoch": 1.6, + "grad_norm": 2.5329983234405518, + "learning_rate": 0.00018800488208221963, + "loss": 3.4291, + "step": 1087 + }, + { + "epoch": 1.6, + "grad_norm": 1.993481993675232, + "learning_rate": 0.00018798259874444297, + "loss": 3.4334, + "step": 1088 + }, + { + "epoch": 1.6, + "grad_norm": 6.799105167388916, + "learning_rate": 0.00018796029605108424, + "loss": 3.4107, + "step": 1089 + }, + { + "epoch": 1.6, + "grad_norm": 6.37121057510376, + "learning_rate": 0.00018793797400704985, + "loss": 3.4237, + "step": 1090 + }, + { + "epoch": 1.61, + "grad_norm": 1.2064040899276733, + "learning_rate": 0.0001879156326172505, + "loss": 3.4487, + "step": 1091 + }, + { + "epoch": 1.61, + "grad_norm": 0.6396639943122864, + "learning_rate": 0.00018789327188660116, + "loss": 3.3509, + "step": 1092 + }, + { + "epoch": 1.61, + "grad_norm": 2.529014825820923, + "learning_rate": 0.00018787089182002105, + "loss": 3.4014, + "step": 1093 + }, + { + "epoch": 1.61, + "grad_norm": 2.6367127895355225, + "learning_rate": 0.00018784849242243362, + "loss": 3.3676, + "step": 1094 + }, + { + "epoch": 1.61, + "grad_norm": 0.968987226486206, + "learning_rate": 0.00018782607369876659, + "loss": 3.4156, + "step": 1095 + }, + { + "epoch": 1.61, + "grad_norm": 5.669402599334717, + "learning_rate": 0.00018780363565395192, + "loss": 3.4195, + "step": 1096 + }, + { + "epoch": 1.61, + "grad_norm": 4.468107223510742, + "learning_rate": 0.00018778117829292587, + "loss": 3.4453, + "step": 1097 + }, + { + "epoch": 1.62, + "grad_norm": 3.340261220932007, + "learning_rate": 0.00018775870162062886, + "loss": 3.4024, + "step": 1098 + }, + { + "epoch": 1.62, + "grad_norm": 2.798548698425293, + "learning_rate": 0.00018773620564200565, + "loss": 3.503, + "step": 1099 + }, + { + "epoch": 1.62, + "grad_norm": 5.848763465881348, + "learning_rate": 0.00018771369036200513, + "loss": 3.3962, + "step": 1100 + }, + { + "epoch": 1.62, + "grad_norm": 6.020981788635254, + "learning_rate": 0.00018769115578558062, + "loss": 3.4708, + "step": 1101 + }, + { + "epoch": 1.62, + "grad_norm": 2.3334991931915283, + "learning_rate": 0.00018766860191768946, + "loss": 3.3861, + "step": 1102 + }, + { + "epoch": 1.62, + "grad_norm": 3.1461942195892334, + "learning_rate": 0.00018764602876329345, + "loss": 3.3828, + "step": 1103 + }, + { + "epoch": 1.62, + "grad_norm": 2.833292007446289, + "learning_rate": 0.00018762343632735848, + "loss": 3.4628, + "step": 1104 + }, + { + "epoch": 1.63, + "grad_norm": 1.1478309631347656, + "learning_rate": 0.00018760082461485476, + "loss": 3.3512, + "step": 1105 + }, + { + "epoch": 1.63, + "grad_norm": 4.6011199951171875, + "learning_rate": 0.00018757819363075665, + "loss": 3.4301, + "step": 1106 + }, + { + "epoch": 1.63, + "grad_norm": 3.997016191482544, + "learning_rate": 0.0001875555433800429, + "loss": 3.4099, + "step": 1107 + }, + { + "epoch": 1.63, + "grad_norm": 4.506150722503662, + "learning_rate": 0.00018753287386769638, + "loss": 3.4169, + "step": 1108 + }, + { + "epoch": 1.63, + "grad_norm": 4.197047710418701, + "learning_rate": 0.0001875101850987042, + "loss": 3.3518, + "step": 1109 + }, + { + "epoch": 1.63, + "grad_norm": 3.504765033721924, + "learning_rate": 0.0001874874770780578, + "loss": 3.3838, + "step": 1110 + }, + { + "epoch": 1.64, + "grad_norm": 2.7405476570129395, + "learning_rate": 0.00018746474981075272, + "loss": 3.376, + "step": 1111 + }, + { + "epoch": 1.64, + "grad_norm": 4.3409037590026855, + "learning_rate": 0.00018744200330178887, + "loss": 3.4184, + "step": 1112 + }, + { + "epoch": 1.64, + "grad_norm": 4.318120002746582, + "learning_rate": 0.0001874192375561703, + "loss": 3.4744, + "step": 1113 + }, + { + "epoch": 1.64, + "grad_norm": 3.5253307819366455, + "learning_rate": 0.00018739645257890533, + "loss": 3.4058, + "step": 1114 + }, + { + "epoch": 1.64, + "grad_norm": 3.368002414703369, + "learning_rate": 0.0001873736483750065, + "loss": 3.4013, + "step": 1115 + }, + { + "epoch": 1.64, + "grad_norm": 3.804229736328125, + "learning_rate": 0.00018735082494949058, + "loss": 3.3625, + "step": 1116 + }, + { + "epoch": 1.64, + "grad_norm": 3.6312084197998047, + "learning_rate": 0.00018732798230737855, + "loss": 3.3568, + "step": 1117 + }, + { + "epoch": 1.65, + "grad_norm": 3.4278273582458496, + "learning_rate": 0.00018730512045369568, + "loss": 3.4077, + "step": 1118 + }, + { + "epoch": 1.65, + "grad_norm": 3.177262306213379, + "learning_rate": 0.0001872822393934714, + "loss": 3.41, + "step": 1119 + }, + { + "epoch": 1.65, + "grad_norm": 3.8335347175598145, + "learning_rate": 0.00018725933913173938, + "loss": 3.3968, + "step": 1120 + }, + { + "epoch": 1.65, + "grad_norm": 3.43869948387146, + "learning_rate": 0.00018723641967353756, + "loss": 3.4295, + "step": 1121 + }, + { + "epoch": 1.65, + "grad_norm": 4.323538780212402, + "learning_rate": 0.00018721348102390803, + "loss": 3.3734, + "step": 1122 + }, + { + "epoch": 1.65, + "grad_norm": 3.6614975929260254, + "learning_rate": 0.00018719052318789717, + "loss": 3.3522, + "step": 1123 + }, + { + "epoch": 1.65, + "grad_norm": 3.442986249923706, + "learning_rate": 0.0001871675461705555, + "loss": 3.3743, + "step": 1124 + }, + { + "epoch": 1.66, + "grad_norm": 3.7068862915039062, + "learning_rate": 0.00018714454997693783, + "loss": 3.3611, + "step": 1125 + }, + { + "epoch": 1.66, + "grad_norm": 3.6128504276275635, + "learning_rate": 0.00018712153461210326, + "loss": 3.3678, + "step": 1126 + }, + { + "epoch": 1.66, + "grad_norm": 3.2057278156280518, + "learning_rate": 0.00018709850008111488, + "loss": 3.4118, + "step": 1127 + }, + { + "epoch": 1.66, + "grad_norm": 4.726739406585693, + "learning_rate": 0.0001870754463890402, + "loss": 3.3669, + "step": 1128 + }, + { + "epoch": 1.66, + "grad_norm": 4.606038570404053, + "learning_rate": 0.00018705237354095085, + "loss": 3.3316, + "step": 1129 + }, + { + "epoch": 1.66, + "grad_norm": 2.9650778770446777, + "learning_rate": 0.00018702928154192274, + "loss": 3.3614, + "step": 1130 + }, + { + "epoch": 1.66, + "grad_norm": 2.7509312629699707, + "learning_rate": 0.00018700617039703595, + "loss": 3.3797, + "step": 1131 + }, + { + "epoch": 1.67, + "grad_norm": 5.575433731079102, + "learning_rate": 0.00018698304011137475, + "loss": 3.384, + "step": 1132 + }, + { + "epoch": 1.67, + "grad_norm": 4.589602947235107, + "learning_rate": 0.00018695989069002764, + "loss": 3.3729, + "step": 1133 + }, + { + "epoch": 1.67, + "grad_norm": 2.780677318572998, + "learning_rate": 0.00018693672213808738, + "loss": 3.3704, + "step": 1134 + }, + { + "epoch": 1.67, + "grad_norm": 1.9864367246627808, + "learning_rate": 0.00018691353446065087, + "loss": 3.3718, + "step": 1135 + }, + { + "epoch": 1.67, + "grad_norm": 5.260591983795166, + "learning_rate": 0.00018689032766281924, + "loss": 3.3856, + "step": 1136 + }, + { + "epoch": 1.67, + "grad_norm": 4.4715447425842285, + "learning_rate": 0.00018686710174969787, + "loss": 3.3433, + "step": 1137 + }, + { + "epoch": 1.68, + "grad_norm": 3.0349228382110596, + "learning_rate": 0.00018684385672639622, + "loss": 3.3361, + "step": 1138 + }, + { + "epoch": 1.68, + "grad_norm": 2.5944900512695312, + "learning_rate": 0.0001868205925980281, + "loss": 3.3379, + "step": 1139 + }, + { + "epoch": 1.68, + "grad_norm": 4.727088928222656, + "learning_rate": 0.00018679730936971148, + "loss": 3.3889, + "step": 1140 + }, + { + "epoch": 1.68, + "grad_norm": 5.072381019592285, + "learning_rate": 0.0001867740070465685, + "loss": 3.4185, + "step": 1141 + }, + { + "epoch": 1.68, + "grad_norm": 2.782468318939209, + "learning_rate": 0.00018675068563372545, + "loss": 3.3635, + "step": 1142 + }, + { + "epoch": 1.68, + "grad_norm": 2.7385001182556152, + "learning_rate": 0.000186727345136313, + "loss": 3.4115, + "step": 1143 + }, + { + "epoch": 1.68, + "grad_norm": 4.811634540557861, + "learning_rate": 0.0001867039855594658, + "loss": 3.4155, + "step": 1144 + }, + { + "epoch": 1.69, + "grad_norm": 4.225504398345947, + "learning_rate": 0.00018668060690832285, + "loss": 3.3729, + "step": 1145 + }, + { + "epoch": 1.69, + "grad_norm": 3.26381254196167, + "learning_rate": 0.0001866572091880273, + "loss": 3.3982, + "step": 1146 + }, + { + "epoch": 1.69, + "grad_norm": 2.842423915863037, + "learning_rate": 0.00018663379240372647, + "loss": 3.387, + "step": 1147 + }, + { + "epoch": 1.69, + "grad_norm": 4.7502217292785645, + "learning_rate": 0.0001866103565605719, + "loss": 3.377, + "step": 1148 + }, + { + "epoch": 1.69, + "grad_norm": 4.280509948730469, + "learning_rate": 0.00018658690166371933, + "loss": 3.3774, + "step": 1149 + }, + { + "epoch": 1.69, + "grad_norm": 3.5164103507995605, + "learning_rate": 0.00018656342771832866, + "loss": 3.357, + "step": 1150 + }, + { + "epoch": 1.69, + "grad_norm": 2.891474485397339, + "learning_rate": 0.00018653993472956404, + "loss": 3.3755, + "step": 1151 + }, + { + "epoch": 1.7, + "grad_norm": 4.281400203704834, + "learning_rate": 0.00018651642270259373, + "loss": 3.3511, + "step": 1152 + }, + { + "epoch": 1.7, + "grad_norm": 4.510589122772217, + "learning_rate": 0.00018649289164259024, + "loss": 3.4113, + "step": 1153 + }, + { + "epoch": 1.7, + "grad_norm": 2.6343142986297607, + "learning_rate": 0.00018646934155473022, + "loss": 3.3291, + "step": 1154 + }, + { + "epoch": 1.7, + "grad_norm": 2.3589959144592285, + "learning_rate": 0.00018644577244419455, + "loss": 3.3638, + "step": 1155 + }, + { + "epoch": 1.7, + "grad_norm": 5.584714889526367, + "learning_rate": 0.00018642218431616828, + "loss": 3.32, + "step": 1156 + }, + { + "epoch": 1.7, + "grad_norm": 4.909949779510498, + "learning_rate": 0.00018639857717584065, + "loss": 3.3947, + "step": 1157 + }, + { + "epoch": 1.7, + "grad_norm": 2.9206135272979736, + "learning_rate": 0.00018637495102840506, + "loss": 3.3826, + "step": 1158 + }, + { + "epoch": 1.71, + "grad_norm": 2.58848237991333, + "learning_rate": 0.0001863513058790591, + "loss": 3.3237, + "step": 1159 + }, + { + "epoch": 1.71, + "grad_norm": 5.167590141296387, + "learning_rate": 0.00018632764173300457, + "loss": 3.3972, + "step": 1160 + }, + { + "epoch": 1.71, + "grad_norm": 5.174654483795166, + "learning_rate": 0.00018630395859544735, + "loss": 3.357, + "step": 1161 + }, + { + "epoch": 1.71, + "grad_norm": 2.867471218109131, + "learning_rate": 0.00018628025647159766, + "loss": 3.3406, + "step": 1162 + }, + { + "epoch": 1.71, + "grad_norm": 2.780794620513916, + "learning_rate": 0.00018625653536666974, + "loss": 3.3527, + "step": 1163 + }, + { + "epoch": 1.71, + "grad_norm": 4.502468585968018, + "learning_rate": 0.00018623279528588214, + "loss": 3.3216, + "step": 1164 + }, + { + "epoch": 1.72, + "grad_norm": 3.549867868423462, + "learning_rate": 0.00018620903623445747, + "loss": 3.3949, + "step": 1165 + }, + { + "epoch": 1.72, + "grad_norm": 3.7050774097442627, + "learning_rate": 0.00018618525821762256, + "loss": 3.395, + "step": 1166 + }, + { + "epoch": 1.72, + "grad_norm": 3.520019769668579, + "learning_rate": 0.00018616146124060846, + "loss": 3.3757, + "step": 1167 + }, + { + "epoch": 1.72, + "grad_norm": 3.870093822479248, + "learning_rate": 0.00018613764530865028, + "loss": 3.3712, + "step": 1168 + }, + { + "epoch": 1.72, + "grad_norm": 3.196948766708374, + "learning_rate": 0.0001861138104269874, + "loss": 3.3337, + "step": 1169 + }, + { + "epoch": 1.72, + "grad_norm": 4.416697025299072, + "learning_rate": 0.00018608995660086337, + "loss": 3.4137, + "step": 1170 + }, + { + "epoch": 1.72, + "grad_norm": 3.9429569244384766, + "learning_rate": 0.00018606608383552582, + "loss": 3.389, + "step": 1171 + }, + { + "epoch": 1.73, + "grad_norm": 3.530494213104248, + "learning_rate": 0.0001860421921362266, + "loss": 3.3752, + "step": 1172 + }, + { + "epoch": 1.73, + "grad_norm": 3.766986608505249, + "learning_rate": 0.00018601828150822177, + "loss": 3.3386, + "step": 1173 + }, + { + "epoch": 1.73, + "grad_norm": 3.8910138607025146, + "learning_rate": 0.00018599435195677146, + "loss": 3.348, + "step": 1174 + }, + { + "epoch": 1.73, + "grad_norm": 3.5032382011413574, + "learning_rate": 0.00018597040348714003, + "loss": 3.408, + "step": 1175 + }, + { + "epoch": 1.73, + "grad_norm": 3.5396668910980225, + "learning_rate": 0.000185946436104596, + "loss": 3.317, + "step": 1176 + }, + { + "epoch": 1.73, + "grad_norm": 3.7018423080444336, + "learning_rate": 0.00018592244981441197, + "loss": 3.3456, + "step": 1177 + }, + { + "epoch": 1.73, + "grad_norm": 3.41996169090271, + "learning_rate": 0.00018589844462186484, + "loss": 3.4147, + "step": 1178 + }, + { + "epoch": 1.74, + "grad_norm": 2.907388210296631, + "learning_rate": 0.00018587442053223557, + "loss": 3.3377, + "step": 1179 + }, + { + "epoch": 1.74, + "grad_norm": 4.651727676391602, + "learning_rate": 0.00018585037755080926, + "loss": 3.3401, + "step": 1180 + }, + { + "epoch": 1.74, + "grad_norm": 4.285563945770264, + "learning_rate": 0.00018582631568287523, + "loss": 3.3671, + "step": 1181 + }, + { + "epoch": 1.74, + "grad_norm": 3.3015687465667725, + "learning_rate": 0.00018580223493372695, + "loss": 3.3049, + "step": 1182 + }, + { + "epoch": 1.74, + "grad_norm": 2.8395988941192627, + "learning_rate": 0.000185778135308662, + "loss": 3.3588, + "step": 1183 + }, + { + "epoch": 1.74, + "eval_loss": 3.1614086627960205, + "eval_runtime": 1168.1043, + "eval_samples_per_second": 14.675, + "eval_steps_per_second": 0.918, + "step": 1183 + }, + { + "epoch": 1.74, + "grad_norm": 4.551041603088379, + "learning_rate": 0.00018575401681298216, + "loss": 3.3579, + "step": 1184 + }, + { + "epoch": 1.74, + "grad_norm": 4.050691604614258, + "learning_rate": 0.00018572987945199328, + "loss": 3.3174, + "step": 1185 + }, + { + "epoch": 1.75, + "grad_norm": 3.483586072921753, + "learning_rate": 0.0001857057232310055, + "loss": 3.3689, + "step": 1186 + }, + { + "epoch": 1.75, + "grad_norm": 3.430351734161377, + "learning_rate": 0.00018568154815533293, + "loss": 3.3624, + "step": 1187 + }, + { + "epoch": 1.75, + "grad_norm": 4.206818103790283, + "learning_rate": 0.00018565735423029404, + "loss": 3.3365, + "step": 1188 + }, + { + "epoch": 1.75, + "grad_norm": 3.6076526641845703, + "learning_rate": 0.00018563314146121122, + "loss": 3.3728, + "step": 1189 + }, + { + "epoch": 1.75, + "grad_norm": 3.844933271408081, + "learning_rate": 0.0001856089098534112, + "loss": 3.3577, + "step": 1190 + }, + { + "epoch": 1.75, + "grad_norm": 3.1116933822631836, + "learning_rate": 0.00018558465941222472, + "loss": 3.3242, + "step": 1191 + }, + { + "epoch": 1.76, + "grad_norm": 4.367002010345459, + "learning_rate": 0.00018556039014298672, + "loss": 3.3382, + "step": 1192 + }, + { + "epoch": 1.76, + "grad_norm": 4.229283332824707, + "learning_rate": 0.0001855361020510363, + "loss": 3.3318, + "step": 1193 + }, + { + "epoch": 1.76, + "grad_norm": 3.6189987659454346, + "learning_rate": 0.00018551179514171668, + "loss": 3.3879, + "step": 1194 + }, + { + "epoch": 1.76, + "grad_norm": 3.045738935470581, + "learning_rate": 0.0001854874694203752, + "loss": 3.3407, + "step": 1195 + }, + { + "epoch": 1.76, + "grad_norm": 4.931478977203369, + "learning_rate": 0.00018546312489236335, + "loss": 3.3428, + "step": 1196 + }, + { + "epoch": 1.76, + "grad_norm": 4.312923908233643, + "learning_rate": 0.00018543876156303676, + "loss": 3.3768, + "step": 1197 + }, + { + "epoch": 1.76, + "grad_norm": 3.236755132675171, + "learning_rate": 0.00018541437943775523, + "loss": 3.3637, + "step": 1198 + }, + { + "epoch": 1.77, + "grad_norm": 2.9794223308563232, + "learning_rate": 0.00018538997852188266, + "loss": 3.2961, + "step": 1199 + }, + { + "epoch": 1.77, + "grad_norm": 4.777314186096191, + "learning_rate": 0.00018536555882078706, + "loss": 3.3838, + "step": 1200 + }, + { + "epoch": 1.77, + "grad_norm": 4.084963321685791, + "learning_rate": 0.0001853411203398406, + "loss": 3.3525, + "step": 1201 + }, + { + "epoch": 1.77, + "grad_norm": 3.598912477493286, + "learning_rate": 0.0001853166630844196, + "loss": 3.3362, + "step": 1202 + }, + { + "epoch": 1.77, + "grad_norm": 3.44163179397583, + "learning_rate": 0.0001852921870599045, + "loss": 3.4129, + "step": 1203 + }, + { + "epoch": 1.77, + "grad_norm": 3.8749027252197266, + "learning_rate": 0.00018526769227167984, + "loss": 3.3069, + "step": 1204 + }, + { + "epoch": 1.77, + "grad_norm": 3.8452203273773193, + "learning_rate": 0.00018524317872513433, + "loss": 3.3084, + "step": 1205 + }, + { + "epoch": 1.78, + "grad_norm": 3.8901963233947754, + "learning_rate": 0.00018521864642566076, + "loss": 3.322, + "step": 1206 + }, + { + "epoch": 1.78, + "grad_norm": 3.340991973876953, + "learning_rate": 0.00018519409537865608, + "loss": 3.3373, + "step": 1207 + }, + { + "epoch": 1.78, + "grad_norm": 4.287864685058594, + "learning_rate": 0.0001851695255895214, + "loss": 3.2985, + "step": 1208 + }, + { + "epoch": 1.78, + "grad_norm": 3.954977512359619, + "learning_rate": 0.00018514493706366184, + "loss": 3.3286, + "step": 1209 + }, + { + "epoch": 1.78, + "grad_norm": 3.5455312728881836, + "learning_rate": 0.00018512032980648674, + "loss": 3.3191, + "step": 1210 + }, + { + "epoch": 1.78, + "grad_norm": 3.2507898807525635, + "learning_rate": 0.00018509570382340955, + "loss": 3.3356, + "step": 1211 + }, + { + "epoch": 1.79, + "grad_norm": 4.234015464782715, + "learning_rate": 0.0001850710591198478, + "loss": 3.365, + "step": 1212 + }, + { + "epoch": 1.79, + "grad_norm": 3.4031906127929688, + "learning_rate": 0.00018504639570122316, + "loss": 3.3405, + "step": 1213 + }, + { + "epoch": 1.79, + "grad_norm": 3.9651551246643066, + "learning_rate": 0.00018502171357296144, + "loss": 3.3953, + "step": 1214 + }, + { + "epoch": 1.79, + "grad_norm": 3.6755764484405518, + "learning_rate": 0.00018499701274049248, + "loss": 3.3102, + "step": 1215 + }, + { + "epoch": 1.79, + "grad_norm": 3.6327264308929443, + "learning_rate": 0.00018497229320925038, + "loss": 3.318, + "step": 1216 + }, + { + "epoch": 1.79, + "grad_norm": 3.4605491161346436, + "learning_rate": 0.0001849475549846732, + "loss": 3.3059, + "step": 1217 + }, + { + "epoch": 1.79, + "grad_norm": 3.7450873851776123, + "learning_rate": 0.00018492279807220326, + "loss": 3.3303, + "step": 1218 + }, + { + "epoch": 1.8, + "grad_norm": 3.5053107738494873, + "learning_rate": 0.00018489802247728684, + "loss": 3.3211, + "step": 1219 + }, + { + "epoch": 1.8, + "grad_norm": 3.7314960956573486, + "learning_rate": 0.00018487322820537444, + "loss": 3.3359, + "step": 1220 + }, + { + "epoch": 1.8, + "grad_norm": 3.397026777267456, + "learning_rate": 0.00018484841526192065, + "loss": 3.3199, + "step": 1221 + }, + { + "epoch": 1.8, + "grad_norm": 3.9689042568206787, + "learning_rate": 0.00018482358365238413, + "loss": 3.3249, + "step": 1222 + }, + { + "epoch": 1.8, + "grad_norm": 3.4028615951538086, + "learning_rate": 0.00018479873338222763, + "loss": 3.3234, + "step": 1223 + }, + { + "epoch": 1.8, + "grad_norm": 4.0719780921936035, + "learning_rate": 0.00018477386445691813, + "loss": 3.3755, + "step": 1224 + }, + { + "epoch": 1.8, + "grad_norm": 3.8761250972747803, + "learning_rate": 0.00018474897688192655, + "loss": 3.3313, + "step": 1225 + }, + { + "epoch": 1.81, + "grad_norm": 3.763307809829712, + "learning_rate": 0.00018472407066272804, + "loss": 3.3112, + "step": 1226 + }, + { + "epoch": 1.81, + "grad_norm": 3.3675754070281982, + "learning_rate": 0.00018469914580480177, + "loss": 3.373, + "step": 1227 + }, + { + "epoch": 1.81, + "grad_norm": 3.7994701862335205, + "learning_rate": 0.00018467420231363103, + "loss": 3.3034, + "step": 1228 + }, + { + "epoch": 1.81, + "grad_norm": 3.433627128601074, + "learning_rate": 0.00018464924019470327, + "loss": 3.2695, + "step": 1229 + }, + { + "epoch": 1.81, + "grad_norm": 3.7266974449157715, + "learning_rate": 0.00018462425945350996, + "loss": 3.3498, + "step": 1230 + }, + { + "epoch": 1.81, + "grad_norm": 3.44728422164917, + "learning_rate": 0.00018459926009554666, + "loss": 3.3682, + "step": 1231 + }, + { + "epoch": 1.81, + "grad_norm": 4.175528049468994, + "learning_rate": 0.00018457424212631312, + "loss": 3.3374, + "step": 1232 + }, + { + "epoch": 1.82, + "grad_norm": 4.267003059387207, + "learning_rate": 0.0001845492055513131, + "loss": 3.2985, + "step": 1233 + }, + { + "epoch": 1.82, + "grad_norm": 3.53112530708313, + "learning_rate": 0.0001845241503760545, + "loss": 3.3654, + "step": 1234 + }, + { + "epoch": 1.82, + "grad_norm": 3.1011736392974854, + "learning_rate": 0.00018449907660604927, + "loss": 3.3169, + "step": 1235 + }, + { + "epoch": 1.82, + "grad_norm": 4.862052917480469, + "learning_rate": 0.0001844739842468135, + "loss": 3.329, + "step": 1236 + }, + { + "epoch": 1.82, + "grad_norm": 4.767108917236328, + "learning_rate": 0.00018444887330386727, + "loss": 3.3546, + "step": 1237 + }, + { + "epoch": 1.82, + "grad_norm": 3.3005921840667725, + "learning_rate": 0.00018442374378273488, + "loss": 3.338, + "step": 1238 + }, + { + "epoch": 1.83, + "grad_norm": 3.2703089714050293, + "learning_rate": 0.00018439859568894463, + "loss": 3.281, + "step": 1239 + }, + { + "epoch": 1.83, + "grad_norm": 4.4580888748168945, + "learning_rate": 0.00018437342902802897, + "loss": 3.3748, + "step": 1240 + }, + { + "epoch": 1.83, + "grad_norm": 3.4954566955566406, + "learning_rate": 0.00018434824380552433, + "loss": 3.3291, + "step": 1241 + }, + { + "epoch": 1.83, + "grad_norm": 3.5984458923339844, + "learning_rate": 0.00018432304002697137, + "loss": 3.3834, + "step": 1242 + }, + { + "epoch": 1.83, + "grad_norm": 3.217467784881592, + "learning_rate": 0.00018429781769791468, + "loss": 3.3814, + "step": 1243 + }, + { + "epoch": 1.83, + "grad_norm": 4.149362087249756, + "learning_rate": 0.00018427257682390304, + "loss": 3.3808, + "step": 1244 + }, + { + "epoch": 1.83, + "grad_norm": 3.7322371006011963, + "learning_rate": 0.0001842473174104893, + "loss": 3.3546, + "step": 1245 + }, + { + "epoch": 1.84, + "grad_norm": 3.667970895767212, + "learning_rate": 0.0001842220394632303, + "loss": 3.3144, + "step": 1246 + }, + { + "epoch": 1.84, + "grad_norm": 3.353289842605591, + "learning_rate": 0.00018419674298768705, + "loss": 3.367, + "step": 1247 + }, + { + "epoch": 1.84, + "grad_norm": 4.078193664550781, + "learning_rate": 0.00018417142798942462, + "loss": 3.3284, + "step": 1248 + }, + { + "epoch": 1.84, + "grad_norm": 3.791130304336548, + "learning_rate": 0.0001841460944740121, + "loss": 3.3223, + "step": 1249 + }, + { + "epoch": 1.84, + "grad_norm": 3.9178853034973145, + "learning_rate": 0.00018412074244702274, + "loss": 3.4204, + "step": 1250 + }, + { + "epoch": 1.84, + "grad_norm": 3.1923162937164307, + "learning_rate": 0.00018409537191403378, + "loss": 3.2976, + "step": 1251 + }, + { + "epoch": 1.84, + "grad_norm": 4.827831745147705, + "learning_rate": 0.00018406998288062658, + "loss": 3.3697, + "step": 1252 + }, + { + "epoch": 1.85, + "grad_norm": 5.54097843170166, + "learning_rate": 0.00018404457535238657, + "loss": 3.3319, + "step": 1253 + }, + { + "epoch": 1.85, + "grad_norm": 2.249396562576294, + "learning_rate": 0.00018401914933490322, + "loss": 3.3106, + "step": 1254 + }, + { + "epoch": 1.85, + "grad_norm": 1.4117510318756104, + "learning_rate": 0.0001839937048337701, + "loss": 3.3664, + "step": 1255 + }, + { + "epoch": 1.85, + "grad_norm": 5.818370819091797, + "learning_rate": 0.0001839682418545848, + "loss": 3.3226, + "step": 1256 + }, + { + "epoch": 1.85, + "grad_norm": 4.127283573150635, + "learning_rate": 0.00018394276040294903, + "loss": 3.2937, + "step": 1257 + }, + { + "epoch": 1.85, + "grad_norm": 3.759086847305298, + "learning_rate": 0.00018391726048446852, + "loss": 3.3379, + "step": 1258 + }, + { + "epoch": 1.85, + "grad_norm": 4.152595520019531, + "learning_rate": 0.0001838917421047531, + "loss": 3.3433, + "step": 1259 + }, + { + "epoch": 1.86, + "grad_norm": 2.377365827560425, + "learning_rate": 0.00018386620526941662, + "loss": 3.3103, + "step": 1260 + }, + { + "epoch": 1.86, + "grad_norm": 1.4664210081100464, + "learning_rate": 0.00018384064998407706, + "loss": 3.3038, + "step": 1261 + }, + { + "epoch": 1.86, + "grad_norm": 5.424790859222412, + "learning_rate": 0.00018381507625435633, + "loss": 3.3518, + "step": 1262 + }, + { + "epoch": 1.86, + "grad_norm": 4.665937900543213, + "learning_rate": 0.00018378948408588058, + "loss": 3.3437, + "step": 1263 + }, + { + "epoch": 1.86, + "grad_norm": 2.893569231033325, + "learning_rate": 0.00018376387348427984, + "loss": 3.2693, + "step": 1264 + }, + { + "epoch": 1.86, + "grad_norm": 3.674450159072876, + "learning_rate": 0.0001837382444551883, + "loss": 3.3324, + "step": 1265 + }, + { + "epoch": 1.87, + "grad_norm": 2.98838210105896, + "learning_rate": 0.00018371259700424416, + "loss": 3.3242, + "step": 1266 + }, + { + "epoch": 1.87, + "grad_norm": 2.417462110519409, + "learning_rate": 0.00018368693113708968, + "loss": 3.3771, + "step": 1267 + }, + { + "epoch": 1.87, + "grad_norm": 4.9014458656311035, + "learning_rate": 0.00018366124685937122, + "loss": 3.3235, + "step": 1268 + }, + { + "epoch": 1.87, + "grad_norm": 4.8541693687438965, + "learning_rate": 0.00018363554417673915, + "loss": 3.3149, + "step": 1269 + }, + { + "epoch": 1.87, + "grad_norm": 3.2948014736175537, + "learning_rate": 0.0001836098230948478, + "loss": 3.3211, + "step": 1270 + }, + { + "epoch": 1.87, + "grad_norm": 3.0249054431915283, + "learning_rate": 0.00018358408361935573, + "loss": 3.3073, + "step": 1271 + }, + { + "epoch": 1.87, + "grad_norm": 4.5272722244262695, + "learning_rate": 0.00018355832575592541, + "loss": 3.2809, + "step": 1272 + }, + { + "epoch": 1.88, + "grad_norm": 3.790753126144409, + "learning_rate": 0.0001835325495102234, + "loss": 3.3502, + "step": 1273 + }, + { + "epoch": 1.88, + "grad_norm": 3.7797255516052246, + "learning_rate": 0.00018350675488792035, + "loss": 3.2818, + "step": 1274 + }, + { + "epoch": 1.88, + "grad_norm": 3.5615875720977783, + "learning_rate": 0.00018348094189469082, + "loss": 3.3617, + "step": 1275 + }, + { + "epoch": 1.88, + "grad_norm": 3.750861167907715, + "learning_rate": 0.00018345511053621355, + "loss": 3.3649, + "step": 1276 + }, + { + "epoch": 1.88, + "grad_norm": 3.202655076980591, + "learning_rate": 0.00018342926081817127, + "loss": 3.3013, + "step": 1277 + }, + { + "epoch": 1.88, + "grad_norm": 3.6577436923980713, + "learning_rate": 0.0001834033927462507, + "loss": 3.3074, + "step": 1278 + }, + { + "epoch": 1.88, + "grad_norm": 3.164114236831665, + "learning_rate": 0.0001833775063261427, + "loss": 3.3211, + "step": 1279 + }, + { + "epoch": 1.89, + "grad_norm": 3.7693588733673096, + "learning_rate": 0.00018335160156354206, + "loss": 3.3229, + "step": 1280 + }, + { + "epoch": 1.89, + "grad_norm": 3.5220518112182617, + "learning_rate": 0.00018332567846414768, + "loss": 3.304, + "step": 1281 + }, + { + "epoch": 1.89, + "grad_norm": 3.884568691253662, + "learning_rate": 0.00018329973703366242, + "loss": 3.2905, + "step": 1282 + }, + { + "epoch": 1.89, + "grad_norm": 3.471947431564331, + "learning_rate": 0.0001832737772777933, + "loss": 3.3356, + "step": 1283 + }, + { + "epoch": 1.89, + "grad_norm": 3.8152236938476562, + "learning_rate": 0.00018324779920225127, + "loss": 3.3036, + "step": 1284 + }, + { + "epoch": 1.89, + "grad_norm": 3.813997507095337, + "learning_rate": 0.0001832218028127513, + "loss": 3.3465, + "step": 1285 + }, + { + "epoch": 1.89, + "grad_norm": 3.5465102195739746, + "learning_rate": 0.00018319578811501244, + "loss": 3.3241, + "step": 1286 + }, + { + "epoch": 1.9, + "grad_norm": 3.25008487701416, + "learning_rate": 0.00018316975511475776, + "loss": 3.3111, + "step": 1287 + }, + { + "epoch": 1.9, + "grad_norm": 4.218858242034912, + "learning_rate": 0.00018314370381771432, + "loss": 3.2914, + "step": 1288 + }, + { + "epoch": 1.9, + "grad_norm": 3.7324070930480957, + "learning_rate": 0.00018311763422961325, + "loss": 3.3188, + "step": 1289 + }, + { + "epoch": 1.9, + "grad_norm": 3.4570472240448, + "learning_rate": 0.00018309154635618965, + "loss": 3.3034, + "step": 1290 + }, + { + "epoch": 1.9, + "grad_norm": 2.9619879722595215, + "learning_rate": 0.00018306544020318274, + "loss": 3.3231, + "step": 1291 + }, + { + "epoch": 1.9, + "grad_norm": 4.235932350158691, + "learning_rate": 0.00018303931577633566, + "loss": 3.3226, + "step": 1292 + }, + { + "epoch": 1.91, + "grad_norm": 3.5511438846588135, + "learning_rate": 0.00018301317308139561, + "loss": 3.3305, + "step": 1293 + }, + { + "epoch": 1.91, + "grad_norm": 3.656703233718872, + "learning_rate": 0.00018298701212411382, + "loss": 3.3261, + "step": 1294 + }, + { + "epoch": 1.91, + "grad_norm": 3.338941812515259, + "learning_rate": 0.00018296083291024552, + "loss": 3.3564, + "step": 1295 + }, + { + "epoch": 1.91, + "grad_norm": 3.9150142669677734, + "learning_rate": 0.00018293463544554993, + "loss": 3.3292, + "step": 1296 + }, + { + "epoch": 1.91, + "grad_norm": 10.699624061584473, + "learning_rate": 0.00018290841973579039, + "loss": 3.3042, + "step": 1297 + }, + { + "epoch": 1.91, + "grad_norm": 1.4264583587646484, + "learning_rate": 0.0001828821857867341, + "loss": 3.3268, + "step": 1298 + }, + { + "epoch": 1.91, + "grad_norm": 6.050600528717041, + "learning_rate": 0.0001828559336041524, + "loss": 3.3629, + "step": 1299 + }, + { + "epoch": 1.92, + "grad_norm": 6.324507236480713, + "learning_rate": 0.00018282966319382057, + "loss": 3.3402, + "step": 1300 + }, + { + "epoch": 1.92, + "grad_norm": 1.820309042930603, + "learning_rate": 0.00018280337456151797, + "loss": 3.2955, + "step": 1301 + }, + { + "epoch": 1.92, + "grad_norm": 7.433422565460205, + "learning_rate": 0.00018277706771302788, + "loss": 3.3309, + "step": 1302 + }, + { + "epoch": 1.92, + "grad_norm": 9.409677505493164, + "learning_rate": 0.00018275074265413762, + "loss": 3.4307, + "step": 1303 + }, + { + "epoch": 1.92, + "grad_norm": 5.268000602722168, + "learning_rate": 0.00018272439939063856, + "loss": 3.4344, + "step": 1304 + }, + { + "epoch": 1.92, + "grad_norm": 5.51171875, + "learning_rate": 0.00018269803792832603, + "loss": 3.3385, + "step": 1305 + }, + { + "epoch": 1.92, + "grad_norm": 5.903563976287842, + "learning_rate": 0.00018267165827299934, + "loss": 3.3243, + "step": 1306 + }, + { + "epoch": 1.93, + "grad_norm": 3.7264065742492676, + "learning_rate": 0.00018264526043046192, + "loss": 3.3759, + "step": 1307 + }, + { + "epoch": 1.93, + "grad_norm": 5.053271293640137, + "learning_rate": 0.00018261884440652106, + "loss": 3.3202, + "step": 1308 + }, + { + "epoch": 1.93, + "grad_norm": 2.33587646484375, + "learning_rate": 0.0001825924102069881, + "loss": 3.2899, + "step": 1309 + }, + { + "epoch": 1.93, + "grad_norm": 4.696453094482422, + "learning_rate": 0.00018256595783767843, + "loss": 3.359, + "step": 1310 + }, + { + "epoch": 1.93, + "grad_norm": 4.715218544006348, + "learning_rate": 0.00018253948730441132, + "loss": 3.3466, + "step": 1311 + }, + { + "epoch": 1.93, + "grad_norm": 2.377490997314453, + "learning_rate": 0.0001825129986130102, + "loss": 3.3496, + "step": 1312 + }, + { + "epoch": 1.94, + "grad_norm": 2.269477367401123, + "learning_rate": 0.00018248649176930235, + "loss": 3.3404, + "step": 1313 + }, + { + "epoch": 1.94, + "grad_norm": 4.577181816101074, + "learning_rate": 0.0001824599667791191, + "loss": 3.3864, + "step": 1314 + }, + { + "epoch": 1.94, + "grad_norm": 4.408148288726807, + "learning_rate": 0.0001824334236482958, + "loss": 3.3405, + "step": 1315 + }, + { + "epoch": 1.94, + "grad_norm": 2.9572770595550537, + "learning_rate": 0.00018240686238267173, + "loss": 3.323, + "step": 1316 + }, + { + "epoch": 1.94, + "grad_norm": 4.4048380851745605, + "learning_rate": 0.0001823802829880902, + "loss": 3.2676, + "step": 1317 + }, + { + "epoch": 1.94, + "grad_norm": 1.4889674186706543, + "learning_rate": 0.0001823536854703985, + "loss": 3.2902, + "step": 1318 + }, + { + "epoch": 1.94, + "grad_norm": 1.6742371320724487, + "learning_rate": 0.00018232706983544792, + "loss": 3.2945, + "step": 1319 + }, + { + "epoch": 1.95, + "grad_norm": 3.9357426166534424, + "learning_rate": 0.0001823004360890937, + "loss": 3.3486, + "step": 1320 + }, + { + "epoch": 1.95, + "grad_norm": 2.596331834793091, + "learning_rate": 0.0001822737842371951, + "loss": 3.3539, + "step": 1321 + }, + { + "epoch": 1.95, + "grad_norm": 4.866861820220947, + "learning_rate": 0.00018224711428561533, + "loss": 3.3449, + "step": 1322 + }, + { + "epoch": 1.95, + "grad_norm": 5.0658955574035645, + "learning_rate": 0.0001822204262402216, + "loss": 3.3107, + "step": 1323 + }, + { + "epoch": 1.95, + "grad_norm": 1.9593414068222046, + "learning_rate": 0.00018219372010688515, + "loss": 3.348, + "step": 1324 + }, + { + "epoch": 1.95, + "grad_norm": 2.256303310394287, + "learning_rate": 0.0001821669958914811, + "loss": 3.3012, + "step": 1325 + }, + { + "epoch": 1.95, + "grad_norm": 3.2346420288085938, + "learning_rate": 0.0001821402535998886, + "loss": 3.3372, + "step": 1326 + }, + { + "epoch": 1.96, + "grad_norm": 1.7795019149780273, + "learning_rate": 0.00018211349323799078, + "loss": 3.3459, + "step": 1327 + }, + { + "epoch": 1.96, + "grad_norm": 7.262651443481445, + "learning_rate": 0.00018208671481167478, + "loss": 3.3454, + "step": 1328 + }, + { + "epoch": 1.96, + "grad_norm": 8.501165390014648, + "learning_rate": 0.00018205991832683164, + "loss": 3.3952, + "step": 1329 + }, + { + "epoch": 1.96, + "grad_norm": 2.0047447681427, + "learning_rate": 0.00018203310378935638, + "loss": 3.3304, + "step": 1330 + }, + { + "epoch": 1.96, + "grad_norm": 6.586772918701172, + "learning_rate": 0.00018200627120514807, + "loss": 3.3235, + "step": 1331 + }, + { + "epoch": 1.96, + "grad_norm": 6.3678083419799805, + "learning_rate": 0.00018197942058010967, + "loss": 3.4105, + "step": 1332 + }, + { + "epoch": 1.96, + "grad_norm": 3.969449520111084, + "learning_rate": 0.00018195255192014816, + "loss": 3.3689, + "step": 1333 + }, + { + "epoch": 1.97, + "grad_norm": 5.26508092880249, + "learning_rate": 0.0001819256652311744, + "loss": 3.3417, + "step": 1334 + }, + { + "epoch": 1.97, + "grad_norm": 5.3484673500061035, + "learning_rate": 0.00018189876051910337, + "loss": 3.32, + "step": 1335 + }, + { + "epoch": 1.97, + "grad_norm": 3.664024829864502, + "learning_rate": 0.00018187183778985388, + "loss": 3.4004, + "step": 1336 + }, + { + "epoch": 1.97, + "grad_norm": 5.414148807525635, + "learning_rate": 0.00018184489704934877, + "loss": 3.2964, + "step": 1337 + }, + { + "epoch": 1.97, + "grad_norm": 1.5016615390777588, + "learning_rate": 0.0001818179383035148, + "loss": 3.3444, + "step": 1338 + }, + { + "epoch": 1.97, + "grad_norm": 7.925120830535889, + "learning_rate": 0.00018179096155828274, + "loss": 3.3881, + "step": 1339 + }, + { + "epoch": 1.98, + "grad_norm": 9.725903511047363, + "learning_rate": 0.00018176396681958724, + "loss": 3.3698, + "step": 1340 + }, + { + "epoch": 1.98, + "grad_norm": 5.562051773071289, + "learning_rate": 0.00018173695409336702, + "loss": 3.3285, + "step": 1341 + }, + { + "epoch": 1.98, + "grad_norm": 4.413424491882324, + "learning_rate": 0.00018170992338556468, + "loss": 3.3209, + "step": 1342 + }, + { + "epoch": 1.98, + "grad_norm": 5.979745864868164, + "learning_rate": 0.00018168287470212677, + "loss": 3.3173, + "step": 1343 + }, + { + "epoch": 1.98, + "grad_norm": 2.968111753463745, + "learning_rate": 0.00018165580804900384, + "loss": 3.3304, + "step": 1344 + }, + { + "epoch": 1.98, + "grad_norm": 6.121416091918945, + "learning_rate": 0.00018162872343215037, + "loss": 3.325, + "step": 1345 + }, + { + "epoch": 1.98, + "grad_norm": 5.006643772125244, + "learning_rate": 0.0001816016208575248, + "loss": 3.3311, + "step": 1346 + }, + { + "epoch": 1.99, + "grad_norm": 2.833677291870117, + "learning_rate": 0.00018157450033108954, + "loss": 3.3484, + "step": 1347 + }, + { + "epoch": 1.99, + "grad_norm": 3.661172389984131, + "learning_rate": 0.00018154736185881088, + "loss": 3.3119, + "step": 1348 + }, + { + "epoch": 1.99, + "grad_norm": 1.4361642599105835, + "learning_rate": 0.0001815202054466591, + "loss": 3.3744, + "step": 1349 + }, + { + "epoch": 1.99, + "grad_norm": 2.236370801925659, + "learning_rate": 0.00018149303110060846, + "loss": 3.285, + "step": 1350 + }, + { + "epoch": 1.99, + "grad_norm": 3.9948387145996094, + "learning_rate": 0.00018146583882663712, + "loss": 3.3107, + "step": 1351 + }, + { + "epoch": 1.99, + "grad_norm": 2.2754180431365967, + "learning_rate": 0.0001814386286307272, + "loss": 3.3646, + "step": 1352 + }, + { + "epoch": 1.99, + "eval_loss": 3.131267786026001, + "eval_runtime": 1163.1, + "eval_samples_per_second": 14.738, + "eval_steps_per_second": 0.922, + "step": 1352 + }, + { + "epoch": 1.99, + "grad_norm": 4.385812282562256, + "learning_rate": 0.00018141140051886474, + "loss": 3.3364, + "step": 1353 + }, + { + "epoch": 2.0, + "grad_norm": 2.7103238105773926, + "learning_rate": 0.0001813841544970398, + "loss": 3.3096, + "step": 1354 + }, + { + "epoch": 2.0, + "grad_norm": 4.108714580535889, + "learning_rate": 0.00018135689057124625, + "loss": 3.3506, + "step": 1355 + }, + { + "epoch": 2.0, + "grad_norm": 3.6100189685821533, + "learning_rate": 0.000181329608747482, + "loss": 3.2999, + "step": 1356 + }, + { + "epoch": 2.0, + "grad_norm": 3.728175640106201, + "learning_rate": 0.0001813023090317489, + "loss": 3.2945, + "step": 1357 + }, + { + "epoch": 2.0, + "grad_norm": 3.231055974960327, + "learning_rate": 0.00018127499143005268, + "loss": 3.3229, + "step": 1358 + }, + { + "epoch": 2.0, + "grad_norm": 3.6905510425567627, + "learning_rate": 0.00018124765594840298, + "loss": 3.3186, + "step": 1359 + }, + { + "epoch": 2.0, + "grad_norm": 3.330751657485962, + "learning_rate": 0.0001812203025928135, + "loss": 3.3526, + "step": 1360 + }, + { + "epoch": 2.01, + "grad_norm": 3.546685218811035, + "learning_rate": 0.00018119293136930175, + "loss": 3.2914, + "step": 1361 + }, + { + "epoch": 2.01, + "grad_norm": 3.202545404434204, + "learning_rate": 0.00018116554228388924, + "loss": 3.3155, + "step": 1362 + }, + { + "epoch": 2.01, + "grad_norm": 3.762091875076294, + "learning_rate": 0.00018113813534260137, + "loss": 3.2653, + "step": 1363 + }, + { + "epoch": 2.01, + "grad_norm": 3.2883477210998535, + "learning_rate": 0.00018111071055146752, + "loss": 3.2802, + "step": 1364 + }, + { + "epoch": 2.01, + "grad_norm": 4.180018424987793, + "learning_rate": 0.0001810832679165209, + "loss": 3.2743, + "step": 1365 + }, + { + "epoch": 2.01, + "grad_norm": 3.009850025177002, + "learning_rate": 0.00018105580744379873, + "loss": 3.3396, + "step": 1366 + }, + { + "epoch": 2.02, + "grad_norm": 3.3239614963531494, + "learning_rate": 0.00018102832913934215, + "loss": 3.302, + "step": 1367 + }, + { + "epoch": 2.0, + "grad_norm": 2.7158830165863037, + "learning_rate": 0.00018100083300919616, + "loss": 3.2927, + "step": 1368 + }, + { + "epoch": 2.0, + "grad_norm": 4.113391399383545, + "learning_rate": 0.00018097331905940977, + "loss": 3.258, + "step": 1369 + }, + { + "epoch": 2.0, + "grad_norm": 4.452334880828857, + "learning_rate": 0.00018094578729603583, + "loss": 3.2988, + "step": 1370 + }, + { + "epoch": 2.0, + "grad_norm": 2.8618204593658447, + "learning_rate": 0.00018091823772513118, + "loss": 3.2938, + "step": 1371 + }, + { + "epoch": 2.01, + "grad_norm": 2.671171188354492, + "learning_rate": 0.00018089067035275652, + "loss": 3.2756, + "step": 1372 + }, + { + "epoch": 2.01, + "grad_norm": 4.292674541473389, + "learning_rate": 0.0001808630851849765, + "loss": 3.2655, + "step": 1373 + }, + { + "epoch": 2.01, + "grad_norm": 4.163233757019043, + "learning_rate": 0.00018083548222785965, + "loss": 3.2825, + "step": 1374 + }, + { + "epoch": 2.01, + "grad_norm": 2.8590521812438965, + "learning_rate": 0.00018080786148747842, + "loss": 3.3127, + "step": 1375 + }, + { + "epoch": 2.01, + "grad_norm": 2.4040653705596924, + "learning_rate": 0.00018078022296990923, + "loss": 3.348, + "step": 1376 + }, + { + "epoch": 2.01, + "grad_norm": 5.037558555603027, + "learning_rate": 0.00018075256668123238, + "loss": 3.275, + "step": 1377 + }, + { + "epoch": 2.02, + "grad_norm": 4.536971569061279, + "learning_rate": 0.000180724892627532, + "loss": 3.2533, + "step": 1378 + }, + { + "epoch": 2.02, + "grad_norm": 2.9320995807647705, + "learning_rate": 0.00018069720081489624, + "loss": 3.3223, + "step": 1379 + }, + { + "epoch": 2.02, + "grad_norm": 2.6839616298675537, + "learning_rate": 0.00018066949124941712, + "loss": 3.3313, + "step": 1380 + }, + { + "epoch": 2.02, + "grad_norm": 3.9813764095306396, + "learning_rate": 0.00018064176393719055, + "loss": 3.2821, + "step": 1381 + }, + { + "epoch": 2.02, + "grad_norm": 3.4426040649414062, + "learning_rate": 0.00018061401888431634, + "loss": 3.3005, + "step": 1382 + }, + { + "epoch": 2.02, + "grad_norm": 3.2250571250915527, + "learning_rate": 0.00018058625609689824, + "loss": 3.2879, + "step": 1383 + }, + { + "epoch": 2.02, + "grad_norm": 2.6666173934936523, + "learning_rate": 0.00018055847558104383, + "loss": 3.2522, + "step": 1384 + }, + { + "epoch": 2.03, + "grad_norm": 4.8121843338012695, + "learning_rate": 0.00018053067734286472, + "loss": 3.3785, + "step": 1385 + }, + { + "epoch": 2.03, + "grad_norm": 4.593588352203369, + "learning_rate": 0.00018050286138847628, + "loss": 3.2831, + "step": 1386 + }, + { + "epoch": 2.03, + "grad_norm": 3.0059878826141357, + "learning_rate": 0.0001804750277239978, + "loss": 3.2606, + "step": 1387 + }, + { + "epoch": 2.03, + "grad_norm": 2.9211039543151855, + "learning_rate": 0.00018044717635555259, + "loss": 3.2735, + "step": 1388 + }, + { + "epoch": 2.03, + "grad_norm": 3.4243338108062744, + "learning_rate": 0.00018041930728926768, + "loss": 3.2794, + "step": 1389 + }, + { + "epoch": 2.03, + "grad_norm": 3.398374557495117, + "learning_rate": 0.00018039142053127412, + "loss": 3.3066, + "step": 1390 + }, + { + "epoch": 2.03, + "grad_norm": 3.260343074798584, + "learning_rate": 0.0001803635160877068, + "loss": 3.3, + "step": 1391 + }, + { + "epoch": 2.04, + "grad_norm": 2.542720079421997, + "learning_rate": 0.00018033559396470454, + "loss": 3.2654, + "step": 1392 + }, + { + "epoch": 2.04, + "grad_norm": 4.721008777618408, + "learning_rate": 0.00018030765416840997, + "loss": 3.3286, + "step": 1393 + }, + { + "epoch": 2.04, + "grad_norm": 4.219196319580078, + "learning_rate": 0.00018027969670496966, + "loss": 3.2769, + "step": 1394 + }, + { + "epoch": 2.04, + "grad_norm": 2.8509461879730225, + "learning_rate": 0.00018025172158053413, + "loss": 3.2763, + "step": 1395 + }, + { + "epoch": 2.04, + "grad_norm": 3.1871495246887207, + "learning_rate": 0.00018022372880125765, + "loss": 3.3038, + "step": 1396 + }, + { + "epoch": 2.04, + "grad_norm": 4.717142105102539, + "learning_rate": 0.00018019571837329847, + "loss": 3.2649, + "step": 1397 + }, + { + "epoch": 2.04, + "grad_norm": 3.9490749835968018, + "learning_rate": 0.00018016769030281872, + "loss": 3.2709, + "step": 1398 + }, + { + "epoch": 2.05, + "grad_norm": 3.04848051071167, + "learning_rate": 0.00018013964459598434, + "loss": 3.2638, + "step": 1399 + }, + { + "epoch": 2.05, + "grad_norm": 2.971536636352539, + "learning_rate": 0.00018011158125896522, + "loss": 3.3796, + "step": 1400 + }, + { + "epoch": 2.05, + "grad_norm": 3.643239736557007, + "learning_rate": 0.0001800835002979351, + "loss": 3.2766, + "step": 1401 + }, + { + "epoch": 2.05, + "grad_norm": 3.0714406967163086, + "learning_rate": 0.00018005540171907162, + "loss": 3.2691, + "step": 1402 + }, + { + "epoch": 2.05, + "grad_norm": 3.928359031677246, + "learning_rate": 0.00018002728552855629, + "loss": 3.2842, + "step": 1403 + }, + { + "epoch": 2.05, + "grad_norm": 3.7628214359283447, + "learning_rate": 0.00017999915173257444, + "loss": 3.3371, + "step": 1404 + }, + { + "epoch": 2.06, + "grad_norm": 3.1093904972076416, + "learning_rate": 0.00017997100033731534, + "loss": 3.2866, + "step": 1405 + }, + { + "epoch": 2.06, + "grad_norm": 2.735902786254883, + "learning_rate": 0.0001799428313489721, + "loss": 3.2598, + "step": 1406 + }, + { + "epoch": 2.06, + "grad_norm": 3.890085458755493, + "learning_rate": 0.00017991464477374173, + "loss": 3.2565, + "step": 1407 + }, + { + "epoch": 2.06, + "grad_norm": 3.5979950428009033, + "learning_rate": 0.00017988644061782508, + "loss": 3.3153, + "step": 1408 + }, + { + "epoch": 2.06, + "grad_norm": 3.224681854248047, + "learning_rate": 0.00017985821888742685, + "loss": 3.2969, + "step": 1409 + }, + { + "epoch": 2.06, + "grad_norm": 3.645235776901245, + "learning_rate": 0.00017982997958875565, + "loss": 3.2823, + "step": 1410 + }, + { + "epoch": 2.06, + "grad_norm": 3.3547754287719727, + "learning_rate": 0.000179801722728024, + "loss": 3.3546, + "step": 1411 + }, + { + "epoch": 2.07, + "grad_norm": 2.8428351879119873, + "learning_rate": 0.00017977344831144807, + "loss": 3.3096, + "step": 1412 + }, + { + "epoch": 2.07, + "grad_norm": 4.488086223602295, + "learning_rate": 0.00017974515634524817, + "loss": 3.2783, + "step": 1413 + }, + { + "epoch": 2.07, + "grad_norm": 4.082129955291748, + "learning_rate": 0.0001797168468356483, + "loss": 3.3131, + "step": 1414 + }, + { + "epoch": 2.07, + "grad_norm": 3.037198305130005, + "learning_rate": 0.00017968851978887636, + "loss": 3.3435, + "step": 1415 + }, + { + "epoch": 2.07, + "grad_norm": 2.8374056816101074, + "learning_rate": 0.0001796601752111641, + "loss": 3.3026, + "step": 1416 + }, + { + "epoch": 2.07, + "grad_norm": 4.270615577697754, + "learning_rate": 0.00017963181310874716, + "loss": 3.3384, + "step": 1417 + }, + { + "epoch": 2.07, + "grad_norm": 4.1699323654174805, + "learning_rate": 0.00017960343348786504, + "loss": 3.2869, + "step": 1418 + }, + { + "epoch": 2.08, + "grad_norm": 2.7383670806884766, + "learning_rate": 0.00017957503635476099, + "loss": 3.2882, + "step": 1419 + }, + { + "epoch": 2.08, + "grad_norm": 2.707303762435913, + "learning_rate": 0.00017954662171568226, + "loss": 3.3616, + "step": 1420 + }, + { + "epoch": 2.08, + "grad_norm": 5.056338787078857, + "learning_rate": 0.00017951818957687981, + "loss": 3.381, + "step": 1421 + }, + { + "epoch": 2.08, + "grad_norm": 4.49586820602417, + "learning_rate": 0.0001794897399446086, + "loss": 3.2837, + "step": 1422 + }, + { + "epoch": 2.08, + "grad_norm": 2.5468156337738037, + "learning_rate": 0.0001794612728251273, + "loss": 3.2547, + "step": 1423 + }, + { + "epoch": 2.08, + "grad_norm": 2.2053098678588867, + "learning_rate": 0.00017943278822469852, + "loss": 3.3216, + "step": 1424 + }, + { + "epoch": 2.09, + "grad_norm": 4.875980854034424, + "learning_rate": 0.00017940428614958865, + "loss": 3.3319, + "step": 1425 + }, + { + "epoch": 2.09, + "grad_norm": 4.466098308563232, + "learning_rate": 0.000179375766606068, + "loss": 3.276, + "step": 1426 + }, + { + "epoch": 2.09, + "grad_norm": 2.8682076930999756, + "learning_rate": 0.0001793472296004106, + "loss": 3.2971, + "step": 1427 + }, + { + "epoch": 2.09, + "grad_norm": 2.2296552658081055, + "learning_rate": 0.00017931867513889448, + "loss": 3.2946, + "step": 1428 + }, + { + "epoch": 2.09, + "grad_norm": 4.954530239105225, + "learning_rate": 0.0001792901032278014, + "loss": 3.269, + "step": 1429 + }, + { + "epoch": 2.09, + "grad_norm": 4.960486888885498, + "learning_rate": 0.00017926151387341697, + "loss": 3.2616, + "step": 1430 + }, + { + "epoch": 2.09, + "grad_norm": 2.2952184677124023, + "learning_rate": 0.0001792329070820307, + "loss": 3.3348, + "step": 1431 + }, + { + "epoch": 2.1, + "grad_norm": 1.9920496940612793, + "learning_rate": 0.00017920428285993583, + "loss": 3.3137, + "step": 1432 + }, + { + "epoch": 2.1, + "grad_norm": 4.211342811584473, + "learning_rate": 0.00017917564121342958, + "loss": 3.3246, + "step": 1433 + }, + { + "epoch": 2.1, + "grad_norm": 3.4940345287323, + "learning_rate": 0.00017914698214881284, + "loss": 3.3421, + "step": 1434 + }, + { + "epoch": 2.1, + "grad_norm": 3.7058966159820557, + "learning_rate": 0.00017911830567239048, + "loss": 3.3377, + "step": 1435 + }, + { + "epoch": 2.1, + "grad_norm": 3.2200000286102295, + "learning_rate": 0.00017908961179047105, + "loss": 3.3125, + "step": 1436 + }, + { + "epoch": 2.1, + "grad_norm": 3.529256582260132, + "learning_rate": 0.00017906090050936712, + "loss": 3.2833, + "step": 1437 + }, + { + "epoch": 2.1, + "grad_norm": 3.189732313156128, + "learning_rate": 0.0001790321718353949, + "loss": 3.2662, + "step": 1438 + }, + { + "epoch": 2.11, + "grad_norm": 3.6067655086517334, + "learning_rate": 0.0001790034257748745, + "loss": 3.2795, + "step": 1439 + }, + { + "epoch": 2.11, + "grad_norm": 3.4113893508911133, + "learning_rate": 0.0001789746623341299, + "loss": 3.2666, + "step": 1440 + }, + { + "epoch": 2.11, + "grad_norm": 3.466991662979126, + "learning_rate": 0.00017894588151948885, + "loss": 3.325, + "step": 1441 + }, + { + "epoch": 2.11, + "grad_norm": 3.057206153869629, + "learning_rate": 0.00017891708333728296, + "loss": 3.2636, + "step": 1442 + }, + { + "epoch": 2.11, + "grad_norm": 3.8873162269592285, + "learning_rate": 0.0001788882677938476, + "loss": 3.3139, + "step": 1443 + }, + { + "epoch": 2.11, + "grad_norm": 3.4515161514282227, + "learning_rate": 0.000178859434895522, + "loss": 3.2884, + "step": 1444 + }, + { + "epoch": 2.11, + "grad_norm": 14.146556854248047, + "learning_rate": 0.00017883058464864923, + "loss": 3.2571, + "step": 1445 + }, + { + "epoch": 2.12, + "grad_norm": 9.11373519897461, + "learning_rate": 0.00017880171705957613, + "loss": 3.295, + "step": 1446 + }, + { + "epoch": 2.12, + "grad_norm": 5.7905049324035645, + "learning_rate": 0.0001787728321346534, + "loss": 3.2532, + "step": 1447 + }, + { + "epoch": 2.12, + "grad_norm": 3.7080447673797607, + "learning_rate": 0.0001787439298802355, + "loss": 3.3201, + "step": 1448 + }, + { + "epoch": 2.12, + "grad_norm": 4.587584018707275, + "learning_rate": 0.00017871501030268072, + "loss": 3.2702, + "step": 1449 + }, + { + "epoch": 2.12, + "grad_norm": 0.9951494932174683, + "learning_rate": 0.0001786860734083512, + "loss": 3.2862, + "step": 1450 + }, + { + "epoch": 2.12, + "grad_norm": 0.8612810373306274, + "learning_rate": 0.00017865711920361292, + "loss": 3.2968, + "step": 1451 + }, + { + "epoch": 2.13, + "grad_norm": 2.446154832839966, + "learning_rate": 0.00017862814769483549, + "loss": 3.2937, + "step": 1452 + }, + { + "epoch": 2.13, + "grad_norm": 1.2420624494552612, + "learning_rate": 0.00017859915888839253, + "loss": 3.2922, + "step": 1453 + }, + { + "epoch": 2.13, + "grad_norm": 1.9885892868041992, + "learning_rate": 0.00017857015279066135, + "loss": 3.3043, + "step": 1454 + }, + { + "epoch": 2.13, + "grad_norm": 2.127197504043579, + "learning_rate": 0.00017854112940802314, + "loss": 3.3503, + "step": 1455 + }, + { + "epoch": 2.13, + "grad_norm": 1.1510310173034668, + "learning_rate": 0.0001785120887468628, + "loss": 3.2314, + "step": 1456 + }, + { + "epoch": 2.13, + "grad_norm": 2.6665754318237305, + "learning_rate": 0.00017848303081356912, + "loss": 3.2788, + "step": 1457 + }, + { + "epoch": 2.13, + "grad_norm": 1.0793408155441284, + "learning_rate": 0.0001784539556145346, + "loss": 3.2623, + "step": 1458 + }, + { + "epoch": 2.14, + "grad_norm": 2.212766647338867, + "learning_rate": 0.00017842486315615563, + "loss": 3.3542, + "step": 1459 + }, + { + "epoch": 2.14, + "grad_norm": 1.8812615871429443, + "learning_rate": 0.00017839575344483238, + "loss": 3.2481, + "step": 1460 + }, + { + "epoch": 2.14, + "grad_norm": 1.537009358406067, + "learning_rate": 0.0001783666264869687, + "loss": 3.2492, + "step": 1461 + }, + { + "epoch": 2.14, + "grad_norm": 1.0844135284423828, + "learning_rate": 0.00017833748228897241, + "loss": 3.3093, + "step": 1462 + }, + { + "epoch": 2.14, + "grad_norm": 1.4660224914550781, + "learning_rate": 0.000178308320857255, + "loss": 3.2598, + "step": 1463 + }, + { + "epoch": 2.14, + "grad_norm": 0.9054508209228516, + "learning_rate": 0.00017827914219823183, + "loss": 3.3132, + "step": 1464 + }, + { + "epoch": 2.14, + "grad_norm": 5.584422588348389, + "learning_rate": 0.00017824994631832197, + "loss": 3.2961, + "step": 1465 + }, + { + "epoch": 2.15, + "grad_norm": 1.9765104055404663, + "learning_rate": 0.0001782207332239483, + "loss": 3.2963, + "step": 1466 + }, + { + "epoch": 2.15, + "grad_norm": 1.4504390954971313, + "learning_rate": 0.00017819150292153757, + "loss": 3.3335, + "step": 1467 + }, + { + "epoch": 2.15, + "grad_norm": 1.2761657238006592, + "learning_rate": 0.0001781622554175202, + "loss": 3.2779, + "step": 1468 + }, + { + "epoch": 2.15, + "grad_norm": 2.1837310791015625, + "learning_rate": 0.00017813299071833046, + "loss": 3.2791, + "step": 1469 + }, + { + "epoch": 2.15, + "grad_norm": 2.7105600833892822, + "learning_rate": 0.00017810370883040636, + "loss": 3.2983, + "step": 1470 + }, + { + "epoch": 2.15, + "grad_norm": 1.1672013998031616, + "learning_rate": 0.00017807440976018977, + "loss": 3.2989, + "step": 1471 + }, + { + "epoch": 2.15, + "grad_norm": 4.486381530761719, + "learning_rate": 0.00017804509351412629, + "loss": 3.3132, + "step": 1472 + }, + { + "epoch": 2.16, + "grad_norm": 3.2585389614105225, + "learning_rate": 0.0001780157600986652, + "loss": 3.3249, + "step": 1473 + }, + { + "epoch": 2.16, + "grad_norm": 3.1671297550201416, + "learning_rate": 0.00017798640952025978, + "loss": 3.31, + "step": 1474 + }, + { + "epoch": 2.16, + "grad_norm": 2.345313310623169, + "learning_rate": 0.0001779570417853669, + "loss": 3.2382, + "step": 1475 + }, + { + "epoch": 2.16, + "grad_norm": 5.773190498352051, + "learning_rate": 0.0001779276569004473, + "loss": 3.2903, + "step": 1476 + }, + { + "epoch": 2.16, + "grad_norm": 4.879696846008301, + "learning_rate": 0.00017789825487196537, + "loss": 3.3107, + "step": 1477 + }, + { + "epoch": 2.16, + "grad_norm": 4.470544815063477, + "learning_rate": 0.00017786883570638942, + "loss": 3.3, + "step": 1478 + }, + { + "epoch": 2.17, + "grad_norm": 5.051728248596191, + "learning_rate": 0.00017783939941019153, + "loss": 3.2964, + "step": 1479 + }, + { + "epoch": 2.17, + "grad_norm": 1.2672648429870605, + "learning_rate": 0.00017780994598984737, + "loss": 3.297, + "step": 1480 + }, + { + "epoch": 2.17, + "grad_norm": 5.323600769042969, + "learning_rate": 0.0001777804754518366, + "loss": 3.2878, + "step": 1481 + }, + { + "epoch": 2.17, + "grad_norm": 4.052299976348877, + "learning_rate": 0.00017775098780264243, + "loss": 3.3135, + "step": 1482 + }, + { + "epoch": 2.17, + "grad_norm": 3.538057565689087, + "learning_rate": 0.00017772148304875205, + "loss": 3.3209, + "step": 1483 + }, + { + "epoch": 2.17, + "grad_norm": 2.5743610858917236, + "learning_rate": 0.0001776919611966563, + "loss": 3.2468, + "step": 1484 + }, + { + "epoch": 2.17, + "grad_norm": 6.518936634063721, + "learning_rate": 0.00017766242225284966, + "loss": 3.2914, + "step": 1485 + }, + { + "epoch": 2.18, + "grad_norm": 5.448997497558594, + "learning_rate": 0.00017763286622383067, + "loss": 3.3127, + "step": 1486 + }, + { + "epoch": 2.18, + "grad_norm": 2.5418717861175537, + "learning_rate": 0.00017760329311610135, + "loss": 3.2913, + "step": 1487 + }, + { + "epoch": 2.18, + "grad_norm": 4.122663974761963, + "learning_rate": 0.00017757370293616762, + "loss": 3.3257, + "step": 1488 + }, + { + "epoch": 2.18, + "grad_norm": 3.3550655841827393, + "learning_rate": 0.00017754409569053917, + "loss": 3.271, + "step": 1489 + }, + { + "epoch": 2.18, + "grad_norm": 2.8279335498809814, + "learning_rate": 0.0001775144713857293, + "loss": 3.2989, + "step": 1490 + }, + { + "epoch": 2.18, + "grad_norm": 2.4097659587860107, + "learning_rate": 0.0001774848300282552, + "loss": 3.2557, + "step": 1491 + }, + { + "epoch": 2.18, + "grad_norm": 2.702836275100708, + "learning_rate": 0.00017745517162463776, + "loss": 3.2719, + "step": 1492 + }, + { + "epoch": 2.19, + "grad_norm": 1.937203288078308, + "learning_rate": 0.00017742549618140167, + "loss": 3.2801, + "step": 1493 + }, + { + "epoch": 2.19, + "grad_norm": 2.2853002548217773, + "learning_rate": 0.00017739580370507532, + "loss": 3.4001, + "step": 1494 + }, + { + "epoch": 2.19, + "grad_norm": 3.39026141166687, + "learning_rate": 0.00017736609420219082, + "loss": 3.2562, + "step": 1495 + }, + { + "epoch": 2.19, + "grad_norm": 2.192948341369629, + "learning_rate": 0.00017733636767928407, + "loss": 3.2548, + "step": 1496 + }, + { + "epoch": 2.19, + "grad_norm": 3.966799259185791, + "learning_rate": 0.00017730662414289472, + "loss": 3.2744, + "step": 1497 + }, + { + "epoch": 2.19, + "grad_norm": 2.6247634887695312, + "learning_rate": 0.00017727686359956615, + "loss": 3.287, + "step": 1498 + }, + { + "epoch": 2.19, + "grad_norm": 2.959963083267212, + "learning_rate": 0.00017724708605584545, + "loss": 3.2864, + "step": 1499 + }, + { + "epoch": 2.2, + "grad_norm": 1.7757418155670166, + "learning_rate": 0.00017721729151828355, + "loss": 3.3003, + "step": 1500 + }, + { + "epoch": 2.2, + "grad_norm": 3.2801480293273926, + "learning_rate": 0.00017718747999343498, + "loss": 3.2662, + "step": 1501 + }, + { + "epoch": 2.2, + "grad_norm": 2.6395089626312256, + "learning_rate": 0.0001771576514878581, + "loss": 3.2991, + "step": 1502 + }, + { + "epoch": 2.2, + "grad_norm": 2.687575340270996, + "learning_rate": 0.00017712780600811497, + "loss": 3.2676, + "step": 1503 + }, + { + "epoch": 2.2, + "grad_norm": 2.702707052230835, + "learning_rate": 0.00017709794356077143, + "loss": 3.2862, + "step": 1504 + }, + { + "epoch": 2.2, + "grad_norm": 2.10098934173584, + "learning_rate": 0.000177068064152397, + "loss": 3.2425, + "step": 1505 + }, + { + "epoch": 2.21, + "grad_norm": 1.3301888704299927, + "learning_rate": 0.0001770381677895649, + "loss": 3.2325, + "step": 1506 + }, + { + "epoch": 2.21, + "grad_norm": 3.560911178588867, + "learning_rate": 0.0001770082544788522, + "loss": 3.2914, + "step": 1507 + }, + { + "epoch": 2.21, + "grad_norm": 2.01891827583313, + "learning_rate": 0.0001769783242268396, + "loss": 3.2213, + "step": 1508 + }, + { + "epoch": 2.21, + "grad_norm": 3.6540088653564453, + "learning_rate": 0.00017694837704011155, + "loss": 3.3066, + "step": 1509 + }, + { + "epoch": 2.21, + "grad_norm": 3.1314940452575684, + "learning_rate": 0.00017691841292525625, + "loss": 3.2576, + "step": 1510 + }, + { + "epoch": 2.21, + "grad_norm": 3.816842555999756, + "learning_rate": 0.0001768884318888656, + "loss": 3.2784, + "step": 1511 + }, + { + "epoch": 2.21, + "grad_norm": 3.5961854457855225, + "learning_rate": 0.0001768584339375352, + "loss": 3.2562, + "step": 1512 + }, + { + "epoch": 2.22, + "grad_norm": 3.9628171920776367, + "learning_rate": 0.0001768284190778644, + "loss": 3.2561, + "step": 1513 + }, + { + "epoch": 2.22, + "grad_norm": 4.071855068206787, + "learning_rate": 0.00017679838731645632, + "loss": 3.2924, + "step": 1514 + }, + { + "epoch": 2.22, + "grad_norm": 2.176236152648926, + "learning_rate": 0.00017676833865991765, + "loss": 3.309, + "step": 1515 + }, + { + "epoch": 2.22, + "grad_norm": 1.6997188329696655, + "learning_rate": 0.000176738273114859, + "loss": 3.2835, + "step": 1516 + }, + { + "epoch": 2.22, + "grad_norm": 3.275249719619751, + "learning_rate": 0.00017670819068789454, + "loss": 3.2506, + "step": 1517 + }, + { + "epoch": 2.22, + "grad_norm": 1.3623771667480469, + "learning_rate": 0.0001766780913856422, + "loss": 3.2054, + "step": 1518 + }, + { + "epoch": 2.22, + "grad_norm": 5.944447994232178, + "learning_rate": 0.00017664797521472364, + "loss": 3.2794, + "step": 1519 + }, + { + "epoch": 2.23, + "grad_norm": 5.1751484870910645, + "learning_rate": 0.0001766178421817642, + "loss": 3.2498, + "step": 1520 + }, + { + "epoch": 2.23, + "grad_norm": 2.410102128982544, + "learning_rate": 0.00017658769229339294, + "loss": 3.2672, + "step": 1521 + }, + { + "epoch": 2.23, + "eval_loss": 3.0885374546051025, + "eval_runtime": 1164.3656, + "eval_samples_per_second": 14.722, + "eval_steps_per_second": 0.921, + "step": 1521 + }, + { + "epoch": 2.23, + "grad_norm": 2.7669081687927246, + "learning_rate": 0.00017655752555624263, + "loss": 3.2525, + "step": 1522 + }, + { + "epoch": 2.23, + "grad_norm": 4.010193347930908, + "learning_rate": 0.00017652734197694977, + "loss": 3.2827, + "step": 1523 + }, + { + "epoch": 2.23, + "grad_norm": 3.233734130859375, + "learning_rate": 0.00017649714156215456, + "loss": 3.2372, + "step": 1524 + }, + { + "epoch": 2.23, + "grad_norm": 3.2297024726867676, + "learning_rate": 0.00017646692431850085, + "loss": 3.2763, + "step": 1525 + }, + { + "epoch": 2.24, + "grad_norm": 2.7911462783813477, + "learning_rate": 0.00017643669025263625, + "loss": 3.2702, + "step": 1526 + }, + { + "epoch": 2.24, + "grad_norm": 2.8759915828704834, + "learning_rate": 0.00017640643937121207, + "loss": 3.2338, + "step": 1527 + }, + { + "epoch": 2.24, + "grad_norm": 2.3804805278778076, + "learning_rate": 0.00017637617168088325, + "loss": 3.2646, + "step": 1528 + }, + { + "epoch": 2.24, + "grad_norm": 4.322963237762451, + "learning_rate": 0.00017634588718830853, + "loss": 3.2743, + "step": 1529 + }, + { + "epoch": 2.24, + "grad_norm": 3.6001713275909424, + "learning_rate": 0.00017631558590015028, + "loss": 3.3172, + "step": 1530 + }, + { + "epoch": 2.24, + "grad_norm": 3.2861557006835938, + "learning_rate": 0.00017628526782307455, + "loss": 3.2723, + "step": 1531 + }, + { + "epoch": 2.24, + "grad_norm": 3.6503138542175293, + "learning_rate": 0.0001762549329637512, + "loss": 3.234, + "step": 1532 + }, + { + "epoch": 2.25, + "grad_norm": 3.1032047271728516, + "learning_rate": 0.00017622458132885357, + "loss": 3.2845, + "step": 1533 + }, + { + "epoch": 2.25, + "grad_norm": 2.5854177474975586, + "learning_rate": 0.00017619421292505892, + "loss": 3.2251, + "step": 1534 + }, + { + "epoch": 2.25, + "grad_norm": 4.162723064422607, + "learning_rate": 0.00017616382775904806, + "loss": 3.2519, + "step": 1535 + }, + { + "epoch": 2.25, + "grad_norm": 3.7609925270080566, + "learning_rate": 0.0001761334258375055, + "loss": 3.334, + "step": 1536 + }, + { + "epoch": 2.25, + "grad_norm": 3.426297664642334, + "learning_rate": 0.00017610300716711948, + "loss": 3.2644, + "step": 1537 + }, + { + "epoch": 2.25, + "grad_norm": 3.2877280712127686, + "learning_rate": 0.0001760725717545819, + "loss": 3.2466, + "step": 1538 + }, + { + "epoch": 2.25, + "grad_norm": 3.2899694442749023, + "learning_rate": 0.00017604211960658837, + "loss": 3.2245, + "step": 1539 + }, + { + "epoch": 2.26, + "grad_norm": 3.202381134033203, + "learning_rate": 0.0001760116507298381, + "loss": 3.2512, + "step": 1540 + }, + { + "epoch": 2.26, + "grad_norm": 9.299320220947266, + "learning_rate": 0.0001759811651310341, + "loss": 3.2295, + "step": 1541 + }, + { + "epoch": 2.26, + "grad_norm": 6.585042476654053, + "learning_rate": 0.000175950662816883, + "loss": 3.2845, + "step": 1542 + }, + { + "epoch": 2.26, + "grad_norm": 3.786369562149048, + "learning_rate": 0.00017592014379409502, + "loss": 3.2796, + "step": 1543 + }, + { + "epoch": 2.26, + "grad_norm": 5.25980281829834, + "learning_rate": 0.00017588960806938424, + "loss": 3.2427, + "step": 1544 + }, + { + "epoch": 2.26, + "grad_norm": 7.450433254241943, + "learning_rate": 0.00017585905564946827, + "loss": 3.3147, + "step": 1545 + }, + { + "epoch": 2.26, + "grad_norm": 3.1463863849639893, + "learning_rate": 0.00017582848654106845, + "loss": 3.2311, + "step": 1546 + }, + { + "epoch": 2.27, + "grad_norm": 5.310190200805664, + "learning_rate": 0.00017579790075090976, + "loss": 3.3092, + "step": 1547 + }, + { + "epoch": 2.27, + "grad_norm": 5.930779933929443, + "learning_rate": 0.0001757672982857209, + "loss": 3.2788, + "step": 1548 + }, + { + "epoch": 2.27, + "grad_norm": 2.841097831726074, + "learning_rate": 0.00017573667915223414, + "loss": 3.1998, + "step": 1549 + }, + { + "epoch": 2.27, + "grad_norm": 7.447371959686279, + "learning_rate": 0.0001757060433571856, + "loss": 3.2633, + "step": 1550 + }, + { + "epoch": 2.27, + "grad_norm": 7.628240585327148, + "learning_rate": 0.00017567539090731482, + "loss": 3.2626, + "step": 1551 + }, + { + "epoch": 2.27, + "grad_norm": 1.3023099899291992, + "learning_rate": 0.0001756447218093652, + "loss": 3.2525, + "step": 1552 + }, + { + "epoch": 2.28, + "grad_norm": 8.924667358398438, + "learning_rate": 0.00017561403607008376, + "loss": 3.2892, + "step": 1553 + }, + { + "epoch": 2.28, + "grad_norm": 11.807168006896973, + "learning_rate": 0.00017558333369622112, + "loss": 3.3097, + "step": 1554 + }, + { + "epoch": 2.28, + "grad_norm": 9.348532676696777, + "learning_rate": 0.0001755526146945316, + "loss": 3.3024, + "step": 1555 + }, + { + "epoch": 2.28, + "grad_norm": 1.3859421014785767, + "learning_rate": 0.00017552187907177315, + "loss": 3.2629, + "step": 1556 + }, + { + "epoch": 2.28, + "grad_norm": 12.93152141571045, + "learning_rate": 0.00017549112683470748, + "loss": 3.3065, + "step": 1557 + }, + { + "epoch": 2.28, + "grad_norm": 18.00539779663086, + "learning_rate": 0.00017546035799009978, + "loss": 3.3714, + "step": 1558 + }, + { + "epoch": 2.28, + "grad_norm": 13.751264572143555, + "learning_rate": 0.00017542957254471908, + "loss": 3.3417, + "step": 1559 + }, + { + "epoch": 2.29, + "grad_norm": 4.594606876373291, + "learning_rate": 0.0001753987705053379, + "loss": 3.343, + "step": 1560 + }, + { + "epoch": 2.29, + "grad_norm": 7.840448379516602, + "learning_rate": 0.0001753679518787325, + "loss": 3.3199, + "step": 1561 + }, + { + "epoch": 2.29, + "grad_norm": 11.789816856384277, + "learning_rate": 0.0001753371166716828, + "loss": 3.3301, + "step": 1562 + }, + { + "epoch": 2.29, + "grad_norm": 11.082632064819336, + "learning_rate": 0.0001753062648909723, + "loss": 3.3078, + "step": 1563 + }, + { + "epoch": 2.29, + "grad_norm": 5.516722679138184, + "learning_rate": 0.0001752753965433882, + "loss": 3.2917, + "step": 1564 + }, + { + "epoch": 2.29, + "grad_norm": 8.504383087158203, + "learning_rate": 0.00017524451163572136, + "loss": 3.3195, + "step": 1565 + }, + { + "epoch": 2.29, + "grad_norm": 14.364538192749023, + "learning_rate": 0.0001752136101747662, + "loss": 3.3471, + "step": 1566 + }, + { + "epoch": 2.3, + "grad_norm": 12.66904067993164, + "learning_rate": 0.00017518269216732087, + "loss": 3.3574, + "step": 1567 + }, + { + "epoch": 2.3, + "grad_norm": 4.23660945892334, + "learning_rate": 0.00017515175762018714, + "loss": 3.3283, + "step": 1568 + }, + { + "epoch": 2.3, + "grad_norm": 6.301276206970215, + "learning_rate": 0.00017512080654017036, + "loss": 3.3064, + "step": 1569 + }, + { + "epoch": 2.3, + "grad_norm": 8.942106246948242, + "learning_rate": 0.00017508983893407958, + "loss": 3.2761, + "step": 1570 + }, + { + "epoch": 2.3, + "grad_norm": 6.445720672607422, + "learning_rate": 0.00017505885480872745, + "loss": 3.3033, + "step": 1571 + }, + { + "epoch": 2.3, + "grad_norm": 3.0577292442321777, + "learning_rate": 0.0001750278541709303, + "loss": 3.2916, + "step": 1572 + }, + { + "epoch": 2.3, + "grad_norm": 5.473748683929443, + "learning_rate": 0.00017499683702750803, + "loss": 3.2609, + "step": 1573 + }, + { + "epoch": 2.31, + "grad_norm": 2.3021624088287354, + "learning_rate": 0.00017496580338528426, + "loss": 3.2478, + "step": 1574 + }, + { + "epoch": 2.31, + "grad_norm": 6.796987056732178, + "learning_rate": 0.00017493475325108611, + "loss": 3.3177, + "step": 1575 + }, + { + "epoch": 2.31, + "grad_norm": 8.425848007202148, + "learning_rate": 0.00017490368663174443, + "loss": 3.2999, + "step": 1576 + }, + { + "epoch": 2.31, + "grad_norm": 4.728596210479736, + "learning_rate": 0.0001748726035340937, + "loss": 3.2693, + "step": 1577 + }, + { + "epoch": 2.31, + "grad_norm": 6.588244915008545, + "learning_rate": 0.00017484150396497193, + "loss": 3.2474, + "step": 1578 + }, + { + "epoch": 2.31, + "grad_norm": 10.9650239944458, + "learning_rate": 0.00017481038793122088, + "loss": 3.3038, + "step": 1579 + }, + { + "epoch": 2.32, + "grad_norm": 5.750149726867676, + "learning_rate": 0.00017477925543968586, + "loss": 3.315, + "step": 1580 + }, + { + "epoch": 2.32, + "grad_norm": 4.661263465881348, + "learning_rate": 0.00017474810649721576, + "loss": 3.2819, + "step": 1581 + }, + { + "epoch": 2.32, + "grad_norm": 6.578607082366943, + "learning_rate": 0.0001747169411106632, + "loss": 3.268, + "step": 1582 + }, + { + "epoch": 2.32, + "grad_norm": 3.776803970336914, + "learning_rate": 0.0001746857592868843, + "loss": 3.2414, + "step": 1583 + }, + { + "epoch": 2.32, + "grad_norm": 5.648257732391357, + "learning_rate": 0.00017465456103273885, + "loss": 3.2601, + "step": 1584 + }, + { + "epoch": 2.32, + "grad_norm": 7.4345011711120605, + "learning_rate": 0.00017462334635509033, + "loss": 3.3107, + "step": 1585 + }, + { + "epoch": 2.32, + "grad_norm": 2.6171629428863525, + "learning_rate": 0.0001745921152608057, + "loss": 3.226, + "step": 1586 + }, + { + "epoch": 2.33, + "grad_norm": 6.692351341247559, + "learning_rate": 0.00017456086775675562, + "loss": 3.283, + "step": 1587 + }, + { + "epoch": 2.33, + "grad_norm": 7.625064373016357, + "learning_rate": 0.00017452960384981432, + "loss": 3.2796, + "step": 1588 + }, + { + "epoch": 2.33, + "grad_norm": 4.043941974639893, + "learning_rate": 0.00017449832354685961, + "loss": 3.2948, + "step": 1589 + }, + { + "epoch": 2.33, + "grad_norm": 6.564191818237305, + "learning_rate": 0.00017446702685477303, + "loss": 3.2824, + "step": 1590 + }, + { + "epoch": 2.33, + "grad_norm": 10.199933052062988, + "learning_rate": 0.00017443571378043959, + "loss": 3.3447, + "step": 1591 + }, + { + "epoch": 2.33, + "grad_norm": 6.0920729637146, + "learning_rate": 0.00017440438433074794, + "loss": 3.2682, + "step": 1592 + }, + { + "epoch": 2.33, + "grad_norm": 3.95531964302063, + "learning_rate": 0.00017437303851259038, + "loss": 3.2439, + "step": 1593 + }, + { + "epoch": 2.34, + "grad_norm": 6.185577869415283, + "learning_rate": 0.0001743416763328628, + "loss": 3.3002, + "step": 1594 + }, + { + "epoch": 2.34, + "grad_norm": 2.757133960723877, + "learning_rate": 0.00017431029779846463, + "loss": 3.2716, + "step": 1595 + }, + { + "epoch": 2.34, + "grad_norm": 8.002574920654297, + "learning_rate": 0.00017427890291629893, + "loss": 3.2844, + "step": 1596 + }, + { + "epoch": 2.34, + "grad_norm": 10.051774024963379, + "learning_rate": 0.0001742474916932724, + "loss": 3.2388, + "step": 1597 + }, + { + "epoch": 2.34, + "grad_norm": 5.608651638031006, + "learning_rate": 0.00017421606413629527, + "loss": 3.226, + "step": 1598 + }, + { + "epoch": 2.34, + "grad_norm": 4.5539231300354, + "learning_rate": 0.00017418462025228145, + "loss": 3.2424, + "step": 1599 + }, + { + "epoch": 2.34, + "grad_norm": 6.649083614349365, + "learning_rate": 0.0001741531600481483, + "loss": 3.2537, + "step": 1600 + }, + { + "epoch": 2.35, + "grad_norm": 3.6664934158325195, + "learning_rate": 0.00017412168353081686, + "loss": 3.243, + "step": 1601 + }, + { + "epoch": 2.35, + "grad_norm": 6.799222469329834, + "learning_rate": 0.00017409019070721183, + "loss": 3.2853, + "step": 1602 + }, + { + "epoch": 2.35, + "grad_norm": 9.875252723693848, + "learning_rate": 0.00017405868158426135, + "loss": 3.2505, + "step": 1603 + }, + { + "epoch": 2.35, + "grad_norm": 4.99692440032959, + "learning_rate": 0.00017402715616889728, + "loss": 3.3176, + "step": 1604 + }, + { + "epoch": 2.35, + "grad_norm": 6.484687328338623, + "learning_rate": 0.00017399561446805492, + "loss": 3.2547, + "step": 1605 + }, + { + "epoch": 2.35, + "grad_norm": 8.336297988891602, + "learning_rate": 0.00017396405648867328, + "loss": 3.2543, + "step": 1606 + }, + { + "epoch": 2.36, + "grad_norm": 5.791290283203125, + "learning_rate": 0.00017393248223769491, + "loss": 3.2538, + "step": 1607 + }, + { + "epoch": 2.36, + "grad_norm": 2.4937000274658203, + "learning_rate": 0.00017390089172206592, + "loss": 3.2665, + "step": 1608 + }, + { + "epoch": 2.36, + "grad_norm": 5.263038635253906, + "learning_rate": 0.00017386928494873602, + "loss": 3.2899, + "step": 1609 + }, + { + "epoch": 2.36, + "grad_norm": 0.9264150857925415, + "learning_rate": 0.00017383766192465846, + "loss": 3.2284, + "step": 1610 + }, + { + "epoch": 2.36, + "grad_norm": 7.343024730682373, + "learning_rate": 0.00017380602265679014, + "loss": 3.3191, + "step": 1611 + }, + { + "epoch": 2.36, + "grad_norm": 8.518827438354492, + "learning_rate": 0.00017377436715209144, + "loss": 3.3296, + "step": 1612 + }, + { + "epoch": 2.36, + "grad_norm": 5.042328357696533, + "learning_rate": 0.00017374269541752638, + "loss": 3.2711, + "step": 1613 + }, + { + "epoch": 2.37, + "grad_norm": 5.329464435577393, + "learning_rate": 0.00017371100746006252, + "loss": 3.2886, + "step": 1614 + }, + { + "epoch": 2.37, + "grad_norm": 8.302837371826172, + "learning_rate": 0.00017367930328667105, + "loss": 3.2565, + "step": 1615 + }, + { + "epoch": 2.37, + "grad_norm": 4.5046539306640625, + "learning_rate": 0.0001736475829043266, + "loss": 3.2309, + "step": 1616 + }, + { + "epoch": 2.37, + "grad_norm": 5.1940789222717285, + "learning_rate": 0.0001736158463200075, + "loss": 3.2687, + "step": 1617 + }, + { + "epoch": 2.37, + "grad_norm": 7.008042335510254, + "learning_rate": 0.00017358409354069554, + "loss": 3.2845, + "step": 1618 + }, + { + "epoch": 2.37, + "grad_norm": 3.637481689453125, + "learning_rate": 0.00017355232457337615, + "loss": 3.2221, + "step": 1619 + }, + { + "epoch": 2.37, + "grad_norm": 6.158731937408447, + "learning_rate": 0.00017352053942503828, + "loss": 3.2727, + "step": 1620 + }, + { + "epoch": 2.38, + "grad_norm": 9.95582103729248, + "learning_rate": 0.00017348873810267443, + "loss": 3.2826, + "step": 1621 + }, + { + "epoch": 2.38, + "grad_norm": 5.530641555786133, + "learning_rate": 0.0001734569206132807, + "loss": 3.291, + "step": 1622 + }, + { + "epoch": 2.38, + "grad_norm": 3.868178367614746, + "learning_rate": 0.00017342508696385674, + "loss": 3.2794, + "step": 1623 + }, + { + "epoch": 2.38, + "grad_norm": 6.0412516593933105, + "learning_rate": 0.00017339323716140573, + "loss": 3.2546, + "step": 1624 + }, + { + "epoch": 2.38, + "grad_norm": 2.217486619949341, + "learning_rate": 0.00017336137121293436, + "loss": 3.2526, + "step": 1625 + }, + { + "epoch": 2.38, + "grad_norm": 8.725723266601562, + "learning_rate": 0.000173329489125453, + "loss": 3.253, + "step": 1626 + }, + { + "epoch": 2.39, + "grad_norm": 12.251547813415527, + "learning_rate": 0.00017329759090597548, + "loss": 3.2593, + "step": 1627 + }, + { + "epoch": 2.39, + "grad_norm": 7.2429070472717285, + "learning_rate": 0.00017326567656151916, + "loss": 3.2555, + "step": 1628 + }, + { + "epoch": 2.39, + "grad_norm": 2.993010997772217, + "learning_rate": 0.000173233746099105, + "loss": 3.2165, + "step": 1629 + }, + { + "epoch": 2.39, + "grad_norm": 5.157336711883545, + "learning_rate": 0.0001732017995257575, + "loss": 3.3013, + "step": 1630 + }, + { + "epoch": 2.39, + "grad_norm": 2.0820674896240234, + "learning_rate": 0.00017316983684850467, + "loss": 3.2088, + "step": 1631 + }, + { + "epoch": 2.39, + "grad_norm": 8.305849075317383, + "learning_rate": 0.0001731378580743781, + "loss": 3.2902, + "step": 1632 + }, + { + "epoch": 2.39, + "grad_norm": 10.2841157913208, + "learning_rate": 0.00017310586321041288, + "loss": 3.2437, + "step": 1633 + }, + { + "epoch": 2.4, + "grad_norm": 4.831589698791504, + "learning_rate": 0.00017307385226364772, + "loss": 3.2177, + "step": 1634 + }, + { + "epoch": 2.4, + "grad_norm": 4.824088096618652, + "learning_rate": 0.00017304182524112478, + "loss": 3.2779, + "step": 1635 + }, + { + "epoch": 2.4, + "grad_norm": 7.442866802215576, + "learning_rate": 0.00017300978214988976, + "loss": 3.2716, + "step": 1636 + }, + { + "epoch": 2.4, + "grad_norm": 4.1812968254089355, + "learning_rate": 0.000172977722996992, + "loss": 3.2746, + "step": 1637 + }, + { + "epoch": 2.4, + "grad_norm": 5.722605228424072, + "learning_rate": 0.00017294564778948422, + "loss": 3.2238, + "step": 1638 + }, + { + "epoch": 2.4, + "grad_norm": 8.271087646484375, + "learning_rate": 0.00017291355653442278, + "loss": 3.3323, + "step": 1639 + }, + { + "epoch": 2.4, + "grad_norm": 3.2383551597595215, + "learning_rate": 0.00017288144923886757, + "loss": 3.2318, + "step": 1640 + }, + { + "epoch": 2.41, + "grad_norm": 6.346744537353516, + "learning_rate": 0.00017284932590988197, + "loss": 3.2569, + "step": 1641 + }, + { + "epoch": 2.41, + "grad_norm": 8.613646507263184, + "learning_rate": 0.00017281718655453286, + "loss": 3.2684, + "step": 1642 + }, + { + "epoch": 2.41, + "grad_norm": 5.7413129806518555, + "learning_rate": 0.00017278503117989073, + "loss": 3.2374, + "step": 1643 + }, + { + "epoch": 2.41, + "grad_norm": 3.2419638633728027, + "learning_rate": 0.0001727528597930295, + "loss": 3.2127, + "step": 1644 + }, + { + "epoch": 2.41, + "grad_norm": 6.177962779998779, + "learning_rate": 0.0001727206724010267, + "loss": 3.2859, + "step": 1645 + }, + { + "epoch": 2.41, + "grad_norm": 1.907907247543335, + "learning_rate": 0.0001726884690109633, + "loss": 3.2221, + "step": 1646 + }, + { + "epoch": 2.41, + "grad_norm": 6.6156415939331055, + "learning_rate": 0.0001726562496299239, + "loss": 3.2339, + "step": 1647 + }, + { + "epoch": 2.42, + "grad_norm": 8.058192253112793, + "learning_rate": 0.00017262401426499646, + "loss": 3.2718, + "step": 1648 + }, + { + "epoch": 2.42, + "grad_norm": 4.197020530700684, + "learning_rate": 0.00017259176292327259, + "loss": 3.3056, + "step": 1649 + }, + { + "epoch": 2.42, + "grad_norm": 5.8328166007995605, + "learning_rate": 0.00017255949561184738, + "loss": 3.2645, + "step": 1650 + }, + { + "epoch": 2.42, + "grad_norm": 9.331326484680176, + "learning_rate": 0.00017252721233781938, + "loss": 3.3117, + "step": 1651 + }, + { + "epoch": 2.42, + "grad_norm": 5.499820232391357, + "learning_rate": 0.00017249491310829078, + "loss": 3.278, + "step": 1652 + }, + { + "epoch": 2.42, + "grad_norm": 4.50565242767334, + "learning_rate": 0.00017246259793036707, + "loss": 3.2662, + "step": 1653 + }, + { + "epoch": 2.43, + "grad_norm": 6.363594055175781, + "learning_rate": 0.00017243026681115748, + "loss": 3.2784, + "step": 1654 + }, + { + "epoch": 2.43, + "grad_norm": 2.7347965240478516, + "learning_rate": 0.00017239791975777458, + "loss": 3.2907, + "step": 1655 + }, + { + "epoch": 2.43, + "grad_norm": 7.49398946762085, + "learning_rate": 0.00017236555677733453, + "loss": 3.3239, + "step": 1656 + }, + { + "epoch": 2.43, + "grad_norm": 10.45259952545166, + "learning_rate": 0.00017233317787695697, + "loss": 3.316, + "step": 1657 + }, + { + "epoch": 2.43, + "grad_norm": 5.672200679779053, + "learning_rate": 0.00017230078306376502, + "loss": 3.2043, + "step": 1658 + }, + { + "epoch": 2.43, + "grad_norm": 4.071951866149902, + "learning_rate": 0.00017226837234488534, + "loss": 3.2218, + "step": 1659 + }, + { + "epoch": 2.43, + "grad_norm": 5.800225257873535, + "learning_rate": 0.00017223594572744806, + "loss": 3.2586, + "step": 1660 + }, + { + "epoch": 2.44, + "grad_norm": 2.592320442199707, + "learning_rate": 0.0001722035032185868, + "loss": 3.244, + "step": 1661 + }, + { + "epoch": 2.44, + "grad_norm": 7.940516471862793, + "learning_rate": 0.00017217104482543876, + "loss": 3.2501, + "step": 1662 + }, + { + "epoch": 2.44, + "grad_norm": 10.571852684020996, + "learning_rate": 0.0001721385705551445, + "loss": 3.2631, + "step": 1663 + }, + { + "epoch": 2.44, + "grad_norm": 5.3643574714660645, + "learning_rate": 0.0001721060804148482, + "loss": 3.2116, + "step": 1664 + }, + { + "epoch": 2.44, + "grad_norm": 4.0979905128479, + "learning_rate": 0.00017207357441169742, + "loss": 3.1886, + "step": 1665 + }, + { + "epoch": 2.44, + "grad_norm": 6.099160194396973, + "learning_rate": 0.0001720410525528433, + "loss": 3.2502, + "step": 1666 + }, + { + "epoch": 2.44, + "grad_norm": 3.1805484294891357, + "learning_rate": 0.00017200851484544037, + "loss": 3.213, + "step": 1667 + }, + { + "epoch": 2.45, + "grad_norm": 7.588128089904785, + "learning_rate": 0.00017197596129664682, + "loss": 3.2935, + "step": 1668 + }, + { + "epoch": 2.45, + "grad_norm": 10.550485610961914, + "learning_rate": 0.0001719433919136241, + "loss": 3.2823, + "step": 1669 + }, + { + "epoch": 2.45, + "grad_norm": 5.52534818649292, + "learning_rate": 0.00017191080670353733, + "loss": 3.2452, + "step": 1670 + }, + { + "epoch": 2.45, + "grad_norm": 4.016864776611328, + "learning_rate": 0.00017187820567355497, + "loss": 3.2412, + "step": 1671 + }, + { + "epoch": 2.45, + "grad_norm": 6.307382583618164, + "learning_rate": 0.0001718455888308491, + "loss": 3.2699, + "step": 1672 + }, + { + "epoch": 2.45, + "grad_norm": 2.715496301651001, + "learning_rate": 0.00017181295618259515, + "loss": 3.2695, + "step": 1673 + }, + { + "epoch": 2.45, + "grad_norm": 7.303908348083496, + "learning_rate": 0.00017178030773597213, + "loss": 3.2392, + "step": 1674 + }, + { + "epoch": 2.46, + "grad_norm": 10.16519546508789, + "learning_rate": 0.00017174764349816245, + "loss": 3.2635, + "step": 1675 + }, + { + "epoch": 2.46, + "grad_norm": 5.080345630645752, + "learning_rate": 0.00017171496347635206, + "loss": 3.2302, + "step": 1676 + }, + { + "epoch": 2.46, + "grad_norm": 5.1449408531188965, + "learning_rate": 0.00017168226767773028, + "loss": 3.2135, + "step": 1677 + }, + { + "epoch": 2.46, + "grad_norm": 7.340461254119873, + "learning_rate": 0.00017164955610949004, + "loss": 3.2561, + "step": 1678 + }, + { + "epoch": 2.46, + "grad_norm": 3.821507453918457, + "learning_rate": 0.00017161682877882758, + "loss": 3.2111, + "step": 1679 + }, + { + "epoch": 2.46, + "grad_norm": 6.017604351043701, + "learning_rate": 0.0001715840856929428, + "loss": 3.3209, + "step": 1680 + }, + { + "epoch": 2.47, + "grad_norm": 8.912531852722168, + "learning_rate": 0.00017155132685903887, + "loss": 3.2441, + "step": 1681 + }, + { + "epoch": 2.47, + "grad_norm": 4.574288368225098, + "learning_rate": 0.00017151855228432253, + "loss": 3.2489, + "step": 1682 + }, + { + "epoch": 2.47, + "grad_norm": 4.964592456817627, + "learning_rate": 0.00017148576197600402, + "loss": 3.2607, + "step": 1683 + }, + { + "epoch": 2.47, + "grad_norm": 7.388319969177246, + "learning_rate": 0.00017145295594129693, + "loss": 3.2537, + "step": 1684 + }, + { + "epoch": 2.47, + "grad_norm": 4.345710277557373, + "learning_rate": 0.0001714201341874184, + "loss": 3.2435, + "step": 1685 + }, + { + "epoch": 2.47, + "grad_norm": 5.151455402374268, + "learning_rate": 0.000171387296721589, + "loss": 3.2198, + "step": 1686 + }, + { + "epoch": 2.47, + "grad_norm": 9.679162979125977, + "learning_rate": 0.0001713544435510327, + "loss": 3.3015, + "step": 1687 + }, + { + "epoch": 2.48, + "grad_norm": 4.5911383628845215, + "learning_rate": 0.00017132157468297704, + "loss": 3.2534, + "step": 1688 + }, + { + "epoch": 2.48, + "grad_norm": 3.5955042839050293, + "learning_rate": 0.00017128869012465292, + "loss": 3.2479, + "step": 1689 + }, + { + "epoch": 2.48, + "grad_norm": 4.769717693328857, + "learning_rate": 0.00017125578988329472, + "loss": 3.2706, + "step": 1690 + }, + { + "epoch": 2.48, + "eval_loss": 3.0677621364593506, + "eval_runtime": 1164.46, + "eval_samples_per_second": 14.721, + "eval_steps_per_second": 0.921, + "step": 1690 + }, + { + "epoch": 2.48, + "grad_norm": 0.9628227949142456, + "learning_rate": 0.0001712228739661403, + "loss": 3.2333, + "step": 1691 + }, + { + "epoch": 2.48, + "grad_norm": 9.899592399597168, + "learning_rate": 0.00017118994238043088, + "loss": 3.287, + "step": 1692 + }, + { + "epoch": 2.48, + "grad_norm": 13.610316276550293, + "learning_rate": 0.00017115699513341127, + "loss": 3.3172, + "step": 1693 + }, + { + "epoch": 2.48, + "grad_norm": 8.333012580871582, + "learning_rate": 0.00017112403223232964, + "loss": 3.2552, + "step": 1694 + }, + { + "epoch": 2.49, + "grad_norm": 1.985851764678955, + "learning_rate": 0.0001710910536844375, + "loss": 3.2494, + "step": 1695 + }, + { + "epoch": 2.49, + "grad_norm": 4.960620403289795, + "learning_rate": 0.00017105805949699, + "loss": 3.2411, + "step": 1696 + }, + { + "epoch": 2.49, + "grad_norm": 2.77993106842041, + "learning_rate": 0.00017102504967724563, + "loss": 3.264, + "step": 1697 + }, + { + "epoch": 2.49, + "grad_norm": 6.050425052642822, + "learning_rate": 0.0001709920242324663, + "loss": 3.2725, + "step": 1698 + }, + { + "epoch": 2.49, + "grad_norm": 6.0771307945251465, + "learning_rate": 0.00017095898316991744, + "loss": 3.199, + "step": 1699 + }, + { + "epoch": 2.49, + "grad_norm": 1.5405139923095703, + "learning_rate": 0.0001709259264968678, + "loss": 3.2523, + "step": 1700 + }, + { + "epoch": 2.49, + "grad_norm": 5.121715068817139, + "learning_rate": 0.00017089285422058968, + "loss": 3.2944, + "step": 1701 + }, + { + "epoch": 2.5, + "grad_norm": 4.806593894958496, + "learning_rate": 0.00017085976634835873, + "loss": 3.2461, + "step": 1702 + }, + { + "epoch": 2.5, + "grad_norm": 3.7719340324401855, + "learning_rate": 0.00017082666288745403, + "loss": 3.2067, + "step": 1703 + }, + { + "epoch": 2.5, + "grad_norm": 3.7723586559295654, + "learning_rate": 0.0001707935438451582, + "loss": 3.2715, + "step": 1704 + }, + { + "epoch": 2.5, + "grad_norm": 2.5075759887695312, + "learning_rate": 0.00017076040922875713, + "loss": 3.2053, + "step": 1705 + }, + { + "epoch": 2.5, + "grad_norm": 2.315262794494629, + "learning_rate": 0.00017072725904554025, + "loss": 3.2529, + "step": 1706 + }, + { + "epoch": 2.5, + "grad_norm": 3.849294900894165, + "learning_rate": 0.0001706940933028004, + "loss": 3.2501, + "step": 1707 + }, + { + "epoch": 2.51, + "grad_norm": 3.2091238498687744, + "learning_rate": 0.00017066091200783375, + "loss": 3.1893, + "step": 1708 + }, + { + "epoch": 2.51, + "grad_norm": 3.587019443511963, + "learning_rate": 0.00017062771516794002, + "loss": 3.2435, + "step": 1709 + }, + { + "epoch": 2.51, + "grad_norm": 2.367382526397705, + "learning_rate": 0.00017059450279042232, + "loss": 3.2453, + "step": 1710 + }, + { + "epoch": 2.51, + "grad_norm": 4.0962700843811035, + "learning_rate": 0.00017056127488258705, + "loss": 3.2043, + "step": 1711 + }, + { + "epoch": 2.51, + "grad_norm": 4.637650966644287, + "learning_rate": 0.0001705280314517442, + "loss": 3.2506, + "step": 1712 + }, + { + "epoch": 2.51, + "grad_norm": 1.244682788848877, + "learning_rate": 0.00017049477250520714, + "loss": 3.2169, + "step": 1713 + }, + { + "epoch": 2.51, + "grad_norm": 0.7056719660758972, + "learning_rate": 0.00017046149805029253, + "loss": 3.212, + "step": 1714 + }, + { + "epoch": 2.52, + "grad_norm": 2.6713907718658447, + "learning_rate": 0.00017042820809432056, + "loss": 3.2665, + "step": 1715 + }, + { + "epoch": 2.52, + "grad_norm": 1.15023934841156, + "learning_rate": 0.00017039490264461477, + "loss": 3.2735, + "step": 1716 + }, + { + "epoch": 2.52, + "grad_norm": 3.00900936126709, + "learning_rate": 0.0001703615817085022, + "loss": 3.2198, + "step": 1717 + }, + { + "epoch": 2.52, + "grad_norm": 3.382124423980713, + "learning_rate": 0.0001703282452933132, + "loss": 3.2279, + "step": 1718 + }, + { + "epoch": 2.52, + "grad_norm": 1.6238938570022583, + "learning_rate": 0.00017029489340638154, + "loss": 3.2521, + "step": 1719 + }, + { + "epoch": 2.52, + "grad_norm": 4.034934997558594, + "learning_rate": 0.00017026152605504442, + "loss": 3.2477, + "step": 1720 + }, + { + "epoch": 2.52, + "grad_norm": 4.463752269744873, + "learning_rate": 0.00017022814324664244, + "loss": 3.2478, + "step": 1721 + }, + { + "epoch": 2.53, + "grad_norm": 2.6497182846069336, + "learning_rate": 0.00017019474498851958, + "loss": 3.2774, + "step": 1722 + }, + { + "epoch": 2.53, + "grad_norm": 2.8634886741638184, + "learning_rate": 0.0001701613312880233, + "loss": 3.2704, + "step": 1723 + }, + { + "epoch": 2.53, + "grad_norm": 3.318505048751831, + "learning_rate": 0.00017012790215250432, + "loss": 3.2681, + "step": 1724 + }, + { + "epoch": 2.53, + "grad_norm": 2.9322783946990967, + "learning_rate": 0.0001700944575893168, + "loss": 3.2489, + "step": 1725 + }, + { + "epoch": 2.53, + "grad_norm": 1.9499863386154175, + "learning_rate": 0.00017006099760581838, + "loss": 3.2197, + "step": 1726 + }, + { + "epoch": 2.53, + "grad_norm": 2.5505218505859375, + "learning_rate": 0.00017002752220937004, + "loss": 3.2399, + "step": 1727 + }, + { + "epoch": 2.53, + "grad_norm": 2.46716570854187, + "learning_rate": 0.0001699940314073361, + "loss": 3.2078, + "step": 1728 + }, + { + "epoch": 2.54, + "grad_norm": 1.8553292751312256, + "learning_rate": 0.00016996052520708434, + "loss": 3.292, + "step": 1729 + }, + { + "epoch": 2.54, + "grad_norm": 2.041597366333008, + "learning_rate": 0.00016992700361598585, + "loss": 3.201, + "step": 1730 + }, + { + "epoch": 2.54, + "grad_norm": 2.173717737197876, + "learning_rate": 0.00016989346664141524, + "loss": 3.2707, + "step": 1731 + }, + { + "epoch": 2.54, + "grad_norm": 1.5214920043945312, + "learning_rate": 0.00016985991429075036, + "loss": 3.2218, + "step": 1732 + }, + { + "epoch": 2.54, + "grad_norm": 6.029373645782471, + "learning_rate": 0.00016982634657137252, + "loss": 3.3117, + "step": 1733 + }, + { + "epoch": 2.54, + "grad_norm": 5.91231107711792, + "learning_rate": 0.00016979276349066638, + "loss": 3.2202, + "step": 1734 + }, + { + "epoch": 2.55, + "grad_norm": 2.8163087368011475, + "learning_rate": 0.00016975916505602, + "loss": 3.2803, + "step": 1735 + }, + { + "epoch": 2.55, + "grad_norm": 3.7836880683898926, + "learning_rate": 0.00016972555127482483, + "loss": 3.2399, + "step": 1736 + }, + { + "epoch": 2.55, + "grad_norm": 3.339012622833252, + "learning_rate": 0.00016969192215447566, + "loss": 3.2619, + "step": 1737 + }, + { + "epoch": 2.55, + "grad_norm": 3.779021739959717, + "learning_rate": 0.00016965827770237067, + "loss": 3.2205, + "step": 1738 + }, + { + "epoch": 2.55, + "grad_norm": 0.7899141311645508, + "learning_rate": 0.00016962461792591137, + "loss": 3.2228, + "step": 1739 + }, + { + "epoch": 2.55, + "grad_norm": 2.307349443435669, + "learning_rate": 0.00016959094283250278, + "loss": 3.2412, + "step": 1740 + }, + { + "epoch": 2.55, + "grad_norm": 1.5662769079208374, + "learning_rate": 0.00016955725242955315, + "loss": 3.257, + "step": 1741 + }, + { + "epoch": 2.56, + "grad_norm": 1.4268426895141602, + "learning_rate": 0.0001695235467244741, + "loss": 3.2718, + "step": 1742 + }, + { + "epoch": 2.56, + "grad_norm": 1.7694865465164185, + "learning_rate": 0.0001694898257246807, + "loss": 3.2414, + "step": 1743 + }, + { + "epoch": 2.56, + "grad_norm": 2.555623769760132, + "learning_rate": 0.00016945608943759138, + "loss": 3.2384, + "step": 1744 + }, + { + "epoch": 2.56, + "grad_norm": 1.4633605480194092, + "learning_rate": 0.0001694223378706278, + "loss": 3.2164, + "step": 1745 + }, + { + "epoch": 2.56, + "grad_norm": 2.0434815883636475, + "learning_rate": 0.0001693885710312152, + "loss": 3.2547, + "step": 1746 + }, + { + "epoch": 2.56, + "grad_norm": 2.4210758209228516, + "learning_rate": 0.00016935478892678195, + "loss": 3.2367, + "step": 1747 + }, + { + "epoch": 2.56, + "grad_norm": 0.8179725408554077, + "learning_rate": 0.00016932099156475995, + "loss": 3.2619, + "step": 1748 + }, + { + "epoch": 2.57, + "grad_norm": 3.3028626441955566, + "learning_rate": 0.00016928717895258437, + "loss": 3.2256, + "step": 1749 + }, + { + "epoch": 2.57, + "grad_norm": 1.852142572402954, + "learning_rate": 0.00016925335109769374, + "loss": 3.2402, + "step": 1750 + }, + { + "epoch": 2.57, + "grad_norm": 2.416368007659912, + "learning_rate": 0.00016921950800753002, + "loss": 3.234, + "step": 1751 + }, + { + "epoch": 2.57, + "grad_norm": 1.5477232933044434, + "learning_rate": 0.00016918564968953842, + "loss": 3.2142, + "step": 1752 + }, + { + "epoch": 2.57, + "grad_norm": 1.5452466011047363, + "learning_rate": 0.00016915177615116754, + "loss": 3.1941, + "step": 1753 + }, + { + "epoch": 2.57, + "grad_norm": 3.213561773300171, + "learning_rate": 0.00016911788739986934, + "loss": 3.2154, + "step": 1754 + }, + { + "epoch": 2.58, + "grad_norm": 1.2881815433502197, + "learning_rate": 0.0001690839834430991, + "loss": 3.2909, + "step": 1755 + }, + { + "epoch": 2.58, + "grad_norm": 3.375774383544922, + "learning_rate": 0.00016905006428831552, + "loss": 3.2096, + "step": 1756 + }, + { + "epoch": 2.58, + "grad_norm": 2.6476569175720215, + "learning_rate": 0.00016901612994298053, + "loss": 3.2082, + "step": 1757 + }, + { + "epoch": 2.58, + "grad_norm": 3.5997047424316406, + "learning_rate": 0.0001689821804145595, + "loss": 3.1989, + "step": 1758 + }, + { + "epoch": 2.58, + "grad_norm": 2.435394763946533, + "learning_rate": 0.00016894821571052104, + "loss": 3.2109, + "step": 1759 + }, + { + "epoch": 2.58, + "grad_norm": 5.273232936859131, + "learning_rate": 0.00016891423583833724, + "loss": 3.2358, + "step": 1760 + }, + { + "epoch": 2.58, + "grad_norm": 4.958769798278809, + "learning_rate": 0.00016888024080548335, + "loss": 3.2315, + "step": 1761 + }, + { + "epoch": 2.59, + "grad_norm": 2.171224594116211, + "learning_rate": 0.00016884623061943813, + "loss": 3.2587, + "step": 1762 + }, + { + "epoch": 2.59, + "grad_norm": 1.9900020360946655, + "learning_rate": 0.00016881220528768356, + "loss": 3.2082, + "step": 1763 + }, + { + "epoch": 2.59, + "grad_norm": 2.3555827140808105, + "learning_rate": 0.00016877816481770496, + "loss": 3.2356, + "step": 1764 + }, + { + "epoch": 2.59, + "grad_norm": 1.4140695333480835, + "learning_rate": 0.00016874410921699105, + "loss": 3.1722, + "step": 1765 + }, + { + "epoch": 2.59, + "grad_norm": 4.323090076446533, + "learning_rate": 0.00016871003849303382, + "loss": 3.2533, + "step": 1766 + }, + { + "epoch": 2.59, + "grad_norm": 2.4270732402801514, + "learning_rate": 0.00016867595265332859, + "loss": 3.2092, + "step": 1767 + }, + { + "epoch": 2.59, + "grad_norm": 3.9084079265594482, + "learning_rate": 0.000168641851705374, + "loss": 3.2595, + "step": 1768 + }, + { + "epoch": 2.6, + "grad_norm": 3.986670970916748, + "learning_rate": 0.00016860773565667206, + "loss": 3.2302, + "step": 1769 + }, + { + "epoch": 2.6, + "grad_norm": 3.4050045013427734, + "learning_rate": 0.0001685736045147281, + "loss": 3.2589, + "step": 1770 + }, + { + "epoch": 2.6, + "grad_norm": 4.226673126220703, + "learning_rate": 0.0001685394582870507, + "loss": 3.235, + "step": 1771 + }, + { + "epoch": 2.6, + "grad_norm": 1.5907903909683228, + "learning_rate": 0.00016850529698115175, + "loss": 3.3035, + "step": 1772 + }, + { + "epoch": 2.6, + "grad_norm": 2.343108654022217, + "learning_rate": 0.00016847112060454665, + "loss": 3.1848, + "step": 1773 + }, + { + "epoch": 2.6, + "grad_norm": 3.3092055320739746, + "learning_rate": 0.00016843692916475384, + "loss": 3.2648, + "step": 1774 + }, + { + "epoch": 2.6, + "grad_norm": 1.7477455139160156, + "learning_rate": 0.00016840272266929523, + "loss": 3.206, + "step": 1775 + }, + { + "epoch": 2.61, + "grad_norm": 4.240764617919922, + "learning_rate": 0.00016836850112569613, + "loss": 3.2177, + "step": 1776 + }, + { + "epoch": 2.61, + "grad_norm": 4.2325758934021, + "learning_rate": 0.0001683342645414849, + "loss": 3.2154, + "step": 1777 + }, + { + "epoch": 2.61, + "grad_norm": 2.993154525756836, + "learning_rate": 0.00016830001292419345, + "loss": 3.1972, + "step": 1778 + }, + { + "epoch": 2.61, + "grad_norm": 3.2484753131866455, + "learning_rate": 0.00016826574628135688, + "loss": 3.232, + "step": 1779 + }, + { + "epoch": 2.61, + "grad_norm": 2.438793420791626, + "learning_rate": 0.00016823146462051364, + "loss": 3.2292, + "step": 1780 + }, + { + "epoch": 2.61, + "grad_norm": 2.0203206539154053, + "learning_rate": 0.0001681971679492054, + "loss": 3.1791, + "step": 1781 + }, + { + "epoch": 2.62, + "grad_norm": 4.8010125160217285, + "learning_rate": 0.0001681628562749773, + "loss": 3.2012, + "step": 1782 + }, + { + "epoch": 2.62, + "grad_norm": 4.596678256988525, + "learning_rate": 0.00016812852960537762, + "loss": 3.2204, + "step": 1783 + }, + { + "epoch": 2.62, + "grad_norm": 2.4350953102111816, + "learning_rate": 0.00016809418794795796, + "loss": 3.2175, + "step": 1784 + }, + { + "epoch": 2.62, + "grad_norm": 15.600502014160156, + "learning_rate": 0.00016805983131027333, + "loss": 3.256, + "step": 1785 + }, + { + "epoch": 2.62, + "grad_norm": 3.7827413082122803, + "learning_rate": 0.00016802545969988195, + "loss": 3.2165, + "step": 1786 + }, + { + "epoch": 2.62, + "grad_norm": 3.0212562084198, + "learning_rate": 0.0001679910731243453, + "loss": 3.2537, + "step": 1787 + }, + { + "epoch": 2.62, + "grad_norm": 2.496250629425049, + "learning_rate": 0.00016795667159122824, + "loss": 3.212, + "step": 1788 + }, + { + "epoch": 2.63, + "grad_norm": 3.222059488296509, + "learning_rate": 0.00016792225510809883, + "loss": 3.2665, + "step": 1789 + }, + { + "epoch": 2.63, + "grad_norm": 3.4902381896972656, + "learning_rate": 0.0001678878236825285, + "loss": 3.2326, + "step": 1790 + }, + { + "epoch": 2.63, + "grad_norm": 0.7913928627967834, + "learning_rate": 0.00016785337732209193, + "loss": 3.1662, + "step": 1791 + }, + { + "epoch": 2.63, + "grad_norm": 3.9567112922668457, + "learning_rate": 0.00016781891603436706, + "loss": 3.2217, + "step": 1792 + }, + { + "epoch": 2.63, + "grad_norm": 1.712324619293213, + "learning_rate": 0.00016778443982693519, + "loss": 3.1855, + "step": 1793 + }, + { + "epoch": 2.63, + "grad_norm": 1.2269927263259888, + "learning_rate": 0.0001677499487073808, + "loss": 3.2306, + "step": 1794 + }, + { + "epoch": 2.63, + "grad_norm": 3.1048622131347656, + "learning_rate": 0.00016771544268329177, + "loss": 3.1975, + "step": 1795 + }, + { + "epoch": 2.64, + "grad_norm": 1.18289315700531, + "learning_rate": 0.00016768092176225915, + "loss": 3.1904, + "step": 1796 + }, + { + "epoch": 2.64, + "grad_norm": 1.7925835847854614, + "learning_rate": 0.0001676463859518773, + "loss": 3.2143, + "step": 1797 + }, + { + "epoch": 2.64, + "grad_norm": 2.179051637649536, + "learning_rate": 0.0001676118352597439, + "loss": 3.2297, + "step": 1798 + }, + { + "epoch": 2.64, + "grad_norm": 1.2693228721618652, + "learning_rate": 0.00016757726969345984, + "loss": 3.2181, + "step": 1799 + }, + { + "epoch": 2.64, + "grad_norm": 5.075618743896484, + "learning_rate": 0.00016754268926062938, + "loss": 3.237, + "step": 1800 + }, + { + "epoch": 2.64, + "grad_norm": 5.278939723968506, + "learning_rate": 0.0001675080939688599, + "loss": 3.2266, + "step": 1801 + }, + { + "epoch": 2.64, + "grad_norm": 2.5977909564971924, + "learning_rate": 0.00016747348382576213, + "loss": 3.2409, + "step": 1802 + }, + { + "epoch": 2.65, + "grad_norm": 3.318488597869873, + "learning_rate": 0.00016743885883895013, + "loss": 3.231, + "step": 1803 + }, + { + "epoch": 2.65, + "grad_norm": 2.22575306892395, + "learning_rate": 0.00016740421901604118, + "loss": 3.2313, + "step": 1804 + }, + { + "epoch": 2.65, + "grad_norm": 2.436943769454956, + "learning_rate": 0.00016736956436465573, + "loss": 3.2268, + "step": 1805 + }, + { + "epoch": 2.65, + "grad_norm": 2.966697931289673, + "learning_rate": 0.0001673348948924176, + "loss": 3.2139, + "step": 1806 + }, + { + "epoch": 2.65, + "grad_norm": 2.458636999130249, + "learning_rate": 0.0001673002106069539, + "loss": 3.2228, + "step": 1807 + }, + { + "epoch": 2.65, + "grad_norm": 3.647672414779663, + "learning_rate": 0.00016726551151589485, + "loss": 3.256, + "step": 1808 + }, + { + "epoch": 2.66, + "grad_norm": 2.713827133178711, + "learning_rate": 0.0001672307976268741, + "loss": 3.2256, + "step": 1809 + }, + { + "epoch": 2.66, + "grad_norm": 3.452815294265747, + "learning_rate": 0.0001671960689475284, + "loss": 3.2875, + "step": 1810 + }, + { + "epoch": 2.66, + "grad_norm": 2.377556085586548, + "learning_rate": 0.00016716132548549786, + "loss": 3.2188, + "step": 1811 + }, + { + "epoch": 2.66, + "grad_norm": 3.791208028793335, + "learning_rate": 0.0001671265672484258, + "loss": 3.234, + "step": 1812 + }, + { + "epoch": 2.66, + "grad_norm": 3.7197184562683105, + "learning_rate": 0.00016709179424395884, + "loss": 3.1895, + "step": 1813 + }, + { + "epoch": 2.66, + "grad_norm": 2.4916012287139893, + "learning_rate": 0.00016705700647974675, + "loss": 3.2053, + "step": 1814 + }, + { + "epoch": 2.66, + "grad_norm": 2.569413185119629, + "learning_rate": 0.00016702220396344267, + "loss": 3.2358, + "step": 1815 + }, + { + "epoch": 2.67, + "grad_norm": 3.520432949066162, + "learning_rate": 0.00016698738670270285, + "loss": 3.2369, + "step": 1816 + }, + { + "epoch": 2.67, + "grad_norm": 2.880115032196045, + "learning_rate": 0.00016695255470518686, + "loss": 3.2232, + "step": 1817 + }, + { + "epoch": 2.67, + "grad_norm": 3.42354416847229, + "learning_rate": 0.00016691770797855758, + "loss": 3.1885, + "step": 1818 + }, + { + "epoch": 2.67, + "grad_norm": 3.31255841255188, + "learning_rate": 0.00016688284653048097, + "loss": 3.21, + "step": 1819 + }, + { + "epoch": 2.67, + "grad_norm": 3.157924175262451, + "learning_rate": 0.00016684797036862635, + "loss": 3.2055, + "step": 1820 + }, + { + "epoch": 2.67, + "grad_norm": 2.820952892303467, + "learning_rate": 0.00016681307950066628, + "loss": 3.1809, + "step": 1821 + }, + { + "epoch": 2.67, + "grad_norm": 3.921417236328125, + "learning_rate": 0.00016677817393427645, + "loss": 3.1902, + "step": 1822 + }, + { + "epoch": 2.68, + "grad_norm": 3.5899341106414795, + "learning_rate": 0.00016674325367713592, + "loss": 3.2222, + "step": 1823 + }, + { + "epoch": 2.68, + "grad_norm": 2.645939350128174, + "learning_rate": 0.00016670831873692683, + "loss": 3.1834, + "step": 1824 + }, + { + "epoch": 2.68, + "grad_norm": 2.4159257411956787, + "learning_rate": 0.00016667336912133468, + "loss": 3.2018, + "step": 1825 + }, + { + "epoch": 2.68, + "grad_norm": 3.56545090675354, + "learning_rate": 0.0001666384048380482, + "loss": 3.2096, + "step": 1826 + }, + { + "epoch": 2.68, + "grad_norm": 2.857130765914917, + "learning_rate": 0.0001666034258947592, + "loss": 3.2102, + "step": 1827 + }, + { + "epoch": 2.68, + "grad_norm": 3.4373629093170166, + "learning_rate": 0.00016656843229916285, + "loss": 3.2017, + "step": 1828 + }, + { + "epoch": 2.68, + "grad_norm": 2.5930848121643066, + "learning_rate": 0.00016653342405895756, + "loss": 3.2265, + "step": 1829 + }, + { + "epoch": 2.69, + "grad_norm": 4.415797710418701, + "learning_rate": 0.0001664984011818448, + "loss": 3.1589, + "step": 1830 + }, + { + "epoch": 2.69, + "grad_norm": 4.656808853149414, + "learning_rate": 0.0001664633636755295, + "loss": 3.2189, + "step": 1831 + }, + { + "epoch": 2.69, + "grad_norm": 1.3080215454101562, + "learning_rate": 0.00016642831154771957, + "loss": 3.2178, + "step": 1832 + }, + { + "epoch": 2.69, + "grad_norm": 1.491684913635254, + "learning_rate": 0.0001663932448061263, + "loss": 3.1922, + "step": 1833 + }, + { + "epoch": 2.69, + "grad_norm": 5.020033359527588, + "learning_rate": 0.0001663581634584641, + "loss": 3.2153, + "step": 1834 + }, + { + "epoch": 2.69, + "grad_norm": 4.078526973724365, + "learning_rate": 0.0001663230675124507, + "loss": 3.2165, + "step": 1835 + }, + { + "epoch": 2.7, + "grad_norm": 2.787774085998535, + "learning_rate": 0.00016628795697580686, + "loss": 3.1903, + "step": 1836 + }, + { + "epoch": 2.7, + "grad_norm": 2.4265925884246826, + "learning_rate": 0.0001662528318562568, + "loss": 3.1732, + "step": 1837 + }, + { + "epoch": 2.7, + "grad_norm": 3.579639434814453, + "learning_rate": 0.00016621769216152766, + "loss": 3.1891, + "step": 1838 + }, + { + "epoch": 2.7, + "grad_norm": 2.984905958175659, + "learning_rate": 0.00016618253789935005, + "loss": 3.2348, + "step": 1839 + }, + { + "epoch": 2.7, + "grad_norm": 4.036933898925781, + "learning_rate": 0.00016614736907745768, + "loss": 3.1677, + "step": 1840 + }, + { + "epoch": 2.7, + "grad_norm": 3.9433419704437256, + "learning_rate": 0.00016611218570358735, + "loss": 3.18, + "step": 1841 + }, + { + "epoch": 2.7, + "grad_norm": 2.285982370376587, + "learning_rate": 0.00016607698778547925, + "loss": 3.2298, + "step": 1842 + }, + { + "epoch": 2.71, + "grad_norm": 2.241262197494507, + "learning_rate": 0.00016604177533087667, + "loss": 3.1903, + "step": 1843 + }, + { + "epoch": 2.71, + "grad_norm": 3.7759506702423096, + "learning_rate": 0.00016600654834752611, + "loss": 3.2093, + "step": 1844 + }, + { + "epoch": 2.71, + "grad_norm": 3.2228505611419678, + "learning_rate": 0.00016597130684317725, + "loss": 3.1934, + "step": 1845 + }, + { + "epoch": 2.71, + "grad_norm": 3.3804237842559814, + "learning_rate": 0.00016593605082558305, + "loss": 3.2345, + "step": 1846 + }, + { + "epoch": 2.71, + "grad_norm": 2.6020681858062744, + "learning_rate": 0.0001659007803024995, + "loss": 3.1868, + "step": 1847 + }, + { + "epoch": 2.71, + "grad_norm": 3.3784382343292236, + "learning_rate": 0.00016586549528168596, + "loss": 3.1865, + "step": 1848 + }, + { + "epoch": 2.71, + "grad_norm": 2.881960391998291, + "learning_rate": 0.0001658301957709049, + "loss": 3.1906, + "step": 1849 + }, + { + "epoch": 2.72, + "grad_norm": 3.543879270553589, + "learning_rate": 0.00016579488177792187, + "loss": 3.2124, + "step": 1850 + }, + { + "epoch": 2.72, + "grad_norm": 3.8492488861083984, + "learning_rate": 0.00016575955331050584, + "loss": 3.3212, + "step": 1851 + }, + { + "epoch": 2.72, + "grad_norm": 2.2801854610443115, + "learning_rate": 0.00016572421037642875, + "loss": 3.2374, + "step": 1852 + }, + { + "epoch": 2.72, + "grad_norm": 2.0372071266174316, + "learning_rate": 0.00016568885298346586, + "loss": 3.2105, + "step": 1853 + }, + { + "epoch": 2.72, + "grad_norm": 4.137518405914307, + "learning_rate": 0.00016565348113939556, + "loss": 3.1718, + "step": 1854 + }, + { + "epoch": 2.72, + "grad_norm": 3.7536251544952393, + "learning_rate": 0.00016561809485199938, + "loss": 3.2118, + "step": 1855 + }, + { + "epoch": 2.73, + "grad_norm": 3.2339396476745605, + "learning_rate": 0.00016558269412906206, + "loss": 3.1746, + "step": 1856 + }, + { + "epoch": 2.73, + "grad_norm": 2.7833776473999023, + "learning_rate": 0.0001655472789783716, + "loss": 3.1876, + "step": 1857 + }, + { + "epoch": 2.73, + "grad_norm": 3.6448328495025635, + "learning_rate": 0.000165511849407719, + "loss": 3.2143, + "step": 1858 + }, + { + "epoch": 2.73, + "grad_norm": 3.4505040645599365, + "learning_rate": 0.0001654764054248986, + "loss": 3.173, + "step": 1859 + }, + { + "epoch": 2.73, + "eval_loss": 3.040992498397827, + "eval_runtime": 1164.9923, + "eval_samples_per_second": 14.714, + "eval_steps_per_second": 0.92, + "step": 1859 + }, + { + "epoch": 2.73, + "grad_norm": 2.8391804695129395, + "learning_rate": 0.0001654409470377078, + "loss": 3.2099, + "step": 1860 + }, + { + "epoch": 2.73, + "grad_norm": 2.7543411254882812, + "learning_rate": 0.0001654054742539472, + "loss": 3.1918, + "step": 1861 + }, + { + "epoch": 2.73, + "grad_norm": 3.5177597999572754, + "learning_rate": 0.00016536998708142062, + "loss": 3.245, + "step": 1862 + }, + { + "epoch": 2.74, + "grad_norm": 3.0664377212524414, + "learning_rate": 0.00016533448552793494, + "loss": 3.1616, + "step": 1863 + }, + { + "epoch": 2.74, + "grad_norm": 2.924795389175415, + "learning_rate": 0.00016529896960130032, + "loss": 3.229, + "step": 1864 + }, + { + "epoch": 2.74, + "grad_norm": 3.104602575302124, + "learning_rate": 0.00016526343930933, + "loss": 3.2294, + "step": 1865 + }, + { + "epoch": 2.74, + "grad_norm": 3.728653907775879, + "learning_rate": 0.0001652278946598404, + "loss": 3.2088, + "step": 1866 + }, + { + "epoch": 2.74, + "grad_norm": 3.854654312133789, + "learning_rate": 0.00016519233566065112, + "loss": 3.1829, + "step": 1867 + }, + { + "epoch": 2.74, + "grad_norm": 2.6493520736694336, + "learning_rate": 0.0001651567623195849, + "loss": 3.1807, + "step": 1868 + }, + { + "epoch": 2.74, + "grad_norm": 2.057978391647339, + "learning_rate": 0.00016512117464446762, + "loss": 3.141, + "step": 1869 + }, + { + "epoch": 2.75, + "grad_norm": 4.454994201660156, + "learning_rate": 0.00016508557264312835, + "loss": 3.1908, + "step": 1870 + }, + { + "epoch": 2.75, + "grad_norm": 4.243129253387451, + "learning_rate": 0.00016504995632339928, + "loss": 3.1819, + "step": 1871 + }, + { + "epoch": 2.75, + "grad_norm": 2.3739356994628906, + "learning_rate": 0.00016501432569311574, + "loss": 3.2002, + "step": 1872 + }, + { + "epoch": 2.75, + "grad_norm": 1.8129799365997314, + "learning_rate": 0.00016497868076011628, + "loss": 3.2084, + "step": 1873 + }, + { + "epoch": 2.75, + "grad_norm": 4.162952899932861, + "learning_rate": 0.0001649430215322425, + "loss": 3.1766, + "step": 1874 + }, + { + "epoch": 2.75, + "grad_norm": 3.205064535140991, + "learning_rate": 0.00016490734801733923, + "loss": 3.1601, + "step": 1875 + }, + { + "epoch": 2.75, + "grad_norm": 3.1646533012390137, + "learning_rate": 0.0001648716602232544, + "loss": 3.2347, + "step": 1876 + }, + { + "epoch": 2.76, + "grad_norm": 3.2153899669647217, + "learning_rate": 0.00016483595815783904, + "loss": 3.2166, + "step": 1877 + }, + { + "epoch": 2.76, + "grad_norm": 2.5029168128967285, + "learning_rate": 0.00016480024182894743, + "loss": 3.2366, + "step": 1878 + }, + { + "epoch": 2.76, + "grad_norm": 1.944775938987732, + "learning_rate": 0.0001647645112444369, + "loss": 3.1608, + "step": 1879 + }, + { + "epoch": 2.76, + "grad_norm": 4.4062700271606445, + "learning_rate": 0.00016472876641216786, + "loss": 3.226, + "step": 1880 + }, + { + "epoch": 2.76, + "grad_norm": 3.8181326389312744, + "learning_rate": 0.00016469300734000403, + "loss": 3.1934, + "step": 1881 + }, + { + "epoch": 2.76, + "grad_norm": 2.7566487789154053, + "learning_rate": 0.00016465723403581218, + "loss": 3.2031, + "step": 1882 + }, + { + "epoch": 2.77, + "grad_norm": 2.438339948654175, + "learning_rate": 0.00016462144650746213, + "loss": 3.2028, + "step": 1883 + }, + { + "epoch": 2.77, + "grad_norm": 3.6031992435455322, + "learning_rate": 0.0001645856447628269, + "loss": 3.2054, + "step": 1884 + }, + { + "epoch": 2.77, + "grad_norm": 2.86798095703125, + "learning_rate": 0.00016454982880978266, + "loss": 3.2189, + "step": 1885 + }, + { + "epoch": 2.77, + "grad_norm": 3.7704923152923584, + "learning_rate": 0.0001645139986562087, + "loss": 3.1854, + "step": 1886 + }, + { + "epoch": 2.77, + "grad_norm": 3.522306442260742, + "learning_rate": 0.0001644781543099874, + "loss": 3.2357, + "step": 1887 + }, + { + "epoch": 2.77, + "grad_norm": 2.8147850036621094, + "learning_rate": 0.00016444229577900423, + "loss": 3.2041, + "step": 1888 + }, + { + "epoch": 2.77, + "grad_norm": 4.882560729980469, + "learning_rate": 0.00016440642307114786, + "loss": 3.1773, + "step": 1889 + }, + { + "epoch": 2.78, + "grad_norm": 1.0316693782806396, + "learning_rate": 0.00016437053619431006, + "loss": 3.195, + "step": 1890 + }, + { + "epoch": 2.78, + "grad_norm": 1.5295097827911377, + "learning_rate": 0.0001643346351563857, + "loss": 3.2313, + "step": 1891 + }, + { + "epoch": 2.78, + "grad_norm": 1.8384897708892822, + "learning_rate": 0.00016429871996527276, + "loss": 3.1694, + "step": 1892 + }, + { + "epoch": 2.78, + "grad_norm": 1.1532273292541504, + "learning_rate": 0.00016426279062887233, + "loss": 3.1872, + "step": 1893 + }, + { + "epoch": 2.78, + "grad_norm": 0.97287917137146, + "learning_rate": 0.0001642268471550886, + "loss": 3.2283, + "step": 1894 + }, + { + "epoch": 2.78, + "grad_norm": 3.258812665939331, + "learning_rate": 0.00016419088955182897, + "loss": 3.2109, + "step": 1895 + }, + { + "epoch": 2.78, + "grad_norm": 3.9376001358032227, + "learning_rate": 0.00016415491782700384, + "loss": 3.213, + "step": 1896 + }, + { + "epoch": 2.79, + "grad_norm": 3.313781261444092, + "learning_rate": 0.00016411893198852668, + "loss": 3.2419, + "step": 1897 + }, + { + "epoch": 2.79, + "grad_norm": 1.2715352773666382, + "learning_rate": 0.00016408293204431425, + "loss": 3.1853, + "step": 1898 + }, + { + "epoch": 2.79, + "grad_norm": 4.407071113586426, + "learning_rate": 0.00016404691800228616, + "loss": 3.1633, + "step": 1899 + }, + { + "epoch": 2.79, + "grad_norm": 2.881059408187866, + "learning_rate": 0.00016401088987036538, + "loss": 3.1711, + "step": 1900 + }, + { + "epoch": 2.79, + "grad_norm": 213.681884765625, + "learning_rate": 0.0001639748476564778, + "loss": 3.219, + "step": 1901 + }, + { + "epoch": 2.79, + "grad_norm": 15.1211576461792, + "learning_rate": 0.00016393879136855248, + "loss": 3.3071, + "step": 1902 + }, + { + "epoch": 2.79, + "grad_norm": 227.103759765625, + "learning_rate": 0.00016390272101452156, + "loss": 3.3632, + "step": 1903 + }, + { + "epoch": 2.8, + "grad_norm": 94.93277740478516, + "learning_rate": 0.00016386663660232026, + "loss": 3.3738, + "step": 1904 + }, + { + "epoch": 2.8, + "grad_norm": 21.8365421295166, + "learning_rate": 0.00016383053813988694, + "loss": 3.4633, + "step": 1905 + }, + { + "epoch": 2.8, + "grad_norm": 461.8182373046875, + "learning_rate": 0.00016379442563516303, + "loss": 3.4446, + "step": 1906 + }, + { + "epoch": 2.8, + "grad_norm": 73.93539428710938, + "learning_rate": 0.00016375829909609296, + "loss": 3.511, + "step": 1907 + }, + { + "epoch": 2.8, + "grad_norm": 26.853099822998047, + "learning_rate": 0.00016372215853062438, + "loss": 3.4775, + "step": 1908 + }, + { + "epoch": 2.8, + "grad_norm": 449.6775207519531, + "learning_rate": 0.000163686003946708, + "loss": 3.5932, + "step": 1909 + }, + { + "epoch": 2.81, + "grad_norm": 1817.4749755859375, + "learning_rate": 0.00016364983535229754, + "loss": 3.6702, + "step": 1910 + }, + { + "epoch": 2.81, + "grad_norm": 431.15777587890625, + "learning_rate": 0.00016361365275534986, + "loss": 4.165, + "step": 1911 + }, + { + "epoch": 2.81, + "grad_norm": 1153.154296875, + "learning_rate": 0.0001635774561638249, + "loss": 4.4609, + "step": 1912 + }, + { + "epoch": 2.81, + "grad_norm": 378.61279296875, + "learning_rate": 0.00016354124558568567, + "loss": 4.9769, + "step": 1913 + }, + { + "epoch": 2.81, + "grad_norm": 228.8455352783203, + "learning_rate": 0.00016350502102889826, + "loss": 5.4494, + "step": 1914 + }, + { + "epoch": 2.81, + "grad_norm": 2911.531005859375, + "learning_rate": 0.0001634687825014318, + "loss": 5.2172, + "step": 1915 + }, + { + "epoch": 2.81, + "grad_norm": 8.003608703613281, + "learning_rate": 0.00016343253001125852, + "loss": 4.3254, + "step": 1916 + }, + { + "epoch": 2.82, + "grad_norm": 122.48676300048828, + "learning_rate": 0.00016339626356635372, + "loss": 3.9329, + "step": 1917 + }, + { + "epoch": 2.82, + "grad_norm": 165.21929931640625, + "learning_rate": 0.00016335998317469586, + "loss": 4.2273, + "step": 1918 + }, + { + "epoch": 2.82, + "grad_norm": 335.0215148925781, + "learning_rate": 0.00016332368884426626, + "loss": 4.668, + "step": 1919 + }, + { + "epoch": 2.82, + "grad_norm": 3772.19091796875, + "learning_rate": 0.00016328738058304956, + "loss": 5.2298, + "step": 1920 + }, + { + "epoch": 2.82, + "grad_norm": 34.06552505493164, + "learning_rate": 0.0001632510583990332, + "loss": 7.4421, + "step": 1921 + }, + { + "epoch": 2.82, + "grad_norm": 3216.72998046875, + "learning_rate": 0.00016321472230020794, + "loss": 7.2781, + "step": 1922 + }, + { + "epoch": 2.82, + "grad_norm": 9.185710906982422, + "learning_rate": 0.0001631783722945674, + "loss": 6.4889, + "step": 1923 + }, + { + "epoch": 2.83, + "grad_norm": 25.194772720336914, + "learning_rate": 0.00016314200839010834, + "loss": 6.3019, + "step": 1924 + }, + { + "epoch": 2.83, + "grad_norm": 35.877498626708984, + "learning_rate": 0.0001631056305948306, + "loss": 6.1576, + "step": 1925 + }, + { + "epoch": 2.83, + "grad_norm": 35.11639404296875, + "learning_rate": 0.00016306923891673707, + "loss": 5.7687, + "step": 1926 + }, + { + "epoch": 2.83, + "grad_norm": 7.56588077545166, + "learning_rate": 0.0001630328333638336, + "loss": 5.4341, + "step": 1927 + }, + { + "epoch": 2.83, + "grad_norm": 11.163580894470215, + "learning_rate": 0.00016299641394412927, + "loss": 5.2751, + "step": 1928 + }, + { + "epoch": 2.83, + "grad_norm": 9.635797500610352, + "learning_rate": 0.00016295998066563603, + "loss": 5.1351, + "step": 1929 + }, + { + "epoch": 2.83, + "grad_norm": 7.929530143737793, + "learning_rate": 0.000162923533536369, + "loss": 4.9651, + "step": 1930 + }, + { + "epoch": 2.84, + "grad_norm": 12.908093452453613, + "learning_rate": 0.00016288707256434623, + "loss": 4.9926, + "step": 1931 + }, + { + "epoch": 2.84, + "grad_norm": 17.120487213134766, + "learning_rate": 0.000162850597757589, + "loss": 4.9808, + "step": 1932 + }, + { + "epoch": 2.84, + "grad_norm": 12.659987449645996, + "learning_rate": 0.00016281410912412144, + "loss": 4.8239, + "step": 1933 + }, + { + "epoch": 2.84, + "grad_norm": 6.449459552764893, + "learning_rate": 0.00016277760667197087, + "loss": 4.7229, + "step": 1934 + }, + { + "epoch": 2.84, + "grad_norm": 11.50796127319336, + "learning_rate": 0.00016274109040916753, + "loss": 4.5966, + "step": 1935 + }, + { + "epoch": 2.84, + "grad_norm": 10.945295333862305, + "learning_rate": 0.00016270456034374474, + "loss": 4.5114, + "step": 1936 + }, + { + "epoch": 2.85, + "grad_norm": 6.565453052520752, + "learning_rate": 0.00016266801648373894, + "loss": 4.4361, + "step": 1937 + }, + { + "epoch": 2.85, + "grad_norm": 6.896590709686279, + "learning_rate": 0.0001626314588371895, + "loss": 4.3267, + "step": 1938 + }, + { + "epoch": 2.85, + "grad_norm": 6.761734485626221, + "learning_rate": 0.00016259488741213884, + "loss": 4.336, + "step": 1939 + }, + { + "epoch": 2.85, + "grad_norm": 7.6577301025390625, + "learning_rate": 0.00016255830221663243, + "loss": 4.2871, + "step": 1940 + }, + { + "epoch": 2.85, + "grad_norm": 5.5717620849609375, + "learning_rate": 0.00016252170325871878, + "loss": 4.2744, + "step": 1941 + }, + { + "epoch": 2.85, + "grad_norm": 6.425532341003418, + "learning_rate": 0.00016248509054644943, + "loss": 4.2368, + "step": 1942 + }, + { + "epoch": 2.85, + "grad_norm": 6.052181720733643, + "learning_rate": 0.00016244846408787895, + "loss": 4.1986, + "step": 1943 + }, + { + "epoch": 2.86, + "grad_norm": 4.308633804321289, + "learning_rate": 0.00016241182389106483, + "loss": 4.2131, + "step": 1944 + }, + { + "epoch": 2.86, + "grad_norm": 3.014502763748169, + "learning_rate": 0.00016237516996406775, + "loss": 4.1368, + "step": 1945 + }, + { + "epoch": 2.86, + "grad_norm": 3.824622869491577, + "learning_rate": 0.00016233850231495126, + "loss": 4.1542, + "step": 1946 + }, + { + "epoch": 2.86, + "grad_norm": 4.700507640838623, + "learning_rate": 0.00016230182095178206, + "loss": 4.032, + "step": 1947 + }, + { + "epoch": 2.86, + "grad_norm": 2.4007976055145264, + "learning_rate": 0.0001622651258826298, + "loss": 4.043, + "step": 1948 + }, + { + "epoch": 2.86, + "grad_norm": 2.202371835708618, + "learning_rate": 0.0001622284171155671, + "loss": 4.0561, + "step": 1949 + }, + { + "epoch": 2.86, + "grad_norm": 3.510922431945801, + "learning_rate": 0.00016219169465866965, + "loss": 4.0186, + "step": 1950 + }, + { + "epoch": 2.87, + "grad_norm": 3.223702907562256, + "learning_rate": 0.0001621549585200162, + "loss": 4.0432, + "step": 1951 + }, + { + "epoch": 2.87, + "grad_norm": 2.4226481914520264, + "learning_rate": 0.00016211820870768842, + "loss": 4.0018, + "step": 1952 + }, + { + "epoch": 2.87, + "grad_norm": 4.14686393737793, + "learning_rate": 0.000162081445229771, + "loss": 3.9202, + "step": 1953 + }, + { + "epoch": 2.87, + "grad_norm": 4.985760688781738, + "learning_rate": 0.0001620446680943517, + "loss": 4.0172, + "step": 1954 + }, + { + "epoch": 2.87, + "grad_norm": 4.1533427238464355, + "learning_rate": 0.00016200787730952124, + "loss": 3.9214, + "step": 1955 + }, + { + "epoch": 2.87, + "grad_norm": 2.108228921890259, + "learning_rate": 0.00016197107288337328, + "loss": 3.9857, + "step": 1956 + }, + { + "epoch": 2.88, + "grad_norm": 2.6513140201568604, + "learning_rate": 0.00016193425482400464, + "loss": 3.9326, + "step": 1957 + }, + { + "epoch": 2.88, + "grad_norm": 3.7441959381103516, + "learning_rate": 0.00016189742313951498, + "loss": 3.8733, + "step": 1958 + }, + { + "epoch": 2.88, + "grad_norm": 2.4456493854522705, + "learning_rate": 0.00016186057783800704, + "loss": 3.9193, + "step": 1959 + }, + { + "epoch": 2.88, + "grad_norm": 2.1966073513031006, + "learning_rate": 0.00016182371892758658, + "loss": 3.9049, + "step": 1960 + }, + { + "epoch": 2.88, + "grad_norm": 3.958984613418579, + "learning_rate": 0.00016178684641636226, + "loss": 3.8627, + "step": 1961 + }, + { + "epoch": 2.88, + "grad_norm": 2.8413755893707275, + "learning_rate": 0.00016174996031244581, + "loss": 3.9005, + "step": 1962 + }, + { + "epoch": 2.88, + "grad_norm": 2.8592469692230225, + "learning_rate": 0.00016171306062395192, + "loss": 3.8878, + "step": 1963 + }, + { + "epoch": 2.89, + "grad_norm": 5.321715831756592, + "learning_rate": 0.00016167614735899827, + "loss": 3.8257, + "step": 1964 + }, + { + "epoch": 2.89, + "grad_norm": 2.1376495361328125, + "learning_rate": 0.00016163922052570554, + "loss": 3.8189, + "step": 1965 + }, + { + "epoch": 2.89, + "grad_norm": 6.535236835479736, + "learning_rate": 0.0001616022801321974, + "loss": 3.8569, + "step": 1966 + }, + { + "epoch": 2.89, + "grad_norm": 4.882155895233154, + "learning_rate": 0.00016156532618660045, + "loss": 3.8699, + "step": 1967 + }, + { + "epoch": 2.89, + "grad_norm": 5.889296531677246, + "learning_rate": 0.00016152835869704432, + "loss": 3.8201, + "step": 1968 + }, + { + "epoch": 2.89, + "grad_norm": 5.7923102378845215, + "learning_rate": 0.00016149137767166165, + "loss": 3.8467, + "step": 1969 + }, + { + "epoch": 2.89, + "grad_norm": 4.793107509613037, + "learning_rate": 0.000161454383118588, + "loss": 3.789, + "step": 1970 + }, + { + "epoch": 2.9, + "grad_norm": 5.585588455200195, + "learning_rate": 0.00016141737504596186, + "loss": 3.8359, + "step": 1971 + }, + { + "epoch": 2.9, + "grad_norm": 1.328306794166565, + "learning_rate": 0.00016138035346192483, + "loss": 3.781, + "step": 1972 + }, + { + "epoch": 2.9, + "grad_norm": 1.4749982357025146, + "learning_rate": 0.00016134331837462142, + "loss": 3.7788, + "step": 1973 + }, + { + "epoch": 2.9, + "grad_norm": 4.906754016876221, + "learning_rate": 0.00016130626979219906, + "loss": 3.8094, + "step": 1974 + }, + { + "epoch": 2.9, + "grad_norm": 2.958904266357422, + "learning_rate": 0.00016126920772280818, + "loss": 3.8173, + "step": 1975 + }, + { + "epoch": 2.9, + "grad_norm": 8.751588821411133, + "learning_rate": 0.00016123213217460224, + "loss": 3.7913, + "step": 1976 + }, + { + "epoch": 2.9, + "grad_norm": 10.918651580810547, + "learning_rate": 0.00016119504315573755, + "loss": 3.7906, + "step": 1977 + }, + { + "epoch": 2.91, + "grad_norm": 2.1029868125915527, + "learning_rate": 0.0001611579406743735, + "loss": 3.8032, + "step": 1978 + }, + { + "epoch": 2.91, + "grad_norm": 9.132828712463379, + "learning_rate": 0.00016112082473867235, + "loss": 3.8242, + "step": 1979 + }, + { + "epoch": 2.91, + "grad_norm": 10.694255828857422, + "learning_rate": 0.0001610836953567994, + "loss": 3.8359, + "step": 1980 + }, + { + "epoch": 2.91, + "grad_norm": 6.331207752227783, + "learning_rate": 0.0001610465525369228, + "loss": 3.7975, + "step": 1981 + }, + { + "epoch": 2.91, + "grad_norm": 7.4574055671691895, + "learning_rate": 0.00016100939628721375, + "loss": 3.7936, + "step": 1982 + }, + { + "epoch": 2.91, + "grad_norm": 12.698988914489746, + "learning_rate": 0.00016097222661584643, + "loss": 3.8415, + "step": 1983 + }, + { + "epoch": 2.92, + "grad_norm": 7.2185750007629395, + "learning_rate": 0.00016093504353099783, + "loss": 3.8274, + "step": 1984 + }, + { + "epoch": 2.92, + "grad_norm": 5.854779243469238, + "learning_rate": 0.00016089784704084804, + "loss": 3.8396, + "step": 1985 + }, + { + "epoch": 2.92, + "grad_norm": 8.324865341186523, + "learning_rate": 0.00016086063715358003, + "loss": 3.8204, + "step": 1986 + }, + { + "epoch": 2.92, + "grad_norm": 3.311805009841919, + "learning_rate": 0.0001608234138773797, + "loss": 3.7652, + "step": 1987 + }, + { + "epoch": 2.92, + "grad_norm": 11.10246467590332, + "learning_rate": 0.00016078617722043595, + "loss": 3.8391, + "step": 1988 + }, + { + "epoch": 2.92, + "grad_norm": 15.776145935058594, + "learning_rate": 0.00016074892719094056, + "loss": 3.799, + "step": 1989 + }, + { + "epoch": 2.92, + "grad_norm": 7.907688617706299, + "learning_rate": 0.00016071166379708833, + "loss": 3.7895, + "step": 1990 + }, + { + "epoch": 2.93, + "grad_norm": 5.6883649826049805, + "learning_rate": 0.00016067438704707695, + "loss": 3.794, + "step": 1991 + }, + { + "epoch": 2.93, + "grad_norm": 8.36189079284668, + "learning_rate": 0.00016063709694910701, + "loss": 3.8036, + "step": 1992 + }, + { + "epoch": 2.93, + "grad_norm": 4.989264488220215, + "learning_rate": 0.00016059979351138213, + "loss": 3.7555, + "step": 1993 + }, + { + "epoch": 2.93, + "grad_norm": 7.9405670166015625, + "learning_rate": 0.0001605624767421088, + "loss": 3.7995, + "step": 1994 + }, + { + "epoch": 2.93, + "grad_norm": 11.919158935546875, + "learning_rate": 0.00016052514664949646, + "loss": 3.8064, + "step": 1995 + }, + { + "epoch": 2.93, + "grad_norm": 15.537467002868652, + "learning_rate": 0.00016048780324175746, + "loss": 3.8284, + "step": 1996 + }, + { + "epoch": 2.93, + "grad_norm": 3.631410837173462, + "learning_rate": 0.00016045044652710712, + "loss": 3.838, + "step": 1997 + }, + { + "epoch": 2.94, + "grad_norm": 2.047104597091675, + "learning_rate": 0.0001604130765137637, + "loss": 3.7354, + "step": 1998 + }, + { + "epoch": 2.94, + "grad_norm": 8.390944480895996, + "learning_rate": 0.00016037569320994827, + "loss": 3.8143, + "step": 1999 + }, + { + "epoch": 2.94, + "grad_norm": 8.165762901306152, + "learning_rate": 0.00016033829662388498, + "loss": 3.7676, + "step": 2000 + }, + { + "epoch": 2.94, + "grad_norm": 2.4152541160583496, + "learning_rate": 0.0001603008867638008, + "loss": 3.7495, + "step": 2001 + }, + { + "epoch": 2.94, + "grad_norm": 5.799648761749268, + "learning_rate": 0.00016026346363792567, + "loss": 3.7996, + "step": 2002 + }, + { + "epoch": 2.94, + "grad_norm": 3.751105785369873, + "learning_rate": 0.00016022602725449238, + "loss": 3.7672, + "step": 2003 + }, + { + "epoch": 2.94, + "grad_norm": 7.3117780685424805, + "learning_rate": 0.0001601885776217367, + "loss": 3.7477, + "step": 2004 + }, + { + "epoch": 2.95, + "grad_norm": 7.469902992248535, + "learning_rate": 0.00016015111474789733, + "loss": 3.7555, + "step": 2005 + }, + { + "epoch": 2.95, + "grad_norm": 2.65592622756958, + "learning_rate": 0.00016011363864121582, + "loss": 3.7347, + "step": 2006 + }, + { + "epoch": 2.95, + "grad_norm": 3.7348546981811523, + "learning_rate": 0.0001600761493099367, + "loss": 3.7731, + "step": 2007 + }, + { + "epoch": 2.95, + "grad_norm": 2.6096608638763428, + "learning_rate": 0.0001600386467623073, + "loss": 3.7667, + "step": 2008 + }, + { + "epoch": 2.95, + "grad_norm": 1.8336734771728516, + "learning_rate": 0.000160001131006578, + "loss": 3.7475, + "step": 2009 + }, + { + "epoch": 2.95, + "grad_norm": 9.613265991210938, + "learning_rate": 0.000159963602051002, + "loss": 3.7442, + "step": 2010 + }, + { + "epoch": 2.96, + "grad_norm": 4.934391021728516, + "learning_rate": 0.00015992605990383538, + "loss": 3.7025, + "step": 2011 + }, + { + "epoch": 2.96, + "grad_norm": 2.538320779800415, + "learning_rate": 0.0001598885045733372, + "loss": 3.7681, + "step": 2012 + }, + { + "epoch": 2.96, + "grad_norm": 2.606849193572998, + "learning_rate": 0.00015985093606776937, + "loss": 3.7237, + "step": 2013 + }, + { + "epoch": 2.96, + "grad_norm": 3.6826469898223877, + "learning_rate": 0.0001598133543953967, + "loss": 3.7382, + "step": 2014 + }, + { + "epoch": 2.96, + "grad_norm": 1.016801118850708, + "learning_rate": 0.0001597757595644869, + "loss": 3.7059, + "step": 2015 + }, + { + "epoch": 2.96, + "grad_norm": 6.855797290802002, + "learning_rate": 0.00015973815158331064, + "loss": 3.7315, + "step": 2016 + }, + { + "epoch": 2.96, + "grad_norm": 4.787756443023682, + "learning_rate": 0.00015970053046014137, + "loss": 3.7273, + "step": 2017 + }, + { + "epoch": 2.97, + "grad_norm": 5.884622573852539, + "learning_rate": 0.00015966289620325546, + "loss": 3.7079, + "step": 2018 + }, + { + "epoch": 2.97, + "grad_norm": 7.18374490737915, + "learning_rate": 0.00015962524882093227, + "loss": 3.7161, + "step": 2019 + }, + { + "epoch": 2.97, + "grad_norm": 1.6866484880447388, + "learning_rate": 0.00015958758832145388, + "loss": 3.6481, + "step": 2020 + }, + { + "epoch": 2.97, + "grad_norm": 10.191508293151855, + "learning_rate": 0.00015954991471310543, + "loss": 3.7034, + "step": 2021 + }, + { + "epoch": 2.97, + "grad_norm": 14.30430793762207, + "learning_rate": 0.0001595122280041748, + "loss": 3.7654, + "step": 2022 + }, + { + "epoch": 2.97, + "grad_norm": 5.003686904907227, + "learning_rate": 0.00015947452820295292, + "loss": 3.7148, + "step": 2023 + }, + { + "epoch": 2.97, + "grad_norm": 6.288287162780762, + "learning_rate": 0.00015943681531773333, + "loss": 3.7528, + "step": 2024 + }, + { + "epoch": 2.98, + "grad_norm": 8.045740127563477, + "learning_rate": 0.00015939908935681273, + "loss": 3.6763, + "step": 2025 + }, + { + "epoch": 2.98, + "grad_norm": 3.2493703365325928, + "learning_rate": 0.0001593613503284906, + "loss": 3.6939, + "step": 2026 + }, + { + "epoch": 2.98, + "grad_norm": 10.84262752532959, + "learning_rate": 0.0001593235982410691, + "loss": 3.7516, + "step": 2027 + }, + { + "epoch": 2.98, + "grad_norm": 15.299811363220215, + "learning_rate": 0.00015928583310285366, + "loss": 3.7319, + "step": 2028 + }, + { + "epoch": 2.98, + "eval_loss": 3.5392327308654785, + "eval_runtime": 1153.2794, + "eval_samples_per_second": 14.864, + "eval_steps_per_second": 0.93, + "step": 2028 + }, + { + "epoch": 2.98, + "grad_norm": 8.231132507324219, + "learning_rate": 0.0001592480549221522, + "loss": 3.7177, + "step": 2029 + }, + { + "epoch": 2.98, + "grad_norm": 5.159106254577637, + "learning_rate": 0.00015921026370727573, + "loss": 3.7017, + "step": 2030 + }, + { + "epoch": 2.98, + "grad_norm": 22.314449310302734, + "learning_rate": 0.00015917245946653807, + "loss": 3.6842, + "step": 2031 + }, + { + "epoch": 2.99, + "grad_norm": 8.433747291564941, + "learning_rate": 0.00015913464220825588, + "loss": 3.6786, + "step": 2032 + }, + { + "epoch": 2.99, + "grad_norm": 3.3623647689819336, + "learning_rate": 0.00015909681194074867, + "loss": 3.678, + "step": 2033 + }, + { + "epoch": 2.99, + "grad_norm": 10.62174129486084, + "learning_rate": 0.00015905896867233894, + "loss": 3.7098, + "step": 2034 + }, + { + "epoch": 2.99, + "grad_norm": 15.299424171447754, + "learning_rate": 0.00015902111241135187, + "loss": 3.6926, + "step": 2035 + }, + { + "epoch": 2.99, + "grad_norm": 9.536619186401367, + "learning_rate": 0.00015898324316611563, + "loss": 3.673, + "step": 2036 + }, + { + "epoch": 2.99, + "grad_norm": 5.059504508972168, + "learning_rate": 0.00015894536094496117, + "loss": 3.6625, + "step": 2037 + }, + { + "epoch": 3.0, + "grad_norm": 7.704652786254883, + "learning_rate": 0.00015890746575622231, + "loss": 3.6632, + "step": 2038 + }, + { + "epoch": 3.0, + "grad_norm": 4.4129557609558105, + "learning_rate": 0.00015886955760823578, + "loss": 3.6315, + "step": 2039 + }, + { + "epoch": 3.0, + "grad_norm": 7.341033935546875, + "learning_rate": 0.0001588316365093411, + "loss": 3.6763, + "step": 2040 + }, + { + "epoch": 3.0, + "grad_norm": 9.556234359741211, + "learning_rate": 0.00015879370246788067, + "loss": 3.6774, + "step": 2041 + }, + { + "epoch": 3.0, + "grad_norm": 3.8044204711914062, + "learning_rate": 0.00015875575549219973, + "loss": 3.6249, + "step": 2042 + }, + { + "epoch": 3.0, + "grad_norm": 7.746762275695801, + "learning_rate": 0.0001587177955906463, + "loss": 3.6366, + "step": 2043 + }, + { + "epoch": 3.0, + "grad_norm": 9.851776123046875, + "learning_rate": 0.0001586798227715714, + "loss": 3.6885, + "step": 2044 + }, + { + "epoch": 3.01, + "grad_norm": 5.628739833831787, + "learning_rate": 0.00015864183704332872, + "loss": 3.6248, + "step": 2045 + }, + { + "epoch": 3.01, + "grad_norm": 6.932154178619385, + "learning_rate": 0.00015860383841427485, + "loss": 3.6499, + "step": 2046 + }, + { + "epoch": 3.01, + "grad_norm": 10.925968170166016, + "learning_rate": 0.00015856582689276934, + "loss": 3.6793, + "step": 2047 + }, + { + "epoch": 3.01, + "grad_norm": 5.439330577850342, + "learning_rate": 0.00015852780248717442, + "loss": 3.654, + "step": 2048 + }, + { + "epoch": 3.01, + "grad_norm": 6.066278457641602, + "learning_rate": 0.0001584897652058551, + "loss": 3.6738, + "step": 2049 + }, + { + "epoch": 3.01, + "grad_norm": 8.065271377563477, + "learning_rate": 0.00015845171505717947, + "loss": 3.659, + "step": 2050 + }, + { + "epoch": 3.01, + "grad_norm": 3.8582706451416016, + "learning_rate": 0.00015841365204951826, + "loss": 3.6458, + "step": 2051 + }, + { + "epoch": 3.02, + "grad_norm": 9.716731071472168, + "learning_rate": 0.00015837557619124507, + "loss": 3.6313, + "step": 2052 + }, + { + "epoch": 3.0, + "grad_norm": 14.822004318237305, + "learning_rate": 0.00015833748749073632, + "loss": 3.6512, + "step": 2053 + }, + { + "epoch": 3.0, + "grad_norm": 8.674092292785645, + "learning_rate": 0.00015829938595637128, + "loss": 3.6251, + "step": 2054 + }, + { + "epoch": 3.0, + "grad_norm": 4.23799467086792, + "learning_rate": 0.00015826127159653203, + "loss": 3.6496, + "step": 2055 + }, + { + "epoch": 3.01, + "grad_norm": 9.582878112792969, + "learning_rate": 0.00015822314441960348, + "loss": 3.5982, + "step": 2056 + }, + { + "epoch": 3.01, + "grad_norm": 5.006598949432373, + "learning_rate": 0.00015818500443397337, + "loss": 3.6347, + "step": 2057 + }, + { + "epoch": 3.01, + "grad_norm": 4.559854984283447, + "learning_rate": 0.0001581468516480322, + "loss": 3.6414, + "step": 2058 + }, + { + "epoch": 3.01, + "grad_norm": 5.798408508300781, + "learning_rate": 0.00015810868607017337, + "loss": 3.6176, + "step": 2059 + }, + { + "epoch": 3.01, + "grad_norm": 1.9246889352798462, + "learning_rate": 0.00015807050770879304, + "loss": 3.6134, + "step": 2060 + }, + { + "epoch": 3.01, + "grad_norm": 1.124200701713562, + "learning_rate": 0.00015803231657229017, + "loss": 3.6295, + "step": 2061 + }, + { + "epoch": 3.01, + "grad_norm": 7.120904445648193, + "learning_rate": 0.00015799411266906657, + "loss": 3.6101, + "step": 2062 + }, + { + "epoch": 3.02, + "grad_norm": 5.709414482116699, + "learning_rate": 0.00015795589600752682, + "loss": 3.6255, + "step": 2063 + }, + { + "epoch": 3.02, + "grad_norm": 3.2987709045410156, + "learning_rate": 0.0001579176665960784, + "loss": 3.6194, + "step": 2064 + }, + { + "epoch": 3.02, + "grad_norm": 4.208385944366455, + "learning_rate": 0.00015787942444313145, + "loss": 3.6357, + "step": 2065 + }, + { + "epoch": 3.02, + "grad_norm": 2.485320568084717, + "learning_rate": 0.00015784116955709897, + "loss": 3.6233, + "step": 2066 + }, + { + "epoch": 3.02, + "grad_norm": 0.9779897332191467, + "learning_rate": 0.0001578029019463969, + "loss": 3.6382, + "step": 2067 + }, + { + "epoch": 3.02, + "grad_norm": 5.236642837524414, + "learning_rate": 0.0001577646216194437, + "loss": 3.5985, + "step": 2068 + }, + { + "epoch": 3.03, + "grad_norm": 4.276678562164307, + "learning_rate": 0.00015772632858466093, + "loss": 3.614, + "step": 2069 + }, + { + "epoch": 3.03, + "grad_norm": 4.390791893005371, + "learning_rate": 0.00015768802285047272, + "loss": 3.6166, + "step": 2070 + }, + { + "epoch": 3.03, + "grad_norm": 4.886763572692871, + "learning_rate": 0.00015764970442530609, + "loss": 3.5902, + "step": 2071 + }, + { + "epoch": 3.03, + "grad_norm": 3.569462776184082, + "learning_rate": 0.00015761137331759084, + "loss": 3.5779, + "step": 2072 + }, + { + "epoch": 3.03, + "grad_norm": 2.935269832611084, + "learning_rate": 0.00015757302953575954, + "loss": 3.6369, + "step": 2073 + }, + { + "epoch": 3.03, + "grad_norm": 4.911141872406006, + "learning_rate": 0.0001575346730882476, + "loss": 3.6034, + "step": 2074 + }, + { + "epoch": 3.03, + "grad_norm": 4.09962797164917, + "learning_rate": 0.00015749630398349316, + "loss": 3.6431, + "step": 2075 + }, + { + "epoch": 3.04, + "grad_norm": 4.247868537902832, + "learning_rate": 0.00015745792222993715, + "loss": 3.6258, + "step": 2076 + }, + { + "epoch": 3.04, + "grad_norm": 3.7722268104553223, + "learning_rate": 0.00015741952783602332, + "loss": 3.597, + "step": 2077 + }, + { + "epoch": 3.04, + "grad_norm": 5.005308151245117, + "learning_rate": 0.00015738112081019817, + "loss": 3.6146, + "step": 2078 + }, + { + "epoch": 3.04, + "grad_norm": 4.696826934814453, + "learning_rate": 0.000157342701160911, + "loss": 3.6477, + "step": 2079 + }, + { + "epoch": 3.04, + "grad_norm": 3.856837034225464, + "learning_rate": 0.00015730426889661386, + "loss": 3.6594, + "step": 2080 + }, + { + "epoch": 3.04, + "grad_norm": 3.4349594116210938, + "learning_rate": 0.0001572658240257616, + "loss": 3.5798, + "step": 2081 + }, + { + "epoch": 3.04, + "grad_norm": 4.7815842628479, + "learning_rate": 0.00015722736655681186, + "loss": 3.5797, + "step": 2082 + }, + { + "epoch": 3.05, + "grad_norm": 3.9778714179992676, + "learning_rate": 0.00015718889649822494, + "loss": 3.6173, + "step": 2083 + }, + { + "epoch": 3.05, + "grad_norm": 3.863435745239258, + "learning_rate": 0.0001571504138584641, + "loss": 3.5669, + "step": 2084 + }, + { + "epoch": 3.05, + "grad_norm": 3.9693548679351807, + "learning_rate": 0.0001571119186459952, + "loss": 3.6072, + "step": 2085 + }, + { + "epoch": 3.05, + "grad_norm": 4.673881530761719, + "learning_rate": 0.00015707341086928698, + "loss": 3.5794, + "step": 2086 + }, + { + "epoch": 3.05, + "grad_norm": 3.680610179901123, + "learning_rate": 0.0001570348905368108, + "loss": 3.5927, + "step": 2087 + }, + { + "epoch": 3.05, + "grad_norm": 4.484524726867676, + "learning_rate": 0.00015699635765704096, + "loss": 3.5588, + "step": 2088 + }, + { + "epoch": 3.05, + "grad_norm": 4.283873558044434, + "learning_rate": 0.00015695781223845441, + "loss": 3.6385, + "step": 2089 + }, + { + "epoch": 3.06, + "grad_norm": 3.3641796112060547, + "learning_rate": 0.00015691925428953089, + "loss": 3.6163, + "step": 2090 + }, + { + "epoch": 3.06, + "grad_norm": 2.774048089981079, + "learning_rate": 0.00015688068381875287, + "loss": 3.6696, + "step": 2091 + }, + { + "epoch": 3.06, + "grad_norm": 5.591458320617676, + "learning_rate": 0.0001568421008346056, + "loss": 3.54, + "step": 2092 + }, + { + "epoch": 3.06, + "grad_norm": 5.354968547821045, + "learning_rate": 0.0001568035053455771, + "loss": 3.5762, + "step": 2093 + }, + { + "epoch": 3.06, + "grad_norm": 2.219583749771118, + "learning_rate": 0.0001567648973601581, + "loss": 3.5527, + "step": 2094 + }, + { + "epoch": 3.06, + "grad_norm": 1.655458688735962, + "learning_rate": 0.0001567262768868421, + "loss": 3.5702, + "step": 2095 + }, + { + "epoch": 3.07, + "grad_norm": 5.801334381103516, + "learning_rate": 0.00015668764393412537, + "loss": 3.5317, + "step": 2096 + }, + { + "epoch": 3.07, + "grad_norm": 5.561006546020508, + "learning_rate": 0.00015664899851050684, + "loss": 3.5875, + "step": 2097 + }, + { + "epoch": 3.07, + "grad_norm": 2.8261959552764893, + "learning_rate": 0.0001566103406244883, + "loss": 3.5284, + "step": 2098 + }, + { + "epoch": 3.07, + "grad_norm": 2.3607285022735596, + "learning_rate": 0.00015657167028457423, + "loss": 3.5596, + "step": 2099 + }, + { + "epoch": 3.07, + "grad_norm": 6.529599666595459, + "learning_rate": 0.00015653298749927178, + "loss": 3.5603, + "step": 2100 + }, + { + "epoch": 3.07, + "grad_norm": 5.787454605102539, + "learning_rate": 0.00015649429227709098, + "loss": 3.5419, + "step": 2101 + }, + { + "epoch": 3.07, + "grad_norm": 1.2445347309112549, + "learning_rate": 0.00015645558462654447, + "loss": 3.5599, + "step": 2102 + }, + { + "epoch": 3.08, + "grad_norm": 1.686257243156433, + "learning_rate": 0.0001564168645561477, + "loss": 3.5051, + "step": 2103 + }, + { + "epoch": 3.08, + "grad_norm": 5.125936985015869, + "learning_rate": 0.00015637813207441882, + "loss": 3.4685, + "step": 2104 + }, + { + "epoch": 3.08, + "grad_norm": 4.658946514129639, + "learning_rate": 0.0001563393871898787, + "loss": 3.5291, + "step": 2105 + }, + { + "epoch": 3.08, + "grad_norm": 4.34171199798584, + "learning_rate": 0.00015630062991105098, + "loss": 3.5044, + "step": 2106 + }, + { + "epoch": 3.08, + "grad_norm": 5.2388458251953125, + "learning_rate": 0.00015626186024646197, + "loss": 3.4543, + "step": 2107 + }, + { + "epoch": 3.08, + "grad_norm": 3.39254093170166, + "learning_rate": 0.00015622307820464077, + "loss": 3.4678, + "step": 2108 + }, + { + "epoch": 3.08, + "grad_norm": 2.4865806102752686, + "learning_rate": 0.00015618428379411917, + "loss": 3.4326, + "step": 2109 + }, + { + "epoch": 3.09, + "grad_norm": 7.326157093048096, + "learning_rate": 0.00015614547702343166, + "loss": 3.4709, + "step": 2110 + }, + { + "epoch": 3.09, + "grad_norm": 7.048543930053711, + "learning_rate": 0.00015610665790111546, + "loss": 3.4498, + "step": 2111 + }, + { + "epoch": 3.09, + "grad_norm": 10.954419136047363, + "learning_rate": 0.00015606782643571055, + "loss": 3.3944, + "step": 2112 + }, + { + "epoch": 3.09, + "grad_norm": 11.594827651977539, + "learning_rate": 0.00015602898263575952, + "loss": 3.4224, + "step": 2113 + }, + { + "epoch": 3.09, + "grad_norm": 6.556940078735352, + "learning_rate": 0.0001559901265098079, + "loss": 3.4079, + "step": 2114 + }, + { + "epoch": 3.09, + "grad_norm": 3.0502734184265137, + "learning_rate": 0.00015595125806640357, + "loss": 3.409, + "step": 2115 + }, + { + "epoch": 3.09, + "grad_norm": 6.4218220710754395, + "learning_rate": 0.0001559123773140975, + "loss": 3.4382, + "step": 2116 + }, + { + "epoch": 3.1, + "grad_norm": 6.8841753005981445, + "learning_rate": 0.00015587348426144312, + "loss": 3.463, + "step": 2117 + }, + { + "epoch": 3.1, + "grad_norm": 35.60420227050781, + "learning_rate": 0.00015583457891699666, + "loss": 3.3878, + "step": 2118 + }, + { + "epoch": 3.1, + "grad_norm": 2.3054988384246826, + "learning_rate": 0.000155795661289317, + "loss": 3.382, + "step": 2119 + }, + { + "epoch": 3.1, + "grad_norm": 6.13869571685791, + "learning_rate": 0.00015575673138696578, + "loss": 3.4213, + "step": 2120 + }, + { + "epoch": 3.1, + "grad_norm": 5.998043537139893, + "learning_rate": 0.00015571778921850732, + "loss": 3.382, + "step": 2121 + }, + { + "epoch": 3.1, + "grad_norm": 1.6560511589050293, + "learning_rate": 0.00015567883479250863, + "loss": 3.4691, + "step": 2122 + }, + { + "epoch": 3.11, + "grad_norm": 2.120764970779419, + "learning_rate": 0.00015563986811753945, + "loss": 3.3711, + "step": 2123 + }, + { + "epoch": 3.11, + "grad_norm": 1.6065012216567993, + "learning_rate": 0.00015560088920217217, + "loss": 3.4219, + "step": 2124 + }, + { + "epoch": 3.11, + "grad_norm": 6.458485126495361, + "learning_rate": 0.00015556189805498188, + "loss": 3.4184, + "step": 2125 + }, + { + "epoch": 3.11, + "grad_norm": 5.5630202293396, + "learning_rate": 0.0001555228946845464, + "loss": 3.4185, + "step": 2126 + }, + { + "epoch": 3.11, + "grad_norm": 4.821037769317627, + "learning_rate": 0.00015548387909944614, + "loss": 3.4003, + "step": 2127 + }, + { + "epoch": 3.11, + "grad_norm": 13.251823425292969, + "learning_rate": 0.00015544485130826433, + "loss": 3.4081, + "step": 2128 + }, + { + "epoch": 3.11, + "grad_norm": 5.775404930114746, + "learning_rate": 0.00015540581131958682, + "loss": 3.3918, + "step": 2129 + }, + { + "epoch": 3.12, + "grad_norm": 6.066825866699219, + "learning_rate": 0.00015536675914200213, + "loss": 3.3863, + "step": 2130 + }, + { + "epoch": 3.12, + "grad_norm": 11.889984130859375, + "learning_rate": 0.0001553276947841015, + "loss": 3.4055, + "step": 2131 + }, + { + "epoch": 3.12, + "grad_norm": 3.267805576324463, + "learning_rate": 0.00015528861825447876, + "loss": 3.4031, + "step": 2132 + }, + { + "epoch": 3.12, + "grad_norm": 11.77675724029541, + "learning_rate": 0.00015524952956173056, + "loss": 3.4159, + "step": 2133 + }, + { + "epoch": 3.12, + "grad_norm": 16.67833137512207, + "learning_rate": 0.0001552104287144561, + "loss": 3.4706, + "step": 2134 + }, + { + "epoch": 3.12, + "grad_norm": 9.038000106811523, + "learning_rate": 0.00015517131572125736, + "loss": 3.4391, + "step": 2135 + }, + { + "epoch": 3.12, + "grad_norm": 7.349591255187988, + "learning_rate": 0.0001551321905907389, + "loss": 3.4181, + "step": 2136 + }, + { + "epoch": 3.13, + "grad_norm": 11.440853118896484, + "learning_rate": 0.00015509305333150797, + "loss": 3.4324, + "step": 2137 + }, + { + "epoch": 3.13, + "grad_norm": 9.080293655395508, + "learning_rate": 0.00015505390395217453, + "loss": 3.4841, + "step": 2138 + }, + { + "epoch": 3.13, + "grad_norm": 2.2852790355682373, + "learning_rate": 0.0001550147424613512, + "loss": 3.35, + "step": 2139 + }, + { + "epoch": 3.13, + "grad_norm": 17.961963653564453, + "learning_rate": 0.00015497556886765316, + "loss": 3.4265, + "step": 2140 + }, + { + "epoch": 3.13, + "grad_norm": 26.195865631103516, + "learning_rate": 0.00015493638317969844, + "loss": 3.6051, + "step": 2141 + }, + { + "epoch": 3.13, + "grad_norm": 19.95352554321289, + "learning_rate": 0.0001548971854061076, + "loss": 3.6381, + "step": 2142 + }, + { + "epoch": 3.13, + "grad_norm": 5.972691535949707, + "learning_rate": 0.00015485797555550386, + "loss": 3.5607, + "step": 2143 + }, + { + "epoch": 3.14, + "grad_norm": 13.078893661499023, + "learning_rate": 0.0001548187536365131, + "loss": 3.4778, + "step": 2144 + }, + { + "epoch": 3.14, + "grad_norm": 15.685088157653809, + "learning_rate": 0.000154779519657764, + "loss": 3.5011, + "step": 2145 + }, + { + "epoch": 3.14, + "grad_norm": 14.251028060913086, + "learning_rate": 0.00015474027362788763, + "loss": 3.6155, + "step": 2146 + }, + { + "epoch": 3.14, + "grad_norm": 10.396077156066895, + "learning_rate": 0.00015470101555551796, + "loss": 3.5646, + "step": 2147 + }, + { + "epoch": 3.14, + "grad_norm": 5.34552526473999, + "learning_rate": 0.0001546617454492914, + "loss": 3.4643, + "step": 2148 + }, + { + "epoch": 3.14, + "grad_norm": 18.76544189453125, + "learning_rate": 0.0001546224633178472, + "loss": 3.4686, + "step": 2149 + }, + { + "epoch": 3.15, + "grad_norm": 32.796051025390625, + "learning_rate": 0.00015458316916982715, + "loss": 3.6171, + "step": 2150 + }, + { + "epoch": 3.15, + "grad_norm": 31.727258682250977, + "learning_rate": 0.00015454386301387565, + "loss": 3.751, + "step": 2151 + }, + { + "epoch": 3.15, + "grad_norm": 18.45448112487793, + "learning_rate": 0.00015450454485863984, + "loss": 3.6405, + "step": 2152 + }, + { + "epoch": 3.15, + "grad_norm": 8.118375778198242, + "learning_rate": 0.00015446521471276946, + "loss": 3.6001, + "step": 2153 + }, + { + "epoch": 3.15, + "grad_norm": 16.14570426940918, + "learning_rate": 0.0001544258725849168, + "loss": 3.559, + "step": 2154 + }, + { + "epoch": 3.15, + "grad_norm": 16.983943939208984, + "learning_rate": 0.00015438651848373696, + "loss": 3.5108, + "step": 2155 + }, + { + "epoch": 3.15, + "grad_norm": 13.206223487854004, + "learning_rate": 0.00015434715241788752, + "loss": 3.4883, + "step": 2156 + }, + { + "epoch": 3.16, + "grad_norm": 8.664280891418457, + "learning_rate": 0.00015430777439602876, + "loss": 3.4346, + "step": 2157 + }, + { + "epoch": 3.16, + "grad_norm": 5.633321762084961, + "learning_rate": 0.00015426838442682362, + "loss": 3.4285, + "step": 2158 + }, + { + "epoch": 3.16, + "grad_norm": 14.557170867919922, + "learning_rate": 0.00015422898251893756, + "loss": 3.4706, + "step": 2159 + }, + { + "epoch": 3.16, + "grad_norm": 19.554941177368164, + "learning_rate": 0.00015418956868103878, + "loss": 3.4929, + "step": 2160 + }, + { + "epoch": 3.16, + "grad_norm": 15.579254150390625, + "learning_rate": 0.00015415014292179805, + "loss": 3.4144, + "step": 2161 + }, + { + "epoch": 3.16, + "grad_norm": 5.079093933105469, + "learning_rate": 0.0001541107052498888, + "loss": 3.3886, + "step": 2162 + }, + { + "epoch": 3.16, + "grad_norm": 10.48103141784668, + "learning_rate": 0.00015407125567398698, + "loss": 3.4392, + "step": 2163 + }, + { + "epoch": 3.17, + "grad_norm": 14.062845230102539, + "learning_rate": 0.00015403179420277131, + "loss": 3.4527, + "step": 2164 + }, + { + "epoch": 3.17, + "grad_norm": 13.237971305847168, + "learning_rate": 0.00015399232084492304, + "loss": 3.4518, + "step": 2165 + }, + { + "epoch": 3.17, + "grad_norm": 7.261989593505859, + "learning_rate": 0.000153952835609126, + "loss": 3.4716, + "step": 2166 + }, + { + "epoch": 3.17, + "grad_norm": 5.766734600067139, + "learning_rate": 0.0001539133385040667, + "loss": 3.3591, + "step": 2167 + }, + { + "epoch": 3.17, + "grad_norm": 10.160755157470703, + "learning_rate": 0.0001538738295384343, + "loss": 3.3638, + "step": 2168 + }, + { + "epoch": 3.17, + "grad_norm": 7.740986347198486, + "learning_rate": 0.00015383430872092037, + "loss": 3.4132, + "step": 2169 + }, + { + "epoch": 3.18, + "grad_norm": 2.3649628162384033, + "learning_rate": 0.00015379477606021933, + "loss": 3.4459, + "step": 2170 + }, + { + "epoch": 3.18, + "grad_norm": 4.513500213623047, + "learning_rate": 0.00015375523156502805, + "loss": 3.4318, + "step": 2171 + }, + { + "epoch": 3.18, + "grad_norm": 3.214402198791504, + "learning_rate": 0.00015371567524404613, + "loss": 3.3651, + "step": 2172 + }, + { + "epoch": 3.18, + "grad_norm": 3.8382275104522705, + "learning_rate": 0.0001536761071059756, + "loss": 3.4077, + "step": 2173 + }, + { + "epoch": 3.18, + "grad_norm": 1.4553401470184326, + "learning_rate": 0.0001536365271595212, + "loss": 3.3301, + "step": 2174 + }, + { + "epoch": 3.18, + "grad_norm": 4.466412544250488, + "learning_rate": 0.00015359693541339037, + "loss": 3.3988, + "step": 2175 + }, + { + "epoch": 3.18, + "grad_norm": 3.6478259563446045, + "learning_rate": 0.00015355733187629288, + "loss": 3.3779, + "step": 2176 + }, + { + "epoch": 3.19, + "grad_norm": 5.013825416564941, + "learning_rate": 0.0001535177165569413, + "loss": 3.413, + "step": 2177 + }, + { + "epoch": 3.19, + "grad_norm": 4.778167247772217, + "learning_rate": 0.00015347808946405075, + "loss": 3.3425, + "step": 2178 + }, + { + "epoch": 3.19, + "grad_norm": 5.024134159088135, + "learning_rate": 0.00015343845060633893, + "loss": 3.3676, + "step": 2179 + }, + { + "epoch": 3.19, + "grad_norm": 5.648217678070068, + "learning_rate": 0.00015339879999252606, + "loss": 3.3165, + "step": 2180 + }, + { + "epoch": 3.19, + "grad_norm": 2.347975015640259, + "learning_rate": 0.00015335913763133511, + "loss": 3.3588, + "step": 2181 + }, + { + "epoch": 3.19, + "grad_norm": 5.947681427001953, + "learning_rate": 0.00015331946353149144, + "loss": 3.3811, + "step": 2182 + }, + { + "epoch": 3.19, + "grad_norm": 5.794849395751953, + "learning_rate": 0.00015327977770172315, + "loss": 3.3751, + "step": 2183 + }, + { + "epoch": 3.2, + "grad_norm": 2.177245855331421, + "learning_rate": 0.0001532400801507608, + "loss": 3.3472, + "step": 2184 + }, + { + "epoch": 3.2, + "grad_norm": 4.221487045288086, + "learning_rate": 0.00015320037088733764, + "loss": 3.3766, + "step": 2185 + }, + { + "epoch": 3.2, + "grad_norm": 3.649068832397461, + "learning_rate": 0.00015316064992018945, + "loss": 3.3633, + "step": 2186 + }, + { + "epoch": 3.2, + "grad_norm": 3.780611038208008, + "learning_rate": 0.00015312091725805448, + "loss": 3.3371, + "step": 2187 + }, + { + "epoch": 3.2, + "grad_norm": 1.3155494928359985, + "learning_rate": 0.00015308117290967373, + "loss": 3.3083, + "step": 2188 + }, + { + "epoch": 3.2, + "grad_norm": 1.961712121963501, + "learning_rate": 0.0001530414168837907, + "loss": 3.3465, + "step": 2189 + }, + { + "epoch": 3.2, + "grad_norm": 2.569746494293213, + "learning_rate": 0.00015300164918915138, + "loss": 3.3853, + "step": 2190 + }, + { + "epoch": 3.21, + "grad_norm": 1.9574949741363525, + "learning_rate": 0.00015296186983450446, + "loss": 3.3444, + "step": 2191 + }, + { + "epoch": 3.21, + "grad_norm": 2.3811464309692383, + "learning_rate": 0.00015292207882860114, + "loss": 3.316, + "step": 2192 + }, + { + "epoch": 3.21, + "grad_norm": 1.6529579162597656, + "learning_rate": 0.00015288227618019513, + "loss": 3.3269, + "step": 2193 + }, + { + "epoch": 3.21, + "grad_norm": 1.4792758226394653, + "learning_rate": 0.00015284246189804272, + "loss": 3.3406, + "step": 2194 + }, + { + "epoch": 3.21, + "grad_norm": 1.3875869512557983, + "learning_rate": 0.0001528026359909029, + "loss": 3.3492, + "step": 2195 + }, + { + "epoch": 3.21, + "grad_norm": 3.4284157752990723, + "learning_rate": 0.00015276279846753696, + "loss": 3.3176, + "step": 2196 + }, + { + "epoch": 3.22, + "grad_norm": 1.68557608127594, + "learning_rate": 0.000152722949336709, + "loss": 3.3142, + "step": 2197 + }, + { + "epoch": 3.22, + "eval_loss": 3.1609973907470703, + "eval_runtime": 1163.3297, + "eval_samples_per_second": 14.735, + "eval_steps_per_second": 0.921, + "step": 2197 + }, + { + "epoch": 3.22, + "grad_norm": 4.416086673736572, + "learning_rate": 0.0001526830886071855, + "loss": 3.3143, + "step": 2198 + }, + { + "epoch": 3.22, + "grad_norm": 2.9269638061523438, + "learning_rate": 0.0001526432162877356, + "loss": 3.3417, + "step": 2199 + }, + { + "epoch": 3.22, + "grad_norm": 3.496257781982422, + "learning_rate": 0.00015260333238713088, + "loss": 3.3356, + "step": 2200 + }, + { + "epoch": 3.22, + "grad_norm": 3.9582598209381104, + "learning_rate": 0.00015256343691414558, + "loss": 3.3688, + "step": 2201 + }, + { + "epoch": 3.22, + "grad_norm": 3.0187695026397705, + "learning_rate": 0.0001525235298775564, + "loss": 3.3381, + "step": 2202 + }, + { + "epoch": 3.22, + "grad_norm": 1.4246793985366821, + "learning_rate": 0.0001524836112861427, + "loss": 3.3228, + "step": 2203 + }, + { + "epoch": 3.23, + "grad_norm": 2.6875085830688477, + "learning_rate": 0.00015244368114868625, + "loss": 3.2918, + "step": 2204 + }, + { + "epoch": 3.23, + "grad_norm": 2.1740024089813232, + "learning_rate": 0.00015240373947397133, + "loss": 3.2933, + "step": 2205 + }, + { + "epoch": 3.23, + "grad_norm": 5.561516284942627, + "learning_rate": 0.000152363786270785, + "loss": 3.3344, + "step": 2206 + }, + { + "epoch": 3.23, + "grad_norm": 2.453502655029297, + "learning_rate": 0.00015232382154791655, + "loss": 3.3204, + "step": 2207 + }, + { + "epoch": 3.23, + "grad_norm": 3.4560389518737793, + "learning_rate": 0.0001522838453141581, + "loss": 3.3054, + "step": 2208 + }, + { + "epoch": 3.23, + "grad_norm": 1.4101018905639648, + "learning_rate": 0.000152243857578304, + "loss": 3.3298, + "step": 2209 + }, + { + "epoch": 3.23, + "grad_norm": 1.3143606185913086, + "learning_rate": 0.0001522038583491514, + "loss": 3.3294, + "step": 2210 + }, + { + "epoch": 3.24, + "grad_norm": 2.1382877826690674, + "learning_rate": 0.00015216384763549977, + "loss": 3.2964, + "step": 2211 + }, + { + "epoch": 3.24, + "grad_norm": 3.5953447818756104, + "learning_rate": 0.00015212382544615126, + "loss": 3.3138, + "step": 2212 + }, + { + "epoch": 3.24, + "grad_norm": 2.013397216796875, + "learning_rate": 0.0001520837917899105, + "loss": 3.3318, + "step": 2213 + }, + { + "epoch": 3.24, + "grad_norm": 5.66317892074585, + "learning_rate": 0.0001520437466755846, + "loss": 3.3226, + "step": 2214 + }, + { + "epoch": 3.24, + "grad_norm": 5.973227024078369, + "learning_rate": 0.00015200369011198319, + "loss": 3.2917, + "step": 2215 + }, + { + "epoch": 3.24, + "grad_norm": 1.0823651552200317, + "learning_rate": 0.00015196362210791845, + "loss": 3.2837, + "step": 2216 + }, + { + "epoch": 3.24, + "grad_norm": 2.2828619480133057, + "learning_rate": 0.00015192354267220512, + "loss": 3.2994, + "step": 2217 + }, + { + "epoch": 3.25, + "grad_norm": 3.3102993965148926, + "learning_rate": 0.00015188345181366036, + "loss": 3.3674, + "step": 2218 + }, + { + "epoch": 3.25, + "grad_norm": 5.152593612670898, + "learning_rate": 0.00015184334954110394, + "loss": 3.3411, + "step": 2219 + }, + { + "epoch": 3.25, + "grad_norm": 6.140212535858154, + "learning_rate": 0.00015180323586335806, + "loss": 3.3188, + "step": 2220 + }, + { + "epoch": 3.25, + "grad_norm": 2.0337228775024414, + "learning_rate": 0.0001517631107892474, + "loss": 3.2678, + "step": 2221 + }, + { + "epoch": 3.25, + "grad_norm": 4.024622917175293, + "learning_rate": 0.0001517229743275993, + "loss": 3.2708, + "step": 2222 + }, + { + "epoch": 3.25, + "grad_norm": 2.378216505050659, + "learning_rate": 0.00015168282648724351, + "loss": 3.3462, + "step": 2223 + }, + { + "epoch": 3.26, + "grad_norm": 4.922878742218018, + "learning_rate": 0.0001516426672770122, + "loss": 3.291, + "step": 2224 + }, + { + "epoch": 3.26, + "grad_norm": 4.67563533782959, + "learning_rate": 0.00015160249670574026, + "loss": 3.2711, + "step": 2225 + }, + { + "epoch": 3.26, + "grad_norm": 4.26483154296875, + "learning_rate": 0.0001515623147822648, + "loss": 3.2862, + "step": 2226 + }, + { + "epoch": 3.26, + "grad_norm": 6.510219097137451, + "learning_rate": 0.00015152212151542567, + "loss": 3.3276, + "step": 2227 + }, + { + "epoch": 3.26, + "grad_norm": 4.469378471374512, + "learning_rate": 0.0001514819169140651, + "loss": 3.3276, + "step": 2228 + }, + { + "epoch": 3.26, + "grad_norm": 4.4898681640625, + "learning_rate": 0.0001514417009870278, + "loss": 3.2907, + "step": 2229 + }, + { + "epoch": 3.26, + "grad_norm": 5.1355485916137695, + "learning_rate": 0.00015140147374316106, + "loss": 3.3146, + "step": 2230 + }, + { + "epoch": 3.27, + "grad_norm": 3.954402208328247, + "learning_rate": 0.00015136123519131456, + "loss": 3.2888, + "step": 2231 + }, + { + "epoch": 3.27, + "grad_norm": 8.745506286621094, + "learning_rate": 0.00015132098534034052, + "loss": 3.3684, + "step": 2232 + }, + { + "epoch": 3.27, + "grad_norm": 7.295302391052246, + "learning_rate": 0.00015128072419909368, + "loss": 3.3378, + "step": 2233 + }, + { + "epoch": 3.27, + "grad_norm": 1.574733853340149, + "learning_rate": 0.00015124045177643112, + "loss": 3.3003, + "step": 2234 + }, + { + "epoch": 3.27, + "grad_norm": 6.663206100463867, + "learning_rate": 0.0001512001680812126, + "loss": 3.3466, + "step": 2235 + }, + { + "epoch": 3.27, + "grad_norm": 7.278547286987305, + "learning_rate": 0.00015115987312230026, + "loss": 3.2945, + "step": 2236 + }, + { + "epoch": 3.27, + "grad_norm": 3.385538101196289, + "learning_rate": 0.00015111956690855863, + "loss": 3.2958, + "step": 2237 + }, + { + "epoch": 3.28, + "grad_norm": 7.496275424957275, + "learning_rate": 0.00015107924944885486, + "loss": 3.3214, + "step": 2238 + }, + { + "epoch": 3.28, + "grad_norm": 25.01158332824707, + "learning_rate": 0.00015103892075205862, + "loss": 3.4204, + "step": 2239 + }, + { + "epoch": 3.28, + "grad_norm": 9.22580337524414, + "learning_rate": 0.00015099858082704182, + "loss": 3.2895, + "step": 2240 + }, + { + "epoch": 3.28, + "grad_norm": 1.5853315591812134, + "learning_rate": 0.00015095822968267896, + "loss": 3.3388, + "step": 2241 + }, + { + "epoch": 3.28, + "grad_norm": 8.119154930114746, + "learning_rate": 0.00015091786732784716, + "loss": 3.35, + "step": 2242 + }, + { + "epoch": 3.28, + "grad_norm": 10.604775428771973, + "learning_rate": 0.00015087749377142574, + "loss": 3.3851, + "step": 2243 + }, + { + "epoch": 3.28, + "grad_norm": 7.424951076507568, + "learning_rate": 0.00015083710902229664, + "loss": 3.3619, + "step": 2244 + }, + { + "epoch": 3.29, + "grad_norm": 5.7252373695373535, + "learning_rate": 0.0001507967130893443, + "loss": 3.342, + "step": 2245 + }, + { + "epoch": 3.29, + "grad_norm": 8.934366226196289, + "learning_rate": 0.00015075630598145547, + "loss": 3.3553, + "step": 2246 + }, + { + "epoch": 3.29, + "grad_norm": 6.990628719329834, + "learning_rate": 0.00015071588770751947, + "loss": 3.3318, + "step": 2247 + }, + { + "epoch": 3.29, + "grad_norm": 19.209877014160156, + "learning_rate": 0.00015067545827642808, + "loss": 3.3181, + "step": 2248 + }, + { + "epoch": 3.29, + "grad_norm": 6.441047191619873, + "learning_rate": 0.00015063501769707548, + "loss": 3.3515, + "step": 2249 + }, + { + "epoch": 3.29, + "grad_norm": 5.99086332321167, + "learning_rate": 0.00015059456597835835, + "loss": 3.334, + "step": 2250 + }, + { + "epoch": 3.3, + "grad_norm": 2.6446921825408936, + "learning_rate": 0.00015055410312917573, + "loss": 3.332, + "step": 2251 + }, + { + "epoch": 3.3, + "grad_norm": 266.819091796875, + "learning_rate": 0.00015051362915842923, + "loss": 3.3605, + "step": 2252 + }, + { + "epoch": 3.3, + "grad_norm": 16.00495719909668, + "learning_rate": 0.00015047314407502283, + "loss": 3.3521, + "step": 2253 + }, + { + "epoch": 3.3, + "grad_norm": 8.721566200256348, + "learning_rate": 0.00015043264788786305, + "loss": 3.3846, + "step": 2254 + }, + { + "epoch": 3.3, + "grad_norm": 3.7059578895568848, + "learning_rate": 0.00015039214060585864, + "loss": 3.39, + "step": 2255 + }, + { + "epoch": 3.3, + "grad_norm": 5.986879825592041, + "learning_rate": 0.00015035162223792106, + "loss": 3.3791, + "step": 2256 + }, + { + "epoch": 3.3, + "grad_norm": 4.337857723236084, + "learning_rate": 0.00015031109279296402, + "loss": 3.3309, + "step": 2257 + }, + { + "epoch": 3.31, + "grad_norm": 4.526264190673828, + "learning_rate": 0.00015027055227990373, + "loss": 3.3768, + "step": 2258 + }, + { + "epoch": 3.31, + "grad_norm": 7.14890718460083, + "learning_rate": 0.00015023000070765884, + "loss": 3.3677, + "step": 2259 + }, + { + "epoch": 3.31, + "grad_norm": 3.9721760749816895, + "learning_rate": 0.0001501894380851504, + "loss": 3.3136, + "step": 2260 + }, + { + "epoch": 3.31, + "grad_norm": 5.246433734893799, + "learning_rate": 0.00015014886442130197, + "loss": 3.3263, + "step": 2261 + }, + { + "epoch": 3.31, + "grad_norm": 7.421927452087402, + "learning_rate": 0.00015010827972503944, + "loss": 3.3386, + "step": 2262 + }, + { + "epoch": 3.31, + "grad_norm": 3.354271173477173, + "learning_rate": 0.00015006768400529117, + "loss": 3.3509, + "step": 2263 + }, + { + "epoch": 3.31, + "grad_norm": 5.732051849365234, + "learning_rate": 0.00015002707727098798, + "loss": 3.3885, + "step": 2264 + }, + { + "epoch": 3.32, + "grad_norm": 6.248003005981445, + "learning_rate": 0.00014998645953106303, + "loss": 3.3291, + "step": 2265 + }, + { + "epoch": 3.32, + "grad_norm": 2.3988747596740723, + "learning_rate": 0.000149945830794452, + "loss": 3.2951, + "step": 2266 + }, + { + "epoch": 3.32, + "grad_norm": 4.345824241638184, + "learning_rate": 0.00014990519107009292, + "loss": 3.3256, + "step": 2267 + }, + { + "epoch": 3.32, + "grad_norm": 4.299257278442383, + "learning_rate": 0.0001498645403669262, + "loss": 3.3154, + "step": 2268 + }, + { + "epoch": 3.32, + "grad_norm": 4.104548931121826, + "learning_rate": 0.00014982387869389485, + "loss": 3.3248, + "step": 2269 + }, + { + "epoch": 3.32, + "grad_norm": 3.4004967212677, + "learning_rate": 0.00014978320605994405, + "loss": 3.3147, + "step": 2270 + }, + { + "epoch": 3.33, + "grad_norm": 3.225266456604004, + "learning_rate": 0.00014974252247402156, + "loss": 3.2917, + "step": 2271 + }, + { + "epoch": 3.33, + "grad_norm": 3.6412038803100586, + "learning_rate": 0.00014970182794507748, + "loss": 3.3557, + "step": 2272 + }, + { + "epoch": 3.33, + "grad_norm": 3.6593449115753174, + "learning_rate": 0.00014966112248206436, + "loss": 3.3048, + "step": 2273 + }, + { + "epoch": 3.33, + "grad_norm": 3.3542096614837646, + "learning_rate": 0.0001496204060939371, + "loss": 3.2807, + "step": 2274 + }, + { + "epoch": 3.33, + "grad_norm": 3.2932636737823486, + "learning_rate": 0.000149579678789653, + "loss": 3.3457, + "step": 2275 + }, + { + "epoch": 3.33, + "grad_norm": 2.769178867340088, + "learning_rate": 0.00014953894057817188, + "loss": 3.3023, + "step": 2276 + }, + { + "epoch": 3.33, + "grad_norm": 3.352829694747925, + "learning_rate": 0.0001494981914684558, + "loss": 3.2836, + "step": 2277 + }, + { + "epoch": 3.34, + "grad_norm": 1.4692974090576172, + "learning_rate": 0.00014945743146946937, + "loss": 3.354, + "step": 2278 + }, + { + "epoch": 3.34, + "grad_norm": 5.136416912078857, + "learning_rate": 0.0001494166605901794, + "loss": 3.2915, + "step": 2279 + }, + { + "epoch": 3.34, + "grad_norm": 4.615678310394287, + "learning_rate": 0.00014937587883955534, + "loss": 3.2766, + "step": 2280 + }, + { + "epoch": 3.34, + "grad_norm": 1.0332159996032715, + "learning_rate": 0.00014933508622656878, + "loss": 3.3291, + "step": 2281 + }, + { + "epoch": 3.34, + "grad_norm": 1.1297160387039185, + "learning_rate": 0.00014929428276019388, + "loss": 3.3465, + "step": 2282 + }, + { + "epoch": 3.34, + "grad_norm": 3.5306038856506348, + "learning_rate": 0.00014925346844940718, + "loss": 3.2554, + "step": 2283 + }, + { + "epoch": 3.34, + "grad_norm": 9.56828498840332, + "learning_rate": 0.00014921264330318746, + "loss": 3.3203, + "step": 2284 + }, + { + "epoch": 3.35, + "grad_norm": 2.0608508586883545, + "learning_rate": 0.00014917180733051602, + "loss": 3.2899, + "step": 2285 + }, + { + "epoch": 3.35, + "grad_norm": 1.6151994466781616, + "learning_rate": 0.00014913096054037654, + "loss": 3.3209, + "step": 2286 + }, + { + "epoch": 3.35, + "grad_norm": 2.8889124393463135, + "learning_rate": 0.00014909010294175497, + "loss": 3.3144, + "step": 2287 + }, + { + "epoch": 3.35, + "grad_norm": 0.650810182094574, + "learning_rate": 0.00014904923454363975, + "loss": 3.2888, + "step": 2288 + }, + { + "epoch": 3.35, + "grad_norm": 2.751948833465576, + "learning_rate": 0.00014900835535502165, + "loss": 3.3317, + "step": 2289 + }, + { + "epoch": 3.35, + "grad_norm": 1.9524742364883423, + "learning_rate": 0.0001489674653848938, + "loss": 3.2798, + "step": 2290 + }, + { + "epoch": 3.35, + "grad_norm": 0.6142035126686096, + "learning_rate": 0.00014892656464225175, + "loss": 3.2485, + "step": 2291 + }, + { + "epoch": 3.36, + "grad_norm": 1.9580166339874268, + "learning_rate": 0.00014888565313609333, + "loss": 3.3353, + "step": 2292 + }, + { + "epoch": 3.36, + "grad_norm": 4.1013898849487305, + "learning_rate": 0.0001488447308754189, + "loss": 3.3178, + "step": 2293 + }, + { + "epoch": 3.36, + "grad_norm": 1.7551053762435913, + "learning_rate": 0.00014880379786923098, + "loss": 3.2874, + "step": 2294 + }, + { + "epoch": 3.36, + "grad_norm": 5.447327136993408, + "learning_rate": 0.00014876285412653456, + "loss": 3.2928, + "step": 2295 + }, + { + "epoch": 3.36, + "grad_norm": 5.519541263580322, + "learning_rate": 0.00014872189965633707, + "loss": 3.3418, + "step": 2296 + }, + { + "epoch": 3.36, + "grad_norm": 2.044977903366089, + "learning_rate": 0.0001486809344676482, + "loss": 3.3401, + "step": 2297 + }, + { + "epoch": 3.37, + "grad_norm": 5.334053993225098, + "learning_rate": 0.00014863995856947995, + "loss": 3.3156, + "step": 2298 + }, + { + "epoch": 3.37, + "grad_norm": 3.4951860904693604, + "learning_rate": 0.0001485989719708468, + "loss": 3.2812, + "step": 2299 + }, + { + "epoch": 3.37, + "grad_norm": 5.83620548248291, + "learning_rate": 0.00014855797468076552, + "loss": 3.3016, + "step": 2300 + }, + { + "epoch": 3.37, + "grad_norm": 6.08860969543457, + "learning_rate": 0.00014851696670825523, + "loss": 3.2941, + "step": 2301 + }, + { + "epoch": 3.37, + "grad_norm": 7.569385051727295, + "learning_rate": 0.00014847594806233737, + "loss": 3.3478, + "step": 2302 + }, + { + "epoch": 3.37, + "grad_norm": 5.372185230255127, + "learning_rate": 0.00014843491875203584, + "loss": 3.2692, + "step": 2303 + }, + { + "epoch": 3.37, + "grad_norm": 4.030975341796875, + "learning_rate": 0.0001483938787863768, + "loss": 3.2826, + "step": 2304 + }, + { + "epoch": 3.38, + "grad_norm": 6.85429048538208, + "learning_rate": 0.00014835282817438875, + "loss": 3.2932, + "step": 2305 + }, + { + "epoch": 3.38, + "grad_norm": 6.535613059997559, + "learning_rate": 0.0001483117669251025, + "loss": 3.3158, + "step": 2306 + }, + { + "epoch": 3.38, + "grad_norm": 3.962482452392578, + "learning_rate": 0.00014827069504755132, + "loss": 3.308, + "step": 2307 + }, + { + "epoch": 3.38, + "grad_norm": 6.860799312591553, + "learning_rate": 0.00014822961255077073, + "loss": 3.3145, + "step": 2308 + }, + { + "epoch": 3.38, + "grad_norm": 9.554657936096191, + "learning_rate": 0.0001481885194437986, + "loss": 3.3366, + "step": 2309 + }, + { + "epoch": 3.38, + "grad_norm": 4.180676460266113, + "learning_rate": 0.00014814741573567514, + "loss": 3.3096, + "step": 2310 + }, + { + "epoch": 3.38, + "grad_norm": 6.105428218841553, + "learning_rate": 0.0001481063014354429, + "loss": 3.2919, + "step": 2311 + }, + { + "epoch": 3.39, + "grad_norm": 7.584985733032227, + "learning_rate": 0.00014806517655214672, + "loss": 3.2756, + "step": 2312 + }, + { + "epoch": 3.39, + "grad_norm": 5.3676300048828125, + "learning_rate": 0.00014802404109483383, + "loss": 3.2969, + "step": 2313 + }, + { + "epoch": 3.39, + "grad_norm": 3.991481065750122, + "learning_rate": 0.00014798289507255373, + "loss": 3.2808, + "step": 2314 + }, + { + "epoch": 3.39, + "grad_norm": 6.212815761566162, + "learning_rate": 0.00014794173849435826, + "loss": 3.2809, + "step": 2315 + }, + { + "epoch": 3.39, + "grad_norm": 1.0110559463500977, + "learning_rate": 0.00014790057136930166, + "loss": 3.3016, + "step": 2316 + }, + { + "epoch": 3.39, + "grad_norm": 9.187312126159668, + "learning_rate": 0.00014785939370644036, + "loss": 3.275, + "step": 2317 + }, + { + "epoch": 3.39, + "grad_norm": 12.635149955749512, + "learning_rate": 0.00014781820551483315, + "loss": 3.3793, + "step": 2318 + }, + { + "epoch": 3.4, + "grad_norm": 9.69156551361084, + "learning_rate": 0.00014777700680354124, + "loss": 3.3063, + "step": 2319 + }, + { + "epoch": 3.4, + "grad_norm": 1.8201349973678589, + "learning_rate": 0.000147735797581628, + "loss": 3.3122, + "step": 2320 + }, + { + "epoch": 3.4, + "grad_norm": 11.608341217041016, + "learning_rate": 0.0001476945778581592, + "loss": 3.341, + "step": 2321 + }, + { + "epoch": 3.4, + "grad_norm": 13.871359825134277, + "learning_rate": 0.0001476533476422029, + "loss": 3.3459, + "step": 2322 + }, + { + "epoch": 3.4, + "grad_norm": 9.090968132019043, + "learning_rate": 0.00014761210694282947, + "loss": 3.369, + "step": 2323 + }, + { + "epoch": 3.4, + "grad_norm": 4.659329414367676, + "learning_rate": 0.0001475708557691116, + "loss": 3.417, + "step": 2324 + }, + { + "epoch": 3.41, + "grad_norm": 6.619305610656738, + "learning_rate": 0.00014752959413012428, + "loss": 3.3524, + "step": 2325 + }, + { + "epoch": 3.41, + "grad_norm": 9.944043159484863, + "learning_rate": 0.00014748832203494478, + "loss": 3.3326, + "step": 2326 + }, + { + "epoch": 3.41, + "grad_norm": 5.2585062980651855, + "learning_rate": 0.00014744703949265268, + "loss": 3.3292, + "step": 2327 + }, + { + "epoch": 3.41, + "grad_norm": 5.1725311279296875, + "learning_rate": 0.00014740574651232987, + "loss": 3.3172, + "step": 2328 + }, + { + "epoch": 3.41, + "grad_norm": 30.300024032592773, + "learning_rate": 0.00014736444310306052, + "loss": 3.3064, + "step": 2329 + }, + { + "epoch": 3.41, + "grad_norm": 9.257378578186035, + "learning_rate": 0.00014732312927393112, + "loss": 3.2872, + "step": 2330 + }, + { + "epoch": 3.41, + "grad_norm": 2.5596399307250977, + "learning_rate": 0.0001472818050340304, + "loss": 3.2914, + "step": 2331 + }, + { + "epoch": 3.42, + "grad_norm": 6.8813934326171875, + "learning_rate": 0.00014724047039244947, + "loss": 3.3698, + "step": 2332 + }, + { + "epoch": 3.42, + "grad_norm": 10.194779396057129, + "learning_rate": 0.00014719912535828166, + "loss": 3.2775, + "step": 2333 + }, + { + "epoch": 3.42, + "grad_norm": 5.2471489906311035, + "learning_rate": 0.00014715776994062254, + "loss": 3.3149, + "step": 2334 + }, + { + "epoch": 3.42, + "grad_norm": 3.753305196762085, + "learning_rate": 0.00014711640414857008, + "loss": 3.3024, + "step": 2335 + }, + { + "epoch": 3.42, + "grad_norm": 5.786775588989258, + "learning_rate": 0.00014707502799122447, + "loss": 3.2799, + "step": 2336 + }, + { + "epoch": 3.42, + "grad_norm": 1.7370816469192505, + "learning_rate": 0.00014703364147768822, + "loss": 3.2881, + "step": 2337 + }, + { + "epoch": 3.42, + "grad_norm": 7.590539932250977, + "learning_rate": 0.000146992244617066, + "loss": 3.3002, + "step": 2338 + }, + { + "epoch": 3.43, + "grad_norm": 9.67332935333252, + "learning_rate": 0.00014695083741846492, + "loss": 3.3037, + "step": 2339 + }, + { + "epoch": 3.43, + "grad_norm": 5.325073719024658, + "learning_rate": 0.00014690941989099426, + "loss": 3.3241, + "step": 2340 + }, + { + "epoch": 3.43, + "grad_norm": 7.251718044281006, + "learning_rate": 0.0001468679920437656, + "loss": 3.316, + "step": 2341 + }, + { + "epoch": 3.43, + "grad_norm": 8.890804290771484, + "learning_rate": 0.00014682655388589274, + "loss": 3.3177, + "step": 2342 + }, + { + "epoch": 3.43, + "grad_norm": 3.9834563732147217, + "learning_rate": 0.0001467851054264919, + "loss": 3.3002, + "step": 2343 + }, + { + "epoch": 3.43, + "grad_norm": 5.180391788482666, + "learning_rate": 0.0001467436466746814, + "loss": 3.3166, + "step": 2344 + }, + { + "epoch": 3.43, + "grad_norm": 9.531121253967285, + "learning_rate": 0.00014670217763958188, + "loss": 3.2482, + "step": 2345 + }, + { + "epoch": 3.44, + "grad_norm": 4.674380302429199, + "learning_rate": 0.00014666069833031624, + "loss": 3.2703, + "step": 2346 + }, + { + "epoch": 3.44, + "grad_norm": 3.9839441776275635, + "learning_rate": 0.00014661920875600973, + "loss": 3.2701, + "step": 2347 + }, + { + "epoch": 3.44, + "grad_norm": 4.408015251159668, + "learning_rate": 0.00014657770892578968, + "loss": 3.2791, + "step": 2348 + }, + { + "epoch": 3.44, + "grad_norm": 2.449519395828247, + "learning_rate": 0.0001465361988487858, + "loss": 3.2654, + "step": 2349 + }, + { + "epoch": 3.44, + "grad_norm": 1.4051802158355713, + "learning_rate": 0.00014649467853413008, + "loss": 3.3101, + "step": 2350 + }, + { + "epoch": 3.44, + "grad_norm": 6.514969825744629, + "learning_rate": 0.00014645314799095665, + "loss": 3.2282, + "step": 2351 + }, + { + "epoch": 3.45, + "grad_norm": 5.476304054260254, + "learning_rate": 0.000146411607228402, + "loss": 3.3043, + "step": 2352 + }, + { + "epoch": 3.45, + "grad_norm": 2.443798780441284, + "learning_rate": 0.00014637005625560477, + "loss": 3.3045, + "step": 2353 + }, + { + "epoch": 3.45, + "grad_norm": 6.676492214202881, + "learning_rate": 0.00014632849508170595, + "loss": 3.2794, + "step": 2354 + }, + { + "epoch": 3.45, + "grad_norm": 2.0837247371673584, + "learning_rate": 0.0001462869237158487, + "loss": 3.3134, + "step": 2355 + }, + { + "epoch": 3.45, + "grad_norm": 7.334787368774414, + "learning_rate": 0.00014624534216717841, + "loss": 3.2919, + "step": 2356 + }, + { + "epoch": 3.45, + "grad_norm": 8.694489479064941, + "learning_rate": 0.00014620375044484277, + "loss": 3.2639, + "step": 2357 + }, + { + "epoch": 3.45, + "grad_norm": 2.46984601020813, + "learning_rate": 0.00014616214855799167, + "loss": 3.2571, + "step": 2358 + }, + { + "epoch": 3.46, + "grad_norm": 7.231738567352295, + "learning_rate": 0.00014612053651577727, + "loss": 3.3468, + "step": 2359 + }, + { + "epoch": 3.46, + "grad_norm": 7.998284816741943, + "learning_rate": 0.0001460789143273539, + "loss": 3.3764, + "step": 2360 + }, + { + "epoch": 3.46, + "grad_norm": 4.452177047729492, + "learning_rate": 0.00014603728200187824, + "loss": 3.3293, + "step": 2361 + }, + { + "epoch": 3.46, + "grad_norm": 9.913199424743652, + "learning_rate": 0.00014599563954850901, + "loss": 3.2631, + "step": 2362 + }, + { + "epoch": 3.46, + "grad_norm": 15.90157413482666, + "learning_rate": 0.00014595398697640733, + "loss": 3.3812, + "step": 2363 + }, + { + "epoch": 3.46, + "grad_norm": 12.660530090332031, + "learning_rate": 0.0001459123242947365, + "loss": 3.371, + "step": 2364 + }, + { + "epoch": 3.46, + "grad_norm": 5.289348125457764, + "learning_rate": 0.00014587065151266203, + "loss": 3.3892, + "step": 2365 + }, + { + "epoch": 3.47, + "grad_norm": 9.648274421691895, + "learning_rate": 0.00014582896863935162, + "loss": 3.2931, + "step": 2366 + }, + { + "epoch": 3.47, + "eval_loss": 3.133899211883545, + "eval_runtime": 1162.2679, + "eval_samples_per_second": 14.749, + "eval_steps_per_second": 0.922, + "step": 2366 + }, + { + "epoch": 3.47, + "grad_norm": 9.562708854675293, + "learning_rate": 0.00014578727568397522, + "loss": 3.3485, + "step": 2367 + }, + { + "epoch": 3.47, + "grad_norm": 19.97218894958496, + "learning_rate": 0.00014574557265570505, + "loss": 3.3403, + "step": 2368 + }, + { + "epoch": 3.47, + "grad_norm": 6.222564220428467, + "learning_rate": 0.0001457038595637155, + "loss": 3.3743, + "step": 2369 + }, + { + "epoch": 3.47, + "grad_norm": 7.945571422576904, + "learning_rate": 0.00014566213641718307, + "loss": 3.2679, + "step": 2370 + }, + { + "epoch": 3.47, + "grad_norm": 8.883635520935059, + "learning_rate": 0.0001456204032252867, + "loss": 3.2295, + "step": 2371 + }, + { + "epoch": 3.48, + "grad_norm": 5.227753639221191, + "learning_rate": 0.00014557865999720735, + "loss": 3.2934, + "step": 2372 + }, + { + "epoch": 3.48, + "grad_norm": 8.865499496459961, + "learning_rate": 0.0001455369067421282, + "loss": 3.3526, + "step": 2373 + }, + { + "epoch": 3.48, + "grad_norm": 10.483139991760254, + "learning_rate": 0.00014549514346923477, + "loss": 3.3058, + "step": 2374 + }, + { + "epoch": 3.48, + "grad_norm": 12.922052383422852, + "learning_rate": 0.00014545337018771472, + "loss": 3.2366, + "step": 2375 + }, + { + "epoch": 3.48, + "grad_norm": 2.131599187850952, + "learning_rate": 0.00014541158690675778, + "loss": 3.2613, + "step": 2376 + }, + { + "epoch": 3.48, + "grad_norm": 10.010235786437988, + "learning_rate": 0.0001453697936355561, + "loss": 3.2698, + "step": 2377 + }, + { + "epoch": 3.48, + "grad_norm": 14.034173011779785, + "learning_rate": 0.00014532799038330385, + "loss": 3.3691, + "step": 2378 + }, + { + "epoch": 3.49, + "grad_norm": 9.930380821228027, + "learning_rate": 0.0001452861771591975, + "loss": 3.304, + "step": 2379 + }, + { + "epoch": 3.49, + "grad_norm": 7.6726555824279785, + "learning_rate": 0.00014524435397243567, + "loss": 3.2806, + "step": 2380 + }, + { + "epoch": 3.49, + "grad_norm": 8.018184661865234, + "learning_rate": 0.00014520252083221916, + "loss": 3.2673, + "step": 2381 + }, + { + "epoch": 3.49, + "grad_norm": 8.354979515075684, + "learning_rate": 0.00014516067774775101, + "loss": 3.295, + "step": 2382 + }, + { + "epoch": 3.49, + "grad_norm": 3.118138313293457, + "learning_rate": 0.00014511882472823643, + "loss": 3.2593, + "step": 2383 + }, + { + "epoch": 3.49, + "grad_norm": 9.12159538269043, + "learning_rate": 0.00014507696178288278, + "loss": 3.2893, + "step": 2384 + }, + { + "epoch": 3.49, + "grad_norm": 13.92984390258789, + "learning_rate": 0.0001450350889208996, + "loss": 3.3191, + "step": 2385 + }, + { + "epoch": 3.5, + "grad_norm": 9.545883178710938, + "learning_rate": 0.0001449932061514987, + "loss": 3.2753, + "step": 2386 + }, + { + "epoch": 3.5, + "grad_norm": 1.6055418252944946, + "learning_rate": 0.00014495131348389394, + "loss": 3.2373, + "step": 2387 + }, + { + "epoch": 3.5, + "grad_norm": 8.169912338256836, + "learning_rate": 0.00014490941092730152, + "loss": 3.2748, + "step": 2388 + }, + { + "epoch": 3.5, + "grad_norm": 8.912496566772461, + "learning_rate": 0.00014486749849093963, + "loss": 3.302, + "step": 2389 + }, + { + "epoch": 3.5, + "grad_norm": 6.068753242492676, + "learning_rate": 0.00014482557618402875, + "loss": 3.2805, + "step": 2390 + }, + { + "epoch": 3.5, + "grad_norm": 4.4562578201293945, + "learning_rate": 0.00014478364401579154, + "loss": 3.2488, + "step": 2391 + }, + { + "epoch": 3.5, + "grad_norm": 4.690038204193115, + "learning_rate": 0.00014474170199545275, + "loss": 3.2896, + "step": 2392 + }, + { + "epoch": 3.51, + "grad_norm": 3.260173797607422, + "learning_rate": 0.0001446997501322394, + "loss": 3.2497, + "step": 2393 + }, + { + "epoch": 3.51, + "grad_norm": 3.311274528503418, + "learning_rate": 0.0001446577884353806, + "loss": 3.2359, + "step": 2394 + }, + { + "epoch": 3.51, + "grad_norm": 2.5797505378723145, + "learning_rate": 0.00014461581691410758, + "loss": 3.3372, + "step": 2395 + }, + { + "epoch": 3.51, + "grad_norm": 1.9033737182617188, + "learning_rate": 0.00014457383557765386, + "loss": 3.2308, + "step": 2396 + }, + { + "epoch": 3.51, + "grad_norm": 3.9633514881134033, + "learning_rate": 0.000144531844435255, + "loss": 3.2787, + "step": 2397 + }, + { + "epoch": 3.51, + "grad_norm": 3.756439685821533, + "learning_rate": 0.00014448984349614886, + "loss": 3.3238, + "step": 2398 + }, + { + "epoch": 3.52, + "grad_norm": 3.5643348693847656, + "learning_rate": 0.00014444783276957524, + "loss": 3.2315, + "step": 2399 + }, + { + "epoch": 3.52, + "grad_norm": 4.421493053436279, + "learning_rate": 0.00014440581226477634, + "loss": 3.277, + "step": 2400 + }, + { + "epoch": 3.52, + "grad_norm": 4.980103969573975, + "learning_rate": 0.00014436378199099634, + "loss": 3.248, + "step": 2401 + }, + { + "epoch": 3.52, + "grad_norm": 7.962924003601074, + "learning_rate": 0.0001443217419574816, + "loss": 3.2628, + "step": 2402 + }, + { + "epoch": 3.52, + "grad_norm": 3.279404401779175, + "learning_rate": 0.00014427969217348065, + "loss": 3.2662, + "step": 2403 + }, + { + "epoch": 3.52, + "grad_norm": 6.002499580383301, + "learning_rate": 0.00014423763264824417, + "loss": 3.2498, + "step": 2404 + }, + { + "epoch": 3.52, + "grad_norm": 8.197328567504883, + "learning_rate": 0.00014419556339102503, + "loss": 3.3432, + "step": 2405 + }, + { + "epoch": 3.53, + "grad_norm": 3.395263910293579, + "learning_rate": 0.00014415348441107806, + "loss": 3.2308, + "step": 2406 + }, + { + "epoch": 3.53, + "grad_norm": 6.569970607757568, + "learning_rate": 0.00014411139571766042, + "loss": 3.2445, + "step": 2407 + }, + { + "epoch": 3.53, + "grad_norm": 9.0120849609375, + "learning_rate": 0.0001440692973200314, + "loss": 3.2689, + "step": 2408 + }, + { + "epoch": 3.53, + "grad_norm": 5.56130313873291, + "learning_rate": 0.00014402718922745228, + "loss": 3.2685, + "step": 2409 + }, + { + "epoch": 3.53, + "grad_norm": 4.367111682891846, + "learning_rate": 0.00014398507144918656, + "loss": 3.3052, + "step": 2410 + }, + { + "epoch": 3.53, + "grad_norm": 6.1577558517456055, + "learning_rate": 0.00014394294399449996, + "loss": 3.2687, + "step": 2411 + }, + { + "epoch": 3.53, + "grad_norm": 14.60132122039795, + "learning_rate": 0.00014390080687266013, + "loss": 3.2542, + "step": 2412 + }, + { + "epoch": 3.54, + "grad_norm": 19.244531631469727, + "learning_rate": 0.000143858660092937, + "loss": 3.2623, + "step": 2413 + }, + { + "epoch": 3.54, + "grad_norm": 5.000961780548096, + "learning_rate": 0.0001438165036646026, + "loss": 3.2887, + "step": 2414 + }, + { + "epoch": 3.54, + "grad_norm": 1.7813776731491089, + "learning_rate": 0.000143774337596931, + "loss": 3.2522, + "step": 2415 + }, + { + "epoch": 3.54, + "grad_norm": 11.567626953125, + "learning_rate": 0.00014373216189919851, + "loss": 3.3463, + "step": 2416 + }, + { + "epoch": 3.54, + "grad_norm": 17.07219696044922, + "learning_rate": 0.0001436899765806835, + "loss": 3.3653, + "step": 2417 + }, + { + "epoch": 3.54, + "grad_norm": 10.967621803283691, + "learning_rate": 0.00014364778165066642, + "loss": 3.339, + "step": 2418 + }, + { + "epoch": 3.54, + "grad_norm": 5.372880935668945, + "learning_rate": 0.00014360557711842991, + "loss": 3.3025, + "step": 2419 + }, + { + "epoch": 3.55, + "grad_norm": 8.31612777709961, + "learning_rate": 0.00014356336299325864, + "loss": 3.2886, + "step": 2420 + }, + { + "epoch": 3.55, + "grad_norm": 6.364083290100098, + "learning_rate": 0.00014352113928443947, + "loss": 3.3376, + "step": 2421 + }, + { + "epoch": 3.55, + "grad_norm": 39.3469123840332, + "learning_rate": 0.00014347890600126134, + "loss": 3.2798, + "step": 2422 + }, + { + "epoch": 3.55, + "grad_norm": 15.578664779663086, + "learning_rate": 0.00014343666315301526, + "loss": 3.3638, + "step": 2423 + }, + { + "epoch": 3.55, + "grad_norm": 15.445240020751953, + "learning_rate": 0.0001433944107489944, + "loss": 3.3672, + "step": 2424 + }, + { + "epoch": 3.55, + "grad_norm": 13.184659957885742, + "learning_rate": 0.00014335214879849394, + "loss": 3.3172, + "step": 2425 + }, + { + "epoch": 3.56, + "grad_norm": 4.225596904754639, + "learning_rate": 0.0001433098773108113, + "loss": 3.2598, + "step": 2426 + }, + { + "epoch": 3.56, + "grad_norm": 9.756654739379883, + "learning_rate": 0.00014326759629524594, + "loss": 3.3621, + "step": 2427 + }, + { + "epoch": 3.56, + "grad_norm": 14.042159080505371, + "learning_rate": 0.00014322530576109933, + "loss": 3.3233, + "step": 2428 + }, + { + "epoch": 3.56, + "grad_norm": 14.112271308898926, + "learning_rate": 0.00014318300571767513, + "loss": 3.3309, + "step": 2429 + }, + { + "epoch": 3.56, + "grad_norm": 9.754487037658691, + "learning_rate": 0.00014314069617427907, + "loss": 3.3182, + "step": 2430 + }, + { + "epoch": 3.56, + "grad_norm": 2.0633256435394287, + "learning_rate": 0.00014309837714021902, + "loss": 3.2493, + "step": 2431 + }, + { + "epoch": 3.56, + "grad_norm": 8.613134384155273, + "learning_rate": 0.00014305604862480478, + "loss": 3.3103, + "step": 2432 + }, + { + "epoch": 3.57, + "grad_norm": 9.88028335571289, + "learning_rate": 0.0001430137106373484, + "loss": 3.3556, + "step": 2433 + }, + { + "epoch": 3.57, + "grad_norm": 8.193672180175781, + "learning_rate": 0.00014297136318716395, + "loss": 3.3321, + "step": 2434 + }, + { + "epoch": 3.57, + "grad_norm": 4.98108434677124, + "learning_rate": 0.00014292900628356758, + "loss": 3.265, + "step": 2435 + }, + { + "epoch": 3.57, + "grad_norm": 5.972259521484375, + "learning_rate": 0.0001428866399358775, + "loss": 3.2759, + "step": 2436 + }, + { + "epoch": 3.57, + "grad_norm": 4.662635803222656, + "learning_rate": 0.00014284426415341408, + "loss": 3.3012, + "step": 2437 + }, + { + "epoch": 3.57, + "grad_norm": 4.476476192474365, + "learning_rate": 0.00014280187894549968, + "loss": 3.2919, + "step": 2438 + }, + { + "epoch": 3.57, + "grad_norm": 3.099634885787964, + "learning_rate": 0.00014275948432145876, + "loss": 3.3264, + "step": 2439 + }, + { + "epoch": 3.58, + "grad_norm": 6.200466632843018, + "learning_rate": 0.00014271708029061785, + "loss": 3.2982, + "step": 2440 + }, + { + "epoch": 3.58, + "grad_norm": 3.2733237743377686, + "learning_rate": 0.00014267466686230553, + "loss": 3.2932, + "step": 2441 + }, + { + "epoch": 3.58, + "grad_norm": 3.1728217601776123, + "learning_rate": 0.0001426322440458525, + "loss": 3.2532, + "step": 2442 + }, + { + "epoch": 3.58, + "grad_norm": 3.4622151851654053, + "learning_rate": 0.0001425898118505915, + "loss": 3.2794, + "step": 2443 + }, + { + "epoch": 3.58, + "grad_norm": 7.931596755981445, + "learning_rate": 0.0001425473702858573, + "loss": 3.2692, + "step": 2444 + }, + { + "epoch": 3.58, + "grad_norm": 3.586261749267578, + "learning_rate": 0.00014250491936098678, + "loss": 3.2432, + "step": 2445 + }, + { + "epoch": 3.58, + "grad_norm": 3.096931219100952, + "learning_rate": 0.00014246245908531882, + "loss": 3.2589, + "step": 2446 + }, + { + "epoch": 3.59, + "grad_norm": 3.957979202270508, + "learning_rate": 0.0001424199894681945, + "loss": 3.278, + "step": 2447 + }, + { + "epoch": 3.59, + "grad_norm": 0.8693408370018005, + "learning_rate": 0.00014237751051895676, + "loss": 3.2366, + "step": 2448 + }, + { + "epoch": 3.59, + "grad_norm": 2.3588359355926514, + "learning_rate": 0.0001423350222469507, + "loss": 3.2469, + "step": 2449 + }, + { + "epoch": 3.59, + "grad_norm": 1.2426791191101074, + "learning_rate": 0.00014229252466152345, + "loss": 3.2929, + "step": 2450 + }, + { + "epoch": 3.59, + "grad_norm": 2.7103374004364014, + "learning_rate": 0.00014225001777202424, + "loss": 3.248, + "step": 2451 + }, + { + "epoch": 3.59, + "grad_norm": 2.5343198776245117, + "learning_rate": 0.00014220750158780424, + "loss": 3.257, + "step": 2452 + }, + { + "epoch": 3.6, + "grad_norm": 1.3891563415527344, + "learning_rate": 0.00014216497611821676, + "loss": 3.2348, + "step": 2453 + }, + { + "epoch": 3.6, + "grad_norm": 3.7499053478240967, + "learning_rate": 0.0001421224413726171, + "loss": 3.2513, + "step": 2454 + }, + { + "epoch": 3.6, + "grad_norm": 1.5479190349578857, + "learning_rate": 0.00014207989736036267, + "loss": 3.2785, + "step": 2455 + }, + { + "epoch": 3.6, + "grad_norm": 2.783564567565918, + "learning_rate": 0.0001420373440908128, + "loss": 3.292, + "step": 2456 + }, + { + "epoch": 3.6, + "grad_norm": 3.4745991230010986, + "learning_rate": 0.00014199478157332895, + "loss": 3.2399, + "step": 2457 + }, + { + "epoch": 3.6, + "grad_norm": 2.5730254650115967, + "learning_rate": 0.0001419522098172746, + "loss": 3.257, + "step": 2458 + }, + { + "epoch": 3.6, + "grad_norm": 3.455461025238037, + "learning_rate": 0.00014190962883201523, + "loss": 3.225, + "step": 2459 + }, + { + "epoch": 3.61, + "grad_norm": 4.234341621398926, + "learning_rate": 0.0001418670386269184, + "loss": 3.2997, + "step": 2460 + }, + { + "epoch": 3.61, + "grad_norm": 9.0831937789917, + "learning_rate": 0.00014182443921135362, + "loss": 3.221, + "step": 2461 + }, + { + "epoch": 3.61, + "grad_norm": 3.933647871017456, + "learning_rate": 0.00014178183059469253, + "loss": 3.2355, + "step": 2462 + }, + { + "epoch": 3.61, + "grad_norm": 1.695810079574585, + "learning_rate": 0.00014173921278630872, + "loss": 3.2724, + "step": 2463 + }, + { + "epoch": 3.61, + "grad_norm": 2.3507513999938965, + "learning_rate": 0.0001416965857955778, + "loss": 3.273, + "step": 2464 + }, + { + "epoch": 3.61, + "grad_norm": 2.5299062728881836, + "learning_rate": 0.00014165394963187747, + "loss": 3.2366, + "step": 2465 + }, + { + "epoch": 3.61, + "grad_norm": 0.9136064052581787, + "learning_rate": 0.00014161130430458737, + "loss": 3.2318, + "step": 2466 + }, + { + "epoch": 3.62, + "grad_norm": 2.159097671508789, + "learning_rate": 0.00014156864982308917, + "loss": 3.2139, + "step": 2467 + }, + { + "epoch": 3.62, + "grad_norm": 2.3743743896484375, + "learning_rate": 0.0001415259861967666, + "loss": 3.2442, + "step": 2468 + }, + { + "epoch": 3.62, + "grad_norm": 1.7210527658462524, + "learning_rate": 0.00014148331343500538, + "loss": 3.2565, + "step": 2469 + }, + { + "epoch": 3.62, + "grad_norm": 4.262208938598633, + "learning_rate": 0.0001414406315471932, + "loss": 3.2542, + "step": 2470 + }, + { + "epoch": 3.62, + "grad_norm": 3.7595646381378174, + "learning_rate": 0.00014139794054271977, + "loss": 3.2048, + "step": 2471 + }, + { + "epoch": 3.62, + "grad_norm": 3.6839802265167236, + "learning_rate": 0.00014135524043097693, + "loss": 3.2873, + "step": 2472 + }, + { + "epoch": 3.63, + "grad_norm": 3.4619247913360596, + "learning_rate": 0.0001413125312213583, + "loss": 3.2562, + "step": 2473 + }, + { + "epoch": 3.63, + "grad_norm": 3.164567708969116, + "learning_rate": 0.00014126981292325965, + "loss": 3.2521, + "step": 2474 + }, + { + "epoch": 3.63, + "grad_norm": 3.066319227218628, + "learning_rate": 0.00014122708554607875, + "loss": 3.2671, + "step": 2475 + }, + { + "epoch": 3.63, + "grad_norm": 2.757352590560913, + "learning_rate": 0.00014118434909921534, + "loss": 3.2061, + "step": 2476 + }, + { + "epoch": 3.63, + "grad_norm": 1.9947091341018677, + "learning_rate": 0.0001411416035920711, + "loss": 3.2391, + "step": 2477 + }, + { + "epoch": 3.63, + "grad_norm": 4.6580986976623535, + "learning_rate": 0.0001410988490340498, + "loss": 3.252, + "step": 2478 + }, + { + "epoch": 3.63, + "grad_norm": 4.087027072906494, + "learning_rate": 0.00014105608543455716, + "loss": 3.2371, + "step": 2479 + }, + { + "epoch": 3.64, + "grad_norm": 3.322829008102417, + "learning_rate": 0.0001410133128030009, + "loss": 3.2261, + "step": 2480 + }, + { + "epoch": 3.64, + "grad_norm": 3.1107428073883057, + "learning_rate": 0.00014097053114879061, + "loss": 3.2418, + "step": 2481 + }, + { + "epoch": 3.64, + "grad_norm": 2.6654446125030518, + "learning_rate": 0.00014092774048133809, + "loss": 3.232, + "step": 2482 + }, + { + "epoch": 3.64, + "grad_norm": 1.8028963804244995, + "learning_rate": 0.00014088494081005695, + "loss": 3.2122, + "step": 2483 + }, + { + "epoch": 3.64, + "grad_norm": 9.199921607971191, + "learning_rate": 0.00014084213214436277, + "loss": 3.2409, + "step": 2484 + }, + { + "epoch": 3.64, + "grad_norm": 7.354410648345947, + "learning_rate": 0.00014079931449367328, + "loss": 3.2151, + "step": 2485 + }, + { + "epoch": 3.64, + "grad_norm": 2.6317131519317627, + "learning_rate": 0.00014075648786740806, + "loss": 3.2449, + "step": 2486 + }, + { + "epoch": 3.65, + "grad_norm": 6.4712042808532715, + "learning_rate": 0.00014071365227498857, + "loss": 3.2384, + "step": 2487 + }, + { + "epoch": 3.65, + "grad_norm": 7.566592216491699, + "learning_rate": 0.00014067080772583846, + "loss": 3.2208, + "step": 2488 + }, + { + "epoch": 3.65, + "grad_norm": 3.7115042209625244, + "learning_rate": 0.00014062795422938325, + "loss": 3.2403, + "step": 2489 + }, + { + "epoch": 3.65, + "grad_norm": 7.227043151855469, + "learning_rate": 0.00014058509179505036, + "loss": 3.2533, + "step": 2490 + }, + { + "epoch": 3.65, + "grad_norm": 10.836502075195312, + "learning_rate": 0.00014054222043226926, + "loss": 3.3034, + "step": 2491 + }, + { + "epoch": 3.65, + "grad_norm": 8.85554313659668, + "learning_rate": 0.00014049934015047136, + "loss": 3.2838, + "step": 2492 + }, + { + "epoch": 3.65, + "grad_norm": 7.335647106170654, + "learning_rate": 0.00014045645095909004, + "loss": 3.2506, + "step": 2493 + }, + { + "epoch": 3.66, + "grad_norm": 6.849621295928955, + "learning_rate": 0.00014041355286756066, + "loss": 3.266, + "step": 2494 + }, + { + "epoch": 3.66, + "grad_norm": 5.7214837074279785, + "learning_rate": 0.00014037064588532047, + "loss": 3.229, + "step": 2495 + }, + { + "epoch": 3.66, + "grad_norm": 1.1197165250778198, + "learning_rate": 0.00014032773002180875, + "loss": 3.189, + "step": 2496 + }, + { + "epoch": 3.66, + "grad_norm": 2.7683522701263428, + "learning_rate": 0.00014028480528646667, + "loss": 3.2357, + "step": 2497 + }, + { + "epoch": 3.66, + "grad_norm": 10.467253684997559, + "learning_rate": 0.0001402418716887374, + "loss": 3.2702, + "step": 2498 + }, + { + "epoch": 3.66, + "grad_norm": 5.283757209777832, + "learning_rate": 0.0001401989292380661, + "loss": 3.2404, + "step": 2499 + }, + { + "epoch": 3.67, + "grad_norm": 3.8525490760803223, + "learning_rate": 0.00014015597794389973, + "loss": 3.2352, + "step": 2500 + }, + { + "epoch": 3.67, + "grad_norm": 4.830957412719727, + "learning_rate": 0.0001401130178156873, + "loss": 3.2432, + "step": 2501 + }, + { + "epoch": 3.67, + "grad_norm": 28.55122184753418, + "learning_rate": 0.0001400700488628798, + "loss": 3.2176, + "step": 2502 + }, + { + "epoch": 3.67, + "grad_norm": 6.477649211883545, + "learning_rate": 0.0001400270710949301, + "loss": 3.1952, + "step": 2503 + }, + { + "epoch": 3.67, + "grad_norm": 4.75806188583374, + "learning_rate": 0.00013998408452129302, + "loss": 3.2384, + "step": 2504 + }, + { + "epoch": 3.67, + "grad_norm": 5.451016426086426, + "learning_rate": 0.00013994108915142526, + "loss": 3.2404, + "step": 2505 + }, + { + "epoch": 3.67, + "grad_norm": 2.510890245437622, + "learning_rate": 0.0001398980849947856, + "loss": 3.221, + "step": 2506 + }, + { + "epoch": 3.68, + "grad_norm": 6.08433198928833, + "learning_rate": 0.0001398550720608346, + "loss": 3.2386, + "step": 2507 + }, + { + "epoch": 3.68, + "grad_norm": 7.88702392578125, + "learning_rate": 0.00013981205035903485, + "loss": 3.2582, + "step": 2508 + }, + { + "epoch": 3.68, + "grad_norm": 3.3581769466400146, + "learning_rate": 0.00013976901989885086, + "loss": 3.2437, + "step": 2509 + }, + { + "epoch": 3.68, + "grad_norm": 5.539381980895996, + "learning_rate": 0.000139725980689749, + "loss": 3.2349, + "step": 2510 + }, + { + "epoch": 3.68, + "grad_norm": 6.059652328491211, + "learning_rate": 0.0001396829327411976, + "loss": 3.3016, + "step": 2511 + }, + { + "epoch": 3.68, + "grad_norm": 1.3593266010284424, + "learning_rate": 0.00013963987606266693, + "loss": 3.2587, + "step": 2512 + }, + { + "epoch": 3.68, + "grad_norm": 9.557854652404785, + "learning_rate": 0.0001395968106636292, + "loss": 3.2695, + "step": 2513 + }, + { + "epoch": 3.69, + "grad_norm": 26.98328971862793, + "learning_rate": 0.0001395537365535585, + "loss": 3.3668, + "step": 2514 + }, + { + "epoch": 3.69, + "grad_norm": 14.079690933227539, + "learning_rate": 0.0001395106537419308, + "loss": 3.2736, + "step": 2515 + }, + { + "epoch": 3.69, + "grad_norm": 5.151083946228027, + "learning_rate": 0.00013946756223822413, + "loss": 3.2461, + "step": 2516 + }, + { + "epoch": 3.69, + "grad_norm": 7.186163902282715, + "learning_rate": 0.00013942446205191824, + "loss": 3.2817, + "step": 2517 + }, + { + "epoch": 3.69, + "grad_norm": 9.277379035949707, + "learning_rate": 0.0001393813531924949, + "loss": 3.2008, + "step": 2518 + }, + { + "epoch": 3.69, + "grad_norm": 19.572731018066406, + "learning_rate": 0.00013933823566943777, + "loss": 3.2131, + "step": 2519 + }, + { + "epoch": 3.69, + "grad_norm": 4.6605329513549805, + "learning_rate": 0.00013929510949223245, + "loss": 3.2512, + "step": 2520 + }, + { + "epoch": 3.7, + "grad_norm": 151.2240753173828, + "learning_rate": 0.00013925197467036635, + "loss": 3.2621, + "step": 2521 + }, + { + "epoch": 3.7, + "grad_norm": 13.225665092468262, + "learning_rate": 0.0001392088312133289, + "loss": 3.2814, + "step": 2522 + }, + { + "epoch": 3.7, + "grad_norm": 14.404492378234863, + "learning_rate": 0.00013916567913061133, + "loss": 3.2903, + "step": 2523 + }, + { + "epoch": 3.7, + "grad_norm": 10.506006240844727, + "learning_rate": 0.00013912251843170687, + "loss": 3.226, + "step": 2524 + }, + { + "epoch": 3.7, + "grad_norm": 3.039416790008545, + "learning_rate": 0.0001390793491261105, + "loss": 3.26, + "step": 2525 + }, + { + "epoch": 3.7, + "grad_norm": 6.364565372467041, + "learning_rate": 0.00013903617122331923, + "loss": 3.1935, + "step": 2526 + }, + { + "epoch": 3.71, + "grad_norm": 3.801405668258667, + "learning_rate": 0.0001389929847328319, + "loss": 3.2204, + "step": 2527 + }, + { + "epoch": 3.71, + "grad_norm": 13.04414176940918, + "learning_rate": 0.00013894978966414926, + "loss": 3.2408, + "step": 2528 + }, + { + "epoch": 3.71, + "grad_norm": 14.216852188110352, + "learning_rate": 0.0001389065860267739, + "loss": 3.2739, + "step": 2529 + }, + { + "epoch": 3.71, + "grad_norm": 10.664838790893555, + "learning_rate": 0.0001388633738302104, + "loss": 3.2684, + "step": 2530 + }, + { + "epoch": 3.71, + "grad_norm": 6.671474456787109, + "learning_rate": 0.00013882015308396508, + "loss": 3.3327, + "step": 2531 + }, + { + "epoch": 3.71, + "grad_norm": 15.43130874633789, + "learning_rate": 0.00013877692379754626, + "loss": 3.2355, + "step": 2532 + }, + { + "epoch": 3.71, + "grad_norm": 44.657169342041016, + "learning_rate": 0.00013873368598046404, + "loss": 3.2309, + "step": 2533 + }, + { + "epoch": 3.72, + "grad_norm": 18.107580184936523, + "learning_rate": 0.00013869043964223051, + "loss": 3.2906, + "step": 2534 + }, + { + "epoch": 3.72, + "grad_norm": 11.789257049560547, + "learning_rate": 0.00013864718479235954, + "loss": 3.3045, + "step": 2535 + }, + { + "epoch": 3.72, + "eval_loss": 3.0709898471832275, + "eval_runtime": 1163.6634, + "eval_samples_per_second": 14.731, + "eval_steps_per_second": 0.921, + "step": 2535 + }, + { + "epoch": 3.72, + "grad_norm": 7.7072319984436035, + "learning_rate": 0.00013860392144036693, + "loss": 3.2131, + "step": 2536 + }, + { + "epoch": 3.72, + "grad_norm": 4.083733081817627, + "learning_rate": 0.0001385606495957703, + "loss": 3.2641, + "step": 2537 + }, + { + "epoch": 3.72, + "grad_norm": 11.785395622253418, + "learning_rate": 0.00013851736926808917, + "loss": 3.2106, + "step": 2538 + }, + { + "epoch": 3.72, + "grad_norm": 13.176704406738281, + "learning_rate": 0.00013847408046684497, + "loss": 3.2857, + "step": 2539 + }, + { + "epoch": 3.72, + "grad_norm": 5.579313278198242, + "learning_rate": 0.00013843078320156088, + "loss": 3.2861, + "step": 2540 + }, + { + "epoch": 3.73, + "grad_norm": 4.398528099060059, + "learning_rate": 0.0001383874774817621, + "loss": 3.225, + "step": 2541 + }, + { + "epoch": 3.73, + "grad_norm": 6.542925834655762, + "learning_rate": 0.00013834416331697544, + "loss": 3.204, + "step": 2542 + }, + { + "epoch": 3.73, + "grad_norm": 9.607292175292969, + "learning_rate": 0.00013830084071672985, + "loss": 3.2284, + "step": 2543 + }, + { + "epoch": 3.73, + "grad_norm": 2.1741347312927246, + "learning_rate": 0.00013825750969055604, + "loss": 3.2318, + "step": 2544 + }, + { + "epoch": 3.73, + "grad_norm": 2.4551427364349365, + "learning_rate": 0.00013821417024798642, + "loss": 3.2454, + "step": 2545 + }, + { + "epoch": 3.73, + "grad_norm": 1.8988637924194336, + "learning_rate": 0.0001381708223985555, + "loss": 3.2688, + "step": 2546 + }, + { + "epoch": 3.73, + "grad_norm": 1.2011829614639282, + "learning_rate": 0.00013812746615179944, + "loss": 3.2418, + "step": 2547 + }, + { + "epoch": 3.74, + "grad_norm": 2.4243390560150146, + "learning_rate": 0.0001380841015172563, + "loss": 3.2759, + "step": 2548 + }, + { + "epoch": 3.74, + "grad_norm": 0.9519268274307251, + "learning_rate": 0.00013804072850446612, + "loss": 3.2498, + "step": 2549 + }, + { + "epoch": 3.74, + "grad_norm": 5.196446418762207, + "learning_rate": 0.00013799734712297059, + "loss": 3.2358, + "step": 2550 + }, + { + "epoch": 3.74, + "grad_norm": 3.189202308654785, + "learning_rate": 0.00013795395738231333, + "loss": 3.2582, + "step": 2551 + }, + { + "epoch": 3.74, + "grad_norm": 5.616455078125, + "learning_rate": 0.0001379105592920398, + "loss": 3.2066, + "step": 2552 + }, + { + "epoch": 3.74, + "grad_norm": 9.704096794128418, + "learning_rate": 0.0001378671528616973, + "loss": 3.2245, + "step": 2553 + }, + { + "epoch": 3.75, + "grad_norm": 3.5445096492767334, + "learning_rate": 0.00013782373810083493, + "loss": 3.2097, + "step": 2554 + }, + { + "epoch": 3.75, + "grad_norm": 3.731013774871826, + "learning_rate": 0.00013778031501900367, + "loss": 3.2075, + "step": 2555 + }, + { + "epoch": 3.75, + "grad_norm": 8.907066345214844, + "learning_rate": 0.0001377368836257563, + "loss": 3.2249, + "step": 2556 + }, + { + "epoch": 3.75, + "grad_norm": 4.645308017730713, + "learning_rate": 0.00013769344393064743, + "loss": 3.2139, + "step": 2557 + }, + { + "epoch": 3.75, + "grad_norm": 4.715915679931641, + "learning_rate": 0.00013764999594323353, + "loss": 3.2613, + "step": 2558 + }, + { + "epoch": 3.75, + "grad_norm": 5.21357536315918, + "learning_rate": 0.0001376065396730728, + "loss": 3.2247, + "step": 2559 + }, + { + "epoch": 3.75, + "grad_norm": 1.811521053314209, + "learning_rate": 0.00013756307512972538, + "loss": 3.2714, + "step": 2560 + }, + { + "epoch": 3.76, + "grad_norm": 5.335707664489746, + "learning_rate": 0.0001375196023227532, + "loss": 3.2079, + "step": 2561 + }, + { + "epoch": 3.76, + "grad_norm": 3.844973087310791, + "learning_rate": 0.0001374761212617199, + "loss": 3.2411, + "step": 2562 + }, + { + "epoch": 3.76, + "grad_norm": 4.505429267883301, + "learning_rate": 0.00013743263195619115, + "loss": 3.2508, + "step": 2563 + }, + { + "epoch": 3.76, + "grad_norm": 5.144504547119141, + "learning_rate": 0.00013738913441573418, + "loss": 3.2112, + "step": 2564 + }, + { + "epoch": 3.76, + "grad_norm": 23.4456787109375, + "learning_rate": 0.00013734562864991823, + "loss": 3.2106, + "step": 2565 + }, + { + "epoch": 3.76, + "grad_norm": 4.499874591827393, + "learning_rate": 0.0001373021146683142, + "loss": 3.2324, + "step": 2566 + }, + { + "epoch": 3.76, + "grad_norm": 2.9865691661834717, + "learning_rate": 0.000137258592480495, + "loss": 3.2498, + "step": 2567 + }, + { + "epoch": 3.77, + "grad_norm": 3.308457374572754, + "learning_rate": 0.00013721506209603512, + "loss": 3.2132, + "step": 2568 + }, + { + "epoch": 3.77, + "grad_norm": 2.9479405879974365, + "learning_rate": 0.000137171523524511, + "loss": 3.2817, + "step": 2569 + }, + { + "epoch": 3.77, + "grad_norm": 2.7257778644561768, + "learning_rate": 0.0001371279767755008, + "loss": 3.185, + "step": 2570 + }, + { + "epoch": 3.77, + "grad_norm": 1.7737255096435547, + "learning_rate": 0.00013708442185858454, + "loss": 3.222, + "step": 2571 + }, + { + "epoch": 3.77, + "grad_norm": 3.827313184738159, + "learning_rate": 0.000137040858783344, + "loss": 3.1943, + "step": 2572 + }, + { + "epoch": 3.77, + "grad_norm": 2.6353366374969482, + "learning_rate": 0.00013699728755936277, + "loss": 3.2699, + "step": 2573 + }, + { + "epoch": 3.78, + "grad_norm": 4.652822017669678, + "learning_rate": 0.0001369537081962262, + "loss": 3.2313, + "step": 2574 + }, + { + "epoch": 3.78, + "grad_norm": 8.656730651855469, + "learning_rate": 0.0001369101207035215, + "loss": 3.1737, + "step": 2575 + }, + { + "epoch": 3.78, + "grad_norm": 1.588462471961975, + "learning_rate": 0.0001368665250908376, + "loss": 3.2057, + "step": 2576 + }, + { + "epoch": 3.78, + "grad_norm": 5.477829456329346, + "learning_rate": 0.00013682292136776524, + "loss": 3.2045, + "step": 2577 + }, + { + "epoch": 3.78, + "grad_norm": 6.158089637756348, + "learning_rate": 0.00013677930954389698, + "loss": 3.2183, + "step": 2578 + }, + { + "epoch": 3.78, + "grad_norm": 2.4769415855407715, + "learning_rate": 0.00013673568962882705, + "loss": 3.1601, + "step": 2579 + }, + { + "epoch": 3.78, + "grad_norm": 7.979514122009277, + "learning_rate": 0.00013669206163215164, + "loss": 3.2011, + "step": 2580 + }, + { + "epoch": 3.79, + "grad_norm": 11.770708084106445, + "learning_rate": 0.00013664842556346856, + "loss": 3.2215, + "step": 2581 + }, + { + "epoch": 3.79, + "grad_norm": 7.018968105316162, + "learning_rate": 0.00013660478143237746, + "loss": 3.2163, + "step": 2582 + }, + { + "epoch": 3.79, + "grad_norm": 4.448172569274902, + "learning_rate": 0.00013656112924847978, + "loss": 3.2027, + "step": 2583 + }, + { + "epoch": 3.79, + "grad_norm": 6.86191987991333, + "learning_rate": 0.00013651746902137868, + "loss": 3.1812, + "step": 2584 + }, + { + "epoch": 3.79, + "grad_norm": 4.056680202484131, + "learning_rate": 0.0001364738007606791, + "loss": 3.2857, + "step": 2585 + }, + { + "epoch": 3.79, + "grad_norm": 5.344080924987793, + "learning_rate": 0.0001364301244759878, + "loss": 3.255, + "step": 2586 + }, + { + "epoch": 3.79, + "grad_norm": 7.4425554275512695, + "learning_rate": 0.00013638644017691326, + "loss": 3.2032, + "step": 2587 + }, + { + "epoch": 3.8, + "grad_norm": 4.709530830383301, + "learning_rate": 0.00013634274787306577, + "loss": 3.25, + "step": 2588 + }, + { + "epoch": 3.8, + "grad_norm": 7.912161827087402, + "learning_rate": 0.00013629904757405723, + "loss": 3.2362, + "step": 2589 + }, + { + "epoch": 3.8, + "grad_norm": 24.812524795532227, + "learning_rate": 0.00013625533928950153, + "loss": 3.231, + "step": 2590 + }, + { + "epoch": 3.8, + "grad_norm": 9.660811424255371, + "learning_rate": 0.0001362116230290142, + "loss": 3.2175, + "step": 2591 + }, + { + "epoch": 3.8, + "grad_norm": 6.260350704193115, + "learning_rate": 0.00013616789880221242, + "loss": 3.2169, + "step": 2592 + }, + { + "epoch": 3.8, + "grad_norm": 2.6841564178466797, + "learning_rate": 0.00013612416661871533, + "loss": 3.2084, + "step": 2593 + }, + { + "epoch": 3.8, + "grad_norm": 6.153440475463867, + "learning_rate": 0.00013608042648814364, + "loss": 3.2029, + "step": 2594 + }, + { + "epoch": 3.81, + "grad_norm": 2.9339137077331543, + "learning_rate": 0.00013603667842011994, + "loss": 3.1789, + "step": 2595 + }, + { + "epoch": 3.81, + "grad_norm": 5.685370922088623, + "learning_rate": 0.00013599292242426851, + "loss": 3.2023, + "step": 2596 + }, + { + "epoch": 3.81, + "grad_norm": 12.19677734375, + "learning_rate": 0.00013594915851021532, + "loss": 3.2433, + "step": 2597 + }, + { + "epoch": 3.81, + "grad_norm": 5.506028652191162, + "learning_rate": 0.00013590538668758822, + "loss": 3.2088, + "step": 2598 + }, + { + "epoch": 3.81, + "grad_norm": 1.9152709245681763, + "learning_rate": 0.00013586160696601665, + "loss": 3.2067, + "step": 2599 + }, + { + "epoch": 3.81, + "grad_norm": 3.531026601791382, + "learning_rate": 0.00013581781935513188, + "loss": 3.2368, + "step": 2600 + }, + { + "epoch": 3.82, + "grad_norm": 1.68514883518219, + "learning_rate": 0.00013577402386456692, + "loss": 3.2567, + "step": 2601 + }, + { + "epoch": 3.82, + "grad_norm": 2.6391441822052, + "learning_rate": 0.00013573022050395643, + "loss": 3.1804, + "step": 2602 + }, + { + "epoch": 3.82, + "grad_norm": 2.4722888469696045, + "learning_rate": 0.0001356864092829369, + "loss": 3.2318, + "step": 2603 + }, + { + "epoch": 3.82, + "grad_norm": 1.5304458141326904, + "learning_rate": 0.0001356425902111465, + "loss": 3.2145, + "step": 2604 + }, + { + "epoch": 3.82, + "grad_norm": 4.6478705406188965, + "learning_rate": 0.0001355987632982251, + "loss": 3.2201, + "step": 2605 + }, + { + "epoch": 3.82, + "grad_norm": 3.80239200592041, + "learning_rate": 0.00013555492855381432, + "loss": 3.1819, + "step": 2606 + }, + { + "epoch": 3.82, + "grad_norm": 2.0871238708496094, + "learning_rate": 0.0001355110859875576, + "loss": 3.2169, + "step": 2607 + }, + { + "epoch": 3.83, + "grad_norm": 1.9589481353759766, + "learning_rate": 0.00013546723560909992, + "loss": 3.2077, + "step": 2608 + }, + { + "epoch": 3.83, + "grad_norm": 2.8086769580841064, + "learning_rate": 0.00013542337742808807, + "loss": 3.2386, + "step": 2609 + }, + { + "epoch": 3.83, + "grad_norm": 1.3195122480392456, + "learning_rate": 0.0001353795114541706, + "loss": 3.1651, + "step": 2610 + }, + { + "epoch": 3.83, + "grad_norm": 4.268999099731445, + "learning_rate": 0.00013533563769699774, + "loss": 3.2113, + "step": 2611 + }, + { + "epoch": 3.83, + "grad_norm": 2.7233781814575195, + "learning_rate": 0.00013529175616622133, + "loss": 3.2103, + "step": 2612 + }, + { + "epoch": 3.83, + "grad_norm": 3.8539392948150635, + "learning_rate": 0.00013524786687149512, + "loss": 3.2368, + "step": 2613 + }, + { + "epoch": 3.83, + "grad_norm": 3.89041805267334, + "learning_rate": 0.00013520396982247442, + "loss": 3.1865, + "step": 2614 + }, + { + "epoch": 3.84, + "grad_norm": 2.546916961669922, + "learning_rate": 0.0001351600650288163, + "loss": 3.192, + "step": 2615 + }, + { + "epoch": 3.84, + "grad_norm": 1.615480661392212, + "learning_rate": 0.0001351161525001795, + "loss": 3.2953, + "step": 2616 + }, + { + "epoch": 3.84, + "grad_norm": 3.4749913215637207, + "learning_rate": 0.00013507223224622445, + "loss": 3.1994, + "step": 2617 + }, + { + "epoch": 3.84, + "grad_norm": 1.6761186122894287, + "learning_rate": 0.00013502830427661338, + "loss": 3.1836, + "step": 2618 + }, + { + "epoch": 3.84, + "grad_norm": 6.091635704040527, + "learning_rate": 0.00013498436860101011, + "loss": 3.2183, + "step": 2619 + }, + { + "epoch": 3.84, + "grad_norm": 5.926030158996582, + "learning_rate": 0.00013494042522908022, + "loss": 3.2222, + "step": 2620 + }, + { + "epoch": 3.84, + "grad_norm": 12.00096321105957, + "learning_rate": 0.00013489647417049095, + "loss": 3.2073, + "step": 2621 + }, + { + "epoch": 3.85, + "grad_norm": 4.7124762535095215, + "learning_rate": 0.0001348525154349112, + "loss": 3.1859, + "step": 2622 + }, + { + "epoch": 3.85, + "grad_norm": 3.3024585247039795, + "learning_rate": 0.0001348085490320116, + "loss": 3.177, + "step": 2623 + }, + { + "epoch": 3.85, + "grad_norm": 4.006857395172119, + "learning_rate": 0.00013476457497146456, + "loss": 3.2317, + "step": 2624 + }, + { + "epoch": 3.85, + "grad_norm": 4.812809467315674, + "learning_rate": 0.00013472059326294396, + "loss": 3.2637, + "step": 2625 + }, + { + "epoch": 3.85, + "grad_norm": 1.2310017347335815, + "learning_rate": 0.00013467660391612555, + "loss": 3.2366, + "step": 2626 + }, + { + "epoch": 3.85, + "grad_norm": 2.896761894226074, + "learning_rate": 0.00013463260694068666, + "loss": 3.2069, + "step": 2627 + }, + { + "epoch": 3.86, + "grad_norm": 2.2707176208496094, + "learning_rate": 0.00013458860234630632, + "loss": 3.224, + "step": 2628 + }, + { + "epoch": 3.86, + "grad_norm": 2.2991745471954346, + "learning_rate": 0.00013454459014266526, + "loss": 3.2074, + "step": 2629 + }, + { + "epoch": 3.86, + "grad_norm": 2.838233470916748, + "learning_rate": 0.0001345005703394459, + "loss": 3.2019, + "step": 2630 + }, + { + "epoch": 3.86, + "grad_norm": 2.686645030975342, + "learning_rate": 0.0001344565429463323, + "loss": 3.1873, + "step": 2631 + }, + { + "epoch": 3.86, + "grad_norm": 3.610236883163452, + "learning_rate": 0.00013441250797301016, + "loss": 3.1667, + "step": 2632 + }, + { + "epoch": 3.86, + "grad_norm": 3.012873888015747, + "learning_rate": 0.00013436846542916686, + "loss": 3.2006, + "step": 2633 + }, + { + "epoch": 3.86, + "grad_norm": 2.223695755004883, + "learning_rate": 0.00013432441532449152, + "loss": 3.2004, + "step": 2634 + }, + { + "epoch": 3.87, + "grad_norm": 5.295112609863281, + "learning_rate": 0.00013428035766867484, + "loss": 3.1959, + "step": 2635 + }, + { + "epoch": 3.87, + "grad_norm": 1.9531642198562622, + "learning_rate": 0.00013423629247140919, + "loss": 3.1796, + "step": 2636 + }, + { + "epoch": 3.87, + "grad_norm": 11.508848190307617, + "learning_rate": 0.00013419221974238864, + "loss": 3.2325, + "step": 2637 + }, + { + "epoch": 3.87, + "grad_norm": 3.0934455394744873, + "learning_rate": 0.00013414813949130894, + "loss": 3.1904, + "step": 2638 + }, + { + "epoch": 3.87, + "grad_norm": 1.843153476715088, + "learning_rate": 0.00013410405172786734, + "loss": 3.177, + "step": 2639 + }, + { + "epoch": 3.87, + "grad_norm": 3.7216010093688965, + "learning_rate": 0.00013405995646176294, + "loss": 3.2278, + "step": 2640 + }, + { + "epoch": 3.87, + "grad_norm": 4.631261825561523, + "learning_rate": 0.00013401585370269637, + "loss": 3.2772, + "step": 2641 + }, + { + "epoch": 3.88, + "grad_norm": 2.928041696548462, + "learning_rate": 0.00013397174346036996, + "loss": 3.2489, + "step": 2642 + }, + { + "epoch": 3.88, + "grad_norm": 2.3705878257751465, + "learning_rate": 0.00013392762574448766, + "loss": 3.2577, + "step": 2643 + }, + { + "epoch": 3.88, + "grad_norm": 4.091426372528076, + "learning_rate": 0.00013388350056475504, + "loss": 3.1664, + "step": 2644 + }, + { + "epoch": 3.88, + "grad_norm": 1.9827549457550049, + "learning_rate": 0.0001338393679308794, + "loss": 3.1786, + "step": 2645 + }, + { + "epoch": 3.88, + "grad_norm": 4.402117729187012, + "learning_rate": 0.0001337952278525696, + "loss": 3.1966, + "step": 2646 + }, + { + "epoch": 3.88, + "grad_norm": 5.21814489364624, + "learning_rate": 0.00013375108033953614, + "loss": 3.2393, + "step": 2647 + }, + { + "epoch": 3.88, + "grad_norm": 2.4417850971221924, + "learning_rate": 0.0001337069254014912, + "loss": 3.1729, + "step": 2648 + }, + { + "epoch": 3.89, + "grad_norm": 3.521346092224121, + "learning_rate": 0.0001336627630481486, + "loss": 3.1997, + "step": 2649 + }, + { + "epoch": 3.89, + "grad_norm": 4.180556297302246, + "learning_rate": 0.0001336185932892237, + "loss": 3.1878, + "step": 2650 + }, + { + "epoch": 3.89, + "grad_norm": 2.869370460510254, + "learning_rate": 0.00013357441613443354, + "loss": 3.2052, + "step": 2651 + }, + { + "epoch": 3.89, + "grad_norm": 2.097275495529175, + "learning_rate": 0.0001335302315934969, + "loss": 3.1459, + "step": 2652 + }, + { + "epoch": 3.89, + "grad_norm": 3.70109486579895, + "learning_rate": 0.00013348603967613398, + "loss": 3.2187, + "step": 2653 + }, + { + "epoch": 3.89, + "grad_norm": 3.026033878326416, + "learning_rate": 0.00013344184039206677, + "loss": 3.2251, + "step": 2654 + }, + { + "epoch": 3.9, + "grad_norm": 2.825608015060425, + "learning_rate": 0.00013339763375101876, + "loss": 3.1845, + "step": 2655 + }, + { + "epoch": 3.9, + "grad_norm": 2.3940205574035645, + "learning_rate": 0.00013335341976271518, + "loss": 3.1909, + "step": 2656 + }, + { + "epoch": 3.9, + "grad_norm": 3.781881332397461, + "learning_rate": 0.00013330919843688275, + "loss": 3.1861, + "step": 2657 + }, + { + "epoch": 3.9, + "grad_norm": 3.5430831909179688, + "learning_rate": 0.0001332649697832499, + "loss": 3.2879, + "step": 2658 + }, + { + "epoch": 3.9, + "grad_norm": 3.327752113342285, + "learning_rate": 0.00013322073381154663, + "loss": 3.1407, + "step": 2659 + }, + { + "epoch": 3.9, + "grad_norm": 3.319768190383911, + "learning_rate": 0.00013317649053150456, + "loss": 3.2215, + "step": 2660 + }, + { + "epoch": 3.9, + "grad_norm": 2.317915201187134, + "learning_rate": 0.0001331322399528569, + "loss": 3.251, + "step": 2661 + }, + { + "epoch": 3.91, + "grad_norm": 2.999483108520508, + "learning_rate": 0.00013308798208533853, + "loss": 3.2182, + "step": 2662 + }, + { + "epoch": 3.91, + "grad_norm": 4.7933549880981445, + "learning_rate": 0.0001330437169386858, + "loss": 3.2272, + "step": 2663 + }, + { + "epoch": 3.91, + "grad_norm": 4.440158367156982, + "learning_rate": 0.00013299944452263682, + "loss": 3.202, + "step": 2664 + }, + { + "epoch": 3.91, + "grad_norm": 1.8913938999176025, + "learning_rate": 0.0001329551648469312, + "loss": 3.1705, + "step": 2665 + }, + { + "epoch": 3.91, + "grad_norm": 4.039271831512451, + "learning_rate": 0.00013291087792131017, + "loss": 3.213, + "step": 2666 + }, + { + "epoch": 3.91, + "grad_norm": 4.9916276931762695, + "learning_rate": 0.00013286658375551654, + "loss": 3.2316, + "step": 2667 + }, + { + "epoch": 3.91, + "grad_norm": 2.624239206314087, + "learning_rate": 0.00013282228235929474, + "loss": 3.1902, + "step": 2668 + }, + { + "epoch": 3.92, + "grad_norm": 2.3713390827178955, + "learning_rate": 0.0001327779737423908, + "loss": 3.1698, + "step": 2669 + }, + { + "epoch": 3.92, + "grad_norm": 3.9648921489715576, + "learning_rate": 0.0001327336579145523, + "loss": 3.1811, + "step": 2670 + }, + { + "epoch": 3.92, + "grad_norm": 3.6182374954223633, + "learning_rate": 0.00013268933488552845, + "loss": 3.2013, + "step": 2671 + }, + { + "epoch": 3.92, + "grad_norm": 1.5024638175964355, + "learning_rate": 0.00013264500466506998, + "loss": 3.1996, + "step": 2672 + }, + { + "epoch": 3.92, + "grad_norm": 1.957371473312378, + "learning_rate": 0.00013260066726292926, + "loss": 3.188, + "step": 2673 + }, + { + "epoch": 3.92, + "grad_norm": 1.9777625799179077, + "learning_rate": 0.00013255632268886023, + "loss": 3.2656, + "step": 2674 + }, + { + "epoch": 3.93, + "grad_norm": 1.5048279762268066, + "learning_rate": 0.0001325119709526184, + "loss": 3.2279, + "step": 2675 + }, + { + "epoch": 3.93, + "grad_norm": 35.774295806884766, + "learning_rate": 0.0001324676120639608, + "loss": 3.2366, + "step": 2676 + }, + { + "epoch": 3.93, + "grad_norm": 3.9443695545196533, + "learning_rate": 0.00013242324603264616, + "loss": 3.1658, + "step": 2677 + }, + { + "epoch": 3.93, + "grad_norm": 2.9231414794921875, + "learning_rate": 0.0001323788728684347, + "loss": 3.2058, + "step": 2678 + }, + { + "epoch": 3.93, + "grad_norm": 4.9291157722473145, + "learning_rate": 0.0001323344925810882, + "loss": 3.1528, + "step": 2679 + }, + { + "epoch": 3.93, + "grad_norm": 8.999177932739258, + "learning_rate": 0.00013229010518037002, + "loss": 3.2864, + "step": 2680 + }, + { + "epoch": 3.93, + "grad_norm": 4.351472854614258, + "learning_rate": 0.0001322457106760451, + "loss": 3.2048, + "step": 2681 + }, + { + "epoch": 3.94, + "grad_norm": 9.449458122253418, + "learning_rate": 0.00013220130907787994, + "loss": 3.2409, + "step": 2682 + }, + { + "epoch": 3.94, + "grad_norm": 2.7245359420776367, + "learning_rate": 0.00013215690039564257, + "loss": 3.1844, + "step": 2683 + }, + { + "epoch": 3.94, + "grad_norm": 1.7337415218353271, + "learning_rate": 0.00013211248463910262, + "loss": 3.1732, + "step": 2684 + }, + { + "epoch": 3.94, + "grad_norm": 2.845367431640625, + "learning_rate": 0.0001320680618180313, + "loss": 3.1608, + "step": 2685 + }, + { + "epoch": 3.94, + "grad_norm": 2.7872724533081055, + "learning_rate": 0.00013202363194220125, + "loss": 3.2057, + "step": 2686 + }, + { + "epoch": 3.94, + "grad_norm": 1.4545655250549316, + "learning_rate": 0.00013197919502138677, + "loss": 3.205, + "step": 2687 + }, + { + "epoch": 3.94, + "grad_norm": 3.010228157043457, + "learning_rate": 0.00013193475106536374, + "loss": 3.1844, + "step": 2688 + }, + { + "epoch": 3.95, + "grad_norm": 2.508854866027832, + "learning_rate": 0.00013189030008390945, + "loss": 3.2108, + "step": 2689 + }, + { + "epoch": 3.95, + "grad_norm": 3.1477272510528564, + "learning_rate": 0.0001318458420868029, + "loss": 3.2018, + "step": 2690 + }, + { + "epoch": 3.95, + "grad_norm": 2.543921947479248, + "learning_rate": 0.0001318013770838245, + "loss": 3.2235, + "step": 2691 + }, + { + "epoch": 3.95, + "grad_norm": 3.577669382095337, + "learning_rate": 0.00013175690508475626, + "loss": 3.2087, + "step": 2692 + }, + { + "epoch": 3.95, + "grad_norm": 2.2817633152008057, + "learning_rate": 0.00013171242609938175, + "loss": 3.1847, + "step": 2693 + }, + { + "epoch": 3.95, + "grad_norm": 7.939913272857666, + "learning_rate": 0.00013166794013748598, + "loss": 3.2269, + "step": 2694 + }, + { + "epoch": 3.95, + "grad_norm": 6.408430576324463, + "learning_rate": 0.00013162344720885565, + "loss": 3.1853, + "step": 2695 + }, + { + "epoch": 3.96, + "grad_norm": 95.63408660888672, + "learning_rate": 0.00013157894732327886, + "loss": 3.2441, + "step": 2696 + }, + { + "epoch": 3.96, + "grad_norm": 76.839111328125, + "learning_rate": 0.00013153444049054523, + "loss": 3.2624, + "step": 2697 + }, + { + "epoch": 3.96, + "grad_norm": 9.534677505493164, + "learning_rate": 0.00013148992672044608, + "loss": 3.2354, + "step": 2698 + }, + { + "epoch": 3.96, + "grad_norm": 29.769962310791016, + "learning_rate": 0.0001314454060227741, + "loss": 3.1857, + "step": 2699 + }, + { + "epoch": 3.96, + "grad_norm": 26.48052215576172, + "learning_rate": 0.00013140087840732355, + "loss": 3.2512, + "step": 2700 + }, + { + "epoch": 3.96, + "grad_norm": 12.384693145751953, + "learning_rate": 0.00013135634388389015, + "loss": 3.2054, + "step": 2701 + }, + { + "epoch": 3.97, + "grad_norm": 4.66050910949707, + "learning_rate": 0.00013131180246227125, + "loss": 3.2012, + "step": 2702 + }, + { + "epoch": 3.97, + "grad_norm": 28.54119300842285, + "learning_rate": 0.00013126725415226563, + "loss": 3.202, + "step": 2703 + }, + { + "epoch": 3.97, + "grad_norm": 69.08728790283203, + "learning_rate": 0.00013122269896367365, + "loss": 3.2423, + "step": 2704 + }, + { + "epoch": 3.97, + "eval_loss": 3.091963529586792, + "eval_runtime": 1165.9245, + "eval_samples_per_second": 14.702, + "eval_steps_per_second": 0.919, + "step": 2704 + }, + { + "epoch": 3.97, + "grad_norm": 22.657764434814453, + "learning_rate": 0.00013117813690629714, + "loss": 3.2586, + "step": 2705 + }, + { + "epoch": 3.97, + "grad_norm": 74.64742279052734, + "learning_rate": 0.00013113356798993947, + "loss": 3.2606, + "step": 2706 + }, + { + "epoch": 3.97, + "grad_norm": 14.374595642089844, + "learning_rate": 0.0001310889922244055, + "loss": 3.2525, + "step": 2707 + }, + { + "epoch": 3.97, + "grad_norm": 5.337686538696289, + "learning_rate": 0.00013104440961950154, + "loss": 3.2355, + "step": 2708 + }, + { + "epoch": 3.98, + "grad_norm": 3.205595016479492, + "learning_rate": 0.00013099982018503558, + "loss": 3.2323, + "step": 2709 + }, + { + "epoch": 3.98, + "grad_norm": 4.343560218811035, + "learning_rate": 0.0001309552239308169, + "loss": 3.2265, + "step": 2710 + }, + { + "epoch": 3.98, + "grad_norm": 68.85613250732422, + "learning_rate": 0.0001309106208666564, + "loss": 3.2412, + "step": 2711 + }, + { + "epoch": 3.98, + "grad_norm": 4.4929633140563965, + "learning_rate": 0.00013086601100236646, + "loss": 3.2243, + "step": 2712 + }, + { + "epoch": 3.98, + "grad_norm": 5.574833393096924, + "learning_rate": 0.000130821394347761, + "loss": 3.2, + "step": 2713 + }, + { + "epoch": 3.98, + "grad_norm": 2.189110040664673, + "learning_rate": 0.00013077677091265528, + "loss": 3.2177, + "step": 2714 + }, + { + "epoch": 3.98, + "grad_norm": 1.2807992696762085, + "learning_rate": 0.00013073214070686626, + "loss": 3.1971, + "step": 2715 + }, + { + "epoch": 3.99, + "grad_norm": 3.459458351135254, + "learning_rate": 0.0001306875037402122, + "loss": 3.186, + "step": 2716 + }, + { + "epoch": 3.99, + "grad_norm": 9.210981369018555, + "learning_rate": 0.000130642860022513, + "loss": 3.1966, + "step": 2717 + }, + { + "epoch": 3.99, + "grad_norm": 2.803443670272827, + "learning_rate": 0.00013059820956358998, + "loss": 3.1927, + "step": 2718 + }, + { + "epoch": 3.99, + "grad_norm": 2.078461170196533, + "learning_rate": 0.00013055355237326588, + "loss": 3.1776, + "step": 2719 + }, + { + "epoch": 3.99, + "grad_norm": 2.4591166973114014, + "learning_rate": 0.00013050888846136502, + "loss": 3.2126, + "step": 2720 + }, + { + "epoch": 3.99, + "grad_norm": 2.6859610080718994, + "learning_rate": 0.00013046421783771318, + "loss": 3.1956, + "step": 2721 + }, + { + "epoch": 3.99, + "grad_norm": 3.064992904663086, + "learning_rate": 0.00013041954051213757, + "loss": 3.25, + "step": 2722 + }, + { + "epoch": 4.0, + "grad_norm": 1.7159329652786255, + "learning_rate": 0.0001303748564944669, + "loss": 3.1734, + "step": 2723 + }, + { + "epoch": 4.0, + "grad_norm": 2.7244114875793457, + "learning_rate": 0.00013033016579453138, + "loss": 3.1743, + "step": 2724 + }, + { + "epoch": 4.0, + "grad_norm": 2.2309274673461914, + "learning_rate": 0.0001302854684221626, + "loss": 3.1751, + "step": 2725 + }, + { + "epoch": 4.0, + "grad_norm": 1.5704296827316284, + "learning_rate": 0.0001302407643871938, + "loss": 3.2018, + "step": 2726 + }, + { + "epoch": 4.0, + "grad_norm": 2.5336227416992188, + "learning_rate": 0.0001301960536994595, + "loss": 3.2139, + "step": 2727 + }, + { + "epoch": 4.0, + "grad_norm": 1.6835451126098633, + "learning_rate": 0.00013015133636879567, + "loss": 3.2198, + "step": 2728 + }, + { + "epoch": 4.01, + "grad_norm": 1.5716129541397095, + "learning_rate": 0.00013010661240503999, + "loss": 3.2108, + "step": 2729 + }, + { + "epoch": 4.01, + "grad_norm": 1.5545756816864014, + "learning_rate": 0.0001300618818180313, + "loss": 3.1592, + "step": 2730 + }, + { + "epoch": 4.01, + "grad_norm": 1.9438145160675049, + "learning_rate": 0.00013001714461761008, + "loss": 3.2145, + "step": 2731 + }, + { + "epoch": 4.01, + "grad_norm": 1.7333067655563354, + "learning_rate": 0.00012997240081361823, + "loss": 3.1362, + "step": 2732 + }, + { + "epoch": 4.01, + "grad_norm": 1.632812261581421, + "learning_rate": 0.00012992765041589904, + "loss": 3.2064, + "step": 2733 + }, + { + "epoch": 4.01, + "grad_norm": 1.1884607076644897, + "learning_rate": 0.00012988289343429733, + "loss": 3.1789, + "step": 2734 + }, + { + "epoch": 4.01, + "grad_norm": 2.9912478923797607, + "learning_rate": 0.00012983812987865935, + "loss": 3.1561, + "step": 2735 + }, + { + "epoch": 4.02, + "grad_norm": 1.3816310167312622, + "learning_rate": 0.00012979335975883276, + "loss": 3.1848, + "step": 2736 + }, + { + "epoch": 4.0, + "grad_norm": 19.01185417175293, + "learning_rate": 0.00012974858308466666, + "loss": 3.1801, + "step": 2737 + }, + { + "epoch": 4.0, + "grad_norm": 5.716503143310547, + "learning_rate": 0.0001297037998660117, + "loss": 3.2216, + "step": 2738 + }, + { + "epoch": 4.0, + "grad_norm": 2.985485792160034, + "learning_rate": 0.00012965901011271978, + "loss": 3.2136, + "step": 2739 + }, + { + "epoch": 4.01, + "grad_norm": 2.645477056503296, + "learning_rate": 0.00012961421383464445, + "loss": 3.176, + "step": 2740 + }, + { + "epoch": 4.01, + "grad_norm": 2.804504871368408, + "learning_rate": 0.0001295694110416405, + "loss": 3.1502, + "step": 2741 + }, + { + "epoch": 4.01, + "grad_norm": 2.257525682449341, + "learning_rate": 0.00012952460174356432, + "loss": 3.1909, + "step": 2742 + }, + { + "epoch": 4.01, + "grad_norm": 2.525448799133301, + "learning_rate": 0.0001294797859502736, + "loss": 3.2058, + "step": 2743 + }, + { + "epoch": 4.01, + "grad_norm": 2.589095115661621, + "learning_rate": 0.00012943496367162754, + "loss": 3.2137, + "step": 2744 + }, + { + "epoch": 4.01, + "grad_norm": 1.6353000402450562, + "learning_rate": 0.00012939013491748672, + "loss": 3.2001, + "step": 2745 + }, + { + "epoch": 4.01, + "grad_norm": 2.5494680404663086, + "learning_rate": 0.00012934529969771324, + "loss": 3.1529, + "step": 2746 + }, + { + "epoch": 4.02, + "grad_norm": 2.1689844131469727, + "learning_rate": 0.00012930045802217043, + "loss": 3.1931, + "step": 2747 + }, + { + "epoch": 4.02, + "grad_norm": 3.0633785724639893, + "learning_rate": 0.00012925560990072324, + "loss": 3.1941, + "step": 2748 + }, + { + "epoch": 4.02, + "grad_norm": 2.6445438861846924, + "learning_rate": 0.00012921075534323795, + "loss": 3.1673, + "step": 2749 + }, + { + "epoch": 4.02, + "grad_norm": 1.6328538656234741, + "learning_rate": 0.00012916589435958223, + "loss": 3.1533, + "step": 2750 + }, + { + "epoch": 4.02, + "grad_norm": 7.05519962310791, + "learning_rate": 0.0001291210269596252, + "loss": 3.1762, + "step": 2751 + }, + { + "epoch": 4.02, + "grad_norm": 10.450418472290039, + "learning_rate": 0.0001290761531532374, + "loss": 3.2032, + "step": 2752 + }, + { + "epoch": 4.02, + "grad_norm": 3.726336717605591, + "learning_rate": 0.00012903127295029078, + "loss": 3.1835, + "step": 2753 + }, + { + "epoch": 4.03, + "grad_norm": 2.4860239028930664, + "learning_rate": 0.00012898638636065868, + "loss": 3.1818, + "step": 2754 + }, + { + "epoch": 4.03, + "grad_norm": 2.166196584701538, + "learning_rate": 0.00012894149339421582, + "loss": 3.1609, + "step": 2755 + }, + { + "epoch": 4.03, + "grad_norm": 2.179659128189087, + "learning_rate": 0.00012889659406083837, + "loss": 3.2125, + "step": 2756 + }, + { + "epoch": 4.03, + "grad_norm": 2.2733445167541504, + "learning_rate": 0.0001288516883704039, + "loss": 3.2931, + "step": 2757 + }, + { + "epoch": 4.03, + "grad_norm": 1.3774925470352173, + "learning_rate": 0.00012880677633279132, + "loss": 3.1741, + "step": 2758 + }, + { + "epoch": 4.03, + "grad_norm": 4.090391159057617, + "learning_rate": 0.000128761857957881, + "loss": 3.2181, + "step": 2759 + }, + { + "epoch": 4.04, + "grad_norm": 3.762101411819458, + "learning_rate": 0.00012871693325555472, + "loss": 3.1991, + "step": 2760 + }, + { + "epoch": 4.04, + "grad_norm": 2.435084819793701, + "learning_rate": 0.00012867200223569555, + "loss": 3.1734, + "step": 2761 + }, + { + "epoch": 4.04, + "grad_norm": 11.30912971496582, + "learning_rate": 0.00012862706490818804, + "loss": 3.1755, + "step": 2762 + }, + { + "epoch": 4.04, + "grad_norm": 3.355214834213257, + "learning_rate": 0.0001285821212829181, + "loss": 3.1718, + "step": 2763 + }, + { + "epoch": 4.04, + "grad_norm": 2.4179513454437256, + "learning_rate": 0.00012853717136977305, + "loss": 3.1722, + "step": 2764 + }, + { + "epoch": 4.04, + "grad_norm": 2.18635630607605, + "learning_rate": 0.0001284922151786415, + "loss": 3.1982, + "step": 2765 + }, + { + "epoch": 4.04, + "grad_norm": 2.3029189109802246, + "learning_rate": 0.0001284472527194136, + "loss": 3.2427, + "step": 2766 + }, + { + "epoch": 4.05, + "grad_norm": 2.3949828147888184, + "learning_rate": 0.00012840228400198071, + "loss": 3.1944, + "step": 2767 + }, + { + "epoch": 4.05, + "grad_norm": 2.5206806659698486, + "learning_rate": 0.0001283573090362357, + "loss": 3.1899, + "step": 2768 + }, + { + "epoch": 4.05, + "grad_norm": 1.8850032091140747, + "learning_rate": 0.00012831232783207277, + "loss": 3.2212, + "step": 2769 + }, + { + "epoch": 4.05, + "grad_norm": 2.58017897605896, + "learning_rate": 0.0001282673403993874, + "loss": 3.2026, + "step": 2770 + }, + { + "epoch": 4.05, + "grad_norm": 2.990920305252075, + "learning_rate": 0.00012822234674807662, + "loss": 3.1852, + "step": 2771 + }, + { + "epoch": 4.05, + "grad_norm": 0.5992396473884583, + "learning_rate": 0.00012817734688803868, + "loss": 3.1736, + "step": 2772 + }, + { + "epoch": 4.05, + "grad_norm": 2.821469306945801, + "learning_rate": 0.0001281323408291733, + "loss": 3.1999, + "step": 2773 + }, + { + "epoch": 4.06, + "grad_norm": 15.83332633972168, + "learning_rate": 0.00012808732858138146, + "loss": 3.1844, + "step": 2774 + }, + { + "epoch": 4.06, + "grad_norm": 2.910820960998535, + "learning_rate": 0.0001280423101545655, + "loss": 3.1587, + "step": 2775 + }, + { + "epoch": 4.06, + "grad_norm": 17.933792114257812, + "learning_rate": 0.00012799728555862934, + "loss": 3.2201, + "step": 2776 + }, + { + "epoch": 4.06, + "grad_norm": 2.721127986907959, + "learning_rate": 0.0001279522548034779, + "loss": 3.1507, + "step": 2777 + }, + { + "epoch": 4.06, + "grad_norm": 1.5368844270706177, + "learning_rate": 0.00012790721789901775, + "loss": 3.1902, + "step": 2778 + }, + { + "epoch": 4.06, + "grad_norm": 3.0113532543182373, + "learning_rate": 0.00012786217485515674, + "loss": 3.149, + "step": 2779 + }, + { + "epoch": 4.06, + "grad_norm": 19.78055191040039, + "learning_rate": 0.00012781712568180393, + "loss": 3.1755, + "step": 2780 + }, + { + "epoch": 4.07, + "grad_norm": 2.508270025253296, + "learning_rate": 0.0001277720703888699, + "loss": 3.1915, + "step": 2781 + }, + { + "epoch": 4.07, + "grad_norm": 2.167954206466675, + "learning_rate": 0.00012772700898626652, + "loss": 3.2054, + "step": 2782 + }, + { + "epoch": 4.07, + "grad_norm": 4.016536235809326, + "learning_rate": 0.00012768194148390696, + "loss": 3.195, + "step": 2783 + }, + { + "epoch": 4.07, + "grad_norm": 3.441601276397705, + "learning_rate": 0.0001276368678917058, + "loss": 3.15, + "step": 2784 + }, + { + "epoch": 4.07, + "grad_norm": 7.370872974395752, + "learning_rate": 0.0001275917882195789, + "loss": 3.1633, + "step": 2785 + }, + { + "epoch": 4.07, + "grad_norm": 2.296081781387329, + "learning_rate": 0.00012754670247744354, + "loss": 3.2345, + "step": 2786 + }, + { + "epoch": 4.08, + "grad_norm": 2.009742259979248, + "learning_rate": 0.00012750161067521826, + "loss": 3.1938, + "step": 2787 + }, + { + "epoch": 4.08, + "grad_norm": 4.299360752105713, + "learning_rate": 0.00012745651282282288, + "loss": 3.2213, + "step": 2788 + }, + { + "epoch": 4.08, + "grad_norm": 3.636892080307007, + "learning_rate": 0.00012741140893017872, + "loss": 3.1713, + "step": 2789 + }, + { + "epoch": 4.08, + "grad_norm": 1.2036387920379639, + "learning_rate": 0.0001273662990072083, + "loss": 3.2107, + "step": 2790 + }, + { + "epoch": 4.08, + "grad_norm": 3.7308552265167236, + "learning_rate": 0.0001273211830638355, + "loss": 3.1955, + "step": 2791 + }, + { + "epoch": 4.08, + "grad_norm": 9.248310089111328, + "learning_rate": 0.00012727606110998549, + "loss": 3.1917, + "step": 2792 + }, + { + "epoch": 4.08, + "grad_norm": 15.789203643798828, + "learning_rate": 0.00012723093315558488, + "loss": 3.1611, + "step": 2793 + }, + { + "epoch": 4.09, + "grad_norm": 3.4315154552459717, + "learning_rate": 0.00012718579921056143, + "loss": 3.2364, + "step": 2794 + }, + { + "epoch": 4.09, + "grad_norm": 2.061835289001465, + "learning_rate": 0.00012714065928484436, + "loss": 3.2079, + "step": 2795 + }, + { + "epoch": 4.09, + "grad_norm": 2.4200713634490967, + "learning_rate": 0.00012709551338836413, + "loss": 3.1846, + "step": 2796 + }, + { + "epoch": 4.09, + "grad_norm": 3.561788558959961, + "learning_rate": 0.00012705036153105257, + "loss": 3.2151, + "step": 2797 + }, + { + "epoch": 4.09, + "grad_norm": 2.967207908630371, + "learning_rate": 0.00012700520372284273, + "loss": 3.2036, + "step": 2798 + }, + { + "epoch": 4.09, + "grad_norm": 3.490912437438965, + "learning_rate": 0.00012696003997366906, + "loss": 3.1893, + "step": 2799 + }, + { + "epoch": 4.09, + "grad_norm": 2.477809429168701, + "learning_rate": 0.00012691487029346728, + "loss": 3.1701, + "step": 2800 + }, + { + "epoch": 4.1, + "grad_norm": 3.558258295059204, + "learning_rate": 0.0001268696946921744, + "loss": 3.2343, + "step": 2801 + }, + { + "epoch": 4.1, + "grad_norm": 3.1721646785736084, + "learning_rate": 0.00012682451317972874, + "loss": 3.201, + "step": 2802 + }, + { + "epoch": 4.1, + "grad_norm": 2.562236785888672, + "learning_rate": 0.00012677932576606998, + "loss": 3.2291, + "step": 2803 + }, + { + "epoch": 4.1, + "grad_norm": 9.012580871582031, + "learning_rate": 0.00012673413246113905, + "loss": 3.2074, + "step": 2804 + }, + { + "epoch": 4.1, + "grad_norm": 2.1724753379821777, + "learning_rate": 0.00012668893327487805, + "loss": 3.2104, + "step": 2805 + }, + { + "epoch": 4.1, + "grad_norm": 1.8097708225250244, + "learning_rate": 0.00012664372821723064, + "loss": 3.1534, + "step": 2806 + }, + { + "epoch": 4.11, + "grad_norm": 2.409672498703003, + "learning_rate": 0.00012659851729814162, + "loss": 3.1748, + "step": 2807 + }, + { + "epoch": 4.11, + "grad_norm": 2.8095180988311768, + "learning_rate": 0.000126553300527557, + "loss": 3.1996, + "step": 2808 + }, + { + "epoch": 4.11, + "grad_norm": 2.353957176208496, + "learning_rate": 0.00012650807791542422, + "loss": 3.1945, + "step": 2809 + }, + { + "epoch": 4.11, + "grad_norm": 3.013678789138794, + "learning_rate": 0.00012646284947169195, + "loss": 3.1839, + "step": 2810 + }, + { + "epoch": 4.11, + "grad_norm": 2.154859781265259, + "learning_rate": 0.00012641761520631016, + "loss": 3.1541, + "step": 2811 + }, + { + "epoch": 4.11, + "grad_norm": 2.0929133892059326, + "learning_rate": 0.00012637237512923006, + "loss": 3.1655, + "step": 2812 + }, + { + "epoch": 4.11, + "grad_norm": 1.2277342081069946, + "learning_rate": 0.0001263271292504042, + "loss": 3.1933, + "step": 2813 + }, + { + "epoch": 4.12, + "grad_norm": 2.6938316822052, + "learning_rate": 0.0001262818775797863, + "loss": 3.1441, + "step": 2814 + }, + { + "epoch": 4.12, + "grad_norm": 1.9652117490768433, + "learning_rate": 0.00012623662012733146, + "loss": 3.1616, + "step": 2815 + }, + { + "epoch": 4.12, + "grad_norm": 1.6476244926452637, + "learning_rate": 0.00012619135690299603, + "loss": 3.1839, + "step": 2816 + }, + { + "epoch": 4.12, + "grad_norm": 1.534227967262268, + "learning_rate": 0.00012614608791673763, + "loss": 3.1646, + "step": 2817 + }, + { + "epoch": 4.12, + "grad_norm": 2.850703477859497, + "learning_rate": 0.00012610081317851511, + "loss": 3.207, + "step": 2818 + }, + { + "epoch": 4.12, + "grad_norm": 2.1084752082824707, + "learning_rate": 0.00012605553269828858, + "loss": 3.1688, + "step": 2819 + }, + { + "epoch": 4.12, + "grad_norm": 2.303504705429077, + "learning_rate": 0.0001260102464860195, + "loss": 3.229, + "step": 2820 + }, + { + "epoch": 4.13, + "grad_norm": 1.5470285415649414, + "learning_rate": 0.0001259649545516705, + "loss": 3.1814, + "step": 2821 + }, + { + "epoch": 4.13, + "grad_norm": 1.4991244077682495, + "learning_rate": 0.0001259196569052055, + "loss": 3.1927, + "step": 2822 + }, + { + "epoch": 4.13, + "grad_norm": 1.342314600944519, + "learning_rate": 0.0001258743535565897, + "loss": 3.1743, + "step": 2823 + }, + { + "epoch": 4.13, + "grad_norm": 2.662546396255493, + "learning_rate": 0.0001258290445157895, + "loss": 3.1884, + "step": 2824 + }, + { + "epoch": 4.13, + "grad_norm": 1.1698952913284302, + "learning_rate": 0.00012578372979277265, + "loss": 3.1841, + "step": 2825 + }, + { + "epoch": 4.13, + "grad_norm": 2.8144288063049316, + "learning_rate": 0.000125738409397508, + "loss": 3.1634, + "step": 2826 + }, + { + "epoch": 4.13, + "grad_norm": 1.0438013076782227, + "learning_rate": 0.00012569308333996578, + "loss": 3.1828, + "step": 2827 + }, + { + "epoch": 4.14, + "grad_norm": 5.697520732879639, + "learning_rate": 0.0001256477516301174, + "loss": 3.1897, + "step": 2828 + }, + { + "epoch": 4.14, + "grad_norm": 6.039661884307861, + "learning_rate": 0.00012560241427793555, + "loss": 3.2308, + "step": 2829 + }, + { + "epoch": 4.14, + "grad_norm": 51.954402923583984, + "learning_rate": 0.00012555707129339417, + "loss": 3.1693, + "step": 2830 + }, + { + "epoch": 4.14, + "grad_norm": 5.388270378112793, + "learning_rate": 0.00012551172268646834, + "loss": 3.2463, + "step": 2831 + }, + { + "epoch": 4.14, + "grad_norm": 4.189896583557129, + "learning_rate": 0.00012546636846713454, + "loss": 3.209, + "step": 2832 + }, + { + "epoch": 4.14, + "grad_norm": 2.421431064605713, + "learning_rate": 0.0001254210086453703, + "loss": 3.2312, + "step": 2833 + }, + { + "epoch": 4.15, + "grad_norm": 4.315710544586182, + "learning_rate": 0.00012537564323115454, + "loss": 3.2011, + "step": 2834 + }, + { + "epoch": 4.15, + "grad_norm": 1.7199636697769165, + "learning_rate": 0.00012533027223446733, + "loss": 3.1634, + "step": 2835 + }, + { + "epoch": 4.15, + "grad_norm": 4.567407608032227, + "learning_rate": 0.00012528489566528996, + "loss": 3.1999, + "step": 2836 + }, + { + "epoch": 4.15, + "grad_norm": 5.22501277923584, + "learning_rate": 0.00012523951353360504, + "loss": 3.1743, + "step": 2837 + }, + { + "epoch": 4.15, + "grad_norm": 2.197282314300537, + "learning_rate": 0.00012519412584939627, + "loss": 3.2332, + "step": 2838 + }, + { + "epoch": 4.15, + "grad_norm": 5.91029167175293, + "learning_rate": 0.00012514873262264864, + "loss": 3.1713, + "step": 2839 + }, + { + "epoch": 4.15, + "grad_norm": 6.8516716957092285, + "learning_rate": 0.0001251033338633484, + "loss": 3.1908, + "step": 2840 + }, + { + "epoch": 4.16, + "grad_norm": 3.321195125579834, + "learning_rate": 0.00012505792958148292, + "loss": 3.1693, + "step": 2841 + }, + { + "epoch": 4.16, + "grad_norm": 11.536995887756348, + "learning_rate": 0.00012501251978704088, + "loss": 3.1626, + "step": 2842 + }, + { + "epoch": 4.16, + "grad_norm": 9.267814636230469, + "learning_rate": 0.0001249671044900121, + "loss": 3.1807, + "step": 2843 + }, + { + "epoch": 4.16, + "grad_norm": 8.332723617553711, + "learning_rate": 0.00012492168370038765, + "loss": 3.1934, + "step": 2844 + }, + { + "epoch": 4.16, + "grad_norm": 3.4367361068725586, + "learning_rate": 0.00012487625742815985, + "loss": 3.1797, + "step": 2845 + }, + { + "epoch": 4.16, + "grad_norm": 6.636847019195557, + "learning_rate": 0.00012483082568332207, + "loss": 3.1851, + "step": 2846 + }, + { + "epoch": 4.16, + "grad_norm": 8.669164657592773, + "learning_rate": 0.00012478538847586907, + "loss": 3.2061, + "step": 2847 + }, + { + "epoch": 4.17, + "grad_norm": 6.34977912902832, + "learning_rate": 0.00012473994581579673, + "loss": 3.196, + "step": 2848 + }, + { + "epoch": 4.17, + "grad_norm": 4.2629780769348145, + "learning_rate": 0.00012469449771310208, + "loss": 3.2128, + "step": 2849 + }, + { + "epoch": 4.17, + "grad_norm": 5.895789623260498, + "learning_rate": 0.00012464904417778345, + "loss": 3.2769, + "step": 2850 + }, + { + "epoch": 4.17, + "grad_norm": 7.333654403686523, + "learning_rate": 0.0001246035852198403, + "loss": 3.2108, + "step": 2851 + }, + { + "epoch": 4.17, + "grad_norm": 2.965632915496826, + "learning_rate": 0.00012455812084927325, + "loss": 3.2012, + "step": 2852 + }, + { + "epoch": 4.17, + "grad_norm": 8.51198673248291, + "learning_rate": 0.0001245126510760842, + "loss": 3.2149, + "step": 2853 + }, + { + "epoch": 4.17, + "grad_norm": 7.021212100982666, + "learning_rate": 0.00012446717591027624, + "loss": 3.2624, + "step": 2854 + }, + { + "epoch": 4.18, + "grad_norm": 1.8178749084472656, + "learning_rate": 0.0001244216953618535, + "loss": 3.2099, + "step": 2855 + }, + { + "epoch": 4.18, + "grad_norm": 11.300747871398926, + "learning_rate": 0.00012437620944082145, + "loss": 3.1795, + "step": 2856 + }, + { + "epoch": 4.18, + "grad_norm": 9.994318962097168, + "learning_rate": 0.0001243307181571867, + "loss": 3.2285, + "step": 2857 + }, + { + "epoch": 4.18, + "grad_norm": 11.500404357910156, + "learning_rate": 0.00012428522152095703, + "loss": 3.1869, + "step": 2858 + }, + { + "epoch": 4.18, + "grad_norm": 5.326147556304932, + "learning_rate": 0.00012423971954214137, + "loss": 3.2226, + "step": 2859 + }, + { + "epoch": 4.18, + "grad_norm": 5.258158206939697, + "learning_rate": 0.00012419421223074982, + "loss": 3.1848, + "step": 2860 + }, + { + "epoch": 4.19, + "grad_norm": 7.557288646697998, + "learning_rate": 0.00012414869959679375, + "loss": 3.2594, + "step": 2861 + }, + { + "epoch": 4.19, + "grad_norm": 19.032909393310547, + "learning_rate": 0.0001241031816502856, + "loss": 3.2464, + "step": 2862 + }, + { + "epoch": 4.19, + "grad_norm": 9.38792896270752, + "learning_rate": 0.00012405765840123903, + "loss": 3.4169, + "step": 2863 + }, + { + "epoch": 4.19, + "grad_norm": 58.51255798339844, + "learning_rate": 0.0001240121298596688, + "loss": 3.3102, + "step": 2864 + }, + { + "epoch": 4.19, + "grad_norm": 9.871623992919922, + "learning_rate": 0.00012396659603559099, + "loss": 3.3069, + "step": 2865 + }, + { + "epoch": 4.19, + "grad_norm": 18.70452117919922, + "learning_rate": 0.00012392105693902262, + "loss": 3.2645, + "step": 2866 + }, + { + "epoch": 4.19, + "grad_norm": 6.495476245880127, + "learning_rate": 0.00012387551257998206, + "loss": 3.2188, + "step": 2867 + }, + { + "epoch": 4.2, + "grad_norm": 4.516813278198242, + "learning_rate": 0.00012382996296848875, + "loss": 3.2188, + "step": 2868 + }, + { + "epoch": 4.2, + "grad_norm": 9.024648666381836, + "learning_rate": 0.00012378440811456324, + "loss": 3.251, + "step": 2869 + }, + { + "epoch": 4.2, + "grad_norm": 10.055442810058594, + "learning_rate": 0.00012373884802822737, + "loss": 3.2687, + "step": 2870 + }, + { + "epoch": 4.2, + "grad_norm": 5.984691619873047, + "learning_rate": 0.00012369328271950405, + "loss": 3.2402, + "step": 2871 + }, + { + "epoch": 4.2, + "grad_norm": 4.42611026763916, + "learning_rate": 0.00012364771219841727, + "loss": 3.2241, + "step": 2872 + }, + { + "epoch": 4.2, + "grad_norm": 5.240525722503662, + "learning_rate": 0.0001236021364749923, + "loss": 3.2565, + "step": 2873 + }, + { + "epoch": 4.2, + "eval_loss": 3.031118154525757, + "eval_runtime": 1166.2283, + "eval_samples_per_second": 14.699, + "eval_steps_per_second": 0.919, + "step": 2873 + }, + { + "epoch": 4.2, + "grad_norm": 3.474372148513794, + "learning_rate": 0.00012355655555925548, + "loss": 3.1972, + "step": 2874 + }, + { + "epoch": 4.21, + "grad_norm": 3.6792542934417725, + "learning_rate": 0.00012351096946123431, + "loss": 3.1889, + "step": 2875 + }, + { + "epoch": 4.21, + "grad_norm": 3.1051828861236572, + "learning_rate": 0.00012346537819095743, + "loss": 3.206, + "step": 2876 + }, + { + "epoch": 4.21, + "grad_norm": 2.708695411682129, + "learning_rate": 0.00012341978175845462, + "loss": 3.2389, + "step": 2877 + }, + { + "epoch": 4.21, + "grad_norm": 2.4177584648132324, + "learning_rate": 0.00012337418017375676, + "loss": 3.1948, + "step": 2878 + }, + { + "epoch": 4.21, + "grad_norm": 3.8860323429107666, + "learning_rate": 0.00012332857344689587, + "loss": 3.2037, + "step": 2879 + }, + { + "epoch": 4.21, + "grad_norm": 4.779682159423828, + "learning_rate": 0.0001232829615879052, + "loss": 3.1901, + "step": 2880 + }, + { + "epoch": 4.21, + "grad_norm": 2.2922635078430176, + "learning_rate": 0.00012323734460681902, + "loss": 3.1367, + "step": 2881 + }, + { + "epoch": 4.22, + "grad_norm": 2.1851584911346436, + "learning_rate": 0.00012319172251367276, + "loss": 3.1782, + "step": 2882 + }, + { + "epoch": 4.22, + "grad_norm": 3.3458876609802246, + "learning_rate": 0.00012314609531850293, + "loss": 3.2361, + "step": 2883 + }, + { + "epoch": 4.22, + "grad_norm": 3.9705002307891846, + "learning_rate": 0.00012310046303134732, + "loss": 3.2227, + "step": 2884 + }, + { + "epoch": 4.22, + "grad_norm": 2.2800686359405518, + "learning_rate": 0.0001230548256622446, + "loss": 3.2052, + "step": 2885 + }, + { + "epoch": 4.22, + "grad_norm": 4.4462199211120605, + "learning_rate": 0.00012300918322123475, + "loss": 3.2142, + "step": 2886 + }, + { + "epoch": 4.22, + "grad_norm": 3.9958441257476807, + "learning_rate": 0.0001229635357183588, + "loss": 3.1989, + "step": 2887 + }, + { + "epoch": 4.23, + "grad_norm": 3.798068046569824, + "learning_rate": 0.00012291788316365888, + "loss": 3.1956, + "step": 2888 + }, + { + "epoch": 4.23, + "grad_norm": 5.135157108306885, + "learning_rate": 0.00012287222556717827, + "loss": 3.1854, + "step": 2889 + }, + { + "epoch": 4.23, + "grad_norm": 2.207172393798828, + "learning_rate": 0.0001228265629389613, + "loss": 3.1583, + "step": 2890 + }, + { + "epoch": 4.23, + "grad_norm": 8.646849632263184, + "learning_rate": 0.00012278089528905346, + "loss": 3.245, + "step": 2891 + }, + { + "epoch": 4.23, + "grad_norm": 7.601872444152832, + "learning_rate": 0.00012273522262750132, + "loss": 3.1968, + "step": 2892 + }, + { + "epoch": 4.23, + "grad_norm": 6.095008373260498, + "learning_rate": 0.00012268954496435258, + "loss": 3.1753, + "step": 2893 + }, + { + "epoch": 4.23, + "grad_norm": 3.311551094055176, + "learning_rate": 0.000122643862309656, + "loss": 3.2052, + "step": 2894 + }, + { + "epoch": 4.24, + "grad_norm": 6.019943714141846, + "learning_rate": 0.0001225981746734615, + "loss": 3.2064, + "step": 2895 + }, + { + "epoch": 4.24, + "grad_norm": 5.016195774078369, + "learning_rate": 0.00012255248206581996, + "loss": 3.2018, + "step": 2896 + }, + { + "epoch": 4.24, + "grad_norm": 2.2277472019195557, + "learning_rate": 0.00012250678449678354, + "loss": 3.1877, + "step": 2897 + }, + { + "epoch": 4.24, + "grad_norm": 5.388427734375, + "learning_rate": 0.00012246108197640538, + "loss": 3.2087, + "step": 2898 + }, + { + "epoch": 4.24, + "grad_norm": 3.7437267303466797, + "learning_rate": 0.0001224153745147397, + "loss": 3.1554, + "step": 2899 + }, + { + "epoch": 4.24, + "grad_norm": 2.8456854820251465, + "learning_rate": 0.0001223696621218418, + "loss": 3.1912, + "step": 2900 + }, + { + "epoch": 4.24, + "grad_norm": 2.9107625484466553, + "learning_rate": 0.0001223239448077682, + "loss": 3.1832, + "step": 2901 + }, + { + "epoch": 4.25, + "grad_norm": 2.9731972217559814, + "learning_rate": 0.00012227822258257636, + "loss": 3.1781, + "step": 2902 + }, + { + "epoch": 4.25, + "grad_norm": 3.2368147373199463, + "learning_rate": 0.00012223249545632482, + "loss": 3.1863, + "step": 2903 + }, + { + "epoch": 4.25, + "grad_norm": 1.3652539253234863, + "learning_rate": 0.00012218676343907328, + "loss": 3.1765, + "step": 2904 + }, + { + "epoch": 4.25, + "grad_norm": 6.007261753082275, + "learning_rate": 0.00012214102654088248, + "loss": 3.2073, + "step": 2905 + }, + { + "epoch": 4.25, + "grad_norm": 6.385366439819336, + "learning_rate": 0.00012209528477181422, + "loss": 3.1974, + "step": 2906 + }, + { + "epoch": 4.25, + "grad_norm": 1.5009405612945557, + "learning_rate": 0.00012204953814193139, + "loss": 3.1815, + "step": 2907 + }, + { + "epoch": 4.26, + "grad_norm": 4.717186450958252, + "learning_rate": 0.00012200378666129793, + "loss": 3.166, + "step": 2908 + }, + { + "epoch": 4.26, + "grad_norm": 4.571866035461426, + "learning_rate": 0.0001219580303399789, + "loss": 3.2043, + "step": 2909 + }, + { + "epoch": 4.26, + "grad_norm": 6.416868686676025, + "learning_rate": 0.0001219122691880403, + "loss": 3.1969, + "step": 2910 + }, + { + "epoch": 4.26, + "grad_norm": 4.813213348388672, + "learning_rate": 0.00012186650321554935, + "loss": 3.1706, + "step": 2911 + }, + { + "epoch": 4.26, + "grad_norm": 3.790310859680176, + "learning_rate": 0.00012182073243257426, + "loss": 3.178, + "step": 2912 + }, + { + "epoch": 4.26, + "grad_norm": 4.808967590332031, + "learning_rate": 0.00012177495684918423, + "loss": 3.2266, + "step": 2913 + }, + { + "epoch": 4.26, + "grad_norm": 8.275779724121094, + "learning_rate": 0.00012172917647544964, + "loss": 3.1736, + "step": 2914 + }, + { + "epoch": 4.27, + "grad_norm": 4.079070568084717, + "learning_rate": 0.00012168339132144186, + "loss": 3.1918, + "step": 2915 + }, + { + "epoch": 4.27, + "grad_norm": 2.108287811279297, + "learning_rate": 0.0001216376013972333, + "loss": 3.1543, + "step": 2916 + }, + { + "epoch": 4.27, + "grad_norm": 3.337580442428589, + "learning_rate": 0.00012159180671289743, + "loss": 3.1937, + "step": 2917 + }, + { + "epoch": 4.27, + "grad_norm": 2.6380503177642822, + "learning_rate": 0.0001215460072785088, + "loss": 3.1794, + "step": 2918 + }, + { + "epoch": 4.27, + "grad_norm": 2.371419906616211, + "learning_rate": 0.00012150020310414296, + "loss": 3.1682, + "step": 2919 + }, + { + "epoch": 4.27, + "grad_norm": 1.647721767425537, + "learning_rate": 0.00012145439419987652, + "loss": 3.1618, + "step": 2920 + }, + { + "epoch": 4.27, + "grad_norm": 0.846366822719574, + "learning_rate": 0.00012140858057578713, + "loss": 3.1883, + "step": 2921 + }, + { + "epoch": 4.28, + "grad_norm": 1.4289370775222778, + "learning_rate": 0.00012136276224195348, + "loss": 3.2119, + "step": 2922 + }, + { + "epoch": 4.28, + "grad_norm": 2.510512351989746, + "learning_rate": 0.00012131693920845533, + "loss": 3.1849, + "step": 2923 + }, + { + "epoch": 4.28, + "grad_norm": 1.3627164363861084, + "learning_rate": 0.0001212711114853734, + "loss": 3.1851, + "step": 2924 + }, + { + "epoch": 4.28, + "grad_norm": 2.383141279220581, + "learning_rate": 0.00012122527908278949, + "loss": 3.1599, + "step": 2925 + }, + { + "epoch": 4.28, + "grad_norm": 2.2744383811950684, + "learning_rate": 0.00012117944201078647, + "loss": 3.1814, + "step": 2926 + }, + { + "epoch": 4.28, + "grad_norm": 2.6775460243225098, + "learning_rate": 0.00012113360027944806, + "loss": 3.1495, + "step": 2927 + }, + { + "epoch": 4.28, + "grad_norm": 2.249178647994995, + "learning_rate": 0.00012108775389885929, + "loss": 3.1721, + "step": 2928 + }, + { + "epoch": 4.29, + "grad_norm": 1.3241461515426636, + "learning_rate": 0.00012104190287910597, + "loss": 3.1563, + "step": 2929 + }, + { + "epoch": 4.29, + "grad_norm": 3.977029323577881, + "learning_rate": 0.00012099604723027502, + "loss": 3.254, + "step": 2930 + }, + { + "epoch": 4.29, + "grad_norm": 3.9783012866973877, + "learning_rate": 0.00012095018696245441, + "loss": 3.1861, + "step": 2931 + }, + { + "epoch": 4.29, + "grad_norm": 1.687117338180542, + "learning_rate": 0.00012090432208573306, + "loss": 3.1259, + "step": 2932 + }, + { + "epoch": 4.29, + "grad_norm": 10.139452934265137, + "learning_rate": 0.00012085845261020095, + "loss": 3.1579, + "step": 2933 + }, + { + "epoch": 4.29, + "grad_norm": 4.6292500495910645, + "learning_rate": 0.00012081257854594906, + "loss": 3.2254, + "step": 2934 + }, + { + "epoch": 4.3, + "grad_norm": 3.5405800342559814, + "learning_rate": 0.00012076669990306934, + "loss": 3.1327, + "step": 2935 + }, + { + "epoch": 4.3, + "grad_norm": 1.5894381999969482, + "learning_rate": 0.00012072081669165482, + "loss": 3.1545, + "step": 2936 + }, + { + "epoch": 4.3, + "grad_norm": 3.921757221221924, + "learning_rate": 0.0001206749289217995, + "loss": 3.1646, + "step": 2937 + }, + { + "epoch": 4.3, + "grad_norm": 3.1364986896514893, + "learning_rate": 0.0001206290366035984, + "loss": 3.1835, + "step": 2938 + }, + { + "epoch": 4.3, + "grad_norm": 1.0131809711456299, + "learning_rate": 0.00012058313974714746, + "loss": 3.1446, + "step": 2939 + }, + { + "epoch": 4.3, + "grad_norm": 2.2197444438934326, + "learning_rate": 0.00012053723836254373, + "loss": 3.1805, + "step": 2940 + }, + { + "epoch": 4.3, + "grad_norm": 1.5680094957351685, + "learning_rate": 0.00012049133245988519, + "loss": 3.2005, + "step": 2941 + }, + { + "epoch": 4.31, + "grad_norm": 2.2786076068878174, + "learning_rate": 0.00012044542204927087, + "loss": 3.1663, + "step": 2942 + }, + { + "epoch": 4.31, + "grad_norm": 2.1755716800689697, + "learning_rate": 0.00012039950714080068, + "loss": 3.1709, + "step": 2943 + }, + { + "epoch": 4.31, + "grad_norm": 1.485142469406128, + "learning_rate": 0.00012035358774457563, + "loss": 3.2298, + "step": 2944 + }, + { + "epoch": 4.31, + "grad_norm": 1.9527829885482788, + "learning_rate": 0.00012030766387069771, + "loss": 3.1868, + "step": 2945 + }, + { + "epoch": 4.31, + "grad_norm": 1.5011379718780518, + "learning_rate": 0.00012026173552926981, + "loss": 3.1533, + "step": 2946 + }, + { + "epoch": 4.31, + "grad_norm": 1.8543486595153809, + "learning_rate": 0.00012021580273039585, + "loss": 3.1536, + "step": 2947 + }, + { + "epoch": 4.31, + "grad_norm": 2.353090763092041, + "learning_rate": 0.00012016986548418083, + "loss": 3.1675, + "step": 2948 + }, + { + "epoch": 4.32, + "grad_norm": 3.509500741958618, + "learning_rate": 0.00012012392380073056, + "loss": 3.1963, + "step": 2949 + }, + { + "epoch": 4.32, + "grad_norm": 4.018800258636475, + "learning_rate": 0.00012007797769015192, + "loss": 3.165, + "step": 2950 + }, + { + "epoch": 4.32, + "grad_norm": 2.620981454849243, + "learning_rate": 0.00012003202716255273, + "loss": 3.1379, + "step": 2951 + }, + { + "epoch": 4.32, + "grad_norm": 2501.735107421875, + "learning_rate": 0.00011998607222804181, + "loss": 3.2292, + "step": 2952 + }, + { + "epoch": 4.32, + "grad_norm": 7.561061859130859, + "learning_rate": 0.00011994011289672895, + "loss": 3.1922, + "step": 2953 + }, + { + "epoch": 4.32, + "grad_norm": 7.529016971588135, + "learning_rate": 0.00011989414917872488, + "loss": 3.1894, + "step": 2954 + }, + { + "epoch": 4.32, + "grad_norm": 2.2252986431121826, + "learning_rate": 0.00011984818108414133, + "loss": 3.1745, + "step": 2955 + }, + { + "epoch": 4.33, + "grad_norm": 6.323204040527344, + "learning_rate": 0.00011980220862309097, + "loss": 3.1978, + "step": 2956 + }, + { + "epoch": 4.33, + "grad_norm": 137.4775390625, + "learning_rate": 0.00011975623180568738, + "loss": 3.222, + "step": 2957 + }, + { + "epoch": 4.33, + "grad_norm": 9.54941177368164, + "learning_rate": 0.0001197102506420452, + "loss": 3.1733, + "step": 2958 + }, + { + "epoch": 4.33, + "grad_norm": 7.13044548034668, + "learning_rate": 0.00011966426514228, + "loss": 3.2191, + "step": 2959 + }, + { + "epoch": 4.33, + "grad_norm": 1.8224902153015137, + "learning_rate": 0.00011961827531650824, + "loss": 3.2066, + "step": 2960 + }, + { + "epoch": 4.33, + "grad_norm": 7.74959659576416, + "learning_rate": 0.00011957228117484736, + "loss": 3.2138, + "step": 2961 + }, + { + "epoch": 4.34, + "grad_norm": 8.487194061279297, + "learning_rate": 0.00011952628272741584, + "loss": 3.1951, + "step": 2962 + }, + { + "epoch": 4.34, + "grad_norm": 8.898509979248047, + "learning_rate": 0.00011948027998433296, + "loss": 3.1982, + "step": 2963 + }, + { + "epoch": 4.34, + "grad_norm": 11.637873649597168, + "learning_rate": 0.00011943427295571899, + "loss": 3.2156, + "step": 2964 + }, + { + "epoch": 4.34, + "grad_norm": 7.426854133605957, + "learning_rate": 0.00011938826165169525, + "loss": 3.2193, + "step": 2965 + }, + { + "epoch": 4.34, + "grad_norm": 12.111673355102539, + "learning_rate": 0.00011934224608238389, + "loss": 3.1818, + "step": 2966 + }, + { + "epoch": 4.34, + "grad_norm": 3.1284587383270264, + "learning_rate": 0.00011929622625790802, + "loss": 3.1583, + "step": 2967 + }, + { + "epoch": 4.34, + "grad_norm": 4.960999488830566, + "learning_rate": 0.00011925020218839167, + "loss": 3.1765, + "step": 2968 + }, + { + "epoch": 4.35, + "grad_norm": 7.320558071136475, + "learning_rate": 0.00011920417388395986, + "loss": 3.19, + "step": 2969 + }, + { + "epoch": 4.35, + "grad_norm": 4.303677082061768, + "learning_rate": 0.0001191581413547385, + "loss": 3.1875, + "step": 2970 + }, + { + "epoch": 4.35, + "grad_norm": 4.7796549797058105, + "learning_rate": 0.0001191121046108544, + "loss": 3.1843, + "step": 2971 + }, + { + "epoch": 4.35, + "grad_norm": 5.886425971984863, + "learning_rate": 0.0001190660636624354, + "loss": 3.1532, + "step": 2972 + }, + { + "epoch": 4.35, + "grad_norm": 2.2462821006774902, + "learning_rate": 0.00011902001851961017, + "loss": 3.1873, + "step": 2973 + }, + { + "epoch": 4.35, + "grad_norm": 6.365395545959473, + "learning_rate": 0.00011897396919250831, + "loss": 3.1509, + "step": 2974 + }, + { + "epoch": 4.35, + "grad_norm": 8.911288261413574, + "learning_rate": 0.0001189279156912604, + "loss": 3.1714, + "step": 2975 + }, + { + "epoch": 4.36, + "grad_norm": 5.056528091430664, + "learning_rate": 0.00011888185802599791, + "loss": 3.1708, + "step": 2976 + }, + { + "epoch": 4.36, + "grad_norm": 3.0920491218566895, + "learning_rate": 0.00011883579620685318, + "loss": 3.1315, + "step": 2977 + }, + { + "epoch": 4.36, + "grad_norm": 4.874855041503906, + "learning_rate": 0.00011878973024395952, + "loss": 3.1369, + "step": 2978 + }, + { + "epoch": 4.36, + "grad_norm": 2.600829601287842, + "learning_rate": 0.0001187436601474511, + "loss": 3.1673, + "step": 2979 + }, + { + "epoch": 4.36, + "grad_norm": 5.483844757080078, + "learning_rate": 0.00011869758592746309, + "loss": 3.1439, + "step": 2980 + }, + { + "epoch": 4.36, + "grad_norm": 7.73532772064209, + "learning_rate": 0.00011865150759413147, + "loss": 3.2182, + "step": 2981 + }, + { + "epoch": 4.36, + "grad_norm": 3.6422669887542725, + "learning_rate": 0.00011860542515759318, + "loss": 3.1551, + "step": 2982 + }, + { + "epoch": 4.37, + "grad_norm": 4.651947975158691, + "learning_rate": 0.00011855933862798605, + "loss": 3.1674, + "step": 2983 + }, + { + "epoch": 4.37, + "grad_norm": 6.519864559173584, + "learning_rate": 0.00011851324801544879, + "loss": 3.1561, + "step": 2984 + }, + { + "epoch": 4.37, + "grad_norm": 3.8000521659851074, + "learning_rate": 0.00011846715333012104, + "loss": 3.174, + "step": 2985 + }, + { + "epoch": 4.37, + "grad_norm": 4.087390899658203, + "learning_rate": 0.00011842105458214333, + "loss": 3.1715, + "step": 2986 + }, + { + "epoch": 4.37, + "grad_norm": 5.734311580657959, + "learning_rate": 0.00011837495178165706, + "loss": 3.1486, + "step": 2987 + }, + { + "epoch": 4.37, + "grad_norm": 2.4519381523132324, + "learning_rate": 0.00011832884493880452, + "loss": 3.1528, + "step": 2988 + }, + { + "epoch": 4.38, + "grad_norm": 4.925870895385742, + "learning_rate": 0.00011828273406372894, + "loss": 3.1785, + "step": 2989 + }, + { + "epoch": 4.38, + "grad_norm": 6.140622615814209, + "learning_rate": 0.0001182366191665744, + "loss": 3.207, + "step": 2990 + }, + { + "epoch": 4.38, + "grad_norm": 2.630859851837158, + "learning_rate": 0.00011819050025748583, + "loss": 3.1437, + "step": 2991 + }, + { + "epoch": 4.38, + "grad_norm": 6.897947311401367, + "learning_rate": 0.00011814437734660917, + "loss": 3.2066, + "step": 2992 + }, + { + "epoch": 4.38, + "grad_norm": 10.167567253112793, + "learning_rate": 0.00011809825044409106, + "loss": 3.1581, + "step": 2993 + }, + { + "epoch": 4.38, + "grad_norm": 13.395803451538086, + "learning_rate": 0.00011805211956007914, + "loss": 3.1348, + "step": 2994 + }, + { + "epoch": 4.38, + "grad_norm": 1.3696818351745605, + "learning_rate": 0.00011800598470472189, + "loss": 3.1391, + "step": 2995 + }, + { + "epoch": 4.39, + "grad_norm": 6.5317063331604, + "learning_rate": 0.00011795984588816869, + "loss": 3.1951, + "step": 2996 + }, + { + "epoch": 4.39, + "grad_norm": 63.888736724853516, + "learning_rate": 0.00011791370312056972, + "loss": 3.2487, + "step": 2997 + }, + { + "epoch": 4.39, + "grad_norm": 9.84652042388916, + "learning_rate": 0.00011786755641207614, + "loss": 3.1655, + "step": 2998 + }, + { + "epoch": 4.39, + "grad_norm": 7.442485809326172, + "learning_rate": 0.00011782140577283991, + "loss": 3.1921, + "step": 2999 + }, + { + "epoch": 4.39, + "grad_norm": 1.4993599653244019, + "learning_rate": 0.00011777525121301383, + "loss": 3.1366, + "step": 3000 + }, + { + "epoch": 4.39, + "grad_norm": 23.01314926147461, + "learning_rate": 0.00011772909274275159, + "loss": 3.1806, + "step": 3001 + }, + { + "epoch": 4.39, + "grad_norm": 13.442095756530762, + "learning_rate": 0.00011768293037220778, + "loss": 3.2192, + "step": 3002 + }, + { + "epoch": 4.4, + "grad_norm": 11.32374382019043, + "learning_rate": 0.0001176367641115378, + "loss": 3.1979, + "step": 3003 + }, + { + "epoch": 4.4, + "grad_norm": 4.027917385101318, + "learning_rate": 0.00011759059397089793, + "loss": 3.1963, + "step": 3004 + }, + { + "epoch": 4.4, + "grad_norm": 6.907668113708496, + "learning_rate": 0.00011754441996044523, + "loss": 3.168, + "step": 3005 + }, + { + "epoch": 4.4, + "grad_norm": 8.269980430603027, + "learning_rate": 0.00011749824209033775, + "loss": 3.1789, + "step": 3006 + }, + { + "epoch": 4.4, + "grad_norm": 6.633981227874756, + "learning_rate": 0.00011745206037073426, + "loss": 3.225, + "step": 3007 + }, + { + "epoch": 4.4, + "grad_norm": 5.4061994552612305, + "learning_rate": 0.00011740587481179442, + "loss": 3.2288, + "step": 3008 + }, + { + "epoch": 4.41, + "grad_norm": 5.786009311676025, + "learning_rate": 0.0001173596854236788, + "loss": 3.1826, + "step": 3009 + }, + { + "epoch": 4.41, + "grad_norm": 6.876216888427734, + "learning_rate": 0.00011731349221654875, + "loss": 3.1563, + "step": 3010 + }, + { + "epoch": 4.41, + "grad_norm": 6.093964099884033, + "learning_rate": 0.00011726729520056641, + "loss": 3.1879, + "step": 3011 + }, + { + "epoch": 4.41, + "grad_norm": 5.7057576179504395, + "learning_rate": 0.00011722109438589486, + "loss": 3.2176, + "step": 3012 + }, + { + "epoch": 4.41, + "grad_norm": 6.603635787963867, + "learning_rate": 0.00011717488978269795, + "loss": 3.1905, + "step": 3013 + }, + { + "epoch": 4.41, + "grad_norm": 4.3624067306518555, + "learning_rate": 0.00011712868140114037, + "loss": 3.1549, + "step": 3014 + }, + { + "epoch": 4.41, + "grad_norm": 3.682727098464966, + "learning_rate": 0.00011708246925138771, + "loss": 3.1376, + "step": 3015 + }, + { + "epoch": 4.42, + "grad_norm": 6.1669921875, + "learning_rate": 0.00011703625334360626, + "loss": 3.1513, + "step": 3016 + }, + { + "epoch": 4.42, + "grad_norm": 4.20526123046875, + "learning_rate": 0.00011699003368796328, + "loss": 3.1493, + "step": 3017 + }, + { + "epoch": 4.42, + "grad_norm": 2.534583568572998, + "learning_rate": 0.0001169438102946267, + "loss": 3.1625, + "step": 3018 + }, + { + "epoch": 4.42, + "grad_norm": 2.841315269470215, + "learning_rate": 0.00011689758317376547, + "loss": 3.2036, + "step": 3019 + }, + { + "epoch": 4.42, + "grad_norm": 14.288698196411133, + "learning_rate": 0.00011685135233554917, + "loss": 3.1499, + "step": 3020 + }, + { + "epoch": 4.42, + "grad_norm": 5.600762844085693, + "learning_rate": 0.00011680511779014827, + "loss": 3.1529, + "step": 3021 + }, + { + "epoch": 4.42, + "grad_norm": 2.945816993713379, + "learning_rate": 0.00011675887954773405, + "loss": 3.1832, + "step": 3022 + }, + { + "epoch": 4.43, + "grad_norm": 3.300783395767212, + "learning_rate": 0.0001167126376184787, + "loss": 3.16, + "step": 3023 + }, + { + "epoch": 4.43, + "grad_norm": 10.50096607208252, + "learning_rate": 0.00011666639201255506, + "loss": 3.128, + "step": 3024 + }, + { + "epoch": 4.43, + "grad_norm": 2.324991226196289, + "learning_rate": 0.00011662014274013684, + "loss": 3.1621, + "step": 3025 + }, + { + "epoch": 4.43, + "grad_norm": 3.7734105587005615, + "learning_rate": 0.00011657388981139865, + "loss": 3.1594, + "step": 3026 + }, + { + "epoch": 4.43, + "grad_norm": 3.324181079864502, + "learning_rate": 0.00011652763323651575, + "loss": 3.1495, + "step": 3027 + }, + { + "epoch": 4.43, + "grad_norm": 2.9058284759521484, + "learning_rate": 0.00011648137302566431, + "loss": 3.1342, + "step": 3028 + }, + { + "epoch": 4.43, + "grad_norm": 3.1446614265441895, + "learning_rate": 0.00011643510918902123, + "loss": 3.1408, + "step": 3029 + }, + { + "epoch": 4.44, + "grad_norm": 1.5861114263534546, + "learning_rate": 0.0001163888417367643, + "loss": 3.1833, + "step": 3030 + }, + { + "epoch": 4.44, + "grad_norm": 0.9234637022018433, + "learning_rate": 0.00011634257067907201, + "loss": 3.1692, + "step": 3031 + }, + { + "epoch": 4.44, + "grad_norm": 2.46486759185791, + "learning_rate": 0.00011629629602612368, + "loss": 3.1654, + "step": 3032 + }, + { + "epoch": 4.44, + "grad_norm": 1.57401442527771, + "learning_rate": 0.00011625001778809945, + "loss": 3.1469, + "step": 3033 + }, + { + "epoch": 4.44, + "grad_norm": 3.949265241622925, + "learning_rate": 0.00011620373597518024, + "loss": 3.2028, + "step": 3034 + }, + { + "epoch": 4.44, + "grad_norm": 4.033581256866455, + "learning_rate": 0.00011615745059754766, + "loss": 3.1827, + "step": 3035 + }, + { + "epoch": 4.45, + "grad_norm": 3.1215124130249023, + "learning_rate": 0.00011611116166538425, + "loss": 3.1908, + "step": 3036 + }, + { + "epoch": 4.45, + "grad_norm": 2.8856616020202637, + "learning_rate": 0.00011606486918887327, + "loss": 3.1631, + "step": 3037 + }, + { + "epoch": 4.45, + "grad_norm": 3.271087408065796, + "learning_rate": 0.00011601857317819871, + "loss": 3.1653, + "step": 3038 + }, + { + "epoch": 4.45, + "grad_norm": 3.887037515640259, + "learning_rate": 0.0001159722736435454, + "loss": 3.1656, + "step": 3039 + }, + { + "epoch": 4.45, + "grad_norm": 1.2171587944030762, + "learning_rate": 0.00011592597059509893, + "loss": 3.1612, + "step": 3040 + }, + { + "epoch": 4.45, + "grad_norm": 1.7865351438522339, + "learning_rate": 0.00011587966404304564, + "loss": 3.1736, + "step": 3041 + }, + { + "epoch": 4.45, + "grad_norm": 0.9533948302268982, + "learning_rate": 0.00011583335399757271, + "loss": 3.1167, + "step": 3042 + }, + { + "epoch": 4.45, + "eval_loss": 3.0038862228393555, + "eval_runtime": 1165.2502, + "eval_samples_per_second": 14.711, + "eval_steps_per_second": 0.92, + "step": 3042 + }, + { + "epoch": 4.46, + "grad_norm": 1.679763674736023, + "learning_rate": 0.000115787040468868, + "loss": 3.1827, + "step": 3043 + }, + { + "epoch": 4.46, + "grad_norm": 0.7595570683479309, + "learning_rate": 0.0001157407234671202, + "loss": 3.1615, + "step": 3044 + }, + { + "epoch": 4.46, + "grad_norm": 1.3758699893951416, + "learning_rate": 0.0001156944030025187, + "loss": 3.2312, + "step": 3045 + }, + { + "epoch": 4.46, + "grad_norm": 1.6547698974609375, + "learning_rate": 0.00011564807908525374, + "loss": 3.1552, + "step": 3046 + }, + { + "epoch": 4.46, + "grad_norm": 1.4767943620681763, + "learning_rate": 0.00011560175172551623, + "loss": 3.2091, + "step": 3047 + }, + { + "epoch": 4.46, + "grad_norm": 0.9520362615585327, + "learning_rate": 0.00011555542093349791, + "loss": 3.1592, + "step": 3048 + }, + { + "epoch": 4.46, + "grad_norm": 0.4594898223876953, + "learning_rate": 0.00011550908671939125, + "loss": 3.1426, + "step": 3049 + }, + { + "epoch": 4.47, + "grad_norm": 0.8685529232025146, + "learning_rate": 0.00011546274909338945, + "loss": 3.115, + "step": 3050 + }, + { + "epoch": 4.47, + "grad_norm": 0.986310601234436, + "learning_rate": 0.00011541640806568646, + "loss": 3.1239, + "step": 3051 + }, + { + "epoch": 4.47, + "grad_norm": 2.280745029449463, + "learning_rate": 0.000115370063646477, + "loss": 3.1833, + "step": 3052 + }, + { + "epoch": 4.47, + "grad_norm": 0.696453869342804, + "learning_rate": 0.00011532371584595656, + "loss": 3.1107, + "step": 3053 + }, + { + "epoch": 4.47, + "grad_norm": 3.4933245182037354, + "learning_rate": 0.00011527736467432133, + "loss": 3.1748, + "step": 3054 + }, + { + "epoch": 4.47, + "grad_norm": 2.052901268005371, + "learning_rate": 0.00011523101014176825, + "loss": 3.16, + "step": 3055 + }, + { + "epoch": 4.47, + "grad_norm": 3.9722251892089844, + "learning_rate": 0.00011518465225849502, + "loss": 3.1674, + "step": 3056 + }, + { + "epoch": 4.48, + "grad_norm": 3.952956438064575, + "learning_rate": 0.00011513829103470004, + "loss": 3.1506, + "step": 3057 + }, + { + "epoch": 4.48, + "grad_norm": 2.466555118560791, + "learning_rate": 0.00011509192648058249, + "loss": 3.1843, + "step": 3058 + }, + { + "epoch": 4.48, + "grad_norm": 2.9880712032318115, + "learning_rate": 0.00011504555860634225, + "loss": 3.2101, + "step": 3059 + }, + { + "epoch": 4.48, + "grad_norm": 3.0772719383239746, + "learning_rate": 0.00011499918742217996, + "loss": 3.1815, + "step": 3060 + }, + { + "epoch": 4.48, + "grad_norm": 3.1504437923431396, + "learning_rate": 0.00011495281293829697, + "loss": 3.1493, + "step": 3061 + }, + { + "epoch": 4.48, + "grad_norm": 1.549119472503662, + "learning_rate": 0.00011490643516489532, + "loss": 3.154, + "step": 3062 + }, + { + "epoch": 4.49, + "grad_norm": 5.081751346588135, + "learning_rate": 0.00011486005411217784, + "loss": 3.2329, + "step": 3063 + }, + { + "epoch": 4.49, + "grad_norm": 5.428029537200928, + "learning_rate": 0.00011481366979034806, + "loss": 3.1709, + "step": 3064 + }, + { + "epoch": 4.49, + "grad_norm": 1.1052792072296143, + "learning_rate": 0.00011476728220961015, + "loss": 3.1713, + "step": 3065 + }, + { + "epoch": 4.49, + "grad_norm": 4.415512561798096, + "learning_rate": 0.00011472089138016917, + "loss": 3.1753, + "step": 3066 + }, + { + "epoch": 4.49, + "grad_norm": 4.5699143409729, + "learning_rate": 0.00011467449731223075, + "loss": 3.209, + "step": 3067 + }, + { + "epoch": 4.49, + "grad_norm": 3.3557825088500977, + "learning_rate": 0.00011462810001600123, + "loss": 3.1721, + "step": 3068 + }, + { + "epoch": 4.49, + "grad_norm": 3.846703290939331, + "learning_rate": 0.00011458169950168774, + "loss": 3.1592, + "step": 3069 + }, + { + "epoch": 4.5, + "grad_norm": 3.0023891925811768, + "learning_rate": 0.0001145352957794981, + "loss": 3.1636, + "step": 3070 + }, + { + "epoch": 4.5, + "grad_norm": 6.419353485107422, + "learning_rate": 0.0001144888888596408, + "loss": 3.1472, + "step": 3071 + }, + { + "epoch": 4.5, + "grad_norm": 6.247819900512695, + "learning_rate": 0.00011444247875232503, + "loss": 3.1901, + "step": 3072 + }, + { + "epoch": 4.5, + "grad_norm": 4.539994239807129, + "learning_rate": 0.00011439606546776076, + "loss": 3.1452, + "step": 3073 + }, + { + "epoch": 4.5, + "grad_norm": 3.100914716720581, + "learning_rate": 0.00011434964901615852, + "loss": 3.1582, + "step": 3074 + }, + { + "epoch": 4.5, + "grad_norm": 8.092155456542969, + "learning_rate": 0.0001143032294077297, + "loss": 3.2045, + "step": 3075 + }, + { + "epoch": 4.5, + "grad_norm": 2.9788618087768555, + "learning_rate": 0.00011425680665268626, + "loss": 3.1545, + "step": 3076 + }, + { + "epoch": 4.51, + "grad_norm": 2.915862560272217, + "learning_rate": 0.00011421038076124089, + "loss": 3.1752, + "step": 3077 + }, + { + "epoch": 4.51, + "grad_norm": 3.0964643955230713, + "learning_rate": 0.000114163951743607, + "loss": 3.1225, + "step": 3078 + }, + { + "epoch": 4.51, + "grad_norm": 2.3280699253082275, + "learning_rate": 0.00011411751960999862, + "loss": 3.179, + "step": 3079 + }, + { + "epoch": 4.51, + "grad_norm": 1.9851235151290894, + "learning_rate": 0.00011407108437063057, + "loss": 3.1557, + "step": 3080 + }, + { + "epoch": 4.51, + "grad_norm": 2.6399571895599365, + "learning_rate": 0.00011402464603571825, + "loss": 3.131, + "step": 3081 + }, + { + "epoch": 4.51, + "grad_norm": 10.382826805114746, + "learning_rate": 0.00011397820461547777, + "loss": 3.1541, + "step": 3082 + }, + { + "epoch": 4.51, + "grad_norm": 1.8169869184494019, + "learning_rate": 0.00011393176012012596, + "loss": 3.2169, + "step": 3083 + }, + { + "epoch": 4.52, + "grad_norm": 3.3234314918518066, + "learning_rate": 0.00011388531255988033, + "loss": 3.1631, + "step": 3084 + }, + { + "epoch": 4.52, + "grad_norm": 1.7650176286697388, + "learning_rate": 0.00011383886194495896, + "loss": 3.1561, + "step": 3085 + }, + { + "epoch": 4.52, + "grad_norm": 5.759214878082275, + "learning_rate": 0.00011379240828558072, + "loss": 3.1424, + "step": 3086 + }, + { + "epoch": 4.52, + "grad_norm": 5.468089580535889, + "learning_rate": 0.00011374595159196506, + "loss": 3.1697, + "step": 3087 + }, + { + "epoch": 4.52, + "grad_norm": 25.312021255493164, + "learning_rate": 0.00011369949187433218, + "loss": 3.1632, + "step": 3088 + }, + { + "epoch": 4.52, + "grad_norm": 3.2354679107666016, + "learning_rate": 0.0001136530291429029, + "loss": 3.166, + "step": 3089 + }, + { + "epoch": 4.53, + "grad_norm": 11.292868614196777, + "learning_rate": 0.00011360656340789873, + "loss": 3.1495, + "step": 3090 + }, + { + "epoch": 4.53, + "grad_norm": 9.209776878356934, + "learning_rate": 0.00011356009467954176, + "loss": 3.189, + "step": 3091 + }, + { + "epoch": 4.53, + "grad_norm": 40.07894515991211, + "learning_rate": 0.00011351362296805485, + "loss": 3.1709, + "step": 3092 + }, + { + "epoch": 4.53, + "grad_norm": 4.195234298706055, + "learning_rate": 0.00011346714828366148, + "loss": 3.2041, + "step": 3093 + }, + { + "epoch": 4.53, + "grad_norm": 154.33792114257812, + "learning_rate": 0.0001134206706365857, + "loss": 3.1673, + "step": 3094 + }, + { + "epoch": 4.53, + "grad_norm": 3.392430305480957, + "learning_rate": 0.00011337419003705237, + "loss": 3.1747, + "step": 3095 + }, + { + "epoch": 4.53, + "grad_norm": 156.2117462158203, + "learning_rate": 0.00011332770649528681, + "loss": 3.1426, + "step": 3096 + }, + { + "epoch": 4.54, + "grad_norm": 3.3672966957092285, + "learning_rate": 0.00011328122002151516, + "loss": 3.1428, + "step": 3097 + }, + { + "epoch": 4.54, + "grad_norm": 1.9393489360809326, + "learning_rate": 0.00011323473062596413, + "loss": 3.1491, + "step": 3098 + }, + { + "epoch": 4.54, + "grad_norm": 4.762568473815918, + "learning_rate": 0.00011318823831886103, + "loss": 3.1408, + "step": 3099 + }, + { + "epoch": 4.54, + "grad_norm": 4.439011573791504, + "learning_rate": 0.00011314174311043391, + "loss": 3.149, + "step": 3100 + }, + { + "epoch": 4.54, + "grad_norm": 1099.63525390625, + "learning_rate": 0.00011309524501091135, + "loss": 3.1883, + "step": 3101 + }, + { + "epoch": 4.54, + "grad_norm": 14.831306457519531, + "learning_rate": 0.00011304874403052266, + "loss": 3.2264, + "step": 3102 + }, + { + "epoch": 4.54, + "grad_norm": 1.7882112264633179, + "learning_rate": 0.00011300224017949772, + "loss": 3.1552, + "step": 3103 + }, + { + "epoch": 4.55, + "grad_norm": 5.061419486999512, + "learning_rate": 0.00011295573346806709, + "loss": 3.1629, + "step": 3104 + }, + { + "epoch": 4.55, + "grad_norm": 2.1199474334716797, + "learning_rate": 0.00011290922390646192, + "loss": 3.1672, + "step": 3105 + }, + { + "epoch": 4.55, + "grad_norm": 2.093383312225342, + "learning_rate": 0.000112862711504914, + "loss": 3.1872, + "step": 3106 + }, + { + "epoch": 4.55, + "grad_norm": 1.5643643140792847, + "learning_rate": 0.00011281619627365576, + "loss": 3.2203, + "step": 3107 + }, + { + "epoch": 4.55, + "grad_norm": 3.1214048862457275, + "learning_rate": 0.00011276967822292024, + "loss": 3.1508, + "step": 3108 + }, + { + "epoch": 4.55, + "grad_norm": 17.56413459777832, + "learning_rate": 0.00011272315736294108, + "loss": 3.1855, + "step": 3109 + }, + { + "epoch": 4.56, + "grad_norm": 3.3824844360351562, + "learning_rate": 0.00011267663370395256, + "loss": 3.1449, + "step": 3110 + }, + { + "epoch": 4.56, + "grad_norm": 2.0639045238494873, + "learning_rate": 0.00011263010725618959, + "loss": 3.1227, + "step": 3111 + }, + { + "epoch": 4.56, + "grad_norm": 2.0494930744171143, + "learning_rate": 0.00011258357802988766, + "loss": 3.1386, + "step": 3112 + }, + { + "epoch": 4.56, + "grad_norm": 2.3310718536376953, + "learning_rate": 0.00011253704603528288, + "loss": 3.1649, + "step": 3113 + }, + { + "epoch": 4.56, + "grad_norm": 1.5938396453857422, + "learning_rate": 0.00011249051128261203, + "loss": 3.1524, + "step": 3114 + }, + { + "epoch": 4.56, + "grad_norm": 1.9962538480758667, + "learning_rate": 0.00011244397378211238, + "loss": 3.1949, + "step": 3115 + }, + { + "epoch": 4.56, + "grad_norm": 2.338400363922119, + "learning_rate": 0.00011239743354402188, + "loss": 3.1258, + "step": 3116 + }, + { + "epoch": 4.57, + "grad_norm": 1.9446300268173218, + "learning_rate": 0.00011235089057857911, + "loss": 3.1829, + "step": 3117 + }, + { + "epoch": 4.57, + "grad_norm": 2.5406713485717773, + "learning_rate": 0.00011230434489602317, + "loss": 3.1782, + "step": 3118 + }, + { + "epoch": 4.57, + "grad_norm": 1.6750293970108032, + "learning_rate": 0.00011225779650659383, + "loss": 3.139, + "step": 3119 + }, + { + "epoch": 4.57, + "grad_norm": 10.995355606079102, + "learning_rate": 0.00011221124542053139, + "loss": 3.152, + "step": 3120 + }, + { + "epoch": 4.57, + "grad_norm": 3.1163713932037354, + "learning_rate": 0.00011216469164807679, + "loss": 3.1542, + "step": 3121 + }, + { + "epoch": 4.57, + "grad_norm": 1.6949200630187988, + "learning_rate": 0.00011211813519947159, + "loss": 3.1438, + "step": 3122 + }, + { + "epoch": 4.57, + "grad_norm": 3.045480251312256, + "learning_rate": 0.00011207157608495779, + "loss": 3.1611, + "step": 3123 + }, + { + "epoch": 4.58, + "grad_norm": 2.5979998111724854, + "learning_rate": 0.0001120250143147782, + "loss": 3.1628, + "step": 3124 + }, + { + "epoch": 4.58, + "grad_norm": 2.419813632965088, + "learning_rate": 0.00011197844989917605, + "loss": 3.1876, + "step": 3125 + }, + { + "epoch": 4.58, + "grad_norm": 2.4058735370635986, + "learning_rate": 0.00011193188284839517, + "loss": 3.1617, + "step": 3126 + }, + { + "epoch": 4.58, + "grad_norm": 1.7974401712417603, + "learning_rate": 0.00011188531317268005, + "loss": 3.1276, + "step": 3127 + }, + { + "epoch": 4.58, + "grad_norm": 2.3527932167053223, + "learning_rate": 0.0001118387408822757, + "loss": 3.1843, + "step": 3128 + }, + { + "epoch": 4.58, + "grad_norm": 2.5821757316589355, + "learning_rate": 0.00011179216598742768, + "loss": 3.1492, + "step": 3129 + }, + { + "epoch": 4.58, + "grad_norm": 0.7473565936088562, + "learning_rate": 0.00011174558849838213, + "loss": 3.1384, + "step": 3130 + }, + { + "epoch": 4.59, + "grad_norm": 2.998114585876465, + "learning_rate": 0.00011169900842538588, + "loss": 3.173, + "step": 3131 + }, + { + "epoch": 4.59, + "grad_norm": 8.924659729003906, + "learning_rate": 0.00011165242577868616, + "loss": 3.1929, + "step": 3132 + }, + { + "epoch": 4.59, + "grad_norm": 13.762849807739258, + "learning_rate": 0.00011160584056853082, + "loss": 3.165, + "step": 3133 + }, + { + "epoch": 4.59, + "grad_norm": 2.115877151489258, + "learning_rate": 0.00011155925280516841, + "loss": 3.119, + "step": 3134 + }, + { + "epoch": 4.59, + "grad_norm": 3.3011577129364014, + "learning_rate": 0.00011151266249884783, + "loss": 3.1905, + "step": 3135 + }, + { + "epoch": 4.59, + "grad_norm": 26.584270477294922, + "learning_rate": 0.00011146606965981862, + "loss": 3.2361, + "step": 3136 + }, + { + "epoch": 4.6, + "grad_norm": 1.8690835237503052, + "learning_rate": 0.00011141947429833096, + "loss": 3.1527, + "step": 3137 + }, + { + "epoch": 4.6, + "grad_norm": 3.516594886779785, + "learning_rate": 0.00011137287642463545, + "loss": 3.1821, + "step": 3138 + }, + { + "epoch": 4.6, + "grad_norm": 2.7472705841064453, + "learning_rate": 0.00011132627604898338, + "loss": 3.1594, + "step": 3139 + }, + { + "epoch": 4.6, + "grad_norm": 1.927894115447998, + "learning_rate": 0.00011127967318162644, + "loss": 3.1446, + "step": 3140 + }, + { + "epoch": 4.6, + "grad_norm": 1.823498249053955, + "learning_rate": 0.00011123306783281701, + "loss": 3.1621, + "step": 3141 + }, + { + "epoch": 4.6, + "grad_norm": 2.379281759262085, + "learning_rate": 0.00011118646001280799, + "loss": 3.1441, + "step": 3142 + }, + { + "epoch": 4.6, + "grad_norm": 1.6748161315917969, + "learning_rate": 0.00011113984973185268, + "loss": 3.1687, + "step": 3143 + }, + { + "epoch": 4.61, + "grad_norm": 4.564174652099609, + "learning_rate": 0.00011109323700020508, + "loss": 3.1976, + "step": 3144 + }, + { + "epoch": 4.61, + "grad_norm": 4.5107550621032715, + "learning_rate": 0.00011104662182811974, + "loss": 3.1711, + "step": 3145 + }, + { + "epoch": 4.61, + "grad_norm": 1.1205596923828125, + "learning_rate": 0.00011100000422585158, + "loss": 3.1573, + "step": 3146 + }, + { + "epoch": 4.61, + "grad_norm": 2.339171886444092, + "learning_rate": 0.00011095338420365621, + "loss": 3.1498, + "step": 3147 + }, + { + "epoch": 4.61, + "grad_norm": 2.139634609222412, + "learning_rate": 0.00011090676177178973, + "loss": 3.1493, + "step": 3148 + }, + { + "epoch": 4.61, + "grad_norm": 0.9772034287452698, + "learning_rate": 0.00011086013694050876, + "loss": 3.1551, + "step": 3149 + }, + { + "epoch": 4.61, + "grad_norm": 3.0146749019622803, + "learning_rate": 0.00011081350972007043, + "loss": 3.1186, + "step": 3150 + }, + { + "epoch": 4.62, + "grad_norm": 3.702209711074829, + "learning_rate": 0.00011076688012073244, + "loss": 3.2039, + "step": 3151 + }, + { + "epoch": 4.62, + "grad_norm": 6.443127155303955, + "learning_rate": 0.000110720248152753, + "loss": 3.1794, + "step": 3152 + }, + { + "epoch": 4.62, + "grad_norm": 3.7430996894836426, + "learning_rate": 0.00011067361382639079, + "loss": 3.172, + "step": 3153 + }, + { + "epoch": 4.62, + "grad_norm": 0.7498925924301147, + "learning_rate": 0.00011062697715190507, + "loss": 3.1905, + "step": 3154 + }, + { + "epoch": 4.62, + "grad_norm": 5.21573543548584, + "learning_rate": 0.00011058033813955557, + "loss": 3.1483, + "step": 3155 + }, + { + "epoch": 4.62, + "grad_norm": 5.235692977905273, + "learning_rate": 0.00011053369679960264, + "loss": 3.1197, + "step": 3156 + }, + { + "epoch": 4.62, + "grad_norm": 2.265049934387207, + "learning_rate": 0.00011048705314230696, + "loss": 3.1359, + "step": 3157 + }, + { + "epoch": 4.63, + "grad_norm": 6.700308799743652, + "learning_rate": 0.00011044040717792989, + "loss": 3.141, + "step": 3158 + }, + { + "epoch": 4.63, + "grad_norm": 7.944544792175293, + "learning_rate": 0.00011039375891673322, + "loss": 3.2029, + "step": 3159 + }, + { + "epoch": 4.63, + "grad_norm": 3.51708984375, + "learning_rate": 0.00011034710836897921, + "loss": 3.1747, + "step": 3160 + }, + { + "epoch": 4.63, + "grad_norm": 4.683743000030518, + "learning_rate": 0.00011030045554493075, + "loss": 3.181, + "step": 3161 + }, + { + "epoch": 4.63, + "grad_norm": 7.184266090393066, + "learning_rate": 0.00011025380045485107, + "loss": 3.1523, + "step": 3162 + }, + { + "epoch": 4.63, + "grad_norm": 5.593520164489746, + "learning_rate": 0.00011020714310900401, + "loss": 3.1564, + "step": 3163 + }, + { + "epoch": 4.64, + "grad_norm": 1.7525691986083984, + "learning_rate": 0.0001101604835176539, + "loss": 3.1461, + "step": 3164 + }, + { + "epoch": 4.64, + "grad_norm": 6.076540470123291, + "learning_rate": 0.00011011382169106546, + "loss": 3.1734, + "step": 3165 + }, + { + "epoch": 4.64, + "grad_norm": 7.846095561981201, + "learning_rate": 0.00011006715763950406, + "loss": 3.1622, + "step": 3166 + }, + { + "epoch": 4.64, + "grad_norm": 4.10769510269165, + "learning_rate": 0.00011002049137323543, + "loss": 3.1823, + "step": 3167 + }, + { + "epoch": 4.64, + "grad_norm": 4.750815391540527, + "learning_rate": 0.00010997382290252585, + "loss": 3.1425, + "step": 3168 + }, + { + "epoch": 4.64, + "grad_norm": 4.68991231918335, + "learning_rate": 0.0001099271522376421, + "loss": 3.1231, + "step": 3169 + }, + { + "epoch": 4.64, + "grad_norm": 3.5092368125915527, + "learning_rate": 0.00010988047938885135, + "loss": 3.2153, + "step": 3170 + }, + { + "epoch": 4.65, + "grad_norm": 4.325351715087891, + "learning_rate": 0.00010983380436642136, + "loss": 3.1567, + "step": 3171 + }, + { + "epoch": 4.65, + "grad_norm": 3.288820505142212, + "learning_rate": 0.00010978712718062036, + "loss": 3.1142, + "step": 3172 + }, + { + "epoch": 4.65, + "grad_norm": 3.504354238510132, + "learning_rate": 0.00010974044784171693, + "loss": 3.143, + "step": 3173 + }, + { + "epoch": 4.65, + "grad_norm": 4.500937461853027, + "learning_rate": 0.00010969376635998023, + "loss": 3.1615, + "step": 3174 + }, + { + "epoch": 4.65, + "grad_norm": 39.75580596923828, + "learning_rate": 0.00010964708274567994, + "loss": 3.2307, + "step": 3175 + }, + { + "epoch": 4.65, + "grad_norm": 3.0470621585845947, + "learning_rate": 0.00010960039700908608, + "loss": 3.1374, + "step": 3176 + }, + { + "epoch": 4.65, + "grad_norm": 2.1173510551452637, + "learning_rate": 0.00010955370916046918, + "loss": 3.1752, + "step": 3177 + }, + { + "epoch": 4.66, + "grad_norm": 2.3193938732147217, + "learning_rate": 0.00010950701921010036, + "loss": 3.1474, + "step": 3178 + }, + { + "epoch": 4.66, + "grad_norm": 2.5212008953094482, + "learning_rate": 0.00010946032716825099, + "loss": 3.2541, + "step": 3179 + }, + { + "epoch": 4.66, + "grad_norm": 1.1363469362258911, + "learning_rate": 0.00010941363304519305, + "loss": 3.1424, + "step": 3180 + }, + { + "epoch": 4.66, + "grad_norm": 2.563216209411621, + "learning_rate": 0.00010936693685119892, + "loss": 3.1572, + "step": 3181 + }, + { + "epoch": 4.66, + "grad_norm": 1.520682454109192, + "learning_rate": 0.00010932023859654145, + "loss": 3.1507, + "step": 3182 + }, + { + "epoch": 4.66, + "grad_norm": 3.5463900566101074, + "learning_rate": 0.00010927353829149398, + "loss": 3.1672, + "step": 3183 + }, + { + "epoch": 4.66, + "grad_norm": 4.035703659057617, + "learning_rate": 0.00010922683594633021, + "loss": 3.2053, + "step": 3184 + }, + { + "epoch": 4.67, + "grad_norm": 3.0087568759918213, + "learning_rate": 0.00010918013157132436, + "loss": 3.1566, + "step": 3185 + }, + { + "epoch": 4.67, + "grad_norm": 2.657742500305176, + "learning_rate": 0.00010913342517675113, + "loss": 3.1382, + "step": 3186 + }, + { + "epoch": 4.67, + "grad_norm": 4.097416877746582, + "learning_rate": 0.00010908671677288552, + "loss": 3.1695, + "step": 3187 + }, + { + "epoch": 4.67, + "grad_norm": 3.2829132080078125, + "learning_rate": 0.00010904000637000313, + "loss": 3.2016, + "step": 3188 + }, + { + "epoch": 4.67, + "grad_norm": 0.8120042681694031, + "learning_rate": 0.00010899329397837996, + "loss": 3.1287, + "step": 3189 + }, + { + "epoch": 4.67, + "grad_norm": 1.4076851606369019, + "learning_rate": 0.00010894657960829235, + "loss": 3.1302, + "step": 3190 + }, + { + "epoch": 4.68, + "grad_norm": 2.486541986465454, + "learning_rate": 0.00010889986327001716, + "loss": 3.1751, + "step": 3191 + }, + { + "epoch": 4.68, + "grad_norm": 1.665196418762207, + "learning_rate": 0.00010885314497383175, + "loss": 3.1353, + "step": 3192 + }, + { + "epoch": 4.68, + "grad_norm": 3.312460422515869, + "learning_rate": 0.00010880642473001375, + "loss": 3.1851, + "step": 3193 + }, + { + "epoch": 4.68, + "grad_norm": 6.796298027038574, + "learning_rate": 0.0001087597025488413, + "loss": 3.1513, + "step": 3194 + }, + { + "epoch": 4.68, + "grad_norm": 1.5955018997192383, + "learning_rate": 0.00010871297844059304, + "loss": 3.1609, + "step": 3195 + }, + { + "epoch": 4.68, + "grad_norm": 4.358046531677246, + "learning_rate": 0.00010866625241554789, + "loss": 3.1224, + "step": 3196 + }, + { + "epoch": 4.68, + "grad_norm": 4.621119976043701, + "learning_rate": 0.00010861952448398529, + "loss": 3.1221, + "step": 3197 + }, + { + "epoch": 4.69, + "grad_norm": 1.6193621158599854, + "learning_rate": 0.00010857279465618508, + "loss": 3.1495, + "step": 3198 + }, + { + "epoch": 4.69, + "grad_norm": 3.5418941974639893, + "learning_rate": 0.00010852606294242749, + "loss": 3.182, + "step": 3199 + }, + { + "epoch": 4.69, + "grad_norm": 3.8759021759033203, + "learning_rate": 0.00010847932935299319, + "loss": 3.1323, + "step": 3200 + }, + { + "epoch": 4.69, + "grad_norm": 1.662946343421936, + "learning_rate": 0.00010843259389816325, + "loss": 3.1605, + "step": 3201 + }, + { + "epoch": 4.69, + "grad_norm": 3.667335271835327, + "learning_rate": 0.00010838585658821918, + "loss": 3.1713, + "step": 3202 + }, + { + "epoch": 4.69, + "grad_norm": 4.988526344299316, + "learning_rate": 0.00010833911743344287, + "loss": 3.1812, + "step": 3203 + }, + { + "epoch": 4.69, + "grad_norm": 3.444049596786499, + "learning_rate": 0.00010829237644411659, + "loss": 3.1978, + "step": 3204 + }, + { + "epoch": 4.7, + "grad_norm": 2.074106216430664, + "learning_rate": 0.00010824563363052307, + "loss": 3.1567, + "step": 3205 + }, + { + "epoch": 4.7, + "grad_norm": 5.682570934295654, + "learning_rate": 0.00010819888900294542, + "loss": 3.1777, + "step": 3206 + }, + { + "epoch": 4.7, + "grad_norm": 7.264477729797363, + "learning_rate": 0.00010815214257166712, + "loss": 3.2065, + "step": 3207 + }, + { + "epoch": 4.7, + "grad_norm": 4.073165416717529, + "learning_rate": 0.0001081053943469721, + "loss": 3.1564, + "step": 3208 + }, + { + "epoch": 4.7, + "grad_norm": 3.7732391357421875, + "learning_rate": 0.00010805864433914465, + "loss": 3.1654, + "step": 3209 + }, + { + "epoch": 4.7, + "grad_norm": 4.149001598358154, + "learning_rate": 0.00010801189255846944, + "loss": 3.1563, + "step": 3210 + }, + { + "epoch": 4.71, + "grad_norm": 4.86202335357666, + "learning_rate": 0.00010796513901523155, + "loss": 3.1624, + "step": 3211 + }, + { + "epoch": 4.71, + "eval_loss": 3.0107593536376953, + "eval_runtime": 1169.2042, + "eval_samples_per_second": 14.661, + "eval_steps_per_second": 0.917, + "step": 3211 + }, + { + "epoch": 4.71, + "grad_norm": 4.124433994293213, + "learning_rate": 0.00010791838371971648, + "loss": 3.1329, + "step": 3212 + }, + { + "epoch": 4.71, + "grad_norm": 1.8730932474136353, + "learning_rate": 0.00010787162668221004, + "loss": 3.1344, + "step": 3213 + }, + { + "epoch": 4.71, + "grad_norm": 3.890324831008911, + "learning_rate": 0.00010782486791299849, + "loss": 3.1527, + "step": 3214 + }, + { + "epoch": 4.71, + "grad_norm": 3.587141513824463, + "learning_rate": 0.00010777810742236844, + "loss": 3.1435, + "step": 3215 + }, + { + "epoch": 4.71, + "grad_norm": 1.505163311958313, + "learning_rate": 0.00010773134522060688, + "loss": 3.1474, + "step": 3216 + }, + { + "epoch": 4.71, + "grad_norm": 3.0347487926483154, + "learning_rate": 0.00010768458131800119, + "loss": 3.1363, + "step": 3217 + }, + { + "epoch": 4.72, + "grad_norm": 3.127628803253174, + "learning_rate": 0.0001076378157248391, + "loss": 3.118, + "step": 3218 + }, + { + "epoch": 4.72, + "grad_norm": 2.165717363357544, + "learning_rate": 0.00010759104845140877, + "loss": 3.1966, + "step": 3219 + }, + { + "epoch": 4.72, + "grad_norm": 2.3362936973571777, + "learning_rate": 0.00010754427950799861, + "loss": 3.1494, + "step": 3220 + }, + { + "epoch": 4.72, + "grad_norm": 2.43580961227417, + "learning_rate": 0.00010749750890489754, + "loss": 3.1281, + "step": 3221 + }, + { + "epoch": 4.72, + "grad_norm": 23.91058349609375, + "learning_rate": 0.00010745073665239476, + "loss": 3.1482, + "step": 3222 + }, + { + "epoch": 4.72, + "grad_norm": 3.7118659019470215, + "learning_rate": 0.00010740396276077985, + "loss": 3.1884, + "step": 3223 + }, + { + "epoch": 4.72, + "grad_norm": 3.4350130558013916, + "learning_rate": 0.00010735718724034273, + "loss": 3.1709, + "step": 3224 + }, + { + "epoch": 4.73, + "grad_norm": 0.7842398285865784, + "learning_rate": 0.00010731041010137375, + "loss": 3.162, + "step": 3225 + }, + { + "epoch": 4.73, + "grad_norm": 4.245161533355713, + "learning_rate": 0.00010726363135416351, + "loss": 3.1882, + "step": 3226 + }, + { + "epoch": 4.73, + "grad_norm": 6.0483622550964355, + "learning_rate": 0.00010721685100900308, + "loss": 3.1815, + "step": 3227 + }, + { + "epoch": 4.73, + "grad_norm": 4.542155742645264, + "learning_rate": 0.00010717006907618377, + "loss": 3.1413, + "step": 3228 + }, + { + "epoch": 4.73, + "grad_norm": 1.648128867149353, + "learning_rate": 0.00010712328556599733, + "loss": 3.1296, + "step": 3229 + }, + { + "epoch": 4.73, + "grad_norm": 5.657898902893066, + "learning_rate": 0.00010707650048873579, + "loss": 3.178, + "step": 3230 + }, + { + "epoch": 4.73, + "grad_norm": 5.276024341583252, + "learning_rate": 0.00010702971385469158, + "loss": 3.192, + "step": 3231 + }, + { + "epoch": 4.74, + "grad_norm": 4.145871639251709, + "learning_rate": 0.00010698292567415742, + "loss": 3.134, + "step": 3232 + }, + { + "epoch": 4.74, + "grad_norm": 12.922083854675293, + "learning_rate": 0.00010693613595742644, + "loss": 3.1523, + "step": 3233 + }, + { + "epoch": 4.74, + "grad_norm": 3.349857807159424, + "learning_rate": 0.00010688934471479199, + "loss": 3.1872, + "step": 3234 + }, + { + "epoch": 4.74, + "grad_norm": 2.6229217052459717, + "learning_rate": 0.00010684255195654788, + "loss": 3.1266, + "step": 3235 + }, + { + "epoch": 4.74, + "grad_norm": 3.148838520050049, + "learning_rate": 0.00010679575769298824, + "loss": 3.2267, + "step": 3236 + }, + { + "epoch": 4.74, + "grad_norm": 3.843184471130371, + "learning_rate": 0.00010674896193440742, + "loss": 3.1531, + "step": 3237 + }, + { + "epoch": 4.75, + "grad_norm": 2.107234239578247, + "learning_rate": 0.00010670216469110019, + "loss": 3.1549, + "step": 3238 + }, + { + "epoch": 4.75, + "grad_norm": 1.9575111865997314, + "learning_rate": 0.00010665536597336167, + "loss": 3.1334, + "step": 3239 + }, + { + "epoch": 4.75, + "grad_norm": 2.4358248710632324, + "learning_rate": 0.00010660856579148724, + "loss": 3.1869, + "step": 3240 + }, + { + "epoch": 4.75, + "grad_norm": 1.9233509302139282, + "learning_rate": 0.00010656176415577262, + "loss": 3.1206, + "step": 3241 + }, + { + "epoch": 4.75, + "grad_norm": 2.0048117637634277, + "learning_rate": 0.00010651496107651384, + "loss": 3.1099, + "step": 3242 + }, + { + "epoch": 4.75, + "grad_norm": 4.973776817321777, + "learning_rate": 0.00010646815656400731, + "loss": 3.1621, + "step": 3243 + }, + { + "epoch": 4.75, + "grad_norm": 3.1254444122314453, + "learning_rate": 0.00010642135062854965, + "loss": 3.1564, + "step": 3244 + }, + { + "epoch": 4.76, + "grad_norm": 2.404599666595459, + "learning_rate": 0.00010637454328043793, + "loss": 3.1772, + "step": 3245 + }, + { + "epoch": 4.76, + "grad_norm": 2.372962474822998, + "learning_rate": 0.00010632773452996936, + "loss": 3.1053, + "step": 3246 + }, + { + "epoch": 4.76, + "grad_norm": 1.7531036138534546, + "learning_rate": 0.00010628092438744167, + "loss": 3.1678, + "step": 3247 + }, + { + "epoch": 4.76, + "grad_norm": 15.43344783782959, + "learning_rate": 0.00010623411286315261, + "loss": 3.1616, + "step": 3248 + }, + { + "epoch": 4.76, + "grad_norm": 3.320650577545166, + "learning_rate": 0.00010618729996740056, + "loss": 3.2037, + "step": 3249 + }, + { + "epoch": 4.76, + "grad_norm": 14.557812690734863, + "learning_rate": 0.00010614048571048401, + "loss": 3.221, + "step": 3250 + }, + { + "epoch": 4.76, + "grad_norm": 1.5140577554702759, + "learning_rate": 0.00010609367010270168, + "loss": 3.1666, + "step": 3251 + }, + { + "epoch": 4.77, + "grad_norm": 2.14962100982666, + "learning_rate": 0.00010604685315435281, + "loss": 3.1691, + "step": 3252 + }, + { + "epoch": 4.77, + "grad_norm": 0.9101262092590332, + "learning_rate": 0.0001060000348757368, + "loss": 3.1835, + "step": 3253 + }, + { + "epoch": 4.77, + "grad_norm": 1.9691673517227173, + "learning_rate": 0.00010595321527715327, + "loss": 3.1425, + "step": 3254 + }, + { + "epoch": 4.77, + "grad_norm": 6.10009241104126, + "learning_rate": 0.0001059063943689023, + "loss": 3.163, + "step": 3255 + }, + { + "epoch": 4.77, + "grad_norm": 1.81471848487854, + "learning_rate": 0.00010585957216128415, + "loss": 3.1742, + "step": 3256 + }, + { + "epoch": 4.77, + "grad_norm": 1.8879374265670776, + "learning_rate": 0.0001058127486645994, + "loss": 3.1732, + "step": 3257 + }, + { + "epoch": 4.77, + "grad_norm": 0.9328217506408691, + "learning_rate": 0.0001057659238891489, + "loss": 3.1223, + "step": 3258 + }, + { + "epoch": 4.78, + "grad_norm": 1.236288070678711, + "learning_rate": 0.00010571909784523377, + "loss": 3.114, + "step": 3259 + }, + { + "epoch": 4.78, + "grad_norm": 2.6463773250579834, + "learning_rate": 0.00010567227054315545, + "loss": 3.1444, + "step": 3260 + }, + { + "epoch": 4.78, + "grad_norm": 1.5447609424591064, + "learning_rate": 0.00010562544199321562, + "loss": 3.1348, + "step": 3261 + }, + { + "epoch": 4.78, + "grad_norm": 3.4742603302001953, + "learning_rate": 0.00010557861220571625, + "loss": 3.1773, + "step": 3262 + }, + { + "epoch": 4.78, + "grad_norm": 2.8108527660369873, + "learning_rate": 0.0001055317811909596, + "loss": 3.1871, + "step": 3263 + }, + { + "epoch": 4.78, + "grad_norm": 2.6389920711517334, + "learning_rate": 0.00010548494895924816, + "loss": 3.1609, + "step": 3264 + }, + { + "epoch": 4.79, + "grad_norm": 2.764688491821289, + "learning_rate": 0.00010543811552088468, + "loss": 3.1937, + "step": 3265 + }, + { + "epoch": 4.79, + "grad_norm": 3.0502400398254395, + "learning_rate": 0.00010539128088617225, + "loss": 3.1669, + "step": 3266 + }, + { + "epoch": 4.79, + "grad_norm": 6.166628837585449, + "learning_rate": 0.00010534444506541417, + "loss": 3.142, + "step": 3267 + }, + { + "epoch": 4.79, + "grad_norm": 47.169883728027344, + "learning_rate": 0.00010529760806891395, + "loss": 3.1328, + "step": 3268 + }, + { + "epoch": 4.79, + "grad_norm": 3.0846803188323975, + "learning_rate": 0.0001052507699069755, + "loss": 3.1455, + "step": 3269 + }, + { + "epoch": 4.79, + "grad_norm": 2.2564151287078857, + "learning_rate": 0.00010520393058990285, + "loss": 3.1306, + "step": 3270 + }, + { + "epoch": 4.79, + "grad_norm": 1.4548720121383667, + "learning_rate": 0.00010515709012800033, + "loss": 3.1312, + "step": 3271 + }, + { + "epoch": 4.8, + "grad_norm": 1.8430453538894653, + "learning_rate": 0.00010511024853157255, + "loss": 3.1276, + "step": 3272 + }, + { + "epoch": 4.8, + "grad_norm": 1.2610652446746826, + "learning_rate": 0.00010506340581092433, + "loss": 3.1682, + "step": 3273 + }, + { + "epoch": 4.8, + "grad_norm": 0.8417081832885742, + "learning_rate": 0.00010501656197636078, + "loss": 3.1505, + "step": 3274 + }, + { + "epoch": 4.8, + "grad_norm": 1.3054739236831665, + "learning_rate": 0.00010496971703818721, + "loss": 3.1465, + "step": 3275 + }, + { + "epoch": 4.8, + "grad_norm": 2.4543936252593994, + "learning_rate": 0.00010492287100670919, + "loss": 3.2098, + "step": 3276 + }, + { + "epoch": 4.8, + "grad_norm": 1.5280994176864624, + "learning_rate": 0.00010487602389223257, + "loss": 3.1443, + "step": 3277 + }, + { + "epoch": 4.8, + "grad_norm": 1.9430744647979736, + "learning_rate": 0.00010482917570506334, + "loss": 3.1531, + "step": 3278 + }, + { + "epoch": 4.81, + "grad_norm": 1.8071426153182983, + "learning_rate": 0.00010478232645550782, + "loss": 3.171, + "step": 3279 + }, + { + "epoch": 4.81, + "grad_norm": 3.768815517425537, + "learning_rate": 0.00010473547615387256, + "loss": 3.1586, + "step": 3280 + }, + { + "epoch": 4.81, + "grad_norm": 2.8686270713806152, + "learning_rate": 0.00010468862481046427, + "loss": 3.112, + "step": 3281 + }, + { + "epoch": 4.81, + "grad_norm": 2.201878070831299, + "learning_rate": 0.0001046417724355899, + "loss": 3.1815, + "step": 3282 + }, + { + "epoch": 4.81, + "grad_norm": 1.5182592868804932, + "learning_rate": 0.00010459491903955676, + "loss": 3.1391, + "step": 3283 + }, + { + "epoch": 4.81, + "grad_norm": 23.839717864990234, + "learning_rate": 0.0001045480646326722, + "loss": 3.1331, + "step": 3284 + }, + { + "epoch": 4.81, + "grad_norm": 1.8520891666412354, + "learning_rate": 0.00010450120922524387, + "loss": 3.1622, + "step": 3285 + }, + { + "epoch": 4.82, + "grad_norm": 0.8244103789329529, + "learning_rate": 0.00010445435282757972, + "loss": 3.1308, + "step": 3286 + }, + { + "epoch": 4.82, + "grad_norm": 2.278554677963257, + "learning_rate": 0.00010440749544998778, + "loss": 3.1283, + "step": 3287 + }, + { + "epoch": 4.82, + "grad_norm": 3.664557933807373, + "learning_rate": 0.00010436063710277639, + "loss": 3.1424, + "step": 3288 + }, + { + "epoch": 4.82, + "grad_norm": 1.274968147277832, + "learning_rate": 0.00010431377779625407, + "loss": 3.1358, + "step": 3289 + }, + { + "epoch": 4.82, + "grad_norm": 0.9037652611732483, + "learning_rate": 0.00010426691754072952, + "loss": 3.1329, + "step": 3290 + }, + { + "epoch": 4.82, + "grad_norm": 2.134868860244751, + "learning_rate": 0.00010422005634651173, + "loss": 3.0978, + "step": 3291 + }, + { + "epoch": 4.83, + "grad_norm": 4.81277322769165, + "learning_rate": 0.00010417319422390979, + "loss": 3.118, + "step": 3292 + }, + { + "epoch": 4.83, + "grad_norm": 1.5502432584762573, + "learning_rate": 0.00010412633118323311, + "loss": 3.1258, + "step": 3293 + }, + { + "epoch": 4.83, + "grad_norm": 1.0471981763839722, + "learning_rate": 0.00010407946723479125, + "loss": 3.1163, + "step": 3294 + }, + { + "epoch": 4.83, + "grad_norm": 2.2549943923950195, + "learning_rate": 0.00010403260238889389, + "loss": 3.1441, + "step": 3295 + }, + { + "epoch": 4.83, + "grad_norm": 1.7782256603240967, + "learning_rate": 0.00010398573665585105, + "loss": 3.1251, + "step": 3296 + }, + { + "epoch": 4.83, + "grad_norm": 3.3199939727783203, + "learning_rate": 0.00010393887004597287, + "loss": 3.1731, + "step": 3297 + }, + { + "epoch": 4.83, + "grad_norm": 3.324904441833496, + "learning_rate": 0.00010389200256956968, + "loss": 3.1439, + "step": 3298 + }, + { + "epoch": 4.84, + "grad_norm": 2.5700957775115967, + "learning_rate": 0.00010384513423695196, + "loss": 3.138, + "step": 3299 + }, + { + "epoch": 4.84, + "grad_norm": 3.218172550201416, + "learning_rate": 0.00010379826505843054, + "loss": 3.1335, + "step": 3300 + }, + { + "epoch": 4.84, + "grad_norm": 1.4542341232299805, + "learning_rate": 0.00010375139504431622, + "loss": 3.1655, + "step": 3301 + }, + { + "epoch": 4.84, + "grad_norm": 21.637861251831055, + "learning_rate": 0.00010370452420492012, + "loss": 3.1785, + "step": 3302 + }, + { + "epoch": 4.84, + "grad_norm": 3.2379684448242188, + "learning_rate": 0.00010365765255055358, + "loss": 3.133, + "step": 3303 + }, + { + "epoch": 4.84, + "grad_norm": 1.356064796447754, + "learning_rate": 0.00010361078009152793, + "loss": 3.1157, + "step": 3304 + }, + { + "epoch": 4.84, + "grad_norm": 1.2905539274215698, + "learning_rate": 0.00010356390683815489, + "loss": 3.1529, + "step": 3305 + }, + { + "epoch": 4.85, + "grad_norm": 2.5484352111816406, + "learning_rate": 0.00010351703280074623, + "loss": 3.1459, + "step": 3306 + }, + { + "epoch": 4.85, + "grad_norm": 1.1913775205612183, + "learning_rate": 0.0001034701579896139, + "loss": 3.2115, + "step": 3307 + }, + { + "epoch": 4.85, + "grad_norm": 1.7160303592681885, + "learning_rate": 0.0001034232824150701, + "loss": 3.1572, + "step": 3308 + }, + { + "epoch": 4.85, + "grad_norm": 2.059537887573242, + "learning_rate": 0.0001033764060874271, + "loss": 3.1361, + "step": 3309 + }, + { + "epoch": 4.85, + "grad_norm": 1.258538007736206, + "learning_rate": 0.00010332952901699738, + "loss": 3.1483, + "step": 3310 + }, + { + "epoch": 4.85, + "grad_norm": 2.301053285598755, + "learning_rate": 0.00010328265121409361, + "loss": 3.214, + "step": 3311 + }, + { + "epoch": 4.86, + "grad_norm": 1.366959810256958, + "learning_rate": 0.00010323577268902853, + "loss": 3.1272, + "step": 3312 + }, + { + "epoch": 4.86, + "grad_norm": 1.7083238363265991, + "learning_rate": 0.00010318889345211518, + "loss": 3.1432, + "step": 3313 + }, + { + "epoch": 4.86, + "grad_norm": 1.0292158126831055, + "learning_rate": 0.00010314201351366666, + "loss": 3.16, + "step": 3314 + }, + { + "epoch": 4.86, + "grad_norm": 1.8825881481170654, + "learning_rate": 0.0001030951328839962, + "loss": 3.1021, + "step": 3315 + }, + { + "epoch": 4.86, + "grad_norm": 1.2715975046157837, + "learning_rate": 0.00010304825157341722, + "loss": 3.1305, + "step": 3316 + }, + { + "epoch": 4.86, + "grad_norm": 1.972051978111267, + "learning_rate": 0.00010300136959224335, + "loss": 3.1384, + "step": 3317 + }, + { + "epoch": 4.86, + "grad_norm": 2.031904697418213, + "learning_rate": 0.00010295448695078829, + "loss": 3.1444, + "step": 3318 + }, + { + "epoch": 4.87, + "grad_norm": 1.3081032037734985, + "learning_rate": 0.00010290760365936587, + "loss": 3.1418, + "step": 3319 + }, + { + "epoch": 4.87, + "grad_norm": 1.2461774349212646, + "learning_rate": 0.00010286071972829016, + "loss": 3.1776, + "step": 3320 + }, + { + "epoch": 4.87, + "grad_norm": 614.052978515625, + "learning_rate": 0.00010281383516787528, + "loss": 3.162, + "step": 3321 + }, + { + "epoch": 4.87, + "grad_norm": 16.35662269592285, + "learning_rate": 0.0001027669499884355, + "loss": 3.1498, + "step": 3322 + }, + { + "epoch": 4.87, + "grad_norm": 5.687806606292725, + "learning_rate": 0.00010272006420028528, + "loss": 3.1927, + "step": 3323 + }, + { + "epoch": 4.87, + "grad_norm": 4.659135341644287, + "learning_rate": 0.00010267317781373918, + "loss": 3.1678, + "step": 3324 + }, + { + "epoch": 4.87, + "grad_norm": 6.275466442108154, + "learning_rate": 0.00010262629083911186, + "loss": 3.1716, + "step": 3325 + }, + { + "epoch": 4.88, + "grad_norm": 3.4971659183502197, + "learning_rate": 0.00010257940328671813, + "loss": 3.1567, + "step": 3326 + }, + { + "epoch": 4.88, + "grad_norm": 4.331486225128174, + "learning_rate": 0.00010253251516687303, + "loss": 3.164, + "step": 3327 + }, + { + "epoch": 4.88, + "grad_norm": 2.695127010345459, + "learning_rate": 0.00010248562648989153, + "loss": 3.1222, + "step": 3328 + }, + { + "epoch": 4.88, + "grad_norm": 2.059220314025879, + "learning_rate": 0.00010243873726608881, + "loss": 3.1783, + "step": 3329 + }, + { + "epoch": 4.88, + "grad_norm": 2.7665822505950928, + "learning_rate": 0.0001023918475057803, + "loss": 3.1525, + "step": 3330 + }, + { + "epoch": 4.88, + "grad_norm": 1.4520095586776733, + "learning_rate": 0.00010234495721928137, + "loss": 3.1559, + "step": 3331 + }, + { + "epoch": 4.88, + "grad_norm": 3.7544808387756348, + "learning_rate": 0.00010229806641690754, + "loss": 3.196, + "step": 3332 + }, + { + "epoch": 4.89, + "grad_norm": 4.363980770111084, + "learning_rate": 0.0001022511751089745, + "loss": 3.1835, + "step": 3333 + }, + { + "epoch": 4.89, + "grad_norm": 2.252725124359131, + "learning_rate": 0.00010220428330579804, + "loss": 3.1515, + "step": 3334 + }, + { + "epoch": 4.89, + "grad_norm": 2.367949962615967, + "learning_rate": 0.00010215739101769399, + "loss": 3.1599, + "step": 3335 + }, + { + "epoch": 4.89, + "grad_norm": 2.9788756370544434, + "learning_rate": 0.00010211049825497841, + "loss": 3.147, + "step": 3336 + }, + { + "epoch": 4.89, + "grad_norm": 1.9200280904769897, + "learning_rate": 0.00010206360502796734, + "loss": 3.1723, + "step": 3337 + }, + { + "epoch": 4.89, + "grad_norm": 2.547455072402954, + "learning_rate": 0.00010201671134697699, + "loss": 3.1607, + "step": 3338 + }, + { + "epoch": 4.9, + "grad_norm": 3.103799343109131, + "learning_rate": 0.00010196981722232368, + "loss": 3.1501, + "step": 3339 + }, + { + "epoch": 4.9, + "grad_norm": 0.7216553092002869, + "learning_rate": 0.00010192292266432377, + "loss": 3.1397, + "step": 3340 + }, + { + "epoch": 4.9, + "grad_norm": 3.535027265548706, + "learning_rate": 0.00010187602768329378, + "loss": 3.1029, + "step": 3341 + }, + { + "epoch": 4.9, + "grad_norm": 3.598963975906372, + "learning_rate": 0.00010182913228955025, + "loss": 3.1537, + "step": 3342 + }, + { + "epoch": 4.9, + "grad_norm": 1.5128989219665527, + "learning_rate": 0.00010178223649340985, + "loss": 3.1342, + "step": 3343 + }, + { + "epoch": 4.9, + "grad_norm": 1.6262668371200562, + "learning_rate": 0.00010173534030518941, + "loss": 3.1654, + "step": 3344 + }, + { + "epoch": 4.9, + "grad_norm": 1.0745606422424316, + "learning_rate": 0.00010168844373520572, + "loss": 3.1105, + "step": 3345 + }, + { + "epoch": 4.91, + "grad_norm": 1.7129157781600952, + "learning_rate": 0.00010164154679377569, + "loss": 3.1317, + "step": 3346 + }, + { + "epoch": 4.91, + "grad_norm": 1.3464902639389038, + "learning_rate": 0.00010159464949121641, + "loss": 3.1503, + "step": 3347 + }, + { + "epoch": 4.91, + "grad_norm": 8.157245635986328, + "learning_rate": 0.00010154775183784493, + "loss": 3.1043, + "step": 3348 + }, + { + "epoch": 4.91, + "grad_norm": 4.697030544281006, + "learning_rate": 0.0001015008538439784, + "loss": 3.194, + "step": 3349 + }, + { + "epoch": 4.91, + "grad_norm": 34.21104049682617, + "learning_rate": 0.00010145395551993408, + "loss": 3.171, + "step": 3350 + }, + { + "epoch": 4.91, + "grad_norm": 3.0200319290161133, + "learning_rate": 0.00010140705687602929, + "loss": 3.1502, + "step": 3351 + }, + { + "epoch": 4.91, + "grad_norm": 0.9504959583282471, + "learning_rate": 0.00010136015792258142, + "loss": 3.1142, + "step": 3352 + }, + { + "epoch": 4.92, + "grad_norm": 2.7994866371154785, + "learning_rate": 0.00010131325866990793, + "loss": 3.1242, + "step": 3353 + }, + { + "epoch": 4.92, + "grad_norm": 2.5278077125549316, + "learning_rate": 0.00010126635912832631, + "loss": 3.0951, + "step": 3354 + }, + { + "epoch": 4.92, + "grad_norm": 3.453552007675171, + "learning_rate": 0.00010121945930815422, + "loss": 3.1458, + "step": 3355 + }, + { + "epoch": 4.92, + "grad_norm": 3.2152633666992188, + "learning_rate": 0.00010117255921970921, + "loss": 3.1589, + "step": 3356 + }, + { + "epoch": 4.92, + "grad_norm": 1.388268232345581, + "learning_rate": 0.00010112565887330903, + "loss": 3.1496, + "step": 3357 + }, + { + "epoch": 4.92, + "grad_norm": 3.23463773727417, + "learning_rate": 0.00010107875827927147, + "loss": 3.1184, + "step": 3358 + }, + { + "epoch": 4.92, + "grad_norm": 2.970151662826538, + "learning_rate": 0.00010103185744791431, + "loss": 3.0931, + "step": 3359 + }, + { + "epoch": 4.93, + "grad_norm": 7.270925998687744, + "learning_rate": 0.00010098495638955538, + "loss": 3.1356, + "step": 3360 + }, + { + "epoch": 4.93, + "grad_norm": 18.72118377685547, + "learning_rate": 0.00010093805511451271, + "loss": 3.1538, + "step": 3361 + }, + { + "epoch": 4.93, + "grad_norm": 52.44887161254883, + "learning_rate": 0.00010089115363310417, + "loss": 3.1924, + "step": 3362 + }, + { + "epoch": 4.93, + "grad_norm": 7.027931213378906, + "learning_rate": 0.00010084425195564782, + "loss": 3.1726, + "step": 3363 + }, + { + "epoch": 4.93, + "grad_norm": 17.143966674804688, + "learning_rate": 0.00010079735009246167, + "loss": 3.1921, + "step": 3364 + }, + { + "epoch": 4.93, + "grad_norm": 3.896998167037964, + "learning_rate": 0.00010075044805386386, + "loss": 3.1341, + "step": 3365 + }, + { + "epoch": 4.94, + "grad_norm": 4.068105697631836, + "learning_rate": 0.0001007035458501725, + "loss": 3.1548, + "step": 3366 + }, + { + "epoch": 4.94, + "grad_norm": 3.3968491554260254, + "learning_rate": 0.00010065664349170578, + "loss": 3.146, + "step": 3367 + }, + { + "epoch": 4.94, + "grad_norm": 3.594057083129883, + "learning_rate": 0.00010060974098878189, + "loss": 3.1631, + "step": 3368 + }, + { + "epoch": 4.94, + "grad_norm": 0.987125813961029, + "learning_rate": 0.00010056283835171906, + "loss": 3.1319, + "step": 3369 + }, + { + "epoch": 4.94, + "grad_norm": 4.898683547973633, + "learning_rate": 0.00010051593559083557, + "loss": 3.1316, + "step": 3370 + }, + { + "epoch": 4.94, + "grad_norm": 6.864829063415527, + "learning_rate": 0.0001004690327164497, + "loss": 3.2037, + "step": 3371 + }, + { + "epoch": 4.94, + "grad_norm": 5.136176109313965, + "learning_rate": 0.00010042212973887982, + "loss": 3.1921, + "step": 3372 + }, + { + "epoch": 4.95, + "grad_norm": 2.621640205383301, + "learning_rate": 0.00010037522666844415, + "loss": 3.1716, + "step": 3373 + }, + { + "epoch": 4.95, + "grad_norm": 4.9672746658325195, + "learning_rate": 0.00010032832351546117, + "loss": 3.1374, + "step": 3374 + }, + { + "epoch": 4.95, + "grad_norm": 6.860297203063965, + "learning_rate": 0.00010028142029024923, + "loss": 3.1863, + "step": 3375 + }, + { + "epoch": 4.95, + "grad_norm": 131.51553344726562, + "learning_rate": 0.00010023451700312669, + "loss": 3.1629, + "step": 3376 + }, + { + "epoch": 4.95, + "grad_norm": 4.443822860717773, + "learning_rate": 0.00010018761366441198, + "loss": 3.1513, + "step": 3377 + }, + { + "epoch": 4.95, + "grad_norm": 1.4408992528915405, + "learning_rate": 0.00010014071028442354, + "loss": 3.1317, + "step": 3378 + }, + { + "epoch": 4.95, + "grad_norm": 8.419514656066895, + "learning_rate": 0.00010009380687347975, + "loss": 3.1649, + "step": 3379 + }, + { + "epoch": 4.96, + "grad_norm": 19.037832260131836, + "learning_rate": 0.00010004690344189912, + "loss": 3.1697, + "step": 3380 + }, + { + "epoch": 4.96, + "eval_loss": 3.100813865661621, + "eval_runtime": 1176.1523, + "eval_samples_per_second": 14.575, + "eval_steps_per_second": 0.911, + "step": 3380 + }, + { + "epoch": 4.96, + "grad_norm": 11.52230167388916, + "learning_rate": 0.0001, + "loss": 3.2739, + "step": 3381 + }, + { + "epoch": 4.96, + "grad_norm": 8.755014419555664, + "learning_rate": 9.995309655810093e-05, + "loss": 3.2172, + "step": 3382 + }, + { + "epoch": 4.96, + "grad_norm": 5.496139049530029, + "learning_rate": 9.990619312652026e-05, + "loss": 3.1656, + "step": 3383 + }, + { + "epoch": 4.96, + "grad_norm": 3.737701416015625, + "learning_rate": 9.985928971557649e-05, + "loss": 3.2014, + "step": 3384 + }, + { + "epoch": 4.96, + "grad_norm": 6.020711898803711, + "learning_rate": 9.981238633558805e-05, + "loss": 3.2036, + "step": 3385 + }, + { + "epoch": 4.96, + "grad_norm": 7.874726295471191, + "learning_rate": 9.976548299687332e-05, + "loss": 3.1735, + "step": 3386 + }, + { + "epoch": 4.97, + "grad_norm": 7.075169563293457, + "learning_rate": 9.97185797097508e-05, + "loss": 3.1674, + "step": 3387 + }, + { + "epoch": 4.97, + "grad_norm": 3.235295057296753, + "learning_rate": 9.967167648453885e-05, + "loss": 3.1999, + "step": 3388 + }, + { + "epoch": 4.97, + "grad_norm": 3.599179744720459, + "learning_rate": 9.962477333155585e-05, + "loss": 3.1829, + "step": 3389 + }, + { + "epoch": 4.97, + "grad_norm": 6.6362528800964355, + "learning_rate": 9.95778702611202e-05, + "loss": 3.2165, + "step": 3390 + }, + { + "epoch": 4.97, + "grad_norm": 7.1437153816223145, + "learning_rate": 9.953096728355031e-05, + "loss": 3.1884, + "step": 3391 + }, + { + "epoch": 4.97, + "grad_norm": 4.677809715270996, + "learning_rate": 9.948406440916443e-05, + "loss": 3.1899, + "step": 3392 + }, + { + "epoch": 4.98, + "grad_norm": 1.904357671737671, + "learning_rate": 9.943716164828096e-05, + "loss": 3.1357, + "step": 3393 + }, + { + "epoch": 4.98, + "grad_norm": 5.046501636505127, + "learning_rate": 9.939025901121814e-05, + "loss": 3.1666, + "step": 3394 + }, + { + "epoch": 4.98, + "grad_norm": 6.0771074295043945, + "learning_rate": 9.934335650829425e-05, + "loss": 3.1961, + "step": 3395 + }, + { + "epoch": 4.98, + "grad_norm": 5.269542694091797, + "learning_rate": 9.92964541498275e-05, + "loss": 3.1937, + "step": 3396 + }, + { + "epoch": 4.98, + "grad_norm": 2.9679410457611084, + "learning_rate": 9.924955194613616e-05, + "loss": 3.1376, + "step": 3397 + }, + { + "epoch": 4.98, + "grad_norm": 3.366711139678955, + "learning_rate": 9.920264990753837e-05, + "loss": 3.1116, + "step": 3398 + }, + { + "epoch": 4.98, + "grad_norm": 6.38279390335083, + "learning_rate": 9.91557480443522e-05, + "loss": 3.1385, + "step": 3399 + }, + { + "epoch": 4.99, + "grad_norm": 5.471691608428955, + "learning_rate": 9.910884636689585e-05, + "loss": 3.1874, + "step": 3400 + }, + { + "epoch": 4.99, + "grad_norm": 2.8460099697113037, + "learning_rate": 9.906194488548734e-05, + "loss": 3.1898, + "step": 3401 + }, + { + "epoch": 4.99, + "grad_norm": 3.193862199783325, + "learning_rate": 9.901504361044461e-05, + "loss": 3.1727, + "step": 3402 + }, + { + "epoch": 4.99, + "grad_norm": 4.644516944885254, + "learning_rate": 9.896814255208571e-05, + "loss": 3.1828, + "step": 3403 + }, + { + "epoch": 4.99, + "grad_norm": 4.342918872833252, + "learning_rate": 9.892124172072857e-05, + "loss": 3.1714, + "step": 3404 + }, + { + "epoch": 4.99, + "grad_norm": 2.4274344444274902, + "learning_rate": 9.887434112669097e-05, + "loss": 3.1509, + "step": 3405 + }, + { + "epoch": 4.99, + "grad_norm": 3.326143264770508, + "learning_rate": 9.882744078029081e-05, + "loss": 3.1433, + "step": 3406 + }, + { + "epoch": 5.0, + "grad_norm": 5.951056003570557, + "learning_rate": 9.878054069184583e-05, + "loss": 3.1543, + "step": 3407 + }, + { + "epoch": 5.0, + "grad_norm": 5.0941972732543945, + "learning_rate": 9.873364087167367e-05, + "loss": 3.1558, + "step": 3408 + }, + { + "epoch": 5.0, + "grad_norm": 2.075986623764038, + "learning_rate": 9.868674133009209e-05, + "loss": 3.144, + "step": 3409 + }, + { + "epoch": 5.0, + "grad_norm": 3.3112316131591797, + "learning_rate": 9.863984207741859e-05, + "loss": 3.1444, + "step": 3410 + }, + { + "epoch": 5.0, + "grad_norm": 4.910267353057861, + "learning_rate": 9.85929431239707e-05, + "loss": 3.1823, + "step": 3411 + }, + { + "epoch": 5.0, + "grad_norm": 4.1945624351501465, + "learning_rate": 9.854604448006594e-05, + "loss": 3.1945, + "step": 3412 + }, + { + "epoch": 5.01, + "grad_norm": 1.877642273902893, + "learning_rate": 9.849914615602163e-05, + "loss": 3.1357, + "step": 3413 + }, + { + "epoch": 5.01, + "grad_norm": 3.460303783416748, + "learning_rate": 9.845224816215508e-05, + "loss": 3.1039, + "step": 3414 + }, + { + "epoch": 5.01, + "grad_norm": 5.127685070037842, + "learning_rate": 9.84053505087836e-05, + "loss": 3.1041, + "step": 3415 + }, + { + "epoch": 5.01, + "grad_norm": 3.685288906097412, + "learning_rate": 9.835845320622434e-05, + "loss": 3.1874, + "step": 3416 + }, + { + "epoch": 5.01, + "grad_norm": 1.9846138954162598, + "learning_rate": 9.83115562647943e-05, + "loss": 3.1365, + "step": 3417 + }, + { + "epoch": 5.01, + "grad_norm": 5.992250442504883, + "learning_rate": 9.826465969481061e-05, + "loss": 3.0993, + "step": 3418 + }, + { + "epoch": 5.01, + "grad_norm": 4.008816242218018, + "learning_rate": 9.821776350659017e-05, + "loss": 3.1471, + "step": 3419 + }, + { + "epoch": 5.0, + "grad_norm": 2.797140598297119, + "learning_rate": 9.81708677104498e-05, + "loss": 3.1822, + "step": 3420 + }, + { + "epoch": 5.0, + "grad_norm": 3.2682175636291504, + "learning_rate": 9.812397231670624e-05, + "loss": 3.1501, + "step": 3421 + }, + { + "epoch": 5.0, + "grad_norm": 4.154157638549805, + "learning_rate": 9.807707733567626e-05, + "loss": 3.1572, + "step": 3422 + }, + { + "epoch": 5.01, + "grad_norm": 2.315260648727417, + "learning_rate": 9.803018277767634e-05, + "loss": 3.1305, + "step": 3423 + }, + { + "epoch": 5.01, + "grad_norm": 4.300922870635986, + "learning_rate": 9.798328865302299e-05, + "loss": 3.2001, + "step": 3424 + }, + { + "epoch": 5.01, + "grad_norm": 4.946724891662598, + "learning_rate": 9.793639497203268e-05, + "loss": 3.1431, + "step": 3425 + }, + { + "epoch": 5.01, + "grad_norm": 1.1593897342681885, + "learning_rate": 9.788950174502164e-05, + "loss": 3.1631, + "step": 3426 + }, + { + "epoch": 5.01, + "grad_norm": 88.28909301757812, + "learning_rate": 9.784260898230603e-05, + "loss": 3.1888, + "step": 3427 + }, + { + "epoch": 5.01, + "grad_norm": 7.40468168258667, + "learning_rate": 9.7795716694202e-05, + "loss": 3.1605, + "step": 3428 + }, + { + "epoch": 5.01, + "grad_norm": 8.019392967224121, + "learning_rate": 9.774882489102555e-05, + "loss": 3.2278, + "step": 3429 + }, + { + "epoch": 5.02, + "grad_norm": 5.9179205894470215, + "learning_rate": 9.770193358309248e-05, + "loss": 3.1715, + "step": 3430 + }, + { + "epoch": 5.02, + "grad_norm": 2.9003238677978516, + "learning_rate": 9.765504278071866e-05, + "loss": 3.1651, + "step": 3431 + }, + { + "epoch": 5.02, + "grad_norm": 4.174776554107666, + "learning_rate": 9.760815249421973e-05, + "loss": 3.1573, + "step": 3432 + }, + { + "epoch": 5.02, + "grad_norm": 5.618264198303223, + "learning_rate": 9.756126273391118e-05, + "loss": 3.1799, + "step": 3433 + }, + { + "epoch": 5.02, + "grad_norm": 5.217837333679199, + "learning_rate": 9.75143735101085e-05, + "loss": 3.1674, + "step": 3434 + }, + { + "epoch": 5.02, + "grad_norm": 6.881947994232178, + "learning_rate": 9.746748483312702e-05, + "loss": 3.2233, + "step": 3435 + }, + { + "epoch": 5.02, + "grad_norm": 4.155002117156982, + "learning_rate": 9.742059671328185e-05, + "loss": 3.1715, + "step": 3436 + }, + { + "epoch": 5.03, + "grad_norm": 3.3155972957611084, + "learning_rate": 9.737370916088815e-05, + "loss": 3.1688, + "step": 3437 + }, + { + "epoch": 5.03, + "grad_norm": 4.904897689819336, + "learning_rate": 9.732682218626086e-05, + "loss": 3.1702, + "step": 3438 + }, + { + "epoch": 5.03, + "grad_norm": 27.04771614074707, + "learning_rate": 9.727993579971471e-05, + "loss": 3.1816, + "step": 3439 + }, + { + "epoch": 5.03, + "grad_norm": 3.3083443641662598, + "learning_rate": 9.723305001156452e-05, + "loss": 3.1635, + "step": 3440 + }, + { + "epoch": 5.03, + "grad_norm": 3.0965356826782227, + "learning_rate": 9.718616483212474e-05, + "loss": 3.1631, + "step": 3441 + }, + { + "epoch": 5.03, + "grad_norm": 2.470313310623169, + "learning_rate": 9.713928027170989e-05, + "loss": 3.1788, + "step": 3442 + }, + { + "epoch": 5.04, + "grad_norm": 1.455441951751709, + "learning_rate": 9.709239634063414e-05, + "loss": 3.1543, + "step": 3443 + }, + { + "epoch": 5.04, + "grad_norm": 3.4748833179473877, + "learning_rate": 9.704551304921175e-05, + "loss": 3.1228, + "step": 3444 + }, + { + "epoch": 5.04, + "grad_norm": 3.481116771697998, + "learning_rate": 9.69986304077567e-05, + "loss": 3.1595, + "step": 3445 + }, + { + "epoch": 5.04, + "grad_norm": 4.942490577697754, + "learning_rate": 9.69517484265828e-05, + "loss": 3.1443, + "step": 3446 + }, + { + "epoch": 5.04, + "grad_norm": 3.2956652641296387, + "learning_rate": 9.690486711600384e-05, + "loss": 3.1699, + "step": 3447 + }, + { + "epoch": 5.04, + "grad_norm": 2.4343531131744385, + "learning_rate": 9.68579864863334e-05, + "loss": 3.1488, + "step": 3448 + }, + { + "epoch": 5.04, + "grad_norm": 2.82985782623291, + "learning_rate": 9.681110654788482e-05, + "loss": 3.1032, + "step": 3449 + }, + { + "epoch": 5.05, + "grad_norm": 1.59348464012146, + "learning_rate": 9.676422731097149e-05, + "loss": 3.1256, + "step": 3450 + }, + { + "epoch": 5.05, + "grad_norm": 4.060868740081787, + "learning_rate": 9.671734878590644e-05, + "loss": 3.1592, + "step": 3451 + }, + { + "epoch": 5.05, + "grad_norm": 4.718034744262695, + "learning_rate": 9.667047098300262e-05, + "loss": 3.1666, + "step": 3452 + }, + { + "epoch": 5.05, + "grad_norm": 1.5934293270111084, + "learning_rate": 9.662359391257293e-05, + "loss": 3.1592, + "step": 3453 + }, + { + "epoch": 5.05, + "grad_norm": 4.770664215087891, + "learning_rate": 9.657671758492992e-05, + "loss": 3.1303, + "step": 3454 + }, + { + "epoch": 5.05, + "grad_norm": 5.743010520935059, + "learning_rate": 9.652984201038608e-05, + "loss": 3.1736, + "step": 3455 + }, + { + "epoch": 5.05, + "grad_norm": 22.09770393371582, + "learning_rate": 9.64829671992538e-05, + "loss": 3.1555, + "step": 3456 + }, + { + "epoch": 5.06, + "grad_norm": 4.085333347320557, + "learning_rate": 9.643609316184514e-05, + "loss": 3.1203, + "step": 3457 + }, + { + "epoch": 5.06, + "grad_norm": 3.675048589706421, + "learning_rate": 9.638921990847206e-05, + "loss": 3.128, + "step": 3458 + }, + { + "epoch": 5.06, + "grad_norm": 1.0433051586151123, + "learning_rate": 9.634234744944646e-05, + "loss": 3.1206, + "step": 3459 + }, + { + "epoch": 5.06, + "grad_norm": 4.096065521240234, + "learning_rate": 9.62954757950799e-05, + "loss": 3.1567, + "step": 3460 + }, + { + "epoch": 5.06, + "grad_norm": 5.4872870445251465, + "learning_rate": 9.624860495568379e-05, + "loss": 3.1482, + "step": 3461 + }, + { + "epoch": 5.06, + "grad_norm": 3.236421585083008, + "learning_rate": 9.620173494156949e-05, + "loss": 3.1172, + "step": 3462 + }, + { + "epoch": 5.06, + "grad_norm": 3.7397096157073975, + "learning_rate": 9.615486576304806e-05, + "loss": 3.1485, + "step": 3463 + }, + { + "epoch": 5.07, + "grad_norm": 2.640143632888794, + "learning_rate": 9.610799743043036e-05, + "loss": 3.1444, + "step": 3464 + }, + { + "epoch": 5.07, + "grad_norm": 2.277892589569092, + "learning_rate": 9.606112995402714e-05, + "loss": 3.117, + "step": 3465 + }, + { + "epoch": 5.07, + "grad_norm": 2.7552220821380615, + "learning_rate": 9.601426334414898e-05, + "loss": 3.1557, + "step": 3466 + }, + { + "epoch": 5.07, + "grad_norm": 3.1442368030548096, + "learning_rate": 9.596739761110613e-05, + "loss": 3.1161, + "step": 3467 + }, + { + "epoch": 5.07, + "grad_norm": 1.997355341911316, + "learning_rate": 9.592053276520877e-05, + "loss": 3.1332, + "step": 3468 + }, + { + "epoch": 5.07, + "grad_norm": 1.959526777267456, + "learning_rate": 9.58736688167669e-05, + "loss": 3.141, + "step": 3469 + }, + { + "epoch": 5.08, + "grad_norm": 2.0642502307891846, + "learning_rate": 9.582680577609022e-05, + "loss": 3.1532, + "step": 3470 + }, + { + "epoch": 5.08, + "grad_norm": 1.1112334728240967, + "learning_rate": 9.57799436534883e-05, + "loss": 3.1402, + "step": 3471 + }, + { + "epoch": 5.08, + "grad_norm": 2.348271131515503, + "learning_rate": 9.57330824592705e-05, + "loss": 3.162, + "step": 3472 + }, + { + "epoch": 5.08, + "grad_norm": 1.7725739479064941, + "learning_rate": 9.5686222203746e-05, + "loss": 3.159, + "step": 3473 + }, + { + "epoch": 5.08, + "grad_norm": 1.3065656423568726, + "learning_rate": 9.563936289722363e-05, + "loss": 3.155, + "step": 3474 + }, + { + "epoch": 5.08, + "grad_norm": 2.1432132720947266, + "learning_rate": 9.559250455001223e-05, + "loss": 3.0939, + "step": 3475 + }, + { + "epoch": 5.08, + "grad_norm": 1.2344911098480225, + "learning_rate": 9.554564717242032e-05, + "loss": 3.1094, + "step": 3476 + }, + { + "epoch": 5.09, + "grad_norm": 1.591896653175354, + "learning_rate": 9.549879077475613e-05, + "loss": 3.1632, + "step": 3477 + }, + { + "epoch": 5.09, + "grad_norm": 1.2728568315505981, + "learning_rate": 9.545193536732783e-05, + "loss": 3.1537, + "step": 3478 + }, + { + "epoch": 5.09, + "grad_norm": 0.9794215559959412, + "learning_rate": 9.540508096044329e-05, + "loss": 3.1012, + "step": 3479 + }, + { + "epoch": 5.09, + "grad_norm": 2.092595100402832, + "learning_rate": 9.535822756441011e-05, + "loss": 3.1352, + "step": 3480 + }, + { + "epoch": 5.09, + "grad_norm": 1.6915316581726074, + "learning_rate": 9.531137518953577e-05, + "loss": 3.1332, + "step": 3481 + }, + { + "epoch": 5.09, + "grad_norm": 6.075295925140381, + "learning_rate": 9.526452384612749e-05, + "loss": 3.1654, + "step": 3482 + }, + { + "epoch": 5.09, + "grad_norm": 2.9814789295196533, + "learning_rate": 9.521767354449219e-05, + "loss": 3.1196, + "step": 3483 + }, + { + "epoch": 5.1, + "grad_norm": 1.2538803815841675, + "learning_rate": 9.51708242949367e-05, + "loss": 3.1304, + "step": 3484 + }, + { + "epoch": 5.1, + "grad_norm": 1.274577260017395, + "learning_rate": 9.512397610776747e-05, + "loss": 3.1293, + "step": 3485 + }, + { + "epoch": 5.1, + "grad_norm": 10.467679023742676, + "learning_rate": 9.507712899329079e-05, + "loss": 3.1396, + "step": 3486 + }, + { + "epoch": 5.1, + "grad_norm": 3.3941073417663574, + "learning_rate": 9.503028296181281e-05, + "loss": 3.1455, + "step": 3487 + }, + { + "epoch": 5.1, + "grad_norm": 1.5310273170471191, + "learning_rate": 9.498343802363923e-05, + "loss": 3.141, + "step": 3488 + }, + { + "epoch": 5.1, + "grad_norm": 4.921885967254639, + "learning_rate": 9.493659418907566e-05, + "loss": 3.1257, + "step": 3489 + }, + { + "epoch": 5.1, + "grad_norm": 6.093456268310547, + "learning_rate": 9.488975146842746e-05, + "loss": 3.1542, + "step": 3490 + }, + { + "epoch": 5.11, + "grad_norm": 2.32529616355896, + "learning_rate": 9.484290987199968e-05, + "loss": 3.1527, + "step": 3491 + }, + { + "epoch": 5.11, + "grad_norm": 4.923595905303955, + "learning_rate": 9.47960694100972e-05, + "loss": 3.1667, + "step": 3492 + }, + { + "epoch": 5.11, + "grad_norm": 5.732090950012207, + "learning_rate": 9.474923009302452e-05, + "loss": 3.1218, + "step": 3493 + }, + { + "epoch": 5.11, + "grad_norm": 3.330801486968994, + "learning_rate": 9.470239193108607e-05, + "loss": 3.1252, + "step": 3494 + }, + { + "epoch": 5.11, + "grad_norm": 3.7033092975616455, + "learning_rate": 9.465555493458588e-05, + "loss": 3.1358, + "step": 3495 + }, + { + "epoch": 5.11, + "grad_norm": 6.471573352813721, + "learning_rate": 9.460871911382774e-05, + "loss": 3.1961, + "step": 3496 + }, + { + "epoch": 5.12, + "grad_norm": 6.723712921142578, + "learning_rate": 9.456188447911533e-05, + "loss": 3.0915, + "step": 3497 + }, + { + "epoch": 5.12, + "grad_norm": 1.7002429962158203, + "learning_rate": 9.451505104075188e-05, + "loss": 3.1452, + "step": 3498 + }, + { + "epoch": 5.12, + "grad_norm": 2.4126813411712646, + "learning_rate": 9.44682188090404e-05, + "loss": 3.1457, + "step": 3499 + }, + { + "epoch": 5.12, + "grad_norm": 1.8296704292297363, + "learning_rate": 9.442138779428376e-05, + "loss": 3.1215, + "step": 3500 + }, + { + "epoch": 5.12, + "grad_norm": 1.7919952869415283, + "learning_rate": 9.43745580067844e-05, + "loss": 3.179, + "step": 3501 + }, + { + "epoch": 5.12, + "grad_norm": 1.8230031728744507, + "learning_rate": 9.432772945684454e-05, + "loss": 3.1339, + "step": 3502 + }, + { + "epoch": 5.12, + "grad_norm": 1.5876286029815674, + "learning_rate": 9.428090215476625e-05, + "loss": 3.1248, + "step": 3503 + }, + { + "epoch": 5.13, + "grad_norm": 2.1340315341949463, + "learning_rate": 9.423407611085113e-05, + "loss": 3.1396, + "step": 3504 + }, + { + "epoch": 5.13, + "grad_norm": 1.6990450620651245, + "learning_rate": 9.41872513354006e-05, + "loss": 3.1119, + "step": 3505 + }, + { + "epoch": 5.13, + "grad_norm": 1.792682409286499, + "learning_rate": 9.414042783871586e-05, + "loss": 3.1476, + "step": 3506 + }, + { + "epoch": 5.13, + "grad_norm": 2.685699701309204, + "learning_rate": 9.409360563109773e-05, + "loss": 3.1468, + "step": 3507 + }, + { + "epoch": 5.13, + "grad_norm": 4.750230312347412, + "learning_rate": 9.404678472284673e-05, + "loss": 3.1361, + "step": 3508 + }, + { + "epoch": 5.13, + "grad_norm": 1.0667452812194824, + "learning_rate": 9.399996512426323e-05, + "loss": 3.1625, + "step": 3509 + }, + { + "epoch": 5.13, + "grad_norm": 2.092228889465332, + "learning_rate": 9.395314684564721e-05, + "loss": 3.1709, + "step": 3510 + }, + { + "epoch": 5.14, + "grad_norm": 1.991498589515686, + "learning_rate": 9.390632989729831e-05, + "loss": 3.1122, + "step": 3511 + }, + { + "epoch": 5.14, + "grad_norm": 10.466729164123535, + "learning_rate": 9.385951428951603e-05, + "loss": 3.1214, + "step": 3512 + }, + { + "epoch": 5.14, + "grad_norm": 3.6561026573181152, + "learning_rate": 9.381270003259945e-05, + "loss": 3.1363, + "step": 3513 + }, + { + "epoch": 5.14, + "grad_norm": 1.4475880861282349, + "learning_rate": 9.37658871368474e-05, + "loss": 3.1801, + "step": 3514 + }, + { + "epoch": 5.14, + "grad_norm": 5.58400821685791, + "learning_rate": 9.371907561255837e-05, + "loss": 3.1223, + "step": 3515 + }, + { + "epoch": 5.14, + "grad_norm": 9.835494041442871, + "learning_rate": 9.367226547003065e-05, + "loss": 3.2195, + "step": 3516 + }, + { + "epoch": 5.14, + "grad_norm": 6.080451011657715, + "learning_rate": 9.362545671956214e-05, + "loss": 3.1594, + "step": 3517 + }, + { + "epoch": 5.15, + "grad_norm": 1.3072457313537598, + "learning_rate": 9.357864937145036e-05, + "loss": 3.147, + "step": 3518 + }, + { + "epoch": 5.15, + "grad_norm": 5.248061180114746, + "learning_rate": 9.353184343599273e-05, + "loss": 3.1497, + "step": 3519 + }, + { + "epoch": 5.15, + "grad_norm": 5.6261420249938965, + "learning_rate": 9.34850389234862e-05, + "loss": 3.1824, + "step": 3520 + }, + { + "epoch": 5.15, + "grad_norm": 3.202202320098877, + "learning_rate": 9.343823584422742e-05, + "loss": 3.1609, + "step": 3521 + }, + { + "epoch": 5.15, + "grad_norm": 4.45296573638916, + "learning_rate": 9.339143420851279e-05, + "loss": 3.1792, + "step": 3522 + }, + { + "epoch": 5.15, + "grad_norm": 4.815643310546875, + "learning_rate": 9.334463402663837e-05, + "loss": 3.2226, + "step": 3523 + }, + { + "epoch": 5.16, + "grad_norm": 12.902685165405273, + "learning_rate": 9.32978353088998e-05, + "loss": 3.1462, + "step": 3524 + }, + { + "epoch": 5.16, + "grad_norm": 3.254309892654419, + "learning_rate": 9.32510380655926e-05, + "loss": 3.1306, + "step": 3525 + }, + { + "epoch": 5.16, + "grad_norm": 8.259904861450195, + "learning_rate": 9.320424230701179e-05, + "loss": 3.2094, + "step": 3526 + }, + { + "epoch": 5.16, + "grad_norm": 2.651756525039673, + "learning_rate": 9.31574480434521e-05, + "loss": 3.1366, + "step": 3527 + }, + { + "epoch": 5.16, + "grad_norm": 4.296948432922363, + "learning_rate": 9.311065528520802e-05, + "loss": 3.1398, + "step": 3528 + }, + { + "epoch": 5.16, + "grad_norm": 4.362511157989502, + "learning_rate": 9.30638640425736e-05, + "loss": 3.163, + "step": 3529 + }, + { + "epoch": 5.16, + "grad_norm": 1.9980674982070923, + "learning_rate": 9.301707432584257e-05, + "loss": 3.1884, + "step": 3530 + }, + { + "epoch": 5.17, + "grad_norm": 4.281768321990967, + "learning_rate": 9.297028614530844e-05, + "loss": 3.1488, + "step": 3531 + }, + { + "epoch": 5.17, + "grad_norm": 5.395291328430176, + "learning_rate": 9.292349951126424e-05, + "loss": 3.1816, + "step": 3532 + }, + { + "epoch": 5.17, + "grad_norm": 1.9070907831192017, + "learning_rate": 9.287671443400268e-05, + "loss": 3.1082, + "step": 3533 + }, + { + "epoch": 5.17, + "grad_norm": 5.979832649230957, + "learning_rate": 9.282993092381625e-05, + "loss": 3.1332, + "step": 3534 + }, + { + "epoch": 5.17, + "grad_norm": 7.1100382804870605, + "learning_rate": 9.278314899099694e-05, + "loss": 3.2056, + "step": 3535 + }, + { + "epoch": 5.17, + "grad_norm": 22.313732147216797, + "learning_rate": 9.273636864583647e-05, + "loss": 3.1634, + "step": 3536 + }, + { + "epoch": 5.17, + "grad_norm": 11.737483978271484, + "learning_rate": 9.268958989862628e-05, + "loss": 3.1481, + "step": 3537 + }, + { + "epoch": 5.18, + "grad_norm": 2.638352394104004, + "learning_rate": 9.264281275965728e-05, + "loss": 3.1341, + "step": 3538 + }, + { + "epoch": 5.18, + "grad_norm": 2.807657241821289, + "learning_rate": 9.25960372392202e-05, + "loss": 3.1505, + "step": 3539 + }, + { + "epoch": 5.18, + "grad_norm": 20.156600952148438, + "learning_rate": 9.254926334760526e-05, + "loss": 3.1956, + "step": 3540 + }, + { + "epoch": 5.18, + "grad_norm": 2.164813280105591, + "learning_rate": 9.25024910951025e-05, + "loss": 3.1844, + "step": 3541 + }, + { + "epoch": 5.18, + "grad_norm": 1.7954508066177368, + "learning_rate": 9.245572049200142e-05, + "loss": 3.222, + "step": 3542 + }, + { + "epoch": 5.18, + "grad_norm": 4.100772380828857, + "learning_rate": 9.240895154859125e-05, + "loss": 3.1239, + "step": 3543 + }, + { + "epoch": 5.19, + "grad_norm": 3.8500773906707764, + "learning_rate": 9.236218427516092e-05, + "loss": 3.1278, + "step": 3544 + }, + { + "epoch": 5.19, + "grad_norm": 2.076373815536499, + "learning_rate": 9.231541868199883e-05, + "loss": 3.164, + "step": 3545 + }, + { + "epoch": 5.19, + "grad_norm": 1.8136000633239746, + "learning_rate": 9.226865477939313e-05, + "loss": 3.1479, + "step": 3546 + }, + { + "epoch": 5.19, + "grad_norm": 128.6366424560547, + "learning_rate": 9.222189257763158e-05, + "loss": 3.1239, + "step": 3547 + }, + { + "epoch": 5.19, + "grad_norm": 3.7925548553466797, + "learning_rate": 9.217513208700154e-05, + "loss": 3.1402, + "step": 3548 + }, + { + "epoch": 5.19, + "grad_norm": 11.528712272644043, + "learning_rate": 9.212837331778997e-05, + "loss": 3.1434, + "step": 3549 + }, + { + "epoch": 5.19, + "eval_loss": 2.9914677143096924, + "eval_runtime": 1173.7001, + "eval_samples_per_second": 14.605, + "eval_steps_per_second": 0.913, + "step": 3549 + }, + { + "epoch": 5.19, + "grad_norm": 3.5686726570129395, + "learning_rate": 9.208161628028355e-05, + "loss": 3.1025, + "step": 3550 + }, + { + "epoch": 5.2, + "grad_norm": 19.285165786743164, + "learning_rate": 9.203486098476849e-05, + "loss": 3.1957, + "step": 3551 + }, + { + "epoch": 5.2, + "grad_norm": 7.824908256530762, + "learning_rate": 9.19881074415306e-05, + "loss": 3.1752, + "step": 3552 + }, + { + "epoch": 5.2, + "grad_norm": 3.9910645484924316, + "learning_rate": 9.194135566085537e-05, + "loss": 3.1751, + "step": 3553 + }, + { + "epoch": 5.2, + "grad_norm": 32.02030944824219, + "learning_rate": 9.189460565302794e-05, + "loss": 3.1539, + "step": 3554 + }, + { + "epoch": 5.2, + "grad_norm": 10.703150749206543, + "learning_rate": 9.18478574283329e-05, + "loss": 3.1623, + "step": 3555 + }, + { + "epoch": 5.2, + "grad_norm": 37.447715759277344, + "learning_rate": 9.180111099705459e-05, + "loss": 3.1512, + "step": 3556 + }, + { + "epoch": 5.2, + "grad_norm": 4.359159469604492, + "learning_rate": 9.175436636947697e-05, + "loss": 3.1754, + "step": 3557 + }, + { + "epoch": 5.21, + "grad_norm": 11.65587329864502, + "learning_rate": 9.170762355588342e-05, + "loss": 3.1365, + "step": 3558 + }, + { + "epoch": 5.21, + "grad_norm": 6.701794147491455, + "learning_rate": 9.166088256655715e-05, + "loss": 3.1705, + "step": 3559 + }, + { + "epoch": 5.21, + "grad_norm": 12.4017915725708, + "learning_rate": 9.161414341178084e-05, + "loss": 3.1724, + "step": 3560 + }, + { + "epoch": 5.21, + "grad_norm": 2.782519578933716, + "learning_rate": 9.156740610183674e-05, + "loss": 3.1809, + "step": 3561 + }, + { + "epoch": 5.21, + "grad_norm": 3.470179319381714, + "learning_rate": 9.152067064700683e-05, + "loss": 3.1531, + "step": 3562 + }, + { + "epoch": 5.21, + "grad_norm": 12.016058921813965, + "learning_rate": 9.147393705757254e-05, + "loss": 3.1714, + "step": 3563 + }, + { + "epoch": 5.21, + "grad_norm": 8.318673133850098, + "learning_rate": 9.142720534381497e-05, + "loss": 3.178, + "step": 3564 + }, + { + "epoch": 5.22, + "grad_norm": 26.07087516784668, + "learning_rate": 9.138047551601473e-05, + "loss": 3.1636, + "step": 3565 + }, + { + "epoch": 5.22, + "grad_norm": 55.82182312011719, + "learning_rate": 9.133374758445212e-05, + "loss": 3.1598, + "step": 3566 + }, + { + "epoch": 5.22, + "grad_norm": 3.5638935565948486, + "learning_rate": 9.1287021559407e-05, + "loss": 3.1711, + "step": 3567 + }, + { + "epoch": 5.22, + "grad_norm": 4.856345176696777, + "learning_rate": 9.12402974511587e-05, + "loss": 3.1686, + "step": 3568 + }, + { + "epoch": 5.22, + "grad_norm": 6.552631855010986, + "learning_rate": 9.119357526998628e-05, + "loss": 3.1437, + "step": 3569 + }, + { + "epoch": 5.22, + "grad_norm": 62.54035186767578, + "learning_rate": 9.11468550261683e-05, + "loss": 3.1711, + "step": 3570 + }, + { + "epoch": 5.23, + "grad_norm": 19.124649047851562, + "learning_rate": 9.110013672998284e-05, + "loss": 3.2002, + "step": 3571 + }, + { + "epoch": 5.23, + "grad_norm": 14.90050983428955, + "learning_rate": 9.105342039170768e-05, + "loss": 3.2217, + "step": 3572 + }, + { + "epoch": 5.23, + "grad_norm": 4.7844109535217285, + "learning_rate": 9.100670602162009e-05, + "loss": 3.1506, + "step": 3573 + }, + { + "epoch": 5.23, + "grad_norm": 1.9373784065246582, + "learning_rate": 9.095999362999687e-05, + "loss": 3.1288, + "step": 3574 + }, + { + "epoch": 5.23, + "grad_norm": 60.557308197021484, + "learning_rate": 9.091328322711449e-05, + "loss": 3.1368, + "step": 3575 + }, + { + "epoch": 5.23, + "grad_norm": 6.0116658210754395, + "learning_rate": 9.086657482324892e-05, + "loss": 3.1685, + "step": 3576 + }, + { + "epoch": 5.23, + "grad_norm": 7.286896228790283, + "learning_rate": 9.081986842867562e-05, + "loss": 3.1756, + "step": 3577 + }, + { + "epoch": 5.24, + "grad_norm": 2.469630479812622, + "learning_rate": 9.077316405366981e-05, + "loss": 3.1425, + "step": 3578 + }, + { + "epoch": 5.24, + "grad_norm": 3.226400136947632, + "learning_rate": 9.072646170850605e-05, + "loss": 3.1015, + "step": 3579 + }, + { + "epoch": 5.24, + "grad_norm": 19.016260147094727, + "learning_rate": 9.067976140345854e-05, + "loss": 3.1746, + "step": 3580 + }, + { + "epoch": 5.24, + "grad_norm": 5.787333011627197, + "learning_rate": 9.06330631488011e-05, + "loss": 3.1975, + "step": 3581 + }, + { + "epoch": 5.24, + "grad_norm": 5.138857841491699, + "learning_rate": 9.058636695480698e-05, + "loss": 3.181, + "step": 3582 + }, + { + "epoch": 5.24, + "grad_norm": 11.81260871887207, + "learning_rate": 9.053967283174901e-05, + "loss": 3.1447, + "step": 3583 + }, + { + "epoch": 5.24, + "grad_norm": 6.516455173492432, + "learning_rate": 9.049298078989967e-05, + "loss": 3.1348, + "step": 3584 + }, + { + "epoch": 5.25, + "grad_norm": 6.460428237915039, + "learning_rate": 9.044629083953082e-05, + "loss": 3.1878, + "step": 3585 + }, + { + "epoch": 5.25, + "grad_norm": 5.189371585845947, + "learning_rate": 9.039960299091397e-05, + "loss": 3.1585, + "step": 3586 + }, + { + "epoch": 5.25, + "grad_norm": 4.265483379364014, + "learning_rate": 9.035291725432008e-05, + "loss": 3.1688, + "step": 3587 + }, + { + "epoch": 5.25, + "grad_norm": 2.625492572784424, + "learning_rate": 9.030623364001979e-05, + "loss": 3.1438, + "step": 3588 + }, + { + "epoch": 5.25, + "grad_norm": 3.3770155906677246, + "learning_rate": 9.025955215828312e-05, + "loss": 3.1574, + "step": 3589 + }, + { + "epoch": 5.25, + "grad_norm": 21.14669418334961, + "learning_rate": 9.021287281937967e-05, + "loss": 3.1473, + "step": 3590 + }, + { + "epoch": 5.25, + "grad_norm": 30.993927001953125, + "learning_rate": 9.016619563357865e-05, + "loss": 3.2186, + "step": 3591 + }, + { + "epoch": 5.26, + "grad_norm": 5.043474197387695, + "learning_rate": 9.011952061114867e-05, + "loss": 3.1602, + "step": 3592 + }, + { + "epoch": 5.26, + "grad_norm": 17.649402618408203, + "learning_rate": 9.007284776235791e-05, + "loss": 3.139, + "step": 3593 + }, + { + "epoch": 5.26, + "grad_norm": 50.776214599609375, + "learning_rate": 9.002617709747416e-05, + "loss": 3.1279, + "step": 3594 + }, + { + "epoch": 5.26, + "grad_norm": 12.535754203796387, + "learning_rate": 8.997950862676462e-05, + "loss": 3.174, + "step": 3595 + }, + { + "epoch": 5.26, + "grad_norm": 4.332036972045898, + "learning_rate": 8.993284236049598e-05, + "loss": 3.106, + "step": 3596 + }, + { + "epoch": 5.26, + "grad_norm": 1.4636400938034058, + "learning_rate": 8.988617830893455e-05, + "loss": 3.1456, + "step": 3597 + }, + { + "epoch": 5.27, + "grad_norm": 4.1321024894714355, + "learning_rate": 8.983951648234617e-05, + "loss": 3.102, + "step": 3598 + }, + { + "epoch": 5.27, + "grad_norm": 4.079101085662842, + "learning_rate": 8.979285689099601e-05, + "loss": 3.1632, + "step": 3599 + }, + { + "epoch": 5.27, + "grad_norm": 7.1509690284729, + "learning_rate": 8.974619954514895e-05, + "loss": 3.1862, + "step": 3600 + }, + { + "epoch": 5.27, + "grad_norm": 3.1132075786590576, + "learning_rate": 8.96995444550693e-05, + "loss": 3.1587, + "step": 3601 + }, + { + "epoch": 5.27, + "grad_norm": 15.199933052062988, + "learning_rate": 8.965289163102078e-05, + "loss": 3.1385, + "step": 3602 + }, + { + "epoch": 5.27, + "grad_norm": 3.226987600326538, + "learning_rate": 8.960624108326679e-05, + "loss": 3.1793, + "step": 3603 + }, + { + "epoch": 5.27, + "grad_norm": 3.0855495929718018, + "learning_rate": 8.955959282207013e-05, + "loss": 3.121, + "step": 3604 + }, + { + "epoch": 5.28, + "grad_norm": 1.8328437805175781, + "learning_rate": 8.951294685769303e-05, + "loss": 3.1515, + "step": 3605 + }, + { + "epoch": 5.28, + "grad_norm": 2.6738593578338623, + "learning_rate": 8.946630320039737e-05, + "loss": 3.1412, + "step": 3606 + }, + { + "epoch": 5.28, + "grad_norm": 2.790065288543701, + "learning_rate": 8.941966186044444e-05, + "loss": 3.1307, + "step": 3607 + }, + { + "epoch": 5.28, + "grad_norm": 3.6638455390930176, + "learning_rate": 8.937302284809494e-05, + "loss": 3.1155, + "step": 3608 + }, + { + "epoch": 5.28, + "grad_norm": 3.6409106254577637, + "learning_rate": 8.932638617360925e-05, + "loss": 3.1242, + "step": 3609 + }, + { + "epoch": 5.28, + "grad_norm": 2.8414742946624756, + "learning_rate": 8.927975184724704e-05, + "loss": 3.1433, + "step": 3610 + }, + { + "epoch": 5.28, + "grad_norm": 4.518361568450928, + "learning_rate": 8.92331198792676e-05, + "loss": 3.1883, + "step": 3611 + }, + { + "epoch": 5.29, + "grad_norm": 7.363691806793213, + "learning_rate": 8.918649027992958e-05, + "loss": 3.1231, + "step": 3612 + }, + { + "epoch": 5.29, + "grad_norm": 2.302523136138916, + "learning_rate": 8.913986305949125e-05, + "loss": 3.0972, + "step": 3613 + }, + { + "epoch": 5.29, + "grad_norm": 1.6068569421768188, + "learning_rate": 8.909323822821031e-05, + "loss": 3.1498, + "step": 3614 + }, + { + "epoch": 5.29, + "grad_norm": 1.9950542449951172, + "learning_rate": 8.90466157963438e-05, + "loss": 3.1339, + "step": 3615 + }, + { + "epoch": 5.29, + "grad_norm": 1.7338231801986694, + "learning_rate": 8.899999577414845e-05, + "loss": 3.1382, + "step": 3616 + }, + { + "epoch": 5.29, + "grad_norm": 2.231691598892212, + "learning_rate": 8.895337817188031e-05, + "loss": 3.1417, + "step": 3617 + }, + { + "epoch": 5.29, + "grad_norm": 1.9115151166915894, + "learning_rate": 8.890676299979492e-05, + "loss": 3.1338, + "step": 3618 + }, + { + "epoch": 5.3, + "grad_norm": 2.1142683029174805, + "learning_rate": 8.886015026814736e-05, + "loss": 3.1768, + "step": 3619 + }, + { + "epoch": 5.3, + "grad_norm": 2.7301149368286133, + "learning_rate": 8.881353998719206e-05, + "loss": 3.1843, + "step": 3620 + }, + { + "epoch": 5.3, + "grad_norm": 3.0939254760742188, + "learning_rate": 8.876693216718297e-05, + "loss": 3.1582, + "step": 3621 + }, + { + "epoch": 5.3, + "grad_norm": 1.2294076681137085, + "learning_rate": 8.872032681837358e-05, + "loss": 3.1484, + "step": 3622 + }, + { + "epoch": 5.3, + "grad_norm": 2.869823694229126, + "learning_rate": 8.867372395101666e-05, + "loss": 3.1455, + "step": 3623 + }, + { + "epoch": 5.3, + "grad_norm": 3.04744291305542, + "learning_rate": 8.862712357536454e-05, + "loss": 3.151, + "step": 3624 + }, + { + "epoch": 5.31, + "grad_norm": 1.6261351108551025, + "learning_rate": 8.858052570166908e-05, + "loss": 3.1434, + "step": 3625 + }, + { + "epoch": 5.31, + "grad_norm": 2.8542559146881104, + "learning_rate": 8.85339303401814e-05, + "loss": 3.1384, + "step": 3626 + }, + { + "epoch": 5.31, + "grad_norm": 0.9066136479377747, + "learning_rate": 8.848733750115218e-05, + "loss": 3.1576, + "step": 3627 + }, + { + "epoch": 5.31, + "grad_norm": 2.466120719909668, + "learning_rate": 8.84407471948316e-05, + "loss": 3.1478, + "step": 3628 + }, + { + "epoch": 5.31, + "grad_norm": 1.9956951141357422, + "learning_rate": 8.839415943146919e-05, + "loss": 3.1492, + "step": 3629 + }, + { + "epoch": 5.31, + "grad_norm": 10.860211372375488, + "learning_rate": 8.834757422131385e-05, + "loss": 3.198, + "step": 3630 + }, + { + "epoch": 5.31, + "grad_norm": 4.2302327156066895, + "learning_rate": 8.830099157461413e-05, + "loss": 3.146, + "step": 3631 + }, + { + "epoch": 5.32, + "grad_norm": 13.344452857971191, + "learning_rate": 8.825441150161788e-05, + "loss": 3.1234, + "step": 3632 + }, + { + "epoch": 5.32, + "grad_norm": 3.983452558517456, + "learning_rate": 8.820783401257236e-05, + "loss": 3.1775, + "step": 3633 + }, + { + "epoch": 5.32, + "grad_norm": 2.76181960105896, + "learning_rate": 8.816125911772433e-05, + "loss": 3.1776, + "step": 3634 + }, + { + "epoch": 5.32, + "grad_norm": 5.169229030609131, + "learning_rate": 8.811468682731997e-05, + "loss": 3.1174, + "step": 3635 + }, + { + "epoch": 5.32, + "grad_norm": 2.876774549484253, + "learning_rate": 8.806811715160485e-05, + "loss": 3.1506, + "step": 3636 + }, + { + "epoch": 5.32, + "grad_norm": 1.740065336227417, + "learning_rate": 8.802155010082396e-05, + "loss": 3.1375, + "step": 3637 + }, + { + "epoch": 5.32, + "grad_norm": 2.2916555404663086, + "learning_rate": 8.797498568522182e-05, + "loss": 3.1736, + "step": 3638 + }, + { + "epoch": 5.33, + "grad_norm": 13.75604248046875, + "learning_rate": 8.792842391504222e-05, + "loss": 3.1382, + "step": 3639 + }, + { + "epoch": 5.33, + "grad_norm": 8.454766273498535, + "learning_rate": 8.788186480052844e-05, + "loss": 3.1663, + "step": 3640 + }, + { + "epoch": 5.33, + "grad_norm": 3.627807140350342, + "learning_rate": 8.783530835192322e-05, + "loss": 3.1633, + "step": 3641 + }, + { + "epoch": 5.33, + "grad_norm": 1.9736120700836182, + "learning_rate": 8.778875457946866e-05, + "loss": 3.2027, + "step": 3642 + }, + { + "epoch": 5.33, + "grad_norm": 4.044384479522705, + "learning_rate": 8.77422034934062e-05, + "loss": 3.1484, + "step": 3643 + }, + { + "epoch": 5.33, + "grad_norm": 4.4502763748168945, + "learning_rate": 8.769565510397684e-05, + "loss": 3.1589, + "step": 3644 + }, + { + "epoch": 5.34, + "grad_norm": 2.187070369720459, + "learning_rate": 8.764910942142093e-05, + "loss": 3.1593, + "step": 3645 + }, + { + "epoch": 5.34, + "grad_norm": 2.6429078578948975, + "learning_rate": 8.760256645597813e-05, + "loss": 3.1466, + "step": 3646 + }, + { + "epoch": 5.34, + "grad_norm": 1.6300479173660278, + "learning_rate": 8.755602621788764e-05, + "loss": 3.1133, + "step": 3647 + }, + { + "epoch": 5.34, + "grad_norm": 2.9914662837982178, + "learning_rate": 8.750948871738802e-05, + "loss": 3.114, + "step": 3648 + }, + { + "epoch": 5.34, + "grad_norm": 2.315000534057617, + "learning_rate": 8.746295396471711e-05, + "loss": 3.1427, + "step": 3649 + }, + { + "epoch": 5.34, + "grad_norm": 3.0107791423797607, + "learning_rate": 8.741642197011235e-05, + "loss": 3.1085, + "step": 3650 + }, + { + "epoch": 5.34, + "grad_norm": 11.682272911071777, + "learning_rate": 8.736989274381045e-05, + "loss": 3.1428, + "step": 3651 + }, + { + "epoch": 5.35, + "grad_norm": 2.8496906757354736, + "learning_rate": 8.732336629604745e-05, + "loss": 3.1269, + "step": 3652 + }, + { + "epoch": 5.35, + "grad_norm": 2.9771509170532227, + "learning_rate": 8.727684263705895e-05, + "loss": 3.147, + "step": 3653 + }, + { + "epoch": 5.35, + "grad_norm": 2.937957525253296, + "learning_rate": 8.723032177707978e-05, + "loss": 3.1731, + "step": 3654 + }, + { + "epoch": 5.35, + "grad_norm": 3.5484066009521484, + "learning_rate": 8.718380372634422e-05, + "loss": 3.1602, + "step": 3655 + }, + { + "epoch": 5.35, + "grad_norm": 9.137898445129395, + "learning_rate": 8.7137288495086e-05, + "loss": 3.1692, + "step": 3656 + }, + { + "epoch": 5.35, + "grad_norm": 10.832776069641113, + "learning_rate": 8.709077609353809e-05, + "loss": 3.1829, + "step": 3657 + }, + { + "epoch": 5.35, + "grad_norm": 3.5576894283294678, + "learning_rate": 8.704426653193289e-05, + "loss": 3.1439, + "step": 3658 + }, + { + "epoch": 5.36, + "grad_norm": 2.771862745285034, + "learning_rate": 8.699775982050229e-05, + "loss": 3.1435, + "step": 3659 + }, + { + "epoch": 5.36, + "grad_norm": 7.5768046379089355, + "learning_rate": 8.695125596947736e-05, + "loss": 3.1898, + "step": 3660 + }, + { + "epoch": 5.36, + "grad_norm": 3.332085132598877, + "learning_rate": 8.690475498908868e-05, + "loss": 3.1365, + "step": 3661 + }, + { + "epoch": 5.36, + "grad_norm": 8.380770683288574, + "learning_rate": 8.685825688956613e-05, + "loss": 3.1733, + "step": 3662 + }, + { + "epoch": 5.36, + "grad_norm": 9.0435791015625, + "learning_rate": 8.6811761681139e-05, + "loss": 3.1752, + "step": 3663 + }, + { + "epoch": 5.36, + "grad_norm": 12.012619018554688, + "learning_rate": 8.676526937403591e-05, + "loss": 3.1615, + "step": 3664 + }, + { + "epoch": 5.36, + "grad_norm": 6.051881313323975, + "learning_rate": 8.671877997848484e-05, + "loss": 3.1719, + "step": 3665 + }, + { + "epoch": 5.37, + "grad_norm": 5.059371471405029, + "learning_rate": 8.667229350471322e-05, + "loss": 3.1775, + "step": 3666 + }, + { + "epoch": 5.37, + "grad_norm": 9.862431526184082, + "learning_rate": 8.662580996294767e-05, + "loss": 3.1795, + "step": 3667 + }, + { + "epoch": 5.37, + "grad_norm": 15.149982452392578, + "learning_rate": 8.657932936341428e-05, + "loss": 3.2141, + "step": 3668 + }, + { + "epoch": 5.37, + "grad_norm": 9.565960884094238, + "learning_rate": 8.653285171633855e-05, + "loss": 3.1831, + "step": 3669 + }, + { + "epoch": 5.37, + "grad_norm": 19.23344612121582, + "learning_rate": 8.648637703194516e-05, + "loss": 3.1925, + "step": 3670 + }, + { + "epoch": 5.37, + "grad_norm": 6.268587589263916, + "learning_rate": 8.643990532045822e-05, + "loss": 3.1621, + "step": 3671 + }, + { + "epoch": 5.38, + "grad_norm": 56.983970642089844, + "learning_rate": 8.639343659210131e-05, + "loss": 3.1715, + "step": 3672 + }, + { + "epoch": 5.38, + "grad_norm": 50.29884719848633, + "learning_rate": 8.634697085709711e-05, + "loss": 3.1474, + "step": 3673 + }, + { + "epoch": 5.38, + "grad_norm": 13.747417449951172, + "learning_rate": 8.630050812566781e-05, + "loss": 3.1805, + "step": 3674 + }, + { + "epoch": 5.38, + "grad_norm": 20.72515106201172, + "learning_rate": 8.625404840803498e-05, + "loss": 3.1528, + "step": 3675 + }, + { + "epoch": 5.38, + "grad_norm": 12.049095153808594, + "learning_rate": 8.620759171441935e-05, + "loss": 3.181, + "step": 3676 + }, + { + "epoch": 5.38, + "grad_norm": 11.251856803894043, + "learning_rate": 8.616113805504106e-05, + "loss": 3.1962, + "step": 3677 + }, + { + "epoch": 5.38, + "grad_norm": 6.218480587005615, + "learning_rate": 8.61146874401197e-05, + "loss": 3.2132, + "step": 3678 + }, + { + "epoch": 5.39, + "grad_norm": 10.480876922607422, + "learning_rate": 8.606823987987405e-05, + "loss": 3.1913, + "step": 3679 + }, + { + "epoch": 5.39, + "grad_norm": 12.099661827087402, + "learning_rate": 8.602179538452222e-05, + "loss": 3.178, + "step": 3680 + }, + { + "epoch": 5.39, + "grad_norm": 23.720500946044922, + "learning_rate": 8.597535396428177e-05, + "loss": 3.1558, + "step": 3681 + }, + { + "epoch": 5.39, + "grad_norm": 16.246944427490234, + "learning_rate": 8.592891562936947e-05, + "loss": 3.1834, + "step": 3682 + }, + { + "epoch": 5.39, + "grad_norm": 37.614784240722656, + "learning_rate": 8.58824803900014e-05, + "loss": 3.174, + "step": 3683 + }, + { + "epoch": 5.39, + "grad_norm": 9.572596549987793, + "learning_rate": 8.583604825639302e-05, + "loss": 3.1871, + "step": 3684 + }, + { + "epoch": 5.39, + "grad_norm": 6.099634170532227, + "learning_rate": 8.578961923875915e-05, + "loss": 3.1663, + "step": 3685 + }, + { + "epoch": 5.4, + "grad_norm": 18.694616317749023, + "learning_rate": 8.57431933473138e-05, + "loss": 3.1668, + "step": 3686 + }, + { + "epoch": 5.4, + "grad_norm": 133.21104431152344, + "learning_rate": 8.569677059227032e-05, + "loss": 3.1881, + "step": 3687 + }, + { + "epoch": 5.4, + "grad_norm": 33.713504791259766, + "learning_rate": 8.565035098384149e-05, + "loss": 3.1648, + "step": 3688 + }, + { + "epoch": 5.4, + "grad_norm": 9.273075103759766, + "learning_rate": 8.56039345322393e-05, + "loss": 3.2306, + "step": 3689 + }, + { + "epoch": 5.4, + "grad_norm": 18.504985809326172, + "learning_rate": 8.555752124767498e-05, + "loss": 3.2082, + "step": 3690 + }, + { + "epoch": 5.4, + "grad_norm": 13.903569221496582, + "learning_rate": 8.551111114035922e-05, + "loss": 3.1519, + "step": 3691 + }, + { + "epoch": 5.4, + "grad_norm": 48.394412994384766, + "learning_rate": 8.546470422050193e-05, + "loss": 3.1922, + "step": 3692 + }, + { + "epoch": 5.41, + "grad_norm": 9.13818073272705, + "learning_rate": 8.541830049831227e-05, + "loss": 3.2432, + "step": 3693 + }, + { + "epoch": 5.41, + "grad_norm": 7.568233013153076, + "learning_rate": 8.537189998399879e-05, + "loss": 3.2007, + "step": 3694 + }, + { + "epoch": 5.41, + "grad_norm": 110.37165832519531, + "learning_rate": 8.53255026877693e-05, + "loss": 3.1837, + "step": 3695 + }, + { + "epoch": 5.41, + "grad_norm": 8.272992134094238, + "learning_rate": 8.527910861983084e-05, + "loss": 3.169, + "step": 3696 + }, + { + "epoch": 5.41, + "grad_norm": 29.426700592041016, + "learning_rate": 8.523271779038986e-05, + "loss": 3.1793, + "step": 3697 + }, + { + "epoch": 5.41, + "grad_norm": 9.58652400970459, + "learning_rate": 8.518633020965199e-05, + "loss": 3.1662, + "step": 3698 + }, + { + "epoch": 5.42, + "grad_norm": 5.326894283294678, + "learning_rate": 8.513994588782217e-05, + "loss": 3.1817, + "step": 3699 + }, + { + "epoch": 5.42, + "grad_norm": 8.72939395904541, + "learning_rate": 8.509356483510471e-05, + "loss": 3.1888, + "step": 3700 + }, + { + "epoch": 5.42, + "grad_norm": 9.870657920837402, + "learning_rate": 8.504718706170306e-05, + "loss": 3.1772, + "step": 3701 + }, + { + "epoch": 5.42, + "grad_norm": 39.32550811767578, + "learning_rate": 8.500081257782003e-05, + "loss": 3.203, + "step": 3702 + }, + { + "epoch": 5.42, + "grad_norm": 20.613840103149414, + "learning_rate": 8.495444139365776e-05, + "loss": 3.1721, + "step": 3703 + }, + { + "epoch": 5.42, + "grad_norm": 10.104787826538086, + "learning_rate": 8.490807351941753e-05, + "loss": 3.1791, + "step": 3704 + }, + { + "epoch": 5.42, + "grad_norm": 8.861245155334473, + "learning_rate": 8.486170896529996e-05, + "loss": 3.2222, + "step": 3705 + }, + { + "epoch": 5.43, + "grad_norm": 40.4407844543457, + "learning_rate": 8.481534774150502e-05, + "loss": 3.3014, + "step": 3706 + }, + { + "epoch": 5.43, + "grad_norm": 4.760523319244385, + "learning_rate": 8.476898985823177e-05, + "loss": 3.1825, + "step": 3707 + }, + { + "epoch": 5.43, + "grad_norm": 1924.5662841796875, + "learning_rate": 8.472263532567871e-05, + "loss": 3.1802, + "step": 3708 + }, + { + "epoch": 5.43, + "grad_norm": 5.515054702758789, + "learning_rate": 8.467628415404345e-05, + "loss": 3.1833, + "step": 3709 + }, + { + "epoch": 5.43, + "grad_norm": 2.0650458335876465, + "learning_rate": 8.462993635352303e-05, + "loss": 3.184, + "step": 3710 + }, + { + "epoch": 5.43, + "grad_norm": 4.477024078369141, + "learning_rate": 8.458359193431358e-05, + "loss": 3.1489, + "step": 3711 + }, + { + "epoch": 5.43, + "grad_norm": 8.2962646484375, + "learning_rate": 8.453725090661057e-05, + "loss": 3.1318, + "step": 3712 + }, + { + "epoch": 5.44, + "grad_norm": 4.55861234664917, + "learning_rate": 8.449091328060877e-05, + "loss": 3.185, + "step": 3713 + }, + { + "epoch": 5.44, + "grad_norm": 6.914697647094727, + "learning_rate": 8.44445790665021e-05, + "loss": 3.1659, + "step": 3714 + }, + { + "epoch": 5.44, + "grad_norm": 5.971372127532959, + "learning_rate": 8.439824827448376e-05, + "loss": 3.1691, + "step": 3715 + }, + { + "epoch": 5.44, + "grad_norm": 3.358804702758789, + "learning_rate": 8.435192091474628e-05, + "loss": 3.187, + "step": 3716 + }, + { + "epoch": 5.44, + "grad_norm": 50.52218246459961, + "learning_rate": 8.430559699748132e-05, + "loss": 3.1595, + "step": 3717 + }, + { + "epoch": 5.44, + "grad_norm": 4.3211774826049805, + "learning_rate": 8.425927653287981e-05, + "loss": 3.2301, + "step": 3718 + }, + { + "epoch": 5.44, + "eval_loss": 3.003319025039673, + "eval_runtime": 1168.0997, + "eval_samples_per_second": 14.675, + "eval_steps_per_second": 0.918, + "step": 3718 + }, + { + "epoch": 5.44, + "grad_norm": 3.5544116497039795, + "learning_rate": 8.421295953113201e-05, + "loss": 3.1628, + "step": 3719 + }, + { + "epoch": 5.45, + "grad_norm": 4.997739791870117, + "learning_rate": 8.416664600242733e-05, + "loss": 3.1741, + "step": 3720 + }, + { + "epoch": 5.45, + "grad_norm": 2.1352174282073975, + "learning_rate": 8.412033595695437e-05, + "loss": 3.1518, + "step": 3721 + }, + { + "epoch": 5.45, + "grad_norm": 7.084630489349365, + "learning_rate": 8.40740294049011e-05, + "loss": 3.1283, + "step": 3722 + }, + { + "epoch": 5.45, + "grad_norm": 2.8684113025665283, + "learning_rate": 8.402772635645465e-05, + "loss": 3.1823, + "step": 3723 + }, + { + "epoch": 5.45, + "grad_norm": 14.201617240905762, + "learning_rate": 8.398142682180131e-05, + "loss": 3.14, + "step": 3724 + }, + { + "epoch": 5.45, + "grad_norm": 4.209918975830078, + "learning_rate": 8.393513081112674e-05, + "loss": 3.1867, + "step": 3725 + }, + { + "epoch": 5.46, + "grad_norm": 2.907388687133789, + "learning_rate": 8.388883833461578e-05, + "loss": 3.2095, + "step": 3726 + }, + { + "epoch": 5.46, + "grad_norm": 11.771166801452637, + "learning_rate": 8.384254940245233e-05, + "loss": 3.1938, + "step": 3727 + }, + { + "epoch": 5.46, + "grad_norm": 15.321097373962402, + "learning_rate": 8.379626402481977e-05, + "loss": 3.1755, + "step": 3728 + }, + { + "epoch": 5.46, + "grad_norm": 3.204758882522583, + "learning_rate": 8.374998221190056e-05, + "loss": 3.1617, + "step": 3729 + }, + { + "epoch": 5.46, + "grad_norm": 7.3266754150390625, + "learning_rate": 8.37037039738763e-05, + "loss": 3.1647, + "step": 3730 + }, + { + "epoch": 5.46, + "grad_norm": 108.37910461425781, + "learning_rate": 8.365742932092801e-05, + "loss": 3.1807, + "step": 3731 + }, + { + "epoch": 5.46, + "grad_norm": 4.3307623863220215, + "learning_rate": 8.361115826323572e-05, + "loss": 3.2109, + "step": 3732 + }, + { + "epoch": 5.47, + "grad_norm": 1.9860690832138062, + "learning_rate": 8.35648908109788e-05, + "loss": 3.1484, + "step": 3733 + }, + { + "epoch": 5.47, + "grad_norm": 3.039022207260132, + "learning_rate": 8.351862697433573e-05, + "loss": 3.175, + "step": 3734 + }, + { + "epoch": 5.47, + "grad_norm": 7.283559799194336, + "learning_rate": 8.347236676348426e-05, + "loss": 3.219, + "step": 3735 + }, + { + "epoch": 5.47, + "grad_norm": 3.774641513824463, + "learning_rate": 8.342611018860139e-05, + "loss": 3.1932, + "step": 3736 + }, + { + "epoch": 5.47, + "grad_norm": 4.264026641845703, + "learning_rate": 8.337985725986314e-05, + "loss": 3.193, + "step": 3737 + }, + { + "epoch": 5.47, + "grad_norm": 1.858871340751648, + "learning_rate": 8.333360798744496e-05, + "loss": 3.122, + "step": 3738 + }, + { + "epoch": 5.47, + "grad_norm": 4.920361518859863, + "learning_rate": 8.328736238152133e-05, + "loss": 3.1425, + "step": 3739 + }, + { + "epoch": 5.48, + "grad_norm": 4.10701847076416, + "learning_rate": 8.324112045226593e-05, + "loss": 3.1595, + "step": 3740 + }, + { + "epoch": 5.48, + "grad_norm": 5.043720245361328, + "learning_rate": 8.319488220985174e-05, + "loss": 3.1407, + "step": 3741 + }, + { + "epoch": 5.48, + "grad_norm": 3.795624256134033, + "learning_rate": 8.314864766445088e-05, + "loss": 3.2385, + "step": 3742 + }, + { + "epoch": 5.48, + "grad_norm": 3.801236867904663, + "learning_rate": 8.310241682623454e-05, + "loss": 3.152, + "step": 3743 + }, + { + "epoch": 5.48, + "grad_norm": 2.491832971572876, + "learning_rate": 8.30561897053733e-05, + "loss": 3.1737, + "step": 3744 + }, + { + "epoch": 5.48, + "grad_norm": 2.1997487545013428, + "learning_rate": 8.300996631203675e-05, + "loss": 3.1273, + "step": 3745 + }, + { + "epoch": 5.49, + "grad_norm": 2.1604340076446533, + "learning_rate": 8.296374665639373e-05, + "loss": 3.2015, + "step": 3746 + }, + { + "epoch": 5.49, + "grad_norm": 14.758793830871582, + "learning_rate": 8.291753074861233e-05, + "loss": 3.1594, + "step": 3747 + }, + { + "epoch": 5.49, + "grad_norm": 1.7705718278884888, + "learning_rate": 8.287131859885964e-05, + "loss": 3.1615, + "step": 3748 + }, + { + "epoch": 5.49, + "grad_norm": 3.673743724822998, + "learning_rate": 8.282511021730206e-05, + "loss": 3.1665, + "step": 3749 + }, + { + "epoch": 5.49, + "grad_norm": 2.580552101135254, + "learning_rate": 8.277890561410518e-05, + "loss": 3.1972, + "step": 3750 + }, + { + "epoch": 5.49, + "grad_norm": 1.8429394960403442, + "learning_rate": 8.273270479943361e-05, + "loss": 3.1316, + "step": 3751 + }, + { + "epoch": 5.49, + "grad_norm": 1.4963501691818237, + "learning_rate": 8.268650778345126e-05, + "loss": 3.1498, + "step": 3752 + }, + { + "epoch": 5.5, + "grad_norm": 3.3234293460845947, + "learning_rate": 8.26403145763212e-05, + "loss": 3.1475, + "step": 3753 + }, + { + "epoch": 5.5, + "grad_norm": 3.2355213165283203, + "learning_rate": 8.25941251882056e-05, + "loss": 3.1404, + "step": 3754 + }, + { + "epoch": 5.5, + "grad_norm": 1.3712440729141235, + "learning_rate": 8.254793962926579e-05, + "loss": 3.1316, + "step": 3755 + }, + { + "epoch": 5.5, + "grad_norm": 1.9082093238830566, + "learning_rate": 8.250175790966228e-05, + "loss": 3.1388, + "step": 3756 + }, + { + "epoch": 5.5, + "grad_norm": 2.096512794494629, + "learning_rate": 8.245558003955481e-05, + "loss": 3.1753, + "step": 3757 + }, + { + "epoch": 5.5, + "grad_norm": 1.1812965869903564, + "learning_rate": 8.240940602910212e-05, + "loss": 3.1647, + "step": 3758 + }, + { + "epoch": 5.5, + "grad_norm": 2.799182415008545, + "learning_rate": 8.23632358884622e-05, + "loss": 3.1474, + "step": 3759 + }, + { + "epoch": 5.51, + "grad_norm": 2.4367105960845947, + "learning_rate": 8.231706962779222e-05, + "loss": 3.1312, + "step": 3760 + }, + { + "epoch": 5.51, + "grad_norm": 2.1872997283935547, + "learning_rate": 8.227090725724842e-05, + "loss": 3.0935, + "step": 3761 + }, + { + "epoch": 5.51, + "grad_norm": 3.3611397743225098, + "learning_rate": 8.222474878698617e-05, + "loss": 3.0987, + "step": 3762 + }, + { + "epoch": 5.51, + "grad_norm": 1.0651507377624512, + "learning_rate": 8.217859422716012e-05, + "loss": 3.183, + "step": 3763 + }, + { + "epoch": 5.51, + "grad_norm": 0.8456958532333374, + "learning_rate": 8.213244358792389e-05, + "loss": 3.123, + "step": 3764 + }, + { + "epoch": 5.51, + "grad_norm": 0.9663138389587402, + "learning_rate": 8.208629687943029e-05, + "loss": 3.16, + "step": 3765 + }, + { + "epoch": 5.51, + "grad_norm": 135.12335205078125, + "learning_rate": 8.204015411183135e-05, + "loss": 3.1066, + "step": 3766 + }, + { + "epoch": 5.52, + "grad_norm": 1.7855501174926758, + "learning_rate": 8.199401529527816e-05, + "loss": 3.0972, + "step": 3767 + }, + { + "epoch": 5.52, + "grad_norm": 1.187641978263855, + "learning_rate": 8.194788043992089e-05, + "loss": 3.1476, + "step": 3768 + }, + { + "epoch": 5.52, + "grad_norm": 6.5035834312438965, + "learning_rate": 8.190174955590897e-05, + "loss": 3.134, + "step": 3769 + }, + { + "epoch": 5.52, + "grad_norm": 3.347416400909424, + "learning_rate": 8.185562265339088e-05, + "loss": 3.1592, + "step": 3770 + }, + { + "epoch": 5.52, + "grad_norm": 2.2179319858551025, + "learning_rate": 8.180949974251416e-05, + "loss": 3.1656, + "step": 3771 + }, + { + "epoch": 5.52, + "grad_norm": 3.2867071628570557, + "learning_rate": 8.176338083342561e-05, + "loss": 3.1231, + "step": 3772 + }, + { + "epoch": 5.53, + "grad_norm": 4.15543794631958, + "learning_rate": 8.17172659362711e-05, + "loss": 3.1501, + "step": 3773 + }, + { + "epoch": 5.53, + "grad_norm": 3.2263450622558594, + "learning_rate": 8.167115506119549e-05, + "loss": 3.1269, + "step": 3774 + }, + { + "epoch": 5.53, + "grad_norm": 1.2961052656173706, + "learning_rate": 8.162504821834295e-05, + "loss": 3.1204, + "step": 3775 + }, + { + "epoch": 5.53, + "grad_norm": 2.9037585258483887, + "learning_rate": 8.157894541785671e-05, + "loss": 3.1801, + "step": 3776 + }, + { + "epoch": 5.53, + "grad_norm": 2.3933663368225098, + "learning_rate": 8.153284666987897e-05, + "loss": 3.1537, + "step": 3777 + }, + { + "epoch": 5.53, + "grad_norm": 1.6720260381698608, + "learning_rate": 8.148675198455123e-05, + "loss": 3.1501, + "step": 3778 + }, + { + "epoch": 5.53, + "grad_norm": 11.296658515930176, + "learning_rate": 8.144066137201398e-05, + "loss": 3.1364, + "step": 3779 + }, + { + "epoch": 5.54, + "grad_norm": 14.168936729431152, + "learning_rate": 8.139457484240687e-05, + "loss": 3.1629, + "step": 3780 + }, + { + "epoch": 5.54, + "grad_norm": 5.073040008544922, + "learning_rate": 8.134849240586854e-05, + "loss": 3.1765, + "step": 3781 + }, + { + "epoch": 5.54, + "grad_norm": 2.859703779220581, + "learning_rate": 8.130241407253693e-05, + "loss": 3.1252, + "step": 3782 + }, + { + "epoch": 5.54, + "grad_norm": 6.422140121459961, + "learning_rate": 8.125633985254893e-05, + "loss": 3.126, + "step": 3783 + }, + { + "epoch": 5.54, + "grad_norm": 16.3258056640625, + "learning_rate": 8.121026975604052e-05, + "loss": 3.1953, + "step": 3784 + }, + { + "epoch": 5.54, + "grad_norm": 7.476664066314697, + "learning_rate": 8.116420379314684e-05, + "loss": 3.1981, + "step": 3785 + }, + { + "epoch": 5.54, + "grad_norm": 57.822113037109375, + "learning_rate": 8.111814197400213e-05, + "loss": 3.1517, + "step": 3786 + }, + { + "epoch": 5.55, + "grad_norm": 6.158851146697998, + "learning_rate": 8.10720843087396e-05, + "loss": 3.1655, + "step": 3787 + }, + { + "epoch": 5.55, + "grad_norm": 7.291560173034668, + "learning_rate": 8.102603080749171e-05, + "loss": 3.113, + "step": 3788 + }, + { + "epoch": 5.55, + "grad_norm": 6.275900840759277, + "learning_rate": 8.097998148038985e-05, + "loss": 3.1209, + "step": 3789 + }, + { + "epoch": 5.55, + "grad_norm": 4.1805338859558105, + "learning_rate": 8.093393633756459e-05, + "loss": 3.1049, + "step": 3790 + }, + { + "epoch": 5.55, + "grad_norm": 6.27461051940918, + "learning_rate": 8.088789538914561e-05, + "loss": 3.169, + "step": 3791 + }, + { + "epoch": 5.55, + "grad_norm": 63.376373291015625, + "learning_rate": 8.084185864526153e-05, + "loss": 3.2346, + "step": 3792 + }, + { + "epoch": 5.55, + "grad_norm": 17.007471084594727, + "learning_rate": 8.079582611604015e-05, + "loss": 3.1799, + "step": 3793 + }, + { + "epoch": 5.56, + "grad_norm": 3.8963406085968018, + "learning_rate": 8.074979781160836e-05, + "loss": 3.1546, + "step": 3794 + }, + { + "epoch": 5.56, + "grad_norm": 1.5541036128997803, + "learning_rate": 8.070377374209202e-05, + "loss": 3.17, + "step": 3795 + }, + { + "epoch": 5.56, + "grad_norm": 4.511335372924805, + "learning_rate": 8.06577539176161e-05, + "loss": 3.1307, + "step": 3796 + }, + { + "epoch": 5.56, + "grad_norm": 3.2192275524139404, + "learning_rate": 8.061173834830476e-05, + "loss": 3.1262, + "step": 3797 + }, + { + "epoch": 5.56, + "grad_norm": 10.192057609558105, + "learning_rate": 8.056572704428103e-05, + "loss": 3.1341, + "step": 3798 + }, + { + "epoch": 5.56, + "grad_norm": 10.097148895263672, + "learning_rate": 8.051972001566708e-05, + "loss": 3.1627, + "step": 3799 + }, + { + "epoch": 5.57, + "grad_norm": 5.317000865936279, + "learning_rate": 8.04737172725842e-05, + "loss": 3.1695, + "step": 3800 + }, + { + "epoch": 5.57, + "grad_norm": 3.3502695560455322, + "learning_rate": 8.042771882515265e-05, + "loss": 3.1111, + "step": 3801 + }, + { + "epoch": 5.57, + "grad_norm": 3.3459572792053223, + "learning_rate": 8.038172468349178e-05, + "loss": 3.1737, + "step": 3802 + }, + { + "epoch": 5.57, + "grad_norm": 4.4290900230407715, + "learning_rate": 8.033573485772001e-05, + "loss": 3.1335, + "step": 3803 + }, + { + "epoch": 5.57, + "grad_norm": 3.444740056991577, + "learning_rate": 8.028974935795481e-05, + "loss": 3.1575, + "step": 3804 + }, + { + "epoch": 5.57, + "grad_norm": 6.389871120452881, + "learning_rate": 8.024376819431266e-05, + "loss": 3.1654, + "step": 3805 + }, + { + "epoch": 5.57, + "grad_norm": 4.629978179931641, + "learning_rate": 8.019779137690906e-05, + "loss": 3.1554, + "step": 3806 + }, + { + "epoch": 5.58, + "grad_norm": 2.2746245861053467, + "learning_rate": 8.015181891585871e-05, + "loss": 3.1065, + "step": 3807 + }, + { + "epoch": 5.58, + "grad_norm": 4.390033721923828, + "learning_rate": 8.010585082127513e-05, + "loss": 3.142, + "step": 3808 + }, + { + "epoch": 5.58, + "grad_norm": 8.950037956237793, + "learning_rate": 8.005988710327105e-05, + "loss": 3.1759, + "step": 3809 + }, + { + "epoch": 5.58, + "grad_norm": 7.786754608154297, + "learning_rate": 8.001392777195821e-05, + "loss": 3.1518, + "step": 3810 + }, + { + "epoch": 5.58, + "grad_norm": 1.803338646888733, + "learning_rate": 7.996797283744732e-05, + "loss": 3.1452, + "step": 3811 + }, + { + "epoch": 5.58, + "grad_norm": 5.436160087585449, + "learning_rate": 7.992202230984811e-05, + "loss": 3.1284, + "step": 3812 + }, + { + "epoch": 5.58, + "grad_norm": 5.9366912841796875, + "learning_rate": 7.987607619926946e-05, + "loss": 3.1547, + "step": 3813 + }, + { + "epoch": 5.59, + "grad_norm": 5.989798545837402, + "learning_rate": 7.98301345158192e-05, + "loss": 3.1264, + "step": 3814 + }, + { + "epoch": 5.59, + "grad_norm": 2.005276679992676, + "learning_rate": 7.978419726960413e-05, + "loss": 3.1387, + "step": 3815 + }, + { + "epoch": 5.59, + "grad_norm": 4.961897373199463, + "learning_rate": 7.97382644707302e-05, + "loss": 3.1506, + "step": 3816 + }, + { + "epoch": 5.59, + "grad_norm": 5.611112594604492, + "learning_rate": 7.969233612930234e-05, + "loss": 3.1237, + "step": 3817 + }, + { + "epoch": 5.59, + "grad_norm": 2.2314023971557617, + "learning_rate": 7.964641225542438e-05, + "loss": 3.1433, + "step": 3818 + }, + { + "epoch": 5.59, + "grad_norm": 27.2403564453125, + "learning_rate": 7.960049285919933e-05, + "loss": 3.1587, + "step": 3819 + }, + { + "epoch": 5.59, + "grad_norm": 7.609023571014404, + "learning_rate": 7.955457795072918e-05, + "loss": 3.2031, + "step": 3820 + }, + { + "epoch": 5.6, + "grad_norm": 6.942310810089111, + "learning_rate": 7.95086675401148e-05, + "loss": 3.1571, + "step": 3821 + }, + { + "epoch": 5.6, + "grad_norm": 5.55133056640625, + "learning_rate": 7.94627616374563e-05, + "loss": 3.1465, + "step": 3822 + }, + { + "epoch": 5.6, + "grad_norm": 2.6575374603271484, + "learning_rate": 7.941686025285257e-05, + "loss": 3.1384, + "step": 3823 + }, + { + "epoch": 5.6, + "grad_norm": 4.183603763580322, + "learning_rate": 7.937096339640162e-05, + "loss": 3.1551, + "step": 3824 + }, + { + "epoch": 5.6, + "grad_norm": 203.84231567382812, + "learning_rate": 7.932507107820051e-05, + "loss": 3.1756, + "step": 3825 + }, + { + "epoch": 5.6, + "grad_norm": 1.8645808696746826, + "learning_rate": 7.92791833083452e-05, + "loss": 3.1254, + "step": 3826 + }, + { + "epoch": 5.61, + "grad_norm": 3.2966809272766113, + "learning_rate": 7.923330009693071e-05, + "loss": 3.1409, + "step": 3827 + }, + { + "epoch": 5.61, + "grad_norm": 16.701486587524414, + "learning_rate": 7.918742145405098e-05, + "loss": 3.1687, + "step": 3828 + }, + { + "epoch": 5.61, + "grad_norm": 5.320707321166992, + "learning_rate": 7.914154738979908e-05, + "loss": 3.166, + "step": 3829 + }, + { + "epoch": 5.61, + "grad_norm": 3.770569086074829, + "learning_rate": 7.909567791426697e-05, + "loss": 3.129, + "step": 3830 + }, + { + "epoch": 5.61, + "grad_norm": 1.9080885648727417, + "learning_rate": 7.904981303754561e-05, + "loss": 3.1792, + "step": 3831 + }, + { + "epoch": 5.61, + "grad_norm": 2.2089335918426514, + "learning_rate": 7.900395276972502e-05, + "loss": 3.1417, + "step": 3832 + }, + { + "epoch": 5.61, + "grad_norm": 2.3139476776123047, + "learning_rate": 7.895809712089406e-05, + "loss": 3.1278, + "step": 3833 + }, + { + "epoch": 5.62, + "grad_norm": 3.4870378971099854, + "learning_rate": 7.891224610114072e-05, + "loss": 3.1562, + "step": 3834 + }, + { + "epoch": 5.62, + "grad_norm": 4.3670244216918945, + "learning_rate": 7.886639972055195e-05, + "loss": 3.1389, + "step": 3835 + }, + { + "epoch": 5.62, + "grad_norm": 5.9168596267700195, + "learning_rate": 7.882055798921358e-05, + "loss": 3.1104, + "step": 3836 + }, + { + "epoch": 5.62, + "grad_norm": 7.184929370880127, + "learning_rate": 7.87747209172105e-05, + "loss": 3.1555, + "step": 3837 + }, + { + "epoch": 5.62, + "grad_norm": 4.011050701141357, + "learning_rate": 7.872888851462663e-05, + "loss": 3.1192, + "step": 3838 + }, + { + "epoch": 5.62, + "grad_norm": 1.6208895444869995, + "learning_rate": 7.86830607915447e-05, + "loss": 3.1861, + "step": 3839 + }, + { + "epoch": 5.62, + "grad_norm": 1.3240761756896973, + "learning_rate": 7.863723775804651e-05, + "loss": 3.1656, + "step": 3840 + }, + { + "epoch": 5.63, + "grad_norm": 2.556436538696289, + "learning_rate": 7.85914194242129e-05, + "loss": 3.1392, + "step": 3841 + }, + { + "epoch": 5.63, + "grad_norm": 1.8541210889816284, + "learning_rate": 7.854560580012351e-05, + "loss": 3.1391, + "step": 3842 + }, + { + "epoch": 5.63, + "grad_norm": 2.840593099594116, + "learning_rate": 7.849979689585705e-05, + "loss": 3.0979, + "step": 3843 + }, + { + "epoch": 5.63, + "grad_norm": 1.50877845287323, + "learning_rate": 7.845399272149124e-05, + "loss": 3.1231, + "step": 3844 + }, + { + "epoch": 5.63, + "grad_norm": 2.6628737449645996, + "learning_rate": 7.84081932871026e-05, + "loss": 3.1347, + "step": 3845 + }, + { + "epoch": 5.63, + "grad_norm": 33.729000091552734, + "learning_rate": 7.836239860276671e-05, + "loss": 3.1319, + "step": 3846 + }, + { + "epoch": 5.64, + "grad_norm": 13.277280807495117, + "learning_rate": 7.831660867855816e-05, + "loss": 3.1439, + "step": 3847 + }, + { + "epoch": 5.64, + "grad_norm": 1.8134537935256958, + "learning_rate": 7.827082352455039e-05, + "loss": 3.1507, + "step": 3848 + }, + { + "epoch": 5.64, + "grad_norm": 2.101228952407837, + "learning_rate": 7.822504315081578e-05, + "loss": 3.1133, + "step": 3849 + }, + { + "epoch": 5.64, + "grad_norm": 2.533062696456909, + "learning_rate": 7.817926756742576e-05, + "loss": 3.1362, + "step": 3850 + }, + { + "epoch": 5.64, + "grad_norm": 4.869479656219482, + "learning_rate": 7.813349678445067e-05, + "loss": 3.1618, + "step": 3851 + }, + { + "epoch": 5.64, + "grad_norm": 2.6743147373199463, + "learning_rate": 7.808773081195972e-05, + "loss": 3.1278, + "step": 3852 + }, + { + "epoch": 5.64, + "grad_norm": 0.8585074543952942, + "learning_rate": 7.804196966002113e-05, + "loss": 3.1386, + "step": 3853 + }, + { + "epoch": 5.65, + "grad_norm": 7.510640621185303, + "learning_rate": 7.799621333870208e-05, + "loss": 3.1786, + "step": 3854 + }, + { + "epoch": 5.65, + "grad_norm": 2.670872449874878, + "learning_rate": 7.795046185806866e-05, + "loss": 3.1483, + "step": 3855 + }, + { + "epoch": 5.65, + "grad_norm": 22.928964614868164, + "learning_rate": 7.79047152281858e-05, + "loss": 3.1633, + "step": 3856 + }, + { + "epoch": 5.65, + "grad_norm": 7.530433177947998, + "learning_rate": 7.785897345911754e-05, + "loss": 3.1601, + "step": 3857 + }, + { + "epoch": 5.65, + "grad_norm": 7.938250541687012, + "learning_rate": 7.781323656092677e-05, + "loss": 3.1554, + "step": 3858 + }, + { + "epoch": 5.65, + "grad_norm": 2.2199909687042236, + "learning_rate": 7.776750454367521e-05, + "loss": 3.1512, + "step": 3859 + }, + { + "epoch": 5.65, + "grad_norm": 92.11273956298828, + "learning_rate": 7.772177741742369e-05, + "loss": 3.1275, + "step": 3860 + }, + { + "epoch": 5.66, + "grad_norm": 4.906170845031738, + "learning_rate": 7.767605519223183e-05, + "loss": 3.1245, + "step": 3861 + }, + { + "epoch": 5.66, + "grad_norm": 3.870858907699585, + "learning_rate": 7.76303378781582e-05, + "loss": 3.1105, + "step": 3862 + }, + { + "epoch": 5.66, + "grad_norm": 2.2505300045013428, + "learning_rate": 7.758462548526033e-05, + "loss": 3.1065, + "step": 3863 + }, + { + "epoch": 5.66, + "grad_norm": 2.947871208190918, + "learning_rate": 7.753891802359466e-05, + "loss": 3.1869, + "step": 3864 + }, + { + "epoch": 5.66, + "grad_norm": 3.654209613800049, + "learning_rate": 7.749321550321645e-05, + "loss": 3.1427, + "step": 3865 + }, + { + "epoch": 5.66, + "grad_norm": 3.259089469909668, + "learning_rate": 7.744751793418005e-05, + "loss": 3.1221, + "step": 3866 + }, + { + "epoch": 5.66, + "grad_norm": 2.3307595252990723, + "learning_rate": 7.740182532653853e-05, + "loss": 3.1556, + "step": 3867 + }, + { + "epoch": 5.67, + "grad_norm": 6.026763439178467, + "learning_rate": 7.735613769034399e-05, + "loss": 3.1646, + "step": 3868 + }, + { + "epoch": 5.67, + "grad_norm": 9.292561531066895, + "learning_rate": 7.731045503564743e-05, + "loss": 3.1498, + "step": 3869 + }, + { + "epoch": 5.67, + "grad_norm": 4.8578314781188965, + "learning_rate": 7.726477737249869e-05, + "loss": 3.1159, + "step": 3870 + }, + { + "epoch": 5.67, + "grad_norm": 5.385825157165527, + "learning_rate": 7.721910471094653e-05, + "loss": 3.1494, + "step": 3871 + }, + { + "epoch": 5.67, + "grad_norm": 3.9133341312408447, + "learning_rate": 7.717343706103873e-05, + "loss": 3.1402, + "step": 3872 + }, + { + "epoch": 5.67, + "grad_norm": 6.215272903442383, + "learning_rate": 7.712777443282175e-05, + "loss": 3.195, + "step": 3873 + }, + { + "epoch": 5.68, + "grad_norm": 3.6820201873779297, + "learning_rate": 7.708211683634112e-05, + "loss": 3.2287, + "step": 3874 + }, + { + "epoch": 5.68, + "grad_norm": 3.4176886081695557, + "learning_rate": 7.703646428164122e-05, + "loss": 3.1445, + "step": 3875 + }, + { + "epoch": 5.68, + "grad_norm": 8.820064544677734, + "learning_rate": 7.699081677876526e-05, + "loss": 3.2085, + "step": 3876 + }, + { + "epoch": 5.68, + "grad_norm": 17.15256118774414, + "learning_rate": 7.694517433775543e-05, + "loss": 3.1627, + "step": 3877 + }, + { + "epoch": 5.68, + "grad_norm": 1.7407811880111694, + "learning_rate": 7.689953696865271e-05, + "loss": 3.1868, + "step": 3878 + }, + { + "epoch": 5.68, + "grad_norm": 3.167860746383667, + "learning_rate": 7.685390468149707e-05, + "loss": 3.1838, + "step": 3879 + }, + { + "epoch": 5.68, + "grad_norm": 4.515937805175781, + "learning_rate": 7.680827748632728e-05, + "loss": 3.1624, + "step": 3880 + }, + { + "epoch": 5.69, + "grad_norm": 5.7442755699157715, + "learning_rate": 7.676265539318098e-05, + "loss": 3.1957, + "step": 3881 + }, + { + "epoch": 5.69, + "grad_norm": 4.945896625518799, + "learning_rate": 7.671703841209481e-05, + "loss": 3.1714, + "step": 3882 + }, + { + "epoch": 5.69, + "grad_norm": 8.167930603027344, + "learning_rate": 7.667142655310414e-05, + "loss": 3.1944, + "step": 3883 + }, + { + "epoch": 5.69, + "grad_norm": 17.595746994018555, + "learning_rate": 7.662581982624326e-05, + "loss": 3.1853, + "step": 3884 + }, + { + "epoch": 5.69, + "grad_norm": 2.751532554626465, + "learning_rate": 7.658021824154542e-05, + "loss": 3.1348, + "step": 3885 + }, + { + "epoch": 5.69, + "grad_norm": 6.661686897277832, + "learning_rate": 7.653462180904258e-05, + "loss": 3.2022, + "step": 3886 + }, + { + "epoch": 5.69, + "grad_norm": 9.880109786987305, + "learning_rate": 7.648903053876568e-05, + "loss": 3.1686, + "step": 3887 + }, + { + "epoch": 5.69, + "eval_loss": 2.989283800125122, + "eval_runtime": 1170.3187, + "eval_samples_per_second": 14.647, + "eval_steps_per_second": 0.916, + "step": 3887 + }, + { + "epoch": 5.7, + "grad_norm": 3.414788246154785, + "learning_rate": 7.644344444074453e-05, + "loss": 3.1669, + "step": 3888 + }, + { + "epoch": 5.7, + "grad_norm": 3.5142674446105957, + "learning_rate": 7.639786352500773e-05, + "loss": 3.1526, + "step": 3889 + }, + { + "epoch": 5.7, + "grad_norm": 7.385360240936279, + "learning_rate": 7.635228780158274e-05, + "loss": 3.1528, + "step": 3890 + }, + { + "epoch": 5.7, + "grad_norm": 7.439982891082764, + "learning_rate": 7.630671728049598e-05, + "loss": 3.1417, + "step": 3891 + }, + { + "epoch": 5.7, + "grad_norm": 4.842582702636719, + "learning_rate": 7.626115197177266e-05, + "loss": 3.2058, + "step": 3892 + }, + { + "epoch": 5.7, + "grad_norm": 6.158822536468506, + "learning_rate": 7.621559188543677e-05, + "loss": 3.2051, + "step": 3893 + }, + { + "epoch": 5.7, + "grad_norm": 4.262391090393066, + "learning_rate": 7.617003703151129e-05, + "loss": 3.2001, + "step": 3894 + }, + { + "epoch": 5.71, + "grad_norm": 22.497093200683594, + "learning_rate": 7.612448742001797e-05, + "loss": 3.1808, + "step": 3895 + }, + { + "epoch": 5.71, + "grad_norm": 3.1336076259613037, + "learning_rate": 7.607894306097739e-05, + "loss": 3.2113, + "step": 3896 + }, + { + "epoch": 5.71, + "grad_norm": 4.773746967315674, + "learning_rate": 7.603340396440902e-05, + "loss": 3.1906, + "step": 3897 + }, + { + "epoch": 5.71, + "grad_norm": 5.188856601715088, + "learning_rate": 7.59878701403312e-05, + "loss": 3.1663, + "step": 3898 + }, + { + "epoch": 5.71, + "grad_norm": 7.35548734664917, + "learning_rate": 7.594234159876098e-05, + "loss": 3.1467, + "step": 3899 + }, + { + "epoch": 5.71, + "grad_norm": 7.549004077911377, + "learning_rate": 7.589681834971443e-05, + "loss": 3.1928, + "step": 3900 + }, + { + "epoch": 5.72, + "grad_norm": 6.5583086013793945, + "learning_rate": 7.585130040320627e-05, + "loss": 3.2067, + "step": 3901 + }, + { + "epoch": 5.72, + "grad_norm": 14.854689598083496, + "learning_rate": 7.580578776925022e-05, + "loss": 3.2208, + "step": 3902 + }, + { + "epoch": 5.72, + "grad_norm": 11.917444229125977, + "learning_rate": 7.576028045785868e-05, + "loss": 3.1828, + "step": 3903 + }, + { + "epoch": 5.72, + "grad_norm": 6.152881145477295, + "learning_rate": 7.5714778479043e-05, + "loss": 3.204, + "step": 3904 + }, + { + "epoch": 5.72, + "grad_norm": 4.3145294189453125, + "learning_rate": 7.566928184281334e-05, + "loss": 3.2282, + "step": 3905 + }, + { + "epoch": 5.72, + "grad_norm": 3.392911434173584, + "learning_rate": 7.562379055917856e-05, + "loss": 3.1795, + "step": 3906 + }, + { + "epoch": 5.72, + "grad_norm": 5.250206470489502, + "learning_rate": 7.557830463814652e-05, + "loss": 3.1574, + "step": 3907 + }, + { + "epoch": 5.73, + "grad_norm": 11.659444808959961, + "learning_rate": 7.553282408972382e-05, + "loss": 3.1975, + "step": 3908 + }, + { + "epoch": 5.73, + "grad_norm": 6.504120826721191, + "learning_rate": 7.54873489239158e-05, + "loss": 3.2318, + "step": 3909 + }, + { + "epoch": 5.73, + "grad_norm": 3.6852314472198486, + "learning_rate": 7.544187915072676e-05, + "loss": 3.1501, + "step": 3910 + }, + { + "epoch": 5.73, + "grad_norm": 4.429998874664307, + "learning_rate": 7.539641478015976e-05, + "loss": 3.1623, + "step": 3911 + }, + { + "epoch": 5.73, + "grad_norm": 2.792065382003784, + "learning_rate": 7.535095582221656e-05, + "loss": 3.1624, + "step": 3912 + }, + { + "epoch": 5.73, + "grad_norm": 21.118961334228516, + "learning_rate": 7.530550228689794e-05, + "loss": 3.2112, + "step": 3913 + }, + { + "epoch": 5.73, + "grad_norm": 18.30959701538086, + "learning_rate": 7.526005418420332e-05, + "loss": 3.2394, + "step": 3914 + }, + { + "epoch": 5.74, + "grad_norm": 8.257102012634277, + "learning_rate": 7.521461152413092e-05, + "loss": 3.2588, + "step": 3915 + }, + { + "epoch": 5.74, + "grad_norm": 3.5304644107818604, + "learning_rate": 7.516917431667795e-05, + "loss": 3.2331, + "step": 3916 + }, + { + "epoch": 5.74, + "grad_norm": 24.953046798706055, + "learning_rate": 7.512374257184019e-05, + "loss": 3.1763, + "step": 3917 + }, + { + "epoch": 5.74, + "grad_norm": 5.283435821533203, + "learning_rate": 7.507831629961234e-05, + "loss": 3.1631, + "step": 3918 + }, + { + "epoch": 5.74, + "grad_norm": 7.136661052703857, + "learning_rate": 7.503289550998791e-05, + "loss": 3.2551, + "step": 3919 + }, + { + "epoch": 5.74, + "grad_norm": 3.8187716007232666, + "learning_rate": 7.498748021295914e-05, + "loss": 3.1214, + "step": 3920 + }, + { + "epoch": 5.74, + "grad_norm": 10.967880249023438, + "learning_rate": 7.494207041851708e-05, + "loss": 3.144, + "step": 3921 + }, + { + "epoch": 5.75, + "grad_norm": 30.177751541137695, + "learning_rate": 7.489666613665163e-05, + "loss": 3.1466, + "step": 3922 + }, + { + "epoch": 5.75, + "grad_norm": 6.267055511474609, + "learning_rate": 7.48512673773514e-05, + "loss": 3.2192, + "step": 3923 + }, + { + "epoch": 5.75, + "grad_norm": 3.4286611080169678, + "learning_rate": 7.480587415060378e-05, + "loss": 3.1519, + "step": 3924 + }, + { + "epoch": 5.75, + "grad_norm": 9.199522972106934, + "learning_rate": 7.476048646639498e-05, + "loss": 3.1738, + "step": 3925 + }, + { + "epoch": 5.75, + "grad_norm": 9.091236114501953, + "learning_rate": 7.471510433471005e-05, + "loss": 3.1704, + "step": 3926 + }, + { + "epoch": 5.75, + "grad_norm": 7.052969455718994, + "learning_rate": 7.46697277655327e-05, + "loss": 3.1835, + "step": 3927 + }, + { + "epoch": 5.76, + "grad_norm": 30.890663146972656, + "learning_rate": 7.462435676884547e-05, + "loss": 3.1883, + "step": 3928 + }, + { + "epoch": 5.76, + "grad_norm": 3.2805261611938477, + "learning_rate": 7.457899135462974e-05, + "loss": 3.2009, + "step": 3929 + }, + { + "epoch": 5.76, + "grad_norm": 11.584432601928711, + "learning_rate": 7.45336315328655e-05, + "loss": 3.2384, + "step": 3930 + }, + { + "epoch": 5.76, + "grad_norm": 6.337993144989014, + "learning_rate": 7.448827731353166e-05, + "loss": 3.1973, + "step": 3931 + }, + { + "epoch": 5.76, + "grad_norm": 13.461885452270508, + "learning_rate": 7.444292870660586e-05, + "loss": 3.2435, + "step": 3932 + }, + { + "epoch": 5.76, + "grad_norm": 9.56443977355957, + "learning_rate": 7.439758572206448e-05, + "loss": 3.1465, + "step": 3933 + }, + { + "epoch": 5.76, + "grad_norm": 7.9625115394592285, + "learning_rate": 7.43522483698826e-05, + "loss": 3.1635, + "step": 3934 + }, + { + "epoch": 5.77, + "grad_norm": 6.776467800140381, + "learning_rate": 7.430691666003424e-05, + "loss": 3.2405, + "step": 3935 + }, + { + "epoch": 5.77, + "grad_norm": 43.86566162109375, + "learning_rate": 7.426159060249204e-05, + "loss": 3.192, + "step": 3936 + }, + { + "epoch": 5.77, + "grad_norm": 12.768895149230957, + "learning_rate": 7.421627020722738e-05, + "loss": 3.1741, + "step": 3937 + }, + { + "epoch": 5.77, + "grad_norm": 6.572139263153076, + "learning_rate": 7.41709554842105e-05, + "loss": 3.1637, + "step": 3938 + }, + { + "epoch": 5.77, + "grad_norm": 5.105015277862549, + "learning_rate": 7.412564644341033e-05, + "loss": 3.2329, + "step": 3939 + }, + { + "epoch": 5.77, + "grad_norm": 4.994419574737549, + "learning_rate": 7.40803430947945e-05, + "loss": 3.2124, + "step": 3940 + }, + { + "epoch": 5.77, + "grad_norm": 13.615537643432617, + "learning_rate": 7.403504544832951e-05, + "loss": 3.2148, + "step": 3941 + }, + { + "epoch": 5.78, + "grad_norm": 6.730780124664307, + "learning_rate": 7.398975351398053e-05, + "loss": 3.2086, + "step": 3942 + }, + { + "epoch": 5.78, + "grad_norm": 8.399843215942383, + "learning_rate": 7.394446730171141e-05, + "loss": 3.2455, + "step": 3943 + }, + { + "epoch": 5.78, + "grad_norm": 7.892696857452393, + "learning_rate": 7.389918682148491e-05, + "loss": 3.263, + "step": 3944 + }, + { + "epoch": 5.78, + "grad_norm": 3.2410852909088135, + "learning_rate": 7.38539120832624e-05, + "loss": 3.2023, + "step": 3945 + }, + { + "epoch": 5.78, + "grad_norm": 43.567543029785156, + "learning_rate": 7.380864309700396e-05, + "loss": 3.2505, + "step": 3946 + }, + { + "epoch": 5.78, + "grad_norm": 8.303853988647461, + "learning_rate": 7.376337987266856e-05, + "loss": 3.2253, + "step": 3947 + }, + { + "epoch": 5.79, + "grad_norm": 5.137606620788574, + "learning_rate": 7.371812242021374e-05, + "loss": 3.2614, + "step": 3948 + }, + { + "epoch": 5.79, + "grad_norm": 7.571026802062988, + "learning_rate": 7.367287074959586e-05, + "loss": 3.3035, + "step": 3949 + }, + { + "epoch": 5.79, + "grad_norm": 5.098783016204834, + "learning_rate": 7.362762487076995e-05, + "loss": 3.24, + "step": 3950 + }, + { + "epoch": 5.79, + "grad_norm": 5.123687744140625, + "learning_rate": 7.358238479368985e-05, + "loss": 3.2333, + "step": 3951 + }, + { + "epoch": 5.79, + "grad_norm": 9.861023902893066, + "learning_rate": 7.353715052830808e-05, + "loss": 3.2665, + "step": 3952 + }, + { + "epoch": 5.79, + "grad_norm": 30.322845458984375, + "learning_rate": 7.349192208457579e-05, + "loss": 3.2337, + "step": 3953 + }, + { + "epoch": 5.79, + "grad_norm": 10.311440467834473, + "learning_rate": 7.344669947244302e-05, + "loss": 3.2307, + "step": 3954 + }, + { + "epoch": 5.8, + "grad_norm": 13.122282981872559, + "learning_rate": 7.340148270185842e-05, + "loss": 3.2456, + "step": 3955 + }, + { + "epoch": 5.8, + "grad_norm": 16.757612228393555, + "learning_rate": 7.335627178276937e-05, + "loss": 3.3208, + "step": 3956 + }, + { + "epoch": 5.8, + "grad_norm": 36.39468002319336, + "learning_rate": 7.331106672512197e-05, + "loss": 3.2313, + "step": 3957 + }, + { + "epoch": 5.8, + "grad_norm": 39.07672882080078, + "learning_rate": 7.326586753886102e-05, + "loss": 3.3304, + "step": 3958 + }, + { + "epoch": 5.8, + "grad_norm": 67.7435302734375, + "learning_rate": 7.322067423393002e-05, + "loss": 3.3623, + "step": 3959 + }, + { + "epoch": 5.8, + "grad_norm": 102.08399963378906, + "learning_rate": 7.317548682027127e-05, + "loss": 3.3498, + "step": 3960 + }, + { + "epoch": 5.8, + "grad_norm": 68.23622131347656, + "learning_rate": 7.313030530782564e-05, + "loss": 3.4284, + "step": 3961 + }, + { + "epoch": 5.81, + "grad_norm": 54.879310607910156, + "learning_rate": 7.308512970653274e-05, + "loss": 3.4101, + "step": 3962 + }, + { + "epoch": 5.81, + "grad_norm": 154.8038787841797, + "learning_rate": 7.303996002633096e-05, + "loss": 3.5803, + "step": 3963 + }, + { + "epoch": 5.81, + "grad_norm": 228.81173706054688, + "learning_rate": 7.29947962771573e-05, + "loss": 3.489, + "step": 3964 + }, + { + "epoch": 5.81, + "grad_norm": 350.7447509765625, + "learning_rate": 7.294963846894744e-05, + "loss": 3.6323, + "step": 3965 + }, + { + "epoch": 5.81, + "grad_norm": 858.3496704101562, + "learning_rate": 7.290448661163587e-05, + "loss": 4.0812, + "step": 3966 + }, + { + "epoch": 5.81, + "grad_norm": 730.7702026367188, + "learning_rate": 7.285934071515568e-05, + "loss": 4.0206, + "step": 3967 + }, + { + "epoch": 5.81, + "grad_norm": 658.6021118164062, + "learning_rate": 7.281420078943855e-05, + "loss": 4.3312, + "step": 3968 + }, + { + "epoch": 5.82, + "grad_norm": 475.8223876953125, + "learning_rate": 7.276906684441514e-05, + "loss": 4.5484, + "step": 3969 + }, + { + "epoch": 5.82, + "grad_norm": 309.2330322265625, + "learning_rate": 7.272393889001454e-05, + "loss": 4.6918, + "step": 3970 + }, + { + "epoch": 5.82, + "grad_norm": 195.97572326660156, + "learning_rate": 7.267881693616452e-05, + "loss": 4.7661, + "step": 3971 + }, + { + "epoch": 5.82, + "grad_norm": 134.8620147705078, + "learning_rate": 7.263370099279172e-05, + "loss": 4.6478, + "step": 3972 + }, + { + "epoch": 5.82, + "grad_norm": 108.79113006591797, + "learning_rate": 7.25885910698213e-05, + "loss": 4.6325, + "step": 3973 + }, + { + "epoch": 5.82, + "grad_norm": 149.6044921875, + "learning_rate": 7.254348717717714e-05, + "loss": 4.8721, + "step": 3974 + }, + { + "epoch": 5.83, + "grad_norm": 117.6059341430664, + "learning_rate": 7.249838932478177e-05, + "loss": 4.8513, + "step": 3975 + }, + { + "epoch": 5.83, + "grad_norm": 56.039024353027344, + "learning_rate": 7.245329752255647e-05, + "loss": 4.8064, + "step": 3976 + }, + { + "epoch": 5.83, + "grad_norm": 100.5042953491211, + "learning_rate": 7.24082117804211e-05, + "loss": 4.8077, + "step": 3977 + }, + { + "epoch": 5.83, + "grad_norm": 101.55089569091797, + "learning_rate": 7.236313210829419e-05, + "loss": 4.9091, + "step": 3978 + }, + { + "epoch": 5.83, + "grad_norm": 54.18492889404297, + "learning_rate": 7.231805851609306e-05, + "loss": 4.9624, + "step": 3979 + }, + { + "epoch": 5.83, + "grad_norm": 90.85334014892578, + "learning_rate": 7.227299101373352e-05, + "loss": 4.8436, + "step": 3980 + }, + { + "epoch": 5.83, + "grad_norm": 108.68850708007812, + "learning_rate": 7.222792961113012e-05, + "loss": 4.7235, + "step": 3981 + }, + { + "epoch": 5.84, + "grad_norm": 46.6330451965332, + "learning_rate": 7.21828743181961e-05, + "loss": 4.7477, + "step": 3982 + }, + { + "epoch": 5.84, + "grad_norm": 116.49347686767578, + "learning_rate": 7.213782514484333e-05, + "loss": 4.6439, + "step": 3983 + }, + { + "epoch": 5.84, + "grad_norm": 46.60487747192383, + "learning_rate": 7.209278210098224e-05, + "loss": 4.6174, + "step": 3984 + }, + { + "epoch": 5.84, + "grad_norm": 74.72257995605469, + "learning_rate": 7.204774519652211e-05, + "loss": 4.6026, + "step": 3985 + }, + { + "epoch": 5.84, + "grad_norm": 117.54887390136719, + "learning_rate": 7.200271444137072e-05, + "loss": 4.3403, + "step": 3986 + }, + { + "epoch": 5.84, + "grad_norm": 66.1172866821289, + "learning_rate": 7.195768984543448e-05, + "loss": 4.2326, + "step": 3987 + }, + { + "epoch": 5.84, + "grad_norm": 23.685178756713867, + "learning_rate": 7.191267141861858e-05, + "loss": 4.2885, + "step": 3988 + }, + { + "epoch": 5.85, + "grad_norm": 33.490779876708984, + "learning_rate": 7.186765917082674e-05, + "loss": 4.2881, + "step": 3989 + }, + { + "epoch": 5.85, + "grad_norm": 49.65608215332031, + "learning_rate": 7.18226531119613e-05, + "loss": 4.0779, + "step": 3990 + }, + { + "epoch": 5.85, + "grad_norm": 21.95450782775879, + "learning_rate": 7.177765325192338e-05, + "loss": 4.0829, + "step": 3991 + }, + { + "epoch": 5.85, + "grad_norm": 36.09683609008789, + "learning_rate": 7.173265960061261e-05, + "loss": 4.1138, + "step": 3992 + }, + { + "epoch": 5.85, + "grad_norm": 135.8228302001953, + "learning_rate": 7.168767216792724e-05, + "loss": 4.0337, + "step": 3993 + }, + { + "epoch": 5.85, + "grad_norm": 42.24320602416992, + "learning_rate": 7.164269096376432e-05, + "loss": 4.0038, + "step": 3994 + }, + { + "epoch": 5.85, + "grad_norm": 90.61548614501953, + "learning_rate": 7.15977159980193e-05, + "loss": 3.9393, + "step": 3995 + }, + { + "epoch": 5.86, + "grad_norm": 83.08573150634766, + "learning_rate": 7.155274728058645e-05, + "loss": 4.085, + "step": 3996 + }, + { + "epoch": 5.86, + "grad_norm": 51.404388427734375, + "learning_rate": 7.150778482135851e-05, + "loss": 4.025, + "step": 3997 + }, + { + "epoch": 5.86, + "grad_norm": 133.36476135253906, + "learning_rate": 7.146282863022699e-05, + "loss": 4.0341, + "step": 3998 + }, + { + "epoch": 5.86, + "grad_norm": 68.87841033935547, + "learning_rate": 7.141787871708194e-05, + "loss": 4.1013, + "step": 3999 + }, + { + "epoch": 5.86, + "grad_norm": 83.18514251708984, + "learning_rate": 7.137293509181197e-05, + "loss": 4.0575, + "step": 4000 + }, + { + "epoch": 5.86, + "grad_norm": 63.65642166137695, + "learning_rate": 7.132799776430446e-05, + "loss": 4.0338, + "step": 4001 + }, + { + "epoch": 5.87, + "grad_norm": 84.51449584960938, + "learning_rate": 7.128306674444531e-05, + "loss": 4.0289, + "step": 4002 + }, + { + "epoch": 5.87, + "grad_norm": 1112.17431640625, + "learning_rate": 7.123814204211899e-05, + "loss": 3.9574, + "step": 4003 + }, + { + "epoch": 5.87, + "grad_norm": 113.361572265625, + "learning_rate": 7.11932236672087e-05, + "loss": 3.9683, + "step": 4004 + }, + { + "epoch": 5.87, + "grad_norm": 94.9935073852539, + "learning_rate": 7.114831162959614e-05, + "loss": 3.867, + "step": 4005 + }, + { + "epoch": 5.87, + "grad_norm": 99.41561126708984, + "learning_rate": 7.110340593916164e-05, + "loss": 4.0003, + "step": 4006 + }, + { + "epoch": 5.87, + "grad_norm": 68.38880157470703, + "learning_rate": 7.105850660578422e-05, + "loss": 3.8946, + "step": 4007 + }, + { + "epoch": 5.87, + "grad_norm": 48.9935188293457, + "learning_rate": 7.101361363934136e-05, + "loss": 3.9, + "step": 4008 + }, + { + "epoch": 5.88, + "grad_norm": 75.62947082519531, + "learning_rate": 7.096872704970923e-05, + "loss": 3.9334, + "step": 4009 + }, + { + "epoch": 5.88, + "grad_norm": 259.18511962890625, + "learning_rate": 7.092384684676262e-05, + "loss": 3.9539, + "step": 4010 + }, + { + "epoch": 5.88, + "grad_norm": 60.03557586669922, + "learning_rate": 7.087897304037483e-05, + "loss": 3.9378, + "step": 4011 + }, + { + "epoch": 5.88, + "grad_norm": 60.427616119384766, + "learning_rate": 7.083410564041779e-05, + "loss": 3.922, + "step": 4012 + }, + { + "epoch": 5.88, + "grad_norm": 86.16722869873047, + "learning_rate": 7.078924465676208e-05, + "loss": 3.9123, + "step": 4013 + }, + { + "epoch": 5.88, + "grad_norm": 92.25879669189453, + "learning_rate": 7.07443900992768e-05, + "loss": 3.9653, + "step": 4014 + }, + { + "epoch": 5.88, + "grad_norm": 63.97455596923828, + "learning_rate": 7.069954197782958e-05, + "loss": 3.9709, + "step": 4015 + }, + { + "epoch": 5.89, + "grad_norm": 85.49566650390625, + "learning_rate": 7.065470030228679e-05, + "loss": 3.9614, + "step": 4016 + }, + { + "epoch": 5.89, + "grad_norm": 200.93997192382812, + "learning_rate": 7.060986508251329e-05, + "loss": 3.9715, + "step": 4017 + }, + { + "epoch": 5.89, + "grad_norm": 47.623600006103516, + "learning_rate": 7.056503632837247e-05, + "loss": 3.9969, + "step": 4018 + }, + { + "epoch": 5.89, + "grad_norm": 57.71437454223633, + "learning_rate": 7.052021404972641e-05, + "loss": 4.0892, + "step": 4019 + }, + { + "epoch": 5.89, + "grad_norm": 101.89254760742188, + "learning_rate": 7.047539825643572e-05, + "loss": 4.1731, + "step": 4020 + }, + { + "epoch": 5.89, + "grad_norm": 56.065486907958984, + "learning_rate": 7.043058895835953e-05, + "loss": 4.1648, + "step": 4021 + }, + { + "epoch": 5.89, + "grad_norm": 98.78327178955078, + "learning_rate": 7.038578616535556e-05, + "loss": 4.3178, + "step": 4022 + }, + { + "epoch": 5.9, + "grad_norm": 383.0876770019531, + "learning_rate": 7.034098988728024e-05, + "loss": 4.3457, + "step": 4023 + }, + { + "epoch": 5.9, + "grad_norm": 162.98101806640625, + "learning_rate": 7.029620013398837e-05, + "loss": 4.3242, + "step": 4024 + }, + { + "epoch": 5.9, + "grad_norm": 149.046142578125, + "learning_rate": 7.025141691533334e-05, + "loss": 4.4494, + "step": 4025 + }, + { + "epoch": 5.9, + "grad_norm": 135.7252655029297, + "learning_rate": 7.020664024116726e-05, + "loss": 4.487, + "step": 4026 + }, + { + "epoch": 5.9, + "grad_norm": 103.65221405029297, + "learning_rate": 7.016187012134068e-05, + "loss": 4.544, + "step": 4027 + }, + { + "epoch": 5.9, + "grad_norm": 95.76521301269531, + "learning_rate": 7.011710656570267e-05, + "loss": 4.5937, + "step": 4028 + }, + { + "epoch": 5.91, + "grad_norm": 287.5489196777344, + "learning_rate": 7.007234958410098e-05, + "loss": 4.6397, + "step": 4029 + }, + { + "epoch": 5.91, + "grad_norm": 55.86167526245117, + "learning_rate": 7.002759918638183e-05, + "loss": 4.6566, + "step": 4030 + }, + { + "epoch": 5.91, + "grad_norm": 50.3218879699707, + "learning_rate": 6.998285538238993e-05, + "loss": 4.6248, + "step": 4031 + }, + { + "epoch": 5.91, + "grad_norm": 414.169921875, + "learning_rate": 6.993811818196872e-05, + "loss": 4.6006, + "step": 4032 + }, + { + "epoch": 5.91, + "grad_norm": 112.764404296875, + "learning_rate": 6.989338759496006e-05, + "loss": 4.6415, + "step": 4033 + }, + { + "epoch": 5.91, + "grad_norm": 53.906734466552734, + "learning_rate": 6.984866363120432e-05, + "loss": 4.6018, + "step": 4034 + }, + { + "epoch": 5.91, + "grad_norm": 45.21807098388672, + "learning_rate": 6.980394630054054e-05, + "loss": 4.5761, + "step": 4035 + }, + { + "epoch": 5.92, + "grad_norm": 123.97179412841797, + "learning_rate": 6.975923561280624e-05, + "loss": 4.4143, + "step": 4036 + }, + { + "epoch": 5.92, + "grad_norm": 39.991905212402344, + "learning_rate": 6.971453157783739e-05, + "loss": 4.3971, + "step": 4037 + }, + { + "epoch": 5.92, + "grad_norm": 245.3745574951172, + "learning_rate": 6.966983420546866e-05, + "loss": 4.3387, + "step": 4038 + }, + { + "epoch": 5.92, + "grad_norm": 535.9085083007812, + "learning_rate": 6.962514350553313e-05, + "loss": 4.2919, + "step": 4039 + }, + { + "epoch": 5.92, + "grad_norm": 31.66384506225586, + "learning_rate": 6.958045948786245e-05, + "loss": 4.2878, + "step": 4040 + }, + { + "epoch": 5.92, + "grad_norm": 104.09449768066406, + "learning_rate": 6.953578216228685e-05, + "loss": 4.2645, + "step": 4041 + }, + { + "epoch": 5.92, + "grad_norm": 73.07312774658203, + "learning_rate": 6.9491111538635e-05, + "loss": 4.2232, + "step": 4042 + }, + { + "epoch": 5.93, + "grad_norm": 55.163421630859375, + "learning_rate": 6.944644762673413e-05, + "loss": 4.2499, + "step": 4043 + }, + { + "epoch": 5.93, + "grad_norm": 97.98957061767578, + "learning_rate": 6.940179043641005e-05, + "loss": 4.2195, + "step": 4044 + }, + { + "epoch": 5.93, + "grad_norm": 35.54196548461914, + "learning_rate": 6.9357139977487e-05, + "loss": 4.2314, + "step": 4045 + }, + { + "epoch": 5.93, + "grad_norm": 36.159202575683594, + "learning_rate": 6.931249625978782e-05, + "loss": 4.1943, + "step": 4046 + }, + { + "epoch": 5.93, + "grad_norm": 119.87337493896484, + "learning_rate": 6.926785929313376e-05, + "loss": 4.1162, + "step": 4047 + }, + { + "epoch": 5.93, + "grad_norm": 7.6323442459106445, + "learning_rate": 6.922322908734474e-05, + "loss": 4.1434, + "step": 4048 + }, + { + "epoch": 5.94, + "grad_norm": 13.205349922180176, + "learning_rate": 6.917860565223905e-05, + "loss": 4.1026, + "step": 4049 + }, + { + "epoch": 5.94, + "grad_norm": 29.02250099182129, + "learning_rate": 6.913398899763352e-05, + "loss": 4.0863, + "step": 4050 + }, + { + "epoch": 5.94, + "grad_norm": 18.842418670654297, + "learning_rate": 6.908937913334362e-05, + "loss": 4.1056, + "step": 4051 + }, + { + "epoch": 5.94, + "grad_norm": 26.442873001098633, + "learning_rate": 6.904477606918313e-05, + "loss": 4.1195, + "step": 4052 + }, + { + "epoch": 5.94, + "grad_norm": 146.73545837402344, + "learning_rate": 6.900017981496443e-05, + "loss": 4.0448, + "step": 4053 + }, + { + "epoch": 5.94, + "grad_norm": 18.860145568847656, + "learning_rate": 6.895559038049846e-05, + "loss": 4.0507, + "step": 4054 + }, + { + "epoch": 5.94, + "grad_norm": 7.4111504554748535, + "learning_rate": 6.891100777559454e-05, + "loss": 3.9955, + "step": 4055 + }, + { + "epoch": 5.95, + "grad_norm": 2.2253613471984863, + "learning_rate": 6.886643201006052e-05, + "loss": 3.9959, + "step": 4056 + }, + { + "epoch": 5.95, + "eval_loss": 3.7561120986938477, + "eval_runtime": 1174.8961, + "eval_samples_per_second": 14.59, + "eval_steps_per_second": 0.912, + "step": 4056 + }, + { + "epoch": 5.95, + "grad_norm": 2.182352066040039, + "learning_rate": 6.882186309370287e-05, + "loss": 3.9472, + "step": 4057 + }, + { + "epoch": 5.95, + "grad_norm": 1.8885817527770996, + "learning_rate": 6.877730103632638e-05, + "loss": 3.9685, + "step": 4058 + }, + { + "epoch": 5.95, + "grad_norm": 1.6018338203430176, + "learning_rate": 6.87327458477344e-05, + "loss": 3.8923, + "step": 4059 + }, + { + "epoch": 5.95, + "grad_norm": 1.4974128007888794, + "learning_rate": 6.868819753772879e-05, + "loss": 3.8199, + "step": 4060 + }, + { + "epoch": 5.95, + "grad_norm": 1.9076157808303833, + "learning_rate": 6.86436561161099e-05, + "loss": 3.8143, + "step": 4061 + }, + { + "epoch": 5.95, + "grad_norm": 1.2935985326766968, + "learning_rate": 6.859912159267648e-05, + "loss": 3.8232, + "step": 4062 + }, + { + "epoch": 5.96, + "grad_norm": 1.5463148355484009, + "learning_rate": 6.85545939772259e-05, + "loss": 3.7682, + "step": 4063 + }, + { + "epoch": 5.96, + "grad_norm": 1.2979236841201782, + "learning_rate": 6.851007327955393e-05, + "loss": 3.812, + "step": 4064 + }, + { + "epoch": 5.96, + "grad_norm": 1.235023856163025, + "learning_rate": 6.846555950945475e-05, + "loss": 3.6794, + "step": 4065 + }, + { + "epoch": 5.96, + "grad_norm": 1.587446689605713, + "learning_rate": 6.842105267672117e-05, + "loss": 3.6736, + "step": 4066 + }, + { + "epoch": 5.96, + "grad_norm": 1.6601265668869019, + "learning_rate": 6.837655279114439e-05, + "loss": 3.7298, + "step": 4067 + }, + { + "epoch": 5.96, + "grad_norm": 1.1815412044525146, + "learning_rate": 6.833205986251404e-05, + "loss": 3.6491, + "step": 4068 + }, + { + "epoch": 5.96, + "grad_norm": 2.2124860286712646, + "learning_rate": 6.828757390061828e-05, + "loss": 3.6432, + "step": 4069 + }, + { + "epoch": 5.97, + "grad_norm": 1.2953498363494873, + "learning_rate": 6.824309491524377e-05, + "loss": 3.6124, + "step": 4070 + }, + { + "epoch": 5.97, + "grad_norm": 1.5017386674880981, + "learning_rate": 6.819862291617555e-05, + "loss": 3.6377, + "step": 4071 + }, + { + "epoch": 5.97, + "grad_norm": 2.5354573726654053, + "learning_rate": 6.815415791319712e-05, + "loss": 3.6211, + "step": 4072 + }, + { + "epoch": 5.97, + "grad_norm": 2.2560806274414062, + "learning_rate": 6.810969991609055e-05, + "loss": 3.5855, + "step": 4073 + }, + { + "epoch": 5.97, + "grad_norm": 2.3098514080047607, + "learning_rate": 6.806524893463631e-05, + "loss": 3.5944, + "step": 4074 + }, + { + "epoch": 5.97, + "grad_norm": 3.807551860809326, + "learning_rate": 6.802080497861326e-05, + "loss": 3.582, + "step": 4075 + }, + { + "epoch": 5.98, + "grad_norm": 0.8520615100860596, + "learning_rate": 6.797636805779878e-05, + "loss": 3.5581, + "step": 4076 + }, + { + "epoch": 5.98, + "grad_norm": 2.77345871925354, + "learning_rate": 6.793193818196876e-05, + "loss": 3.5948, + "step": 4077 + }, + { + "epoch": 5.98, + "grad_norm": 1.2732380628585815, + "learning_rate": 6.788751536089739e-05, + "loss": 3.5607, + "step": 4078 + }, + { + "epoch": 5.98, + "grad_norm": 4.257369518280029, + "learning_rate": 6.784309960435744e-05, + "loss": 3.588, + "step": 4079 + }, + { + "epoch": 5.98, + "grad_norm": 4.027552604675293, + "learning_rate": 6.77986909221201e-05, + "loss": 3.5961, + "step": 4080 + }, + { + "epoch": 5.98, + "grad_norm": 1.3005938529968262, + "learning_rate": 6.77542893239549e-05, + "loss": 3.52, + "step": 4081 + }, + { + "epoch": 5.98, + "grad_norm": 3.389254570007324, + "learning_rate": 6.770989481963001e-05, + "loss": 3.5313, + "step": 4082 + }, + { + "epoch": 5.99, + "grad_norm": 1.3640910387039185, + "learning_rate": 6.766550741891182e-05, + "loss": 3.5291, + "step": 4083 + }, + { + "epoch": 5.99, + "grad_norm": 5.111477375030518, + "learning_rate": 6.76211271315653e-05, + "loss": 3.5717, + "step": 4084 + }, + { + "epoch": 5.99, + "grad_norm": 5.650928974151611, + "learning_rate": 6.757675396735386e-05, + "loss": 3.5892, + "step": 4085 + }, + { + "epoch": 5.99, + "grad_norm": 2.029646158218384, + "learning_rate": 6.753238793603923e-05, + "loss": 3.5179, + "step": 4086 + }, + { + "epoch": 5.99, + "grad_norm": 5.816488742828369, + "learning_rate": 6.748802904738163e-05, + "loss": 3.5582, + "step": 4087 + }, + { + "epoch": 5.99, + "grad_norm": 7.348140716552734, + "learning_rate": 6.74436773111398e-05, + "loss": 3.5371, + "step": 4088 + }, + { + "epoch": 5.99, + "grad_norm": 6.118741512298584, + "learning_rate": 6.739933273707078e-05, + "loss": 3.5298, + "step": 4089 + }, + { + "epoch": 6.0, + "grad_norm": 0.9806273579597473, + "learning_rate": 6.735499533493003e-05, + "loss": 3.5438, + "step": 4090 + }, + { + "epoch": 6.0, + "grad_norm": 9.882466316223145, + "learning_rate": 6.731066511447159e-05, + "loss": 3.5713, + "step": 4091 + }, + { + "epoch": 6.0, + "grad_norm": 14.419783592224121, + "learning_rate": 6.726634208544773e-05, + "loss": 3.569, + "step": 4092 + }, + { + "epoch": 6.0, + "grad_norm": 11.746726036071777, + "learning_rate": 6.722202625760923e-05, + "loss": 3.6069, + "step": 4093 + }, + { + "epoch": 6.0, + "grad_norm": 5.577601432800293, + "learning_rate": 6.717771764070527e-05, + "loss": 3.5287, + "step": 4094 + }, + { + "epoch": 6.0, + "grad_norm": 5.144962310791016, + "learning_rate": 6.713341624448351e-05, + "loss": 3.5013, + "step": 4095 + }, + { + "epoch": 6.0, + "grad_norm": 8.980729103088379, + "learning_rate": 6.708912207868987e-05, + "loss": 3.5225, + "step": 4096 + }, + { + "epoch": 6.01, + "grad_norm": 9.171212196350098, + "learning_rate": 6.70448351530688e-05, + "loss": 3.537, + "step": 4097 + }, + { + "epoch": 6.01, + "grad_norm": 6.954055309295654, + "learning_rate": 6.70005554773632e-05, + "loss": 3.5512, + "step": 4098 + }, + { + "epoch": 6.01, + "grad_norm": 25.35405731201172, + "learning_rate": 6.695628306131422e-05, + "loss": 3.525, + "step": 4099 + }, + { + "epoch": 6.01, + "grad_norm": 7.4213948249816895, + "learning_rate": 6.691201791466148e-05, + "loss": 3.5411, + "step": 4100 + }, + { + "epoch": 6.01, + "grad_norm": 10.75389289855957, + "learning_rate": 6.68677600471431e-05, + "loss": 3.5158, + "step": 4101 + }, + { + "epoch": 6.01, + "grad_norm": 9.7072114944458, + "learning_rate": 6.682350946849546e-05, + "loss": 3.5728, + "step": 4102 + }, + { + "epoch": 6.02, + "grad_norm": 4.010187149047852, + "learning_rate": 6.677926618845337e-05, + "loss": 3.4962, + "step": 4103 + }, + { + "epoch": 6.0, + "grad_norm": 5.2085371017456055, + "learning_rate": 6.673503021675011e-05, + "loss": 3.5387, + "step": 4104 + }, + { + "epoch": 6.0, + "grad_norm": 7.690248012542725, + "learning_rate": 6.669080156311729e-05, + "loss": 3.5012, + "step": 4105 + }, + { + "epoch": 6.0, + "grad_norm": 7.818577766418457, + "learning_rate": 6.664658023728485e-05, + "loss": 3.5594, + "step": 4106 + }, + { + "epoch": 6.01, + "grad_norm": 4.1020355224609375, + "learning_rate": 6.660236624898124e-05, + "loss": 3.4641, + "step": 4107 + }, + { + "epoch": 6.01, + "grad_norm": 3.3743233680725098, + "learning_rate": 6.655815960793328e-05, + "loss": 3.5123, + "step": 4108 + }, + { + "epoch": 6.01, + "grad_norm": 5.35082483291626, + "learning_rate": 6.651396032386603e-05, + "loss": 3.5047, + "step": 4109 + }, + { + "epoch": 6.01, + "grad_norm": 2.8813364505767822, + "learning_rate": 6.646976840650312e-05, + "loss": 3.4699, + "step": 4110 + }, + { + "epoch": 6.01, + "grad_norm": 3.898869037628174, + "learning_rate": 6.642558386556648e-05, + "loss": 3.4723, + "step": 4111 + }, + { + "epoch": 6.01, + "grad_norm": 3.897127389907837, + "learning_rate": 6.638140671077633e-05, + "loss": 3.4887, + "step": 4112 + }, + { + "epoch": 6.01, + "grad_norm": 0.8849645256996155, + "learning_rate": 6.633723695185141e-05, + "loss": 3.4564, + "step": 4113 + }, + { + "epoch": 6.02, + "grad_norm": 4.971652507781982, + "learning_rate": 6.62930745985088e-05, + "loss": 3.5218, + "step": 4114 + }, + { + "epoch": 6.02, + "grad_norm": 4.998004913330078, + "learning_rate": 6.624891966046385e-05, + "loss": 3.5211, + "step": 4115 + }, + { + "epoch": 6.02, + "grad_norm": 1.827019214630127, + "learning_rate": 6.620477214743042e-05, + "loss": 3.4959, + "step": 4116 + }, + { + "epoch": 6.02, + "grad_norm": 4.872849464416504, + "learning_rate": 6.616063206912063e-05, + "loss": 3.4849, + "step": 4117 + }, + { + "epoch": 6.02, + "grad_norm": 5.961864948272705, + "learning_rate": 6.611649943524498e-05, + "loss": 3.4685, + "step": 4118 + }, + { + "epoch": 6.02, + "grad_norm": 3.4859068393707275, + "learning_rate": 6.607237425551236e-05, + "loss": 3.4841, + "step": 4119 + }, + { + "epoch": 6.03, + "grad_norm": 4.598318099975586, + "learning_rate": 6.602825653963006e-05, + "loss": 3.5138, + "step": 4120 + }, + { + "epoch": 6.03, + "grad_norm": 8.418424606323242, + "learning_rate": 6.598414629730367e-05, + "loss": 3.4861, + "step": 4121 + }, + { + "epoch": 6.03, + "grad_norm": 6.949596881866455, + "learning_rate": 6.594004353823708e-05, + "loss": 3.4545, + "step": 4122 + }, + { + "epoch": 6.03, + "grad_norm": 1.4345308542251587, + "learning_rate": 6.589594827213268e-05, + "loss": 3.4615, + "step": 4123 + }, + { + "epoch": 6.03, + "grad_norm": 5.833983898162842, + "learning_rate": 6.585186050869112e-05, + "loss": 3.4469, + "step": 4124 + }, + { + "epoch": 6.03, + "grad_norm": 8.324283599853516, + "learning_rate": 6.580778025761136e-05, + "loss": 3.5428, + "step": 4125 + }, + { + "epoch": 6.03, + "grad_norm": 7.164588451385498, + "learning_rate": 6.576370752859085e-05, + "loss": 3.44, + "step": 4126 + }, + { + "epoch": 6.04, + "grad_norm": 4.524144172668457, + "learning_rate": 6.57196423313252e-05, + "loss": 3.4798, + "step": 4127 + }, + { + "epoch": 6.04, + "grad_norm": 3.6542019844055176, + "learning_rate": 6.56755846755085e-05, + "loss": 3.4431, + "step": 4128 + }, + { + "epoch": 6.04, + "grad_norm": 6.415909290313721, + "learning_rate": 6.563153457083315e-05, + "loss": 3.4509, + "step": 4129 + }, + { + "epoch": 6.04, + "grad_norm": 4.0182623863220215, + "learning_rate": 6.558749202698988e-05, + "loss": 3.4182, + "step": 4130 + }, + { + "epoch": 6.04, + "grad_norm": 22.318021774291992, + "learning_rate": 6.55434570536677e-05, + "loss": 3.5704, + "step": 4131 + }, + { + "epoch": 6.04, + "grad_norm": 6.184370040893555, + "learning_rate": 6.54994296605541e-05, + "loss": 3.3746, + "step": 4132 + }, + { + "epoch": 6.04, + "grad_norm": 7.076589584350586, + "learning_rate": 6.545540985733473e-05, + "loss": 3.4426, + "step": 4133 + }, + { + "epoch": 6.05, + "grad_norm": 3.5989532470703125, + "learning_rate": 6.541139765369368e-05, + "loss": 3.4139, + "step": 4134 + }, + { + "epoch": 6.05, + "grad_norm": 4.95648717880249, + "learning_rate": 6.536739305931338e-05, + "loss": 3.392, + "step": 4135 + }, + { + "epoch": 6.05, + "grad_norm": 8.156216621398926, + "learning_rate": 6.53233960838745e-05, + "loss": 3.4219, + "step": 4136 + }, + { + "epoch": 6.05, + "grad_norm": 6.330258846282959, + "learning_rate": 6.527940673705605e-05, + "loss": 3.4077, + "step": 4137 + }, + { + "epoch": 6.05, + "grad_norm": 1.3029074668884277, + "learning_rate": 6.523542502853547e-05, + "loss": 3.422, + "step": 4138 + }, + { + "epoch": 6.05, + "grad_norm": 5.435974597930908, + "learning_rate": 6.519145096798842e-05, + "loss": 3.4704, + "step": 4139 + }, + { + "epoch": 6.05, + "grad_norm": 5.678780555725098, + "learning_rate": 6.514748456508885e-05, + "loss": 3.419, + "step": 4140 + }, + { + "epoch": 6.06, + "grad_norm": 3.2500386238098145, + "learning_rate": 6.510352582950909e-05, + "loss": 3.4207, + "step": 4141 + }, + { + "epoch": 6.06, + "grad_norm": 4.012285232543945, + "learning_rate": 6.505957477091981e-05, + "loss": 3.3755, + "step": 4142 + }, + { + "epoch": 6.06, + "grad_norm": 6.388498783111572, + "learning_rate": 6.501563139898991e-05, + "loss": 3.4086, + "step": 4143 + }, + { + "epoch": 6.06, + "grad_norm": 3.951239824295044, + "learning_rate": 6.497169572338661e-05, + "loss": 3.3432, + "step": 4144 + }, + { + "epoch": 6.06, + "grad_norm": 4.527097702026367, + "learning_rate": 6.492776775377556e-05, + "loss": 3.3961, + "step": 4145 + }, + { + "epoch": 6.06, + "grad_norm": 5.017838954925537, + "learning_rate": 6.488384749982053e-05, + "loss": 3.3483, + "step": 4146 + }, + { + "epoch": 6.07, + "grad_norm": 4.438302516937256, + "learning_rate": 6.483993497118371e-05, + "loss": 3.456, + "step": 4147 + }, + { + "epoch": 6.07, + "grad_norm": 0.9947657585144043, + "learning_rate": 6.479603017752559e-05, + "loss": 3.3646, + "step": 4148 + }, + { + "epoch": 6.07, + "grad_norm": 2.749354600906372, + "learning_rate": 6.47521331285049e-05, + "loss": 3.4993, + "step": 4149 + }, + { + "epoch": 6.07, + "grad_norm": 2.3780393600463867, + "learning_rate": 6.470824383377867e-05, + "loss": 3.3783, + "step": 4150 + }, + { + "epoch": 6.07, + "grad_norm": 2.840714454650879, + "learning_rate": 6.46643623030023e-05, + "loss": 3.3901, + "step": 4151 + }, + { + "epoch": 6.07, + "grad_norm": 2.8188705444335938, + "learning_rate": 6.462048854582944e-05, + "loss": 3.3313, + "step": 4152 + }, + { + "epoch": 6.07, + "grad_norm": 0.6622484922409058, + "learning_rate": 6.457662257191195e-05, + "loss": 3.3428, + "step": 4153 + }, + { + "epoch": 6.08, + "grad_norm": 2.970710515975952, + "learning_rate": 6.453276439090012e-05, + "loss": 3.3343, + "step": 4154 + }, + { + "epoch": 6.08, + "grad_norm": 2.88561749458313, + "learning_rate": 6.448891401244244e-05, + "loss": 3.3591, + "step": 4155 + }, + { + "epoch": 6.08, + "grad_norm": 2.3622708320617676, + "learning_rate": 6.444507144618566e-05, + "loss": 3.39, + "step": 4156 + }, + { + "epoch": 6.08, + "grad_norm": 1.4716769456863403, + "learning_rate": 6.440123670177492e-05, + "loss": 3.4068, + "step": 4157 + }, + { + "epoch": 6.08, + "grad_norm": 10.4777193069458, + "learning_rate": 6.435740978885353e-05, + "loss": 3.352, + "step": 4158 + }, + { + "epoch": 6.08, + "grad_norm": 6.331012725830078, + "learning_rate": 6.43135907170631e-05, + "loss": 3.3506, + "step": 4159 + }, + { + "epoch": 6.08, + "grad_norm": 4.355317115783691, + "learning_rate": 6.426977949604358e-05, + "loss": 3.3904, + "step": 4160 + }, + { + "epoch": 6.09, + "grad_norm": 1.3198426961898804, + "learning_rate": 6.422597613543309e-05, + "loss": 3.3279, + "step": 4161 + }, + { + "epoch": 6.09, + "grad_norm": 3.2531580924987793, + "learning_rate": 6.41821806448681e-05, + "loss": 3.427, + "step": 4162 + }, + { + "epoch": 6.09, + "grad_norm": 1.780542254447937, + "learning_rate": 6.413839303398337e-05, + "loss": 3.382, + "step": 4163 + }, + { + "epoch": 6.09, + "grad_norm": 4.514572620391846, + "learning_rate": 6.40946133124118e-05, + "loss": 3.3181, + "step": 4164 + }, + { + "epoch": 6.09, + "grad_norm": 4.696477890014648, + "learning_rate": 6.405084148978471e-05, + "loss": 3.3693, + "step": 4165 + }, + { + "epoch": 6.09, + "grad_norm": 1.9454889297485352, + "learning_rate": 6.400707757573152e-05, + "loss": 3.3362, + "step": 4166 + }, + { + "epoch": 6.1, + "grad_norm": 4.424253463745117, + "learning_rate": 6.396332157988007e-05, + "loss": 3.3042, + "step": 4167 + }, + { + "epoch": 6.1, + "grad_norm": 6.185297012329102, + "learning_rate": 6.39195735118564e-05, + "loss": 3.3837, + "step": 4168 + }, + { + "epoch": 6.1, + "grad_norm": 4.592702865600586, + "learning_rate": 6.387583338128471e-05, + "loss": 3.35, + "step": 4169 + }, + { + "epoch": 6.1, + "grad_norm": 1.6287086009979248, + "learning_rate": 6.38321011977876e-05, + "loss": 3.3304, + "step": 4170 + }, + { + "epoch": 6.1, + "grad_norm": 4.277772903442383, + "learning_rate": 6.378837697098586e-05, + "loss": 3.321, + "step": 4171 + }, + { + "epoch": 6.1, + "grad_norm": 2.2683122158050537, + "learning_rate": 6.374466071049847e-05, + "loss": 3.2938, + "step": 4172 + }, + { + "epoch": 6.1, + "grad_norm": 2.8517701625823975, + "learning_rate": 6.370095242594279e-05, + "loss": 3.338, + "step": 4173 + }, + { + "epoch": 6.11, + "grad_norm": 3.4475059509277344, + "learning_rate": 6.365725212693428e-05, + "loss": 3.3719, + "step": 4174 + }, + { + "epoch": 6.11, + "grad_norm": 1.0694806575775146, + "learning_rate": 6.361355982308675e-05, + "loss": 3.3108, + "step": 4175 + }, + { + "epoch": 6.11, + "grad_norm": 4.983922958374023, + "learning_rate": 6.356987552401222e-05, + "loss": 3.326, + "step": 4176 + }, + { + "epoch": 6.11, + "grad_norm": 5.746216773986816, + "learning_rate": 6.352619923932093e-05, + "loss": 3.3831, + "step": 4177 + }, + { + "epoch": 6.11, + "grad_norm": 2.469836473464966, + "learning_rate": 6.348253097862134e-05, + "loss": 3.3593, + "step": 4178 + }, + { + "epoch": 6.11, + "grad_norm": 3.8957202434539795, + "learning_rate": 6.343887075152024e-05, + "loss": 3.3594, + "step": 4179 + }, + { + "epoch": 6.11, + "grad_norm": 5.375302791595459, + "learning_rate": 6.339521856762254e-05, + "loss": 3.3357, + "step": 4180 + }, + { + "epoch": 6.12, + "grad_norm": 4.205276966094971, + "learning_rate": 6.335157443653142e-05, + "loss": 3.3987, + "step": 4181 + }, + { + "epoch": 6.12, + "grad_norm": 1.7369492053985596, + "learning_rate": 6.330793836784836e-05, + "loss": 3.3606, + "step": 4182 + }, + { + "epoch": 6.12, + "grad_norm": 2.639575481414795, + "learning_rate": 6.326431037117296e-05, + "loss": 3.4109, + "step": 4183 + }, + { + "epoch": 6.12, + "grad_norm": 0.6227443218231201, + "learning_rate": 6.322069045610305e-05, + "loss": 3.3122, + "step": 4184 + }, + { + "epoch": 6.12, + "grad_norm": 2.4166953563690186, + "learning_rate": 6.317707863223477e-05, + "loss": 3.339, + "step": 4185 + }, + { + "epoch": 6.12, + "grad_norm": 1.3330849409103394, + "learning_rate": 6.313347490916243e-05, + "loss": 3.2775, + "step": 4186 + }, + { + "epoch": 6.12, + "grad_norm": 3.1274631023406982, + "learning_rate": 6.308987929647852e-05, + "loss": 3.2931, + "step": 4187 + }, + { + "epoch": 6.13, + "grad_norm": 2.9381067752838135, + "learning_rate": 6.30462918037738e-05, + "loss": 3.347, + "step": 4188 + }, + { + "epoch": 6.13, + "grad_norm": 0.6513715982437134, + "learning_rate": 6.300271244063726e-05, + "loss": 3.3084, + "step": 4189 + }, + { + "epoch": 6.13, + "grad_norm": 1.9384881258010864, + "learning_rate": 6.295914121665602e-05, + "loss": 3.3508, + "step": 4190 + }, + { + "epoch": 6.13, + "grad_norm": 1.33826744556427, + "learning_rate": 6.291557814141547e-05, + "loss": 3.3385, + "step": 4191 + }, + { + "epoch": 6.13, + "grad_norm": 1.159730076789856, + "learning_rate": 6.287202322449922e-05, + "loss": 3.3025, + "step": 4192 + }, + { + "epoch": 6.13, + "grad_norm": 15.805403709411621, + "learning_rate": 6.282847647548904e-05, + "loss": 3.3027, + "step": 4193 + }, + { + "epoch": 6.14, + "grad_norm": 4.59617280960083, + "learning_rate": 6.278493790396489e-05, + "loss": 3.3538, + "step": 4194 + }, + { + "epoch": 6.14, + "grad_norm": 2.9616920948028564, + "learning_rate": 6.274140751950501e-05, + "loss": 3.3287, + "step": 4195 + }, + { + "epoch": 6.14, + "grad_norm": 2.299812078475952, + "learning_rate": 6.269788533168581e-05, + "loss": 3.3451, + "step": 4196 + }, + { + "epoch": 6.14, + "grad_norm": 2.9177210330963135, + "learning_rate": 6.26543713500818e-05, + "loss": 3.3583, + "step": 4197 + }, + { + "epoch": 6.14, + "grad_norm": 0.9003336429595947, + "learning_rate": 6.261086558426586e-05, + "loss": 3.2672, + "step": 4198 + }, + { + "epoch": 6.14, + "grad_norm": 6.0708184242248535, + "learning_rate": 6.25673680438089e-05, + "loss": 3.2929, + "step": 4199 + }, + { + "epoch": 6.14, + "grad_norm": 6.111046314239502, + "learning_rate": 6.25238787382801e-05, + "loss": 3.3031, + "step": 4200 + }, + { + "epoch": 6.15, + "grad_norm": 4.454471588134766, + "learning_rate": 6.248039767724683e-05, + "loss": 3.3293, + "step": 4201 + }, + { + "epoch": 6.15, + "grad_norm": 0.9689366221427917, + "learning_rate": 6.243692487027464e-05, + "loss": 3.2844, + "step": 4202 + }, + { + "epoch": 6.15, + "grad_norm": 4.885942459106445, + "learning_rate": 6.239346032692721e-05, + "loss": 3.3346, + "step": 4203 + }, + { + "epoch": 6.15, + "grad_norm": 3.6569879055023193, + "learning_rate": 6.23500040567665e-05, + "loss": 3.2936, + "step": 4204 + }, + { + "epoch": 6.15, + "grad_norm": 1.7212425470352173, + "learning_rate": 6.23065560693526e-05, + "loss": 3.3124, + "step": 4205 + }, + { + "epoch": 6.15, + "grad_norm": 2.9631359577178955, + "learning_rate": 6.226311637424371e-05, + "loss": 3.3716, + "step": 4206 + }, + { + "epoch": 6.15, + "grad_norm": 1.4877088069915771, + "learning_rate": 6.221968498099636e-05, + "loss": 3.2881, + "step": 4207 + }, + { + "epoch": 6.16, + "grad_norm": 5.8026533126831055, + "learning_rate": 6.21762618991651e-05, + "loss": 3.2705, + "step": 4208 + }, + { + "epoch": 6.16, + "grad_norm": 6.365312576293945, + "learning_rate": 6.213284713830271e-05, + "loss": 3.3174, + "step": 4209 + }, + { + "epoch": 6.16, + "grad_norm": 5.353916645050049, + "learning_rate": 6.208944070796022e-05, + "loss": 3.3226, + "step": 4210 + }, + { + "epoch": 6.16, + "grad_norm": 1.4095226526260376, + "learning_rate": 6.204604261768669e-05, + "loss": 3.3037, + "step": 4211 + }, + { + "epoch": 6.16, + "grad_norm": 2.51399827003479, + "learning_rate": 6.200265287702942e-05, + "loss": 3.3118, + "step": 4212 + }, + { + "epoch": 6.16, + "grad_norm": 1.4615494012832642, + "learning_rate": 6.19592714955339e-05, + "loss": 3.2963, + "step": 4213 + }, + { + "epoch": 6.16, + "grad_norm": 2.4555823802948, + "learning_rate": 6.191589848274368e-05, + "loss": 3.275, + "step": 4214 + }, + { + "epoch": 6.17, + "grad_norm": 3.085233211517334, + "learning_rate": 6.187253384820061e-05, + "loss": 3.2726, + "step": 4215 + }, + { + "epoch": 6.17, + "grad_norm": 0.8324146270751953, + "learning_rate": 6.182917760144453e-05, + "loss": 3.3158, + "step": 4216 + }, + { + "epoch": 6.17, + "grad_norm": 1.7136584520339966, + "learning_rate": 6.178582975201359e-05, + "loss": 3.2768, + "step": 4217 + }, + { + "epoch": 6.17, + "grad_norm": 1.937159538269043, + "learning_rate": 6.1742490309444e-05, + "loss": 3.3338, + "step": 4218 + }, + { + "epoch": 6.17, + "grad_norm": 1.2840466499328613, + "learning_rate": 6.169915928327012e-05, + "loss": 3.2964, + "step": 4219 + }, + { + "epoch": 6.17, + "grad_norm": 2.5718352794647217, + "learning_rate": 6.165583668302457e-05, + "loss": 3.287, + "step": 4220 + }, + { + "epoch": 6.18, + "grad_norm": 4.60259485244751, + "learning_rate": 6.161252251823795e-05, + "loss": 3.2494, + "step": 4221 + }, + { + "epoch": 6.18, + "grad_norm": 1.160286545753479, + "learning_rate": 6.15692167984391e-05, + "loss": 3.3453, + "step": 4222 + }, + { + "epoch": 6.18, + "grad_norm": 2.802889823913574, + "learning_rate": 6.152591953315505e-05, + "loss": 3.3133, + "step": 4223 + }, + { + "epoch": 6.18, + "grad_norm": 1.775070071220398, + "learning_rate": 6.148263073191082e-05, + "loss": 3.2936, + "step": 4224 + }, + { + "epoch": 6.18, + "grad_norm": 1.9516983032226562, + "learning_rate": 6.14393504042297e-05, + "loss": 3.3066, + "step": 4225 + }, + { + "epoch": 6.18, + "eval_loss": 3.107570171356201, + "eval_runtime": 1177.648, + "eval_samples_per_second": 14.556, + "eval_steps_per_second": 0.91, + "step": 4225 + }, + { + "epoch": 6.18, + "grad_norm": 1.7126461267471313, + "learning_rate": 6.13960785596331e-05, + "loss": 3.2474, + "step": 4226 + }, + { + "epoch": 6.18, + "grad_norm": 1.2491967678070068, + "learning_rate": 6.135281520764049e-05, + "loss": 3.3243, + "step": 4227 + }, + { + "epoch": 6.19, + "grad_norm": 1.4934719800949097, + "learning_rate": 6.130956035776951e-05, + "loss": 3.3423, + "step": 4228 + }, + { + "epoch": 6.19, + "grad_norm": 1.9269001483917236, + "learning_rate": 6.126631401953598e-05, + "loss": 3.2633, + "step": 4229 + }, + { + "epoch": 6.19, + "grad_norm": 1.3831140995025635, + "learning_rate": 6.12230762024538e-05, + "loss": 3.286, + "step": 4230 + }, + { + "epoch": 6.19, + "grad_norm": 2.75412917137146, + "learning_rate": 6.117984691603493e-05, + "loss": 3.2788, + "step": 4231 + }, + { + "epoch": 6.19, + "grad_norm": 3.342359781265259, + "learning_rate": 6.113662616978961e-05, + "loss": 3.2737, + "step": 4232 + }, + { + "epoch": 6.19, + "grad_norm": 0.9600224494934082, + "learning_rate": 6.109341397322611e-05, + "loss": 3.2899, + "step": 4233 + }, + { + "epoch": 6.19, + "grad_norm": 2.350534200668335, + "learning_rate": 6.105021033585074e-05, + "loss": 3.2768, + "step": 4234 + }, + { + "epoch": 6.2, + "grad_norm": 1.1939469575881958, + "learning_rate": 6.10070152671681e-05, + "loss": 3.2706, + "step": 4235 + }, + { + "epoch": 6.2, + "grad_norm": 4.221333980560303, + "learning_rate": 6.096382877668079e-05, + "loss": 3.2853, + "step": 4236 + }, + { + "epoch": 6.2, + "grad_norm": 4.897353649139404, + "learning_rate": 6.092065087388953e-05, + "loss": 3.3084, + "step": 4237 + }, + { + "epoch": 6.2, + "grad_norm": 2.3200554847717285, + "learning_rate": 6.087748156829316e-05, + "loss": 3.2422, + "step": 4238 + }, + { + "epoch": 6.2, + "grad_norm": 4.042237758636475, + "learning_rate": 6.08343208693887e-05, + "loss": 3.2441, + "step": 4239 + }, + { + "epoch": 6.2, + "grad_norm": 4.729053974151611, + "learning_rate": 6.079116878667115e-05, + "loss": 3.2911, + "step": 4240 + }, + { + "epoch": 6.2, + "grad_norm": 3.419166088104248, + "learning_rate": 6.074802532963368e-05, + "loss": 3.2984, + "step": 4241 + }, + { + "epoch": 6.21, + "grad_norm": 2.1573331356048584, + "learning_rate": 6.0704890507767597e-05, + "loss": 3.2494, + "step": 4242 + }, + { + "epoch": 6.21, + "grad_norm": 3.604609489440918, + "learning_rate": 6.0661764330562276e-05, + "loss": 3.2961, + "step": 4243 + }, + { + "epoch": 6.21, + "grad_norm": 1.5325900316238403, + "learning_rate": 6.061864680750513e-05, + "loss": 3.2401, + "step": 4244 + }, + { + "epoch": 6.21, + "grad_norm": 3.4049322605133057, + "learning_rate": 6.0575537948081793e-05, + "loss": 3.292, + "step": 4245 + }, + { + "epoch": 6.21, + "grad_norm": 4.517602920532227, + "learning_rate": 6.053243776177591e-05, + "loss": 3.2811, + "step": 4246 + }, + { + "epoch": 6.21, + "grad_norm": 2.2732179164886475, + "learning_rate": 6.0489346258069177e-05, + "loss": 3.2883, + "step": 4247 + }, + { + "epoch": 6.22, + "grad_norm": 5.137287139892578, + "learning_rate": 6.044626344644151e-05, + "loss": 3.2465, + "step": 4248 + }, + { + "epoch": 6.22, + "grad_norm": 5.510692596435547, + "learning_rate": 6.040318933637081e-05, + "loss": 3.3609, + "step": 4249 + }, + { + "epoch": 6.22, + "grad_norm": 4.384718894958496, + "learning_rate": 6.036012393733306e-05, + "loss": 3.2768, + "step": 4250 + }, + { + "epoch": 6.22, + "grad_norm": 0.5632131695747375, + "learning_rate": 6.031706725880243e-05, + "loss": 3.2637, + "step": 4251 + }, + { + "epoch": 6.22, + "grad_norm": 4.083331108093262, + "learning_rate": 6.027401931025103e-05, + "loss": 3.2579, + "step": 4252 + }, + { + "epoch": 6.22, + "grad_norm": 4.434952735900879, + "learning_rate": 6.0230980101149136e-05, + "loss": 3.2995, + "step": 4253 + }, + { + "epoch": 6.22, + "grad_norm": 1.9891043901443481, + "learning_rate": 6.0187949640965146e-05, + "loss": 3.2652, + "step": 4254 + }, + { + "epoch": 6.23, + "grad_norm": 3.985672950744629, + "learning_rate": 6.014492793916539e-05, + "loss": 3.256, + "step": 4255 + }, + { + "epoch": 6.23, + "grad_norm": 5.051290988922119, + "learning_rate": 6.010191500521439e-05, + "loss": 3.2773, + "step": 4256 + }, + { + "epoch": 6.23, + "grad_norm": 3.2940456867218018, + "learning_rate": 6.005891084857473e-05, + "loss": 3.3033, + "step": 4257 + }, + { + "epoch": 6.23, + "grad_norm": 2.565002202987671, + "learning_rate": 6.001591547870701e-05, + "loss": 3.2838, + "step": 4258 + }, + { + "epoch": 6.23, + "grad_norm": 3.5721094608306885, + "learning_rate": 5.997292890506989e-05, + "loss": 3.2499, + "step": 4259 + }, + { + "epoch": 6.23, + "grad_norm": 1.7328767776489258, + "learning_rate": 5.992995113712019e-05, + "loss": 3.2925, + "step": 4260 + }, + { + "epoch": 6.23, + "grad_norm": 3.9454033374786377, + "learning_rate": 5.9886982184312725e-05, + "loss": 3.2744, + "step": 4261 + }, + { + "epoch": 6.24, + "grad_norm": 4.74951171875, + "learning_rate": 5.984402205610031e-05, + "loss": 3.2632, + "step": 4262 + }, + { + "epoch": 6.24, + "grad_norm": 2.6633002758026123, + "learning_rate": 5.980107076193393e-05, + "loss": 3.2585, + "step": 4263 + }, + { + "epoch": 6.24, + "grad_norm": 2.6646337509155273, + "learning_rate": 5.975812831126261e-05, + "loss": 3.3006, + "step": 4264 + }, + { + "epoch": 6.24, + "grad_norm": 3.7945520877838135, + "learning_rate": 5.971519471353335e-05, + "loss": 3.2882, + "step": 4265 + }, + { + "epoch": 6.24, + "grad_norm": 1.7113827466964722, + "learning_rate": 5.9672269978191266e-05, + "loss": 3.2554, + "step": 4266 + }, + { + "epoch": 6.24, + "grad_norm": 3.288426399230957, + "learning_rate": 5.9629354114679556e-05, + "loss": 3.329, + "step": 4267 + }, + { + "epoch": 6.25, + "grad_norm": 5.059687614440918, + "learning_rate": 5.958644713243937e-05, + "loss": 3.2254, + "step": 4268 + }, + { + "epoch": 6.25, + "grad_norm": 2.444850206375122, + "learning_rate": 5.9543549040909954e-05, + "loss": 3.2804, + "step": 4269 + }, + { + "epoch": 6.25, + "grad_norm": 2.6846628189086914, + "learning_rate": 5.9500659849528664e-05, + "loss": 3.2151, + "step": 4270 + }, + { + "epoch": 6.25, + "grad_norm": 3.8976359367370605, + "learning_rate": 5.945777956773077e-05, + "loss": 3.3193, + "step": 4271 + }, + { + "epoch": 6.25, + "grad_norm": 2.4183270931243896, + "learning_rate": 5.941490820494965e-05, + "loss": 3.261, + "step": 4272 + }, + { + "epoch": 6.25, + "grad_norm": 3.022622585296631, + "learning_rate": 5.937204577061677e-05, + "loss": 3.2569, + "step": 4273 + }, + { + "epoch": 6.25, + "grad_norm": 3.768277168273926, + "learning_rate": 5.932919227416156e-05, + "loss": 3.2582, + "step": 4274 + }, + { + "epoch": 6.26, + "grad_norm": 2.1014842987060547, + "learning_rate": 5.9286347725011424e-05, + "loss": 3.236, + "step": 4275 + }, + { + "epoch": 6.26, + "grad_norm": 2.9817419052124023, + "learning_rate": 5.9243512132591985e-05, + "loss": 3.2667, + "step": 4276 + }, + { + "epoch": 6.26, + "grad_norm": 4.514947891235352, + "learning_rate": 5.9200685506326734e-05, + "loss": 3.3089, + "step": 4277 + }, + { + "epoch": 6.26, + "grad_norm": 2.3111038208007812, + "learning_rate": 5.915786785563722e-05, + "loss": 3.253, + "step": 4278 + }, + { + "epoch": 6.26, + "grad_norm": 2.7890591621398926, + "learning_rate": 5.911505918994309e-05, + "loss": 3.2643, + "step": 4279 + }, + { + "epoch": 6.26, + "grad_norm": 3.4461700916290283, + "learning_rate": 5.907225951866195e-05, + "loss": 3.2625, + "step": 4280 + }, + { + "epoch": 6.26, + "grad_norm": 2.0314648151397705, + "learning_rate": 5.902946885120939e-05, + "loss": 3.2712, + "step": 4281 + }, + { + "epoch": 6.27, + "grad_norm": 2.992061138153076, + "learning_rate": 5.8986687196999135e-05, + "loss": 3.2365, + "step": 4282 + }, + { + "epoch": 6.27, + "grad_norm": 4.312194347381592, + "learning_rate": 5.8943914565442856e-05, + "loss": 3.2867, + "step": 4283 + }, + { + "epoch": 6.27, + "grad_norm": 2.6370973587036133, + "learning_rate": 5.8901150965950194e-05, + "loss": 3.239, + "step": 4284 + }, + { + "epoch": 6.27, + "grad_norm": 3.0366334915161133, + "learning_rate": 5.885839640792892e-05, + "loss": 3.2651, + "step": 4285 + }, + { + "epoch": 6.27, + "grad_norm": 4.975083827972412, + "learning_rate": 5.881565090078469e-05, + "loss": 3.2896, + "step": 4286 + }, + { + "epoch": 6.27, + "grad_norm": 2.4393575191497803, + "learning_rate": 5.8772914453921276e-05, + "loss": 3.2273, + "step": 4287 + }, + { + "epoch": 6.27, + "grad_norm": 2.5122718811035156, + "learning_rate": 5.873018707674036e-05, + "loss": 3.2617, + "step": 4288 + }, + { + "epoch": 6.28, + "grad_norm": 3.912677526473999, + "learning_rate": 5.868746877864173e-05, + "loss": 3.3112, + "step": 4289 + }, + { + "epoch": 6.28, + "grad_norm": 1.8555430173873901, + "learning_rate": 5.864475956902312e-05, + "loss": 3.322, + "step": 4290 + }, + { + "epoch": 6.28, + "grad_norm": 11.267170906066895, + "learning_rate": 5.860205945728021e-05, + "loss": 3.2759, + "step": 4291 + }, + { + "epoch": 6.28, + "grad_norm": 7.718640327453613, + "learning_rate": 5.855936845280682e-05, + "loss": 3.2407, + "step": 4292 + }, + { + "epoch": 6.28, + "grad_norm": 8.495302200317383, + "learning_rate": 5.851668656499465e-05, + "loss": 3.3493, + "step": 4293 + }, + { + "epoch": 6.28, + "grad_norm": 4.235414981842041, + "learning_rate": 5.8474013803233394e-05, + "loss": 3.2929, + "step": 4294 + }, + { + "epoch": 6.29, + "grad_norm": 2.0188095569610596, + "learning_rate": 5.843135017691085e-05, + "loss": 3.2134, + "step": 4295 + }, + { + "epoch": 6.29, + "grad_norm": 4.253573417663574, + "learning_rate": 5.8388695695412666e-05, + "loss": 3.2903, + "step": 4296 + }, + { + "epoch": 6.29, + "grad_norm": 2.978238582611084, + "learning_rate": 5.834605036812255e-05, + "loss": 3.2783, + "step": 4297 + }, + { + "epoch": 6.29, + "grad_norm": 1.3776922225952148, + "learning_rate": 5.830341420442221e-05, + "loss": 3.2773, + "step": 4298 + }, + { + "epoch": 6.29, + "grad_norm": 2.41166353225708, + "learning_rate": 5.82607872136913e-05, + "loss": 3.2214, + "step": 4299 + }, + { + "epoch": 6.29, + "grad_norm": 0.7983728647232056, + "learning_rate": 5.821816940530749e-05, + "loss": 3.3103, + "step": 4300 + }, + { + "epoch": 6.29, + "grad_norm": 1.7378346920013428, + "learning_rate": 5.8175560788646414e-05, + "loss": 3.2548, + "step": 4301 + }, + { + "epoch": 6.3, + "grad_norm": 0.7058059573173523, + "learning_rate": 5.813296137308164e-05, + "loss": 3.2865, + "step": 4302 + }, + { + "epoch": 6.3, + "grad_norm": 2.132636070251465, + "learning_rate": 5.809037116798479e-05, + "loss": 3.2757, + "step": 4303 + }, + { + "epoch": 6.3, + "grad_norm": 1.359208345413208, + "learning_rate": 5.804779018272544e-05, + "loss": 3.2997, + "step": 4304 + }, + { + "epoch": 6.3, + "grad_norm": 2.7147774696350098, + "learning_rate": 5.800521842667107e-05, + "loss": 3.2789, + "step": 4305 + }, + { + "epoch": 6.3, + "grad_norm": 1.9574331045150757, + "learning_rate": 5.7962655909187226e-05, + "loss": 3.2634, + "step": 4306 + }, + { + "epoch": 6.3, + "grad_norm": 1.1457573175430298, + "learning_rate": 5.792010263963735e-05, + "loss": 3.2802, + "step": 4307 + }, + { + "epoch": 6.3, + "grad_norm": 0.9981921315193176, + "learning_rate": 5.7877558627382886e-05, + "loss": 3.3431, + "step": 4308 + }, + { + "epoch": 6.31, + "grad_norm": 1.708836317062378, + "learning_rate": 5.783502388178328e-05, + "loss": 3.2667, + "step": 4309 + }, + { + "epoch": 6.31, + "grad_norm": 1.0355262756347656, + "learning_rate": 5.7792498412195764e-05, + "loss": 3.2917, + "step": 4310 + }, + { + "epoch": 6.31, + "grad_norm": 3.3587546348571777, + "learning_rate": 5.774998222797577e-05, + "loss": 3.2507, + "step": 4311 + }, + { + "epoch": 6.31, + "grad_norm": 3.278871774673462, + "learning_rate": 5.770747533847658e-05, + "loss": 3.2427, + "step": 4312 + }, + { + "epoch": 6.31, + "grad_norm": 0.9090314507484436, + "learning_rate": 5.766497775304931e-05, + "loss": 3.2401, + "step": 4313 + }, + { + "epoch": 6.31, + "grad_norm": 3.0170810222625732, + "learning_rate": 5.7622489481043254e-05, + "loss": 3.2708, + "step": 4314 + }, + { + "epoch": 6.31, + "grad_norm": 3.1690783500671387, + "learning_rate": 5.758001053180555e-05, + "loss": 3.2354, + "step": 4315 + }, + { + "epoch": 6.32, + "grad_norm": 0.4450177550315857, + "learning_rate": 5.753754091468115e-05, + "loss": 3.2435, + "step": 4316 + }, + { + "epoch": 6.32, + "grad_norm": 3.6252622604370117, + "learning_rate": 5.749508063901328e-05, + "loss": 3.2594, + "step": 4317 + }, + { + "epoch": 6.32, + "grad_norm": 5.237750053405762, + "learning_rate": 5.745262971414276e-05, + "loss": 3.2698, + "step": 4318 + }, + { + "epoch": 6.32, + "grad_norm": 1.0414494276046753, + "learning_rate": 5.741018814940853e-05, + "loss": 3.2509, + "step": 4319 + }, + { + "epoch": 6.32, + "grad_norm": 2.6525533199310303, + "learning_rate": 5.736775595414756e-05, + "loss": 3.2609, + "step": 4320 + }, + { + "epoch": 6.32, + "grad_norm": 3.908628225326538, + "learning_rate": 5.7325333137694525e-05, + "loss": 3.2479, + "step": 4321 + }, + { + "epoch": 6.33, + "grad_norm": 1.5807950496673584, + "learning_rate": 5.728291970938218e-05, + "loss": 3.2861, + "step": 4322 + }, + { + "epoch": 6.33, + "grad_norm": 3.6872596740722656, + "learning_rate": 5.72405156785413e-05, + "loss": 3.2316, + "step": 4323 + }, + { + "epoch": 6.33, + "grad_norm": 4.621198654174805, + "learning_rate": 5.719812105450035e-05, + "loss": 3.245, + "step": 4324 + }, + { + "epoch": 6.33, + "grad_norm": 1.5106016397476196, + "learning_rate": 5.715573584658589e-05, + "loss": 3.215, + "step": 4325 + }, + { + "epoch": 6.33, + "grad_norm": 3.5040335655212402, + "learning_rate": 5.711336006412251e-05, + "loss": 3.2626, + "step": 4326 + }, + { + "epoch": 6.33, + "grad_norm": 4.461084842681885, + "learning_rate": 5.707099371643245e-05, + "loss": 3.2804, + "step": 4327 + }, + { + "epoch": 6.33, + "grad_norm": 2.81062912940979, + "learning_rate": 5.7028636812836035e-05, + "loss": 3.2476, + "step": 4328 + }, + { + "epoch": 6.34, + "grad_norm": 2.1609275341033936, + "learning_rate": 5.6986289362651624e-05, + "loss": 3.2922, + "step": 4329 + }, + { + "epoch": 6.34, + "grad_norm": 39.97700881958008, + "learning_rate": 5.694395137519525e-05, + "loss": 3.262, + "step": 4330 + }, + { + "epoch": 6.34, + "grad_norm": 4.451782703399658, + "learning_rate": 5.690162285978102e-05, + "loss": 3.3011, + "step": 4331 + }, + { + "epoch": 6.34, + "grad_norm": 1.8545382022857666, + "learning_rate": 5.6859303825720935e-05, + "loss": 3.2684, + "step": 4332 + }, + { + "epoch": 6.34, + "grad_norm": 3.6993982791900635, + "learning_rate": 5.6816994282324875e-05, + "loss": 3.2664, + "step": 4333 + }, + { + "epoch": 6.34, + "grad_norm": 4.563552379608154, + "learning_rate": 5.67746942389007e-05, + "loss": 3.2004, + "step": 4334 + }, + { + "epoch": 6.34, + "grad_norm": 3.6278369426727295, + "learning_rate": 5.6732403704754086e-05, + "loss": 3.1965, + "step": 4335 + }, + { + "epoch": 6.35, + "grad_norm": 1.3190866708755493, + "learning_rate": 5.66901226891887e-05, + "loss": 3.2505, + "step": 4336 + }, + { + "epoch": 6.35, + "grad_norm": 3.1047582626342773, + "learning_rate": 5.664785120150607e-05, + "loss": 3.3744, + "step": 4337 + }, + { + "epoch": 6.35, + "grad_norm": 1.4010859727859497, + "learning_rate": 5.660558925100565e-05, + "loss": 3.2555, + "step": 4338 + }, + { + "epoch": 6.35, + "grad_norm": 2.454122304916382, + "learning_rate": 5.656333684698476e-05, + "loss": 3.2401, + "step": 4339 + }, + { + "epoch": 6.35, + "grad_norm": 2.7071664333343506, + "learning_rate": 5.652109399873873e-05, + "loss": 3.2436, + "step": 4340 + }, + { + "epoch": 6.35, + "grad_norm": 0.4050593674182892, + "learning_rate": 5.647886071556054e-05, + "loss": 3.2843, + "step": 4341 + }, + { + "epoch": 6.35, + "grad_norm": 3.2467408180236816, + "learning_rate": 5.6436637006741375e-05, + "loss": 3.2526, + "step": 4342 + }, + { + "epoch": 6.36, + "grad_norm": 3.083259344100952, + "learning_rate": 5.639442288157015e-05, + "loss": 3.2539, + "step": 4343 + }, + { + "epoch": 6.36, + "grad_norm": 1.2270174026489258, + "learning_rate": 5.635221834933359e-05, + "loss": 3.2435, + "step": 4344 + }, + { + "epoch": 6.36, + "grad_norm": 1.6782840490341187, + "learning_rate": 5.63100234193165e-05, + "loss": 3.2838, + "step": 4345 + }, + { + "epoch": 6.36, + "grad_norm": 1.8353314399719238, + "learning_rate": 5.6267838100801516e-05, + "loss": 3.2793, + "step": 4346 + }, + { + "epoch": 6.36, + "grad_norm": 1.642630934715271, + "learning_rate": 5.622566240306899e-05, + "loss": 3.261, + "step": 4347 + }, + { + "epoch": 6.36, + "grad_norm": 1.191734790802002, + "learning_rate": 5.618349633539741e-05, + "loss": 3.2822, + "step": 4348 + }, + { + "epoch": 6.37, + "grad_norm": 0.7267574071884155, + "learning_rate": 5.614133990706304e-05, + "loss": 3.2482, + "step": 4349 + }, + { + "epoch": 6.37, + "grad_norm": 0.9115816950798035, + "learning_rate": 5.6099193127339864e-05, + "loss": 3.2347, + "step": 4350 + }, + { + "epoch": 6.37, + "grad_norm": 1.4531807899475098, + "learning_rate": 5.6057056005500085e-05, + "loss": 3.2186, + "step": 4351 + }, + { + "epoch": 6.37, + "grad_norm": 1.0633835792541504, + "learning_rate": 5.601492855081345e-05, + "loss": 3.2526, + "step": 4352 + }, + { + "epoch": 6.37, + "grad_norm": 0.716886579990387, + "learning_rate": 5.5972810772547724e-05, + "loss": 3.2448, + "step": 4353 + }, + { + "epoch": 6.37, + "grad_norm": 0.9413835406303406, + "learning_rate": 5.5930702679968636e-05, + "loss": 3.2125, + "step": 4354 + }, + { + "epoch": 6.37, + "grad_norm": 1.2338863611221313, + "learning_rate": 5.58886042823396e-05, + "loss": 3.2228, + "step": 4355 + }, + { + "epoch": 6.38, + "grad_norm": 2.2373838424682617, + "learning_rate": 5.584651558892193e-05, + "loss": 3.2777, + "step": 4356 + }, + { + "epoch": 6.38, + "grad_norm": 1.3574821949005127, + "learning_rate": 5.580443660897503e-05, + "loss": 3.2371, + "step": 4357 + }, + { + "epoch": 6.38, + "grad_norm": 1.0699307918548584, + "learning_rate": 5.576236735175584e-05, + "loss": 3.2889, + "step": 4358 + }, + { + "epoch": 6.38, + "grad_norm": 0.737332284450531, + "learning_rate": 5.572030782651938e-05, + "loss": 3.2533, + "step": 4359 + }, + { + "epoch": 6.38, + "grad_norm": 1.1307657957077026, + "learning_rate": 5.567825804251843e-05, + "loss": 3.2832, + "step": 4360 + }, + { + "epoch": 6.38, + "grad_norm": 0.6265879273414612, + "learning_rate": 5.563621800900369e-05, + "loss": 3.2663, + "step": 4361 + }, + { + "epoch": 6.38, + "grad_norm": 2.2958106994628906, + "learning_rate": 5.559418773522367e-05, + "loss": 3.2426, + "step": 4362 + }, + { + "epoch": 6.39, + "grad_norm": 1.0793524980545044, + "learning_rate": 5.555216723042476e-05, + "loss": 3.2336, + "step": 4363 + }, + { + "epoch": 6.39, + "grad_norm": 1.8671746253967285, + "learning_rate": 5.551015650385119e-05, + "loss": 3.2449, + "step": 4364 + }, + { + "epoch": 6.39, + "grad_norm": 1.5960359573364258, + "learning_rate": 5.5468155564745e-05, + "loss": 3.2135, + "step": 4365 + }, + { + "epoch": 6.39, + "grad_norm": 2.3406100273132324, + "learning_rate": 5.542616442234618e-05, + "loss": 3.2034, + "step": 4366 + }, + { + "epoch": 6.39, + "grad_norm": 1.8202327489852905, + "learning_rate": 5.538418308589246e-05, + "loss": 3.2608, + "step": 4367 + }, + { + "epoch": 6.39, + "grad_norm": 1.2679892778396606, + "learning_rate": 5.534221156461945e-05, + "loss": 3.2336, + "step": 4368 + }, + { + "epoch": 6.4, + "grad_norm": 1.0765639543533325, + "learning_rate": 5.530024986776061e-05, + "loss": 3.2034, + "step": 4369 + }, + { + "epoch": 6.4, + "grad_norm": 2.065032720565796, + "learning_rate": 5.5258298004547246e-05, + "loss": 3.2323, + "step": 4370 + }, + { + "epoch": 6.4, + "grad_norm": 1.7677899599075317, + "learning_rate": 5.521635598420847e-05, + "loss": 3.2416, + "step": 4371 + }, + { + "epoch": 6.4, + "grad_norm": 1.2142525911331177, + "learning_rate": 5.5174423815971254e-05, + "loss": 3.2783, + "step": 4372 + }, + { + "epoch": 6.4, + "grad_norm": 1.2158989906311035, + "learning_rate": 5.513250150906039e-05, + "loss": 3.2203, + "step": 4373 + }, + { + "epoch": 6.4, + "grad_norm": 2.0104503631591797, + "learning_rate": 5.509058907269854e-05, + "loss": 3.2627, + "step": 4374 + }, + { + "epoch": 6.4, + "grad_norm": 1.794409155845642, + "learning_rate": 5.504868651610604e-05, + "loss": 3.2247, + "step": 4375 + }, + { + "epoch": 6.41, + "grad_norm": 1.241929531097412, + "learning_rate": 5.5006793848501314e-05, + "loss": 3.2352, + "step": 4376 + }, + { + "epoch": 6.41, + "grad_norm": 1.3682582378387451, + "learning_rate": 5.4964911079100446e-05, + "loss": 3.2834, + "step": 4377 + }, + { + "epoch": 6.41, + "grad_norm": 1.42366623878479, + "learning_rate": 5.492303821711724e-05, + "loss": 3.2356, + "step": 4378 + }, + { + "epoch": 6.41, + "grad_norm": 0.9967645406723022, + "learning_rate": 5.488117527176357e-05, + "loss": 3.2623, + "step": 4379 + }, + { + "epoch": 6.41, + "grad_norm": 1.6339219808578491, + "learning_rate": 5.4839322252249016e-05, + "loss": 3.2162, + "step": 4380 + }, + { + "epoch": 6.41, + "grad_norm": 0.9269053936004639, + "learning_rate": 5.479747916778088e-05, + "loss": 3.2236, + "step": 4381 + }, + { + "epoch": 6.41, + "grad_norm": 3.181241989135742, + "learning_rate": 5.4755646027564336e-05, + "loss": 3.2464, + "step": 4382 + }, + { + "epoch": 6.42, + "grad_norm": 3.343838930130005, + "learning_rate": 5.4713822840802544e-05, + "loss": 3.227, + "step": 4383 + }, + { + "epoch": 6.42, + "grad_norm": 0.8503923416137695, + "learning_rate": 5.467200961669619e-05, + "loss": 3.2975, + "step": 4384 + }, + { + "epoch": 6.42, + "grad_norm": 2.970435380935669, + "learning_rate": 5.4630206364443936e-05, + "loss": 3.2325, + "step": 4385 + }, + { + "epoch": 6.42, + "grad_norm": 3.7299089431762695, + "learning_rate": 5.4588413093242244e-05, + "loss": 3.1893, + "step": 4386 + }, + { + "epoch": 6.42, + "grad_norm": 1.3558520078659058, + "learning_rate": 5.454662981228532e-05, + "loss": 3.2106, + "step": 4387 + }, + { + "epoch": 6.42, + "grad_norm": 3.116187572479248, + "learning_rate": 5.450485653076524e-05, + "loss": 3.2292, + "step": 4388 + }, + { + "epoch": 6.42, + "grad_norm": 3.9433794021606445, + "learning_rate": 5.446309325787181e-05, + "loss": 3.2971, + "step": 4389 + }, + { + "epoch": 6.43, + "grad_norm": 2.001657247543335, + "learning_rate": 5.44213400027927e-05, + "loss": 3.2795, + "step": 4390 + }, + { + "epoch": 6.43, + "grad_norm": 2.557114601135254, + "learning_rate": 5.4379596774713334e-05, + "loss": 3.2177, + "step": 4391 + }, + { + "epoch": 6.43, + "grad_norm": 3.371882677078247, + "learning_rate": 5.4337863582816936e-05, + "loss": 3.2478, + "step": 4392 + }, + { + "epoch": 6.43, + "grad_norm": 1.5067156553268433, + "learning_rate": 5.4296140436284546e-05, + "loss": 3.2362, + "step": 4393 + }, + { + "epoch": 6.43, + "grad_norm": 2.2881274223327637, + "learning_rate": 5.425442734429495e-05, + "loss": 3.2567, + "step": 4394 + }, + { + "epoch": 6.43, + "eval_loss": 3.067939043045044, + "eval_runtime": 1166.4557, + "eval_samples_per_second": 14.696, + "eval_steps_per_second": 0.919, + "step": 4394 + }, + { + "epoch": 6.43, + "grad_norm": 2.7140753269195557, + "learning_rate": 5.4212724316024776e-05, + "loss": 3.1976, + "step": 4395 + }, + { + "epoch": 6.44, + "grad_norm": 1.5862021446228027, + "learning_rate": 5.4171031360648406e-05, + "loss": 3.2628, + "step": 4396 + }, + { + "epoch": 6.44, + "grad_norm": 1.6570472717285156, + "learning_rate": 5.4129348487337995e-05, + "loss": 3.2399, + "step": 4397 + }, + { + "epoch": 6.44, + "grad_norm": 1.0379539728164673, + "learning_rate": 5.408767570526352e-05, + "loss": 3.2495, + "step": 4398 + }, + { + "epoch": 6.44, + "grad_norm": 0.9085617065429688, + "learning_rate": 5.404601302359269e-05, + "loss": 3.2073, + "step": 4399 + }, + { + "epoch": 6.44, + "grad_norm": 1.3575804233551025, + "learning_rate": 5.400436045149102e-05, + "loss": 3.2191, + "step": 4400 + }, + { + "epoch": 6.44, + "grad_norm": 1.4507752656936646, + "learning_rate": 5.39627179981218e-05, + "loss": 3.2083, + "step": 4401 + }, + { + "epoch": 6.44, + "grad_norm": 1.4040168523788452, + "learning_rate": 5.392108567264609e-05, + "loss": 3.2492, + "step": 4402 + }, + { + "epoch": 6.45, + "grad_norm": 1.684368371963501, + "learning_rate": 5.387946348422274e-05, + "loss": 3.2433, + "step": 4403 + }, + { + "epoch": 6.45, + "grad_norm": 1.438843846321106, + "learning_rate": 5.383785144200832e-05, + "loss": 3.2335, + "step": 4404 + }, + { + "epoch": 6.45, + "grad_norm": 1.311325192451477, + "learning_rate": 5.379624955515723e-05, + "loss": 3.2418, + "step": 4405 + }, + { + "epoch": 6.45, + "grad_norm": 1.0702279806137085, + "learning_rate": 5.375465783282163e-05, + "loss": 3.2534, + "step": 4406 + }, + { + "epoch": 6.45, + "grad_norm": 1.4737073183059692, + "learning_rate": 5.371307628415131e-05, + "loss": 3.2167, + "step": 4407 + }, + { + "epoch": 6.45, + "grad_norm": 1.0920416116714478, + "learning_rate": 5.367150491829408e-05, + "loss": 3.2418, + "step": 4408 + }, + { + "epoch": 6.45, + "grad_norm": 2.5086584091186523, + "learning_rate": 5.362994374439525e-05, + "loss": 3.2499, + "step": 4409 + }, + { + "epoch": 6.46, + "grad_norm": 1.8857651948928833, + "learning_rate": 5.3588392771598e-05, + "loss": 3.2232, + "step": 4410 + }, + { + "epoch": 6.46, + "grad_norm": 1.4724599123001099, + "learning_rate": 5.354685200904337e-05, + "loss": 3.2534, + "step": 4411 + }, + { + "epoch": 6.46, + "grad_norm": 1.5440292358398438, + "learning_rate": 5.350532146586996e-05, + "loss": 3.2061, + "step": 4412 + }, + { + "epoch": 6.46, + "grad_norm": 1.186196208000183, + "learning_rate": 5.346380115121419e-05, + "loss": 3.2526, + "step": 4413 + }, + { + "epoch": 6.46, + "grad_norm": 1.4283154010772705, + "learning_rate": 5.342229107421037e-05, + "loss": 3.2498, + "step": 4414 + }, + { + "epoch": 6.46, + "grad_norm": 1.7490146160125732, + "learning_rate": 5.3380791243990325e-05, + "loss": 3.2403, + "step": 4415 + }, + { + "epoch": 6.46, + "grad_norm": 1.5179111957550049, + "learning_rate": 5.333930166968374e-05, + "loss": 3.2239, + "step": 4416 + }, + { + "epoch": 6.47, + "grad_norm": 1.3260568380355835, + "learning_rate": 5.329782236041816e-05, + "loss": 3.2357, + "step": 4417 + }, + { + "epoch": 6.47, + "grad_norm": 0.9309088587760925, + "learning_rate": 5.325635332531864e-05, + "loss": 3.2212, + "step": 4418 + }, + { + "epoch": 6.47, + "grad_norm": 1.8849228620529175, + "learning_rate": 5.321489457350812e-05, + "loss": 3.2158, + "step": 4419 + }, + { + "epoch": 6.47, + "grad_norm": 2.031489610671997, + "learning_rate": 5.3173446114107265e-05, + "loss": 3.229, + "step": 4420 + }, + { + "epoch": 6.47, + "grad_norm": 1.7095611095428467, + "learning_rate": 5.313200795623443e-05, + "loss": 3.2323, + "step": 4421 + }, + { + "epoch": 6.47, + "grad_norm": 2.060396671295166, + "learning_rate": 5.309058010900576e-05, + "loss": 3.2519, + "step": 4422 + }, + { + "epoch": 6.48, + "grad_norm": 0.750653088092804, + "learning_rate": 5.3049162581535094e-05, + "loss": 3.2344, + "step": 4423 + }, + { + "epoch": 6.48, + "grad_norm": 1.6587752103805542, + "learning_rate": 5.300775538293401e-05, + "loss": 3.2506, + "step": 4424 + }, + { + "epoch": 6.48, + "grad_norm": 0.8414655923843384, + "learning_rate": 5.296635852231182e-05, + "loss": 3.2065, + "step": 4425 + }, + { + "epoch": 6.48, + "grad_norm": 0.4840925335884094, + "learning_rate": 5.292497200877553e-05, + "loss": 3.2054, + "step": 4426 + }, + { + "epoch": 6.48, + "grad_norm": 0.6098539233207703, + "learning_rate": 5.288359585142993e-05, + "loss": 3.2024, + "step": 4427 + }, + { + "epoch": 6.48, + "grad_norm": 0.46651723980903625, + "learning_rate": 5.2842230059377486e-05, + "loss": 3.2161, + "step": 4428 + }, + { + "epoch": 6.48, + "grad_norm": 0.508057177066803, + "learning_rate": 5.280087464171838e-05, + "loss": 3.1871, + "step": 4429 + }, + { + "epoch": 6.49, + "grad_norm": 0.5408399701118469, + "learning_rate": 5.275952960755055e-05, + "loss": 3.2624, + "step": 4430 + }, + { + "epoch": 6.49, + "grad_norm": 1.0045307874679565, + "learning_rate": 5.271819496596964e-05, + "loss": 3.2303, + "step": 4431 + }, + { + "epoch": 6.49, + "grad_norm": 0.6911265850067139, + "learning_rate": 5.2676870726068906e-05, + "loss": 3.2087, + "step": 4432 + }, + { + "epoch": 6.49, + "grad_norm": 0.6002917289733887, + "learning_rate": 5.2635556896939486e-05, + "loss": 3.1978, + "step": 4433 + }, + { + "epoch": 6.49, + "grad_norm": 0.9630332589149475, + "learning_rate": 5.259425348767018e-05, + "loss": 3.2238, + "step": 4434 + }, + { + "epoch": 6.49, + "grad_norm": 0.7534378170967102, + "learning_rate": 5.2552960507347325e-05, + "loss": 3.2308, + "step": 4435 + }, + { + "epoch": 6.49, + "grad_norm": 0.9733850359916687, + "learning_rate": 5.251167796505523e-05, + "loss": 3.2392, + "step": 4436 + }, + { + "epoch": 6.5, + "grad_norm": 1.1052712202072144, + "learning_rate": 5.247040586987575e-05, + "loss": 3.2257, + "step": 4437 + }, + { + "epoch": 6.5, + "grad_norm": 0.8000559210777283, + "learning_rate": 5.2429144230888384e-05, + "loss": 3.2837, + "step": 4438 + }, + { + "epoch": 6.5, + "grad_norm": 0.7555025219917297, + "learning_rate": 5.238789305717053e-05, + "loss": 3.2463, + "step": 4439 + }, + { + "epoch": 6.5, + "grad_norm": 1.0577791929244995, + "learning_rate": 5.2346652357797146e-05, + "loss": 3.2328, + "step": 4440 + }, + { + "epoch": 6.5, + "grad_norm": 0.6421762704849243, + "learning_rate": 5.230542214184081e-05, + "loss": 3.2186, + "step": 4441 + }, + { + "epoch": 6.5, + "grad_norm": 1.1943224668502808, + "learning_rate": 5.226420241837204e-05, + "loss": 3.2487, + "step": 4442 + }, + { + "epoch": 6.5, + "grad_norm": 0.6367546916007996, + "learning_rate": 5.22229931964588e-05, + "loss": 3.2938, + "step": 4443 + }, + { + "epoch": 6.51, + "grad_norm": 1.3476903438568115, + "learning_rate": 5.2181794485166824e-05, + "loss": 3.2098, + "step": 4444 + }, + { + "epoch": 6.51, + "grad_norm": 1.686813235282898, + "learning_rate": 5.214060629355968e-05, + "loss": 3.2456, + "step": 4445 + }, + { + "epoch": 6.51, + "grad_norm": 0.49779945611953735, + "learning_rate": 5.209942863069837e-05, + "loss": 3.2537, + "step": 4446 + }, + { + "epoch": 6.51, + "grad_norm": 1.0947519540786743, + "learning_rate": 5.20582615056417e-05, + "loss": 3.3028, + "step": 4447 + }, + { + "epoch": 6.51, + "grad_norm": 0.6501612663269043, + "learning_rate": 5.201710492744629e-05, + "loss": 3.2317, + "step": 4448 + }, + { + "epoch": 6.51, + "grad_norm": 0.4535963535308838, + "learning_rate": 5.19759589051662e-05, + "loss": 3.23, + "step": 4449 + }, + { + "epoch": 6.52, + "grad_norm": 0.5250782370567322, + "learning_rate": 5.193482344785327e-05, + "loss": 3.2435, + "step": 4450 + }, + { + "epoch": 6.52, + "grad_norm": 0.400921493768692, + "learning_rate": 5.1893698564557124e-05, + "loss": 3.2188, + "step": 4451 + }, + { + "epoch": 6.52, + "grad_norm": 0.902998149394989, + "learning_rate": 5.1852584264324866e-05, + "loss": 3.2495, + "step": 4452 + }, + { + "epoch": 6.52, + "grad_norm": 0.7799638509750366, + "learning_rate": 5.1811480556201375e-05, + "loss": 3.234, + "step": 4453 + }, + { + "epoch": 6.52, + "grad_norm": 1.8399096727371216, + "learning_rate": 5.1770387449229276e-05, + "loss": 3.3015, + "step": 4454 + }, + { + "epoch": 6.52, + "grad_norm": 0.94755619764328, + "learning_rate": 5.172930495244869e-05, + "loss": 3.2394, + "step": 4455 + }, + { + "epoch": 6.52, + "grad_norm": 2.66386079788208, + "learning_rate": 5.168823307489751e-05, + "loss": 3.2519, + "step": 4456 + }, + { + "epoch": 6.53, + "grad_norm": 1.893293857574463, + "learning_rate": 5.1647171825611284e-05, + "loss": 3.1926, + "step": 4457 + }, + { + "epoch": 6.53, + "grad_norm": 0.6333227753639221, + "learning_rate": 5.160612121362322e-05, + "loss": 3.2169, + "step": 4458 + }, + { + "epoch": 6.53, + "grad_norm": 1.2479146718978882, + "learning_rate": 5.156508124796415e-05, + "loss": 3.2308, + "step": 4459 + }, + { + "epoch": 6.53, + "grad_norm": 0.3292587399482727, + "learning_rate": 5.152405193766263e-05, + "loss": 3.2359, + "step": 4460 + }, + { + "epoch": 6.53, + "grad_norm": 1.253524899482727, + "learning_rate": 5.1483033291744797e-05, + "loss": 3.2176, + "step": 4461 + }, + { + "epoch": 6.53, + "grad_norm": 0.4894808232784271, + "learning_rate": 5.14420253192345e-05, + "loss": 3.2433, + "step": 4462 + }, + { + "epoch": 6.53, + "grad_norm": 1.1630624532699585, + "learning_rate": 5.140102802915321e-05, + "loss": 3.213, + "step": 4463 + }, + { + "epoch": 6.54, + "grad_norm": 0.8647369742393494, + "learning_rate": 5.136004143052007e-05, + "loss": 3.2736, + "step": 4464 + }, + { + "epoch": 6.54, + "grad_norm": 0.9852306842803955, + "learning_rate": 5.1319065532351865e-05, + "loss": 3.2649, + "step": 4465 + }, + { + "epoch": 6.54, + "grad_norm": 0.7359192371368408, + "learning_rate": 5.1278100343662916e-05, + "loss": 3.2311, + "step": 4466 + }, + { + "epoch": 6.54, + "grad_norm": 0.38752880692481995, + "learning_rate": 5.123714587346543e-05, + "loss": 3.2308, + "step": 4467 + }, + { + "epoch": 6.54, + "grad_norm": 1.2246233224868774, + "learning_rate": 5.119620213076909e-05, + "loss": 3.1778, + "step": 4468 + }, + { + "epoch": 6.54, + "grad_norm": 0.7400122284889221, + "learning_rate": 5.115526912458113e-05, + "loss": 3.2127, + "step": 4469 + }, + { + "epoch": 6.55, + "grad_norm": 0.9233433604240417, + "learning_rate": 5.111434686390666e-05, + "loss": 3.184, + "step": 4470 + }, + { + "epoch": 6.55, + "grad_norm": 1.7230321168899536, + "learning_rate": 5.107343535774831e-05, + "loss": 3.2116, + "step": 4471 + }, + { + "epoch": 6.55, + "grad_norm": 1.6479278802871704, + "learning_rate": 5.1032534615106195e-05, + "loss": 3.2049, + "step": 4472 + }, + { + "epoch": 6.55, + "grad_norm": 2.626024007797241, + "learning_rate": 5.0991644644978356e-05, + "loss": 3.2434, + "step": 4473 + }, + { + "epoch": 6.55, + "grad_norm": 3.1591696739196777, + "learning_rate": 5.095076545636028e-05, + "loss": 3.2116, + "step": 4474 + }, + { + "epoch": 6.55, + "grad_norm": 0.8667658567428589, + "learning_rate": 5.0909897058245025e-05, + "loss": 3.1994, + "step": 4475 + }, + { + "epoch": 6.55, + "grad_norm": 3.626915454864502, + "learning_rate": 5.08690394596235e-05, + "loss": 3.221, + "step": 4476 + }, + { + "epoch": 6.56, + "grad_norm": 4.862388610839844, + "learning_rate": 5.0828192669484e-05, + "loss": 3.2172, + "step": 4477 + }, + { + "epoch": 6.56, + "grad_norm": 2.2208385467529297, + "learning_rate": 5.078735669681257e-05, + "loss": 3.2153, + "step": 4478 + }, + { + "epoch": 6.56, + "grad_norm": 2.200000286102295, + "learning_rate": 5.074653155059286e-05, + "loss": 3.2971, + "step": 4479 + }, + { + "epoch": 6.56, + "grad_norm": 2.9845163822174072, + "learning_rate": 5.0705717239806125e-05, + "loss": 3.2522, + "step": 4480 + }, + { + "epoch": 6.56, + "grad_norm": 1.4984869956970215, + "learning_rate": 5.066491377343126e-05, + "loss": 3.2009, + "step": 4481 + }, + { + "epoch": 6.56, + "grad_norm": 2.6040592193603516, + "learning_rate": 5.0624121160444716e-05, + "loss": 3.1945, + "step": 4482 + }, + { + "epoch": 6.56, + "grad_norm": 3.3462085723876953, + "learning_rate": 5.0583339409820616e-05, + "loss": 3.1927, + "step": 4483 + }, + { + "epoch": 6.57, + "grad_norm": 1.2761520147323608, + "learning_rate": 5.0542568530530676e-05, + "loss": 3.1981, + "step": 4484 + }, + { + "epoch": 6.57, + "grad_norm": 2.611973285675049, + "learning_rate": 5.05018085315442e-05, + "loss": 3.1832, + "step": 4485 + }, + { + "epoch": 6.57, + "grad_norm": 3.3917155265808105, + "learning_rate": 5.046105942182815e-05, + "loss": 3.2319, + "step": 4486 + }, + { + "epoch": 6.57, + "grad_norm": 2.2328946590423584, + "learning_rate": 5.042032121034701e-05, + "loss": 3.2443, + "step": 4487 + }, + { + "epoch": 6.57, + "grad_norm": 2.1951651573181152, + "learning_rate": 5.037959390606294e-05, + "loss": 3.2073, + "step": 4488 + }, + { + "epoch": 6.57, + "grad_norm": 3.5785024166107178, + "learning_rate": 5.033887751793567e-05, + "loss": 3.264, + "step": 4489 + }, + { + "epoch": 6.57, + "grad_norm": 1.865713119506836, + "learning_rate": 5.0298172054922534e-05, + "loss": 3.2455, + "step": 4490 + }, + { + "epoch": 6.58, + "grad_norm": 2.268017292022705, + "learning_rate": 5.0257477525978446e-05, + "loss": 3.2089, + "step": 4491 + }, + { + "epoch": 6.58, + "grad_norm": 2.626833200454712, + "learning_rate": 5.021679394005596e-05, + "loss": 3.2474, + "step": 4492 + }, + { + "epoch": 6.58, + "grad_norm": 1.4748424291610718, + "learning_rate": 5.0176121306105164e-05, + "loss": 3.2171, + "step": 4493 + }, + { + "epoch": 6.58, + "grad_norm": 1.906065583229065, + "learning_rate": 5.0135459633073776e-05, + "loss": 3.2625, + "step": 4494 + }, + { + "epoch": 6.58, + "grad_norm": 2.797215700149536, + "learning_rate": 5.00948089299071e-05, + "loss": 3.2316, + "step": 4495 + }, + { + "epoch": 6.58, + "grad_norm": 0.824268639087677, + "learning_rate": 5.0054169205548016e-05, + "loss": 3.211, + "step": 4496 + }, + { + "epoch": 6.59, + "grad_norm": 2.206386089324951, + "learning_rate": 5.001354046893697e-05, + "loss": 3.2278, + "step": 4497 + }, + { + "epoch": 6.59, + "grad_norm": 2.08886981010437, + "learning_rate": 4.997292272901203e-05, + "loss": 3.2376, + "step": 4498 + }, + { + "epoch": 6.59, + "grad_norm": 0.7808924913406372, + "learning_rate": 4.993231599470887e-05, + "loss": 3.2411, + "step": 4499 + }, + { + "epoch": 6.59, + "grad_norm": 1.8568205833435059, + "learning_rate": 4.9891720274960566e-05, + "loss": 3.2718, + "step": 4500 + }, + { + "epoch": 6.59, + "grad_norm": 1.4308000802993774, + "learning_rate": 4.985113557869803e-05, + "loss": 3.2299, + "step": 4501 + }, + { + "epoch": 6.59, + "grad_norm": 1.8029470443725586, + "learning_rate": 4.981056191484962e-05, + "loss": 3.2458, + "step": 4502 + }, + { + "epoch": 6.59, + "grad_norm": 1.7437525987625122, + "learning_rate": 4.9769999292341194e-05, + "loss": 3.2255, + "step": 4503 + }, + { + "epoch": 6.6, + "grad_norm": 1.1221601963043213, + "learning_rate": 4.9729447720096265e-05, + "loss": 3.1949, + "step": 4504 + }, + { + "epoch": 6.6, + "grad_norm": 1.0662968158721924, + "learning_rate": 4.968890720703602e-05, + "loss": 3.1963, + "step": 4505 + }, + { + "epoch": 6.6, + "grad_norm": 1.716160535812378, + "learning_rate": 4.9648377762078966e-05, + "loss": 3.2037, + "step": 4506 + }, + { + "epoch": 6.6, + "grad_norm": 1.2884470224380493, + "learning_rate": 4.960785939414134e-05, + "loss": 3.2482, + "step": 4507 + }, + { + "epoch": 6.6, + "grad_norm": 2.104020595550537, + "learning_rate": 4.956735211213701e-05, + "loss": 3.2097, + "step": 4508 + }, + { + "epoch": 6.6, + "grad_norm": 2.3521175384521484, + "learning_rate": 4.9526855924977175e-05, + "loss": 3.2618, + "step": 4509 + }, + { + "epoch": 6.6, + "grad_norm": 1.780421257019043, + "learning_rate": 4.9486370841570803e-05, + "loss": 3.2177, + "step": 4510 + }, + { + "epoch": 6.61, + "grad_norm": 1.345507264137268, + "learning_rate": 4.94458968708243e-05, + "loss": 3.1907, + "step": 4511 + }, + { + "epoch": 6.61, + "grad_norm": 1.3987131118774414, + "learning_rate": 4.9405434021641705e-05, + "loss": 3.2361, + "step": 4512 + }, + { + "epoch": 6.61, + "grad_norm": 1.295686960220337, + "learning_rate": 4.9364982302924545e-05, + "loss": 3.2055, + "step": 4513 + }, + { + "epoch": 6.61, + "grad_norm": 1.0267897844314575, + "learning_rate": 4.9324541723571936e-05, + "loss": 3.2089, + "step": 4514 + }, + { + "epoch": 6.61, + "grad_norm": 1.2543052434921265, + "learning_rate": 4.9284112292480535e-05, + "loss": 3.2271, + "step": 4515 + }, + { + "epoch": 6.61, + "grad_norm": 1.0035556554794312, + "learning_rate": 4.924369401854455e-05, + "loss": 3.2686, + "step": 4516 + }, + { + "epoch": 6.61, + "grad_norm": 1.4459877014160156, + "learning_rate": 4.920328691065573e-05, + "loss": 3.2149, + "step": 4517 + }, + { + "epoch": 6.62, + "grad_norm": 1.3623418807983398, + "learning_rate": 4.916289097770337e-05, + "loss": 3.2456, + "step": 4518 + }, + { + "epoch": 6.62, + "grad_norm": 1.6200556755065918, + "learning_rate": 4.9122506228574286e-05, + "loss": 3.1883, + "step": 4519 + }, + { + "epoch": 6.62, + "grad_norm": 1.0919500589370728, + "learning_rate": 4.908213267215287e-05, + "loss": 3.1899, + "step": 4520 + }, + { + "epoch": 6.62, + "grad_norm": 1.162226676940918, + "learning_rate": 4.904177031732103e-05, + "loss": 3.2402, + "step": 4521 + }, + { + "epoch": 6.62, + "grad_norm": 1.1378471851348877, + "learning_rate": 4.900141917295822e-05, + "loss": 3.3377, + "step": 4522 + }, + { + "epoch": 6.62, + "grad_norm": 1.183754563331604, + "learning_rate": 4.896107924794141e-05, + "loss": 3.2368, + "step": 4523 + }, + { + "epoch": 6.63, + "grad_norm": 1.1688711643218994, + "learning_rate": 4.892075055114511e-05, + "loss": 3.2358, + "step": 4524 + }, + { + "epoch": 6.63, + "grad_norm": 0.9402367472648621, + "learning_rate": 4.888043309144138e-05, + "loss": 3.1947, + "step": 4525 + }, + { + "epoch": 6.63, + "grad_norm": 0.9889912009239197, + "learning_rate": 4.8840126877699787e-05, + "loss": 3.2409, + "step": 4526 + }, + { + "epoch": 6.63, + "grad_norm": 1.3844680786132812, + "learning_rate": 4.879983191878742e-05, + "loss": 3.2658, + "step": 4527 + }, + { + "epoch": 6.63, + "grad_norm": 0.8452149033546448, + "learning_rate": 4.875954822356893e-05, + "loss": 3.1865, + "step": 4528 + }, + { + "epoch": 6.63, + "grad_norm": 0.9102594256401062, + "learning_rate": 4.871927580090636e-05, + "loss": 3.2226, + "step": 4529 + }, + { + "epoch": 6.63, + "grad_norm": 0.7820479869842529, + "learning_rate": 4.867901465965949e-05, + "loss": 3.2125, + "step": 4530 + }, + { + "epoch": 6.64, + "grad_norm": 1.0344125032424927, + "learning_rate": 4.863876480868549e-05, + "loss": 3.1863, + "step": 4531 + }, + { + "epoch": 6.64, + "grad_norm": 1.1749505996704102, + "learning_rate": 4.8598526256838947e-05, + "loss": 3.2641, + "step": 4532 + }, + { + "epoch": 6.64, + "grad_norm": 1.117828130722046, + "learning_rate": 4.855829901297223e-05, + "loss": 3.1992, + "step": 4533 + }, + { + "epoch": 6.64, + "grad_norm": 1.255987286567688, + "learning_rate": 4.8518083085934954e-05, + "loss": 3.1951, + "step": 4534 + }, + { + "epoch": 6.64, + "grad_norm": 0.6159536838531494, + "learning_rate": 4.847787848457433e-05, + "loss": 3.2087, + "step": 4535 + }, + { + "epoch": 6.64, + "grad_norm": 1.0251574516296387, + "learning_rate": 4.843768521773523e-05, + "loss": 3.202, + "step": 4536 + }, + { + "epoch": 6.64, + "grad_norm": 1.4189834594726562, + "learning_rate": 4.8397503294259795e-05, + "loss": 3.2329, + "step": 4537 + }, + { + "epoch": 6.65, + "grad_norm": 1.0889549255371094, + "learning_rate": 4.835733272298777e-05, + "loss": 3.2087, + "step": 4538 + }, + { + "epoch": 6.65, + "grad_norm": 1.1167737245559692, + "learning_rate": 4.831717351275653e-05, + "loss": 3.2098, + "step": 4539 + }, + { + "epoch": 6.65, + "grad_norm": 1.141223430633545, + "learning_rate": 4.827702567240072e-05, + "loss": 3.2151, + "step": 4540 + }, + { + "epoch": 6.65, + "grad_norm": 1.1248409748077393, + "learning_rate": 4.823688921075258e-05, + "loss": 3.2147, + "step": 4541 + }, + { + "epoch": 6.65, + "grad_norm": 1.471689224243164, + "learning_rate": 4.8196764136642e-05, + "loss": 3.1923, + "step": 4542 + }, + { + "epoch": 6.65, + "grad_norm": 1.6416572332382202, + "learning_rate": 4.81566504588961e-05, + "loss": 3.2319, + "step": 4543 + }, + { + "epoch": 6.65, + "grad_norm": 1.1048762798309326, + "learning_rate": 4.811654818633966e-05, + "loss": 3.2002, + "step": 4544 + }, + { + "epoch": 6.66, + "grad_norm": 1.3798350095748901, + "learning_rate": 4.807645732779491e-05, + "loss": 3.1903, + "step": 4545 + }, + { + "epoch": 6.66, + "grad_norm": 1.5549540519714355, + "learning_rate": 4.8036377892081565e-05, + "loss": 3.181, + "step": 4546 + }, + { + "epoch": 6.66, + "grad_norm": 0.42195671796798706, + "learning_rate": 4.7996309888016844e-05, + "loss": 3.1872, + "step": 4547 + }, + { + "epoch": 6.66, + "grad_norm": 1.59599769115448, + "learning_rate": 4.795625332441545e-05, + "loss": 3.2153, + "step": 4548 + }, + { + "epoch": 6.66, + "grad_norm": 1.7859076261520386, + "learning_rate": 4.7916208210089516e-05, + "loss": 3.246, + "step": 4549 + }, + { + "epoch": 6.66, + "grad_norm": 1.5909998416900635, + "learning_rate": 4.787617455384874e-05, + "loss": 3.1916, + "step": 4550 + }, + { + "epoch": 6.67, + "grad_norm": 2.1160545349121094, + "learning_rate": 4.783615236450024e-05, + "loss": 3.1881, + "step": 4551 + }, + { + "epoch": 6.67, + "grad_norm": 0.9319918751716614, + "learning_rate": 4.779614165084864e-05, + "loss": 3.2286, + "step": 4552 + }, + { + "epoch": 6.67, + "grad_norm": 2.0976734161376953, + "learning_rate": 4.775614242169606e-05, + "loss": 3.2103, + "step": 4553 + }, + { + "epoch": 6.67, + "grad_norm": 1.695967197418213, + "learning_rate": 4.7716154685841944e-05, + "loss": 3.1638, + "step": 4554 + }, + { + "epoch": 6.67, + "grad_norm": 2.1299004554748535, + "learning_rate": 4.767617845208344e-05, + "loss": 3.21, + "step": 4555 + }, + { + "epoch": 6.67, + "grad_norm": 1.7198915481567383, + "learning_rate": 4.763621372921506e-05, + "loss": 3.2789, + "step": 4556 + }, + { + "epoch": 6.67, + "grad_norm": 0.6392155885696411, + "learning_rate": 4.759626052602866e-05, + "loss": 3.2074, + "step": 4557 + }, + { + "epoch": 6.68, + "grad_norm": 1.7603440284729004, + "learning_rate": 4.7556318851313784e-05, + "loss": 3.1682, + "step": 4558 + }, + { + "epoch": 6.68, + "grad_norm": 0.6075698733329773, + "learning_rate": 4.7516388713857344e-05, + "loss": 3.2082, + "step": 4559 + }, + { + "epoch": 6.68, + "grad_norm": 1.1561111211776733, + "learning_rate": 4.7476470122443576e-05, + "loss": 3.2178, + "step": 4560 + }, + { + "epoch": 6.68, + "grad_norm": 0.6579719185829163, + "learning_rate": 4.743656308585441e-05, + "loss": 3.2283, + "step": 4561 + }, + { + "epoch": 6.68, + "grad_norm": 1.3691036701202393, + "learning_rate": 4.739666761286915e-05, + "loss": 3.2209, + "step": 4562 + }, + { + "epoch": 6.68, + "grad_norm": 0.4905560314655304, + "learning_rate": 4.735678371226441e-05, + "loss": 3.1764, + "step": 4563 + }, + { + "epoch": 6.68, + "eval_loss": 3.0459163188934326, + "eval_runtime": 1166.1122, + "eval_samples_per_second": 14.7, + "eval_steps_per_second": 0.919, + "step": 4563 + }, + { + "epoch": 6.68, + "grad_norm": 0.34124380350112915, + "learning_rate": 4.731691139281449e-05, + "loss": 3.2197, + "step": 4564 + }, + { + "epoch": 6.69, + "grad_norm": 0.7039169073104858, + "learning_rate": 4.7277050663291035e-05, + "loss": 3.2013, + "step": 4565 + }, + { + "epoch": 6.69, + "grad_norm": 1.3341939449310303, + "learning_rate": 4.723720153246303e-05, + "loss": 3.2084, + "step": 4566 + }, + { + "epoch": 6.69, + "grad_norm": 0.4564155638217926, + "learning_rate": 4.719736400909716e-05, + "loss": 3.1865, + "step": 4567 + }, + { + "epoch": 6.69, + "grad_norm": 0.9328960180282593, + "learning_rate": 4.715753810195729e-05, + "loss": 3.2035, + "step": 4568 + }, + { + "epoch": 6.69, + "grad_norm": 0.6796619892120361, + "learning_rate": 4.7117723819804884e-05, + "loss": 3.2024, + "step": 4569 + }, + { + "epoch": 6.69, + "grad_norm": 0.4047408699989319, + "learning_rate": 4.7077921171398895e-05, + "loss": 3.1935, + "step": 4570 + }, + { + "epoch": 6.7, + "grad_norm": 0.6333791017532349, + "learning_rate": 4.7038130165495555e-05, + "loss": 3.1794, + "step": 4571 + }, + { + "epoch": 6.7, + "grad_norm": 0.6958187818527222, + "learning_rate": 4.6998350810848605e-05, + "loss": 3.2401, + "step": 4572 + }, + { + "epoch": 6.7, + "grad_norm": 1.0862935781478882, + "learning_rate": 4.6958583116209344e-05, + "loss": 3.2027, + "step": 4573 + }, + { + "epoch": 6.7, + "grad_norm": 0.5394841432571411, + "learning_rate": 4.691882709032629e-05, + "loss": 3.2092, + "step": 4574 + }, + { + "epoch": 6.7, + "grad_norm": 1.7290090322494507, + "learning_rate": 4.687908274194555e-05, + "loss": 3.2199, + "step": 4575 + }, + { + "epoch": 6.7, + "grad_norm": 1.2097941637039185, + "learning_rate": 4.683935007981061e-05, + "loss": 3.2319, + "step": 4576 + }, + { + "epoch": 6.7, + "grad_norm": 1.8092429637908936, + "learning_rate": 4.6799629112662366e-05, + "loss": 3.1968, + "step": 4577 + }, + { + "epoch": 6.71, + "grad_norm": 1.7616616487503052, + "learning_rate": 4.675991984923921e-05, + "loss": 3.2206, + "step": 4578 + }, + { + "epoch": 6.71, + "grad_norm": 0.6763386726379395, + "learning_rate": 4.672022229827687e-05, + "loss": 3.1857, + "step": 4579 + }, + { + "epoch": 6.71, + "grad_norm": 1.4232481718063354, + "learning_rate": 4.668053646850856e-05, + "loss": 3.184, + "step": 4580 + }, + { + "epoch": 6.71, + "grad_norm": 1.7722804546356201, + "learning_rate": 4.664086236866491e-05, + "loss": 3.225, + "step": 4581 + }, + { + "epoch": 6.71, + "grad_norm": 1.0724995136260986, + "learning_rate": 4.660120000747393e-05, + "loss": 3.1924, + "step": 4582 + }, + { + "epoch": 6.71, + "grad_norm": 0.3242776393890381, + "learning_rate": 4.656154939366109e-05, + "loss": 3.2068, + "step": 4583 + }, + { + "epoch": 6.71, + "grad_norm": 1.4377954006195068, + "learning_rate": 4.652191053594924e-05, + "loss": 3.193, + "step": 4584 + }, + { + "epoch": 6.72, + "grad_norm": 4.462525844573975, + "learning_rate": 4.64822834430587e-05, + "loss": 3.275, + "step": 4585 + }, + { + "epoch": 6.72, + "grad_norm": 1.4812723398208618, + "learning_rate": 4.644266812370714e-05, + "loss": 3.2099, + "step": 4586 + }, + { + "epoch": 6.72, + "grad_norm": 0.6171543002128601, + "learning_rate": 4.640306458660966e-05, + "loss": 3.178, + "step": 4587 + }, + { + "epoch": 6.72, + "grad_norm": 1.2330536842346191, + "learning_rate": 4.636347284047877e-05, + "loss": 3.2013, + "step": 4588 + }, + { + "epoch": 6.72, + "grad_norm": 0.5005393624305725, + "learning_rate": 4.6323892894024414e-05, + "loss": 3.2114, + "step": 4589 + }, + { + "epoch": 6.72, + "grad_norm": 1.613756775856018, + "learning_rate": 4.628432475595393e-05, + "loss": 3.1852, + "step": 4590 + }, + { + "epoch": 6.72, + "grad_norm": 0.8560778498649597, + "learning_rate": 4.6244768434971944e-05, + "loss": 3.2414, + "step": 4591 + }, + { + "epoch": 6.73, + "grad_norm": 1.844510555267334, + "learning_rate": 4.620522393978068e-05, + "loss": 3.1791, + "step": 4592 + }, + { + "epoch": 6.73, + "grad_norm": 1.6803126335144043, + "learning_rate": 4.616569127907968e-05, + "loss": 3.2298, + "step": 4593 + }, + { + "epoch": 6.73, + "grad_norm": 1.2029790878295898, + "learning_rate": 4.6126170461565746e-05, + "loss": 3.2044, + "step": 4594 + }, + { + "epoch": 6.73, + "grad_norm": 0.8809365630149841, + "learning_rate": 4.608666149593328e-05, + "loss": 3.1935, + "step": 4595 + }, + { + "epoch": 6.73, + "grad_norm": 0.6382874846458435, + "learning_rate": 4.604716439087403e-05, + "loss": 3.1965, + "step": 4596 + }, + { + "epoch": 6.73, + "grad_norm": 1.0096112489700317, + "learning_rate": 4.600767915507695e-05, + "loss": 3.1924, + "step": 4597 + }, + { + "epoch": 6.74, + "grad_norm": 0.8474308252334595, + "learning_rate": 4.596820579722867e-05, + "loss": 3.1994, + "step": 4598 + }, + { + "epoch": 6.74, + "grad_norm": 1.0024635791778564, + "learning_rate": 4.592874432601304e-05, + "loss": 3.1834, + "step": 4599 + }, + { + "epoch": 6.74, + "grad_norm": 0.9615989923477173, + "learning_rate": 4.588929475011125e-05, + "loss": 3.1855, + "step": 4600 + }, + { + "epoch": 6.74, + "grad_norm": 0.4216245114803314, + "learning_rate": 4.584985707820198e-05, + "loss": 3.2381, + "step": 4601 + }, + { + "epoch": 6.74, + "grad_norm": 0.5493209362030029, + "learning_rate": 4.581043131896126e-05, + "loss": 3.1567, + "step": 4602 + }, + { + "epoch": 6.74, + "grad_norm": 0.6222708225250244, + "learning_rate": 4.5771017481062475e-05, + "loss": 3.2149, + "step": 4603 + }, + { + "epoch": 6.74, + "grad_norm": 0.5318418741226196, + "learning_rate": 4.573161557317642e-05, + "loss": 3.2158, + "step": 4604 + }, + { + "epoch": 6.75, + "grad_norm": 0.7692480087280273, + "learning_rate": 4.5692225603971264e-05, + "loss": 3.2096, + "step": 4605 + }, + { + "epoch": 6.75, + "grad_norm": 0.5248206257820129, + "learning_rate": 4.56528475821125e-05, + "loss": 3.1933, + "step": 4606 + }, + { + "epoch": 6.75, + "grad_norm": 1.0218887329101562, + "learning_rate": 4.561348151626306e-05, + "loss": 3.1819, + "step": 4607 + }, + { + "epoch": 6.75, + "grad_norm": 2.8102357387542725, + "learning_rate": 4.55741274150832e-05, + "loss": 3.2286, + "step": 4608 + }, + { + "epoch": 6.75, + "grad_norm": 0.5493364334106445, + "learning_rate": 4.5534785287230575e-05, + "loss": 3.1981, + "step": 4609 + }, + { + "epoch": 6.75, + "grad_norm": 0.438005656003952, + "learning_rate": 4.5495455141360165e-05, + "loss": 3.1831, + "step": 4610 + }, + { + "epoch": 6.75, + "grad_norm": 1.3687691688537598, + "learning_rate": 4.5456136986124355e-05, + "loss": 3.1993, + "step": 4611 + }, + { + "epoch": 6.76, + "grad_norm": 0.7376412749290466, + "learning_rate": 4.5416830830172876e-05, + "loss": 3.2093, + "step": 4612 + }, + { + "epoch": 6.76, + "grad_norm": 0.710468590259552, + "learning_rate": 4.53775366821528e-05, + "loss": 3.1834, + "step": 4613 + }, + { + "epoch": 6.76, + "grad_norm": 0.6461208462715149, + "learning_rate": 4.53382545507086e-05, + "loss": 3.2279, + "step": 4614 + }, + { + "epoch": 6.76, + "grad_norm": 0.4084911644458771, + "learning_rate": 4.5298984444482075e-05, + "loss": 3.1921, + "step": 4615 + }, + { + "epoch": 6.76, + "grad_norm": 0.5108605027198792, + "learning_rate": 4.5259726372112376e-05, + "loss": 3.2015, + "step": 4616 + }, + { + "epoch": 6.76, + "grad_norm": 0.36462754011154175, + "learning_rate": 4.522048034223603e-05, + "loss": 3.2099, + "step": 4617 + }, + { + "epoch": 6.76, + "grad_norm": 0.5964351892471313, + "learning_rate": 4.518124636348688e-05, + "loss": 3.223, + "step": 4618 + }, + { + "epoch": 6.77, + "grad_norm": 0.532785177230835, + "learning_rate": 4.514202444449615e-05, + "loss": 3.1739, + "step": 4619 + }, + { + "epoch": 6.77, + "grad_norm": 0.7253513336181641, + "learning_rate": 4.510281459389242e-05, + "loss": 3.202, + "step": 4620 + }, + { + "epoch": 6.77, + "grad_norm": 1.3560603857040405, + "learning_rate": 4.506361682030156e-05, + "loss": 3.2175, + "step": 4621 + }, + { + "epoch": 6.77, + "grad_norm": 0.6363286375999451, + "learning_rate": 4.502443113234688e-05, + "loss": 3.2485, + "step": 4622 + }, + { + "epoch": 6.77, + "grad_norm": 1.2093783617019653, + "learning_rate": 4.498525753864884e-05, + "loss": 3.2012, + "step": 4623 + }, + { + "epoch": 6.77, + "grad_norm": 1.1001776456832886, + "learning_rate": 4.4946096047825515e-05, + "loss": 3.1938, + "step": 4624 + }, + { + "epoch": 6.78, + "grad_norm": 0.4927830398082733, + "learning_rate": 4.490694666849208e-05, + "loss": 3.234, + "step": 4625 + }, + { + "epoch": 6.78, + "grad_norm": 1.6655173301696777, + "learning_rate": 4.486780940926112e-05, + "loss": 3.1661, + "step": 4626 + }, + { + "epoch": 6.78, + "grad_norm": 1.3249220848083496, + "learning_rate": 4.482868427874268e-05, + "loss": 3.2283, + "step": 4627 + }, + { + "epoch": 6.78, + "grad_norm": 2.2860875129699707, + "learning_rate": 4.478957128554392e-05, + "loss": 3.2115, + "step": 4628 + }, + { + "epoch": 6.78, + "grad_norm": 2.1470510959625244, + "learning_rate": 4.4750470438269446e-05, + "loss": 3.2151, + "step": 4629 + }, + { + "epoch": 6.78, + "grad_norm": 1.3575111627578735, + "learning_rate": 4.4711381745521275e-05, + "loss": 3.1938, + "step": 4630 + }, + { + "epoch": 6.78, + "grad_norm": 1.7453752756118774, + "learning_rate": 4.467230521589856e-05, + "loss": 3.1969, + "step": 4631 + }, + { + "epoch": 6.79, + "grad_norm": 1.7523752450942993, + "learning_rate": 4.463324085799786e-05, + "loss": 3.2266, + "step": 4632 + }, + { + "epoch": 6.79, + "grad_norm": 1.433594822883606, + "learning_rate": 4.4594188680413206e-05, + "loss": 3.1962, + "step": 4633 + }, + { + "epoch": 6.79, + "grad_norm": 1.0330843925476074, + "learning_rate": 4.4555148691735685e-05, + "loss": 3.2325, + "step": 4634 + }, + { + "epoch": 6.79, + "grad_norm": 0.9396421313285828, + "learning_rate": 4.451612090055388e-05, + "loss": 3.2274, + "step": 4635 + }, + { + "epoch": 6.79, + "grad_norm": 1.1527962684631348, + "learning_rate": 4.447710531545364e-05, + "loss": 3.2075, + "step": 4636 + }, + { + "epoch": 6.79, + "grad_norm": 0.8900932669639587, + "learning_rate": 4.443810194501813e-05, + "loss": 3.1952, + "step": 4637 + }, + { + "epoch": 6.79, + "grad_norm": 1.3143221139907837, + "learning_rate": 4.4399110797827835e-05, + "loss": 3.2086, + "step": 4638 + }, + { + "epoch": 6.8, + "grad_norm": 1.1299476623535156, + "learning_rate": 4.4360131882460555e-05, + "loss": 3.2, + "step": 4639 + }, + { + "epoch": 6.8, + "grad_norm": 0.8985263109207153, + "learning_rate": 4.432116520749137e-05, + "loss": 3.15, + "step": 4640 + }, + { + "epoch": 6.8, + "grad_norm": 1.0726135969161987, + "learning_rate": 4.4282210781492684e-05, + "loss": 3.2016, + "step": 4641 + }, + { + "epoch": 6.8, + "grad_norm": 0.556115984916687, + "learning_rate": 4.424326861303424e-05, + "loss": 3.2035, + "step": 4642 + }, + { + "epoch": 6.8, + "grad_norm": 1.1372523307800293, + "learning_rate": 4.420433871068302e-05, + "loss": 3.1998, + "step": 4643 + }, + { + "epoch": 6.8, + "grad_norm": 0.9902729988098145, + "learning_rate": 4.416542108300338e-05, + "loss": 3.1661, + "step": 4644 + }, + { + "epoch": 6.8, + "grad_norm": 0.6972740292549133, + "learning_rate": 4.412651573855689e-05, + "loss": 3.191, + "step": 4645 + }, + { + "epoch": 6.81, + "grad_norm": 0.7196264863014221, + "learning_rate": 4.408762268590251e-05, + "loss": 3.1772, + "step": 4646 + }, + { + "epoch": 6.81, + "grad_norm": 0.6130596995353699, + "learning_rate": 4.404874193359646e-05, + "loss": 3.1859, + "step": 4647 + }, + { + "epoch": 6.81, + "grad_norm": 0.8541386723518372, + "learning_rate": 4.4009873490192135e-05, + "loss": 3.1891, + "step": 4648 + }, + { + "epoch": 6.81, + "grad_norm": 0.4068596065044403, + "learning_rate": 4.397101736424046e-05, + "loss": 3.2386, + "step": 4649 + }, + { + "epoch": 6.81, + "grad_norm": 0.35903048515319824, + "learning_rate": 4.3932173564289506e-05, + "loss": 3.2046, + "step": 4650 + }, + { + "epoch": 6.81, + "grad_norm": 1.0451300144195557, + "learning_rate": 4.389334209888455e-05, + "loss": 3.2677, + "step": 4651 + }, + { + "epoch": 6.82, + "grad_norm": 0.3824341595172882, + "learning_rate": 4.385452297656836e-05, + "loss": 3.1955, + "step": 4652 + }, + { + "epoch": 6.82, + "grad_norm": 0.9694157838821411, + "learning_rate": 4.381571620588088e-05, + "loss": 3.2113, + "step": 4653 + }, + { + "epoch": 6.82, + "grad_norm": 0.6231606602668762, + "learning_rate": 4.3776921795359216e-05, + "loss": 3.1968, + "step": 4654 + }, + { + "epoch": 6.82, + "grad_norm": 0.39795351028442383, + "learning_rate": 4.373813975353802e-05, + "loss": 3.1859, + "step": 4655 + }, + { + "epoch": 6.82, + "grad_norm": 0.3494223654270172, + "learning_rate": 4.3699370088949066e-05, + "loss": 3.2194, + "step": 4656 + }, + { + "epoch": 6.82, + "grad_norm": 0.7372296452522278, + "learning_rate": 4.366061281012129e-05, + "loss": 3.2062, + "step": 4657 + }, + { + "epoch": 6.82, + "grad_norm": 1.144978642463684, + "learning_rate": 4.362186792558122e-05, + "loss": 3.1948, + "step": 4658 + }, + { + "epoch": 6.83, + "grad_norm": 0.5863081812858582, + "learning_rate": 4.358313544385233e-05, + "loss": 3.1734, + "step": 4659 + }, + { + "epoch": 6.83, + "grad_norm": 1.334794282913208, + "learning_rate": 4.354441537345553e-05, + "loss": 3.187, + "step": 4660 + }, + { + "epoch": 6.83, + "grad_norm": 0.4816846251487732, + "learning_rate": 4.350570772290906e-05, + "loss": 3.1916, + "step": 4661 + }, + { + "epoch": 6.83, + "grad_norm": 0.8258687853813171, + "learning_rate": 4.3467012500728245e-05, + "loss": 3.2171, + "step": 4662 + }, + { + "epoch": 6.83, + "grad_norm": 0.9091712236404419, + "learning_rate": 4.342832971542578e-05, + "loss": 3.1695, + "step": 4663 + }, + { + "epoch": 6.83, + "grad_norm": 1.3142671585083008, + "learning_rate": 4.3389659375511725e-05, + "loss": 3.1681, + "step": 4664 + }, + { + "epoch": 6.83, + "grad_norm": 2.6121108531951904, + "learning_rate": 4.335100148949318e-05, + "loss": 3.2161, + "step": 4665 + }, + { + "epoch": 6.84, + "grad_norm": 0.6220512986183167, + "learning_rate": 4.3312356065874635e-05, + "loss": 3.1868, + "step": 4666 + }, + { + "epoch": 6.84, + "grad_norm": 1.4593570232391357, + "learning_rate": 4.327372311315792e-05, + "loss": 3.1968, + "step": 4667 + }, + { + "epoch": 6.84, + "grad_norm": 1.4904062747955322, + "learning_rate": 4.323510263984192e-05, + "loss": 3.2013, + "step": 4668 + }, + { + "epoch": 6.84, + "grad_norm": 0.8087517023086548, + "learning_rate": 4.3196494654422916e-05, + "loss": 3.2106, + "step": 4669 + }, + { + "epoch": 6.84, + "grad_norm": 0.5217124223709106, + "learning_rate": 4.315789916539441e-05, + "loss": 3.2188, + "step": 4670 + }, + { + "epoch": 6.84, + "grad_norm": 0.7983100414276123, + "learning_rate": 4.311931618124715e-05, + "loss": 3.217, + "step": 4671 + }, + { + "epoch": 6.84, + "grad_norm": 0.49825963377952576, + "learning_rate": 4.308074571046913e-05, + "loss": 3.2089, + "step": 4672 + }, + { + "epoch": 6.85, + "grad_norm": 0.9408275485038757, + "learning_rate": 4.30421877615456e-05, + "loss": 3.1924, + "step": 4673 + }, + { + "epoch": 6.85, + "grad_norm": 0.6962942481040955, + "learning_rate": 4.300364234295905e-05, + "loss": 3.2022, + "step": 4674 + }, + { + "epoch": 6.85, + "grad_norm": 0.5625227093696594, + "learning_rate": 4.296510946318921e-05, + "loss": 3.2056, + "step": 4675 + }, + { + "epoch": 6.85, + "grad_norm": 0.6208386421203613, + "learning_rate": 4.292658913071306e-05, + "loss": 3.1973, + "step": 4676 + }, + { + "epoch": 6.85, + "grad_norm": 0.5512813925743103, + "learning_rate": 4.2888081354004805e-05, + "loss": 3.2122, + "step": 4677 + }, + { + "epoch": 6.85, + "grad_norm": 1.9432053565979004, + "learning_rate": 4.2849586141535945e-05, + "loss": 3.2239, + "step": 4678 + }, + { + "epoch": 6.86, + "grad_norm": 4.5366291999816895, + "learning_rate": 4.281110350177506e-05, + "loss": 3.1618, + "step": 4679 + }, + { + "epoch": 6.86, + "grad_norm": 0.9245926737785339, + "learning_rate": 4.277263344318817e-05, + "loss": 3.2309, + "step": 4680 + }, + { + "epoch": 6.86, + "grad_norm": 0.3581990897655487, + "learning_rate": 4.2734175974238444e-05, + "loss": 3.1872, + "step": 4681 + }, + { + "epoch": 6.86, + "grad_norm": 0.9905257821083069, + "learning_rate": 4.269573110338615e-05, + "loss": 3.2077, + "step": 4682 + }, + { + "epoch": 6.86, + "grad_norm": 1.2098913192749023, + "learning_rate": 4.265729883908902e-05, + "loss": 3.2249, + "step": 4683 + }, + { + "epoch": 6.86, + "grad_norm": 0.711704671382904, + "learning_rate": 4.2618879189801885e-05, + "loss": 3.1625, + "step": 4684 + }, + { + "epoch": 6.86, + "grad_norm": 2.870743989944458, + "learning_rate": 4.25804721639767e-05, + "loss": 3.1953, + "step": 4685 + }, + { + "epoch": 6.87, + "grad_norm": 1.7720706462860107, + "learning_rate": 4.254207777006287e-05, + "loss": 3.1671, + "step": 4686 + }, + { + "epoch": 6.87, + "grad_norm": 0.8603296875953674, + "learning_rate": 4.25036960165069e-05, + "loss": 3.166, + "step": 4687 + }, + { + "epoch": 6.87, + "grad_norm": 2.4739768505096436, + "learning_rate": 4.246532691175241e-05, + "loss": 3.1946, + "step": 4688 + }, + { + "epoch": 6.87, + "grad_norm": 2.0245025157928467, + "learning_rate": 4.242697046424046e-05, + "loss": 3.2304, + "step": 4689 + }, + { + "epoch": 6.87, + "grad_norm": 0.7027145028114319, + "learning_rate": 4.2388626682409194e-05, + "loss": 3.2526, + "step": 4690 + }, + { + "epoch": 6.87, + "grad_norm": 2.3779094219207764, + "learning_rate": 4.23502955746939e-05, + "loss": 3.2621, + "step": 4691 + }, + { + "epoch": 6.87, + "grad_norm": 1.9404739141464233, + "learning_rate": 4.23119771495273e-05, + "loss": 3.1891, + "step": 4692 + }, + { + "epoch": 6.88, + "grad_norm": 0.6578683257102966, + "learning_rate": 4.227367141533909e-05, + "loss": 3.2023, + "step": 4693 + }, + { + "epoch": 6.88, + "grad_norm": 5.048638820648193, + "learning_rate": 4.2235378380556255e-05, + "loss": 3.136, + "step": 4694 + }, + { + "epoch": 6.88, + "grad_norm": 2.4565985202789307, + "learning_rate": 4.2197098053603136e-05, + "loss": 3.1868, + "step": 4695 + }, + { + "epoch": 6.88, + "grad_norm": 0.770892858505249, + "learning_rate": 4.215883044290103e-05, + "loss": 3.1455, + "step": 4696 + }, + { + "epoch": 6.88, + "grad_norm": 2.0807790756225586, + "learning_rate": 4.2120575556868594e-05, + "loss": 3.1894, + "step": 4697 + }, + { + "epoch": 6.88, + "grad_norm": 2.7063095569610596, + "learning_rate": 4.208233340392164e-05, + "loss": 3.1982, + "step": 4698 + }, + { + "epoch": 6.89, + "grad_norm": 1.1912204027175903, + "learning_rate": 4.2044103992473195e-05, + "loss": 3.1818, + "step": 4699 + }, + { + "epoch": 6.89, + "grad_norm": 2.9936110973358154, + "learning_rate": 4.2005887330933457e-05, + "loss": 3.2109, + "step": 4700 + }, + { + "epoch": 6.89, + "grad_norm": 3.9401204586029053, + "learning_rate": 4.196768342770986e-05, + "loss": 3.2246, + "step": 4701 + }, + { + "epoch": 6.89, + "grad_norm": 2.341156482696533, + "learning_rate": 4.1929492291206983e-05, + "loss": 3.1819, + "step": 4702 + }, + { + "epoch": 6.89, + "grad_norm": 0.7375439405441284, + "learning_rate": 4.1891313929826635e-05, + "loss": 3.2047, + "step": 4703 + }, + { + "epoch": 6.89, + "grad_norm": 1.9721698760986328, + "learning_rate": 4.185314835196781e-05, + "loss": 3.2131, + "step": 4704 + }, + { + "epoch": 6.89, + "grad_norm": 0.8218457102775574, + "learning_rate": 4.181499556602665e-05, + "loss": 3.1891, + "step": 4705 + }, + { + "epoch": 6.9, + "grad_norm": 1.2695006132125854, + "learning_rate": 4.177685558039652e-05, + "loss": 3.1774, + "step": 4706 + }, + { + "epoch": 6.9, + "grad_norm": 1.4132906198501587, + "learning_rate": 4.1738728403467974e-05, + "loss": 3.1856, + "step": 4707 + }, + { + "epoch": 6.9, + "grad_norm": 0.5927780866622925, + "learning_rate": 4.170061404362874e-05, + "loss": 3.2209, + "step": 4708 + }, + { + "epoch": 6.9, + "grad_norm": 1.482879877090454, + "learning_rate": 4.166251250926371e-05, + "loss": 3.1621, + "step": 4709 + }, + { + "epoch": 6.9, + "grad_norm": 0.4204716384410858, + "learning_rate": 4.1624423808754956e-05, + "loss": 3.171, + "step": 4710 + }, + { + "epoch": 6.9, + "grad_norm": 1.5936245918273926, + "learning_rate": 4.158634795048175e-05, + "loss": 3.1746, + "step": 4711 + }, + { + "epoch": 6.9, + "grad_norm": 1.1268365383148193, + "learning_rate": 4.154828494282056e-05, + "loss": 3.1932, + "step": 4712 + }, + { + "epoch": 6.91, + "grad_norm": 1.2134582996368408, + "learning_rate": 4.151023479414489e-05, + "loss": 3.2064, + "step": 4713 + }, + { + "epoch": 6.91, + "grad_norm": 1.2098625898361206, + "learning_rate": 4.147219751282562e-05, + "loss": 3.2025, + "step": 4714 + }, + { + "epoch": 6.91, + "grad_norm": 4.699478626251221, + "learning_rate": 4.143417310723069e-05, + "loss": 3.1962, + "step": 4715 + }, + { + "epoch": 6.91, + "grad_norm": 1.845900297164917, + "learning_rate": 4.139616158572512e-05, + "loss": 3.1635, + "step": 4716 + }, + { + "epoch": 6.91, + "grad_norm": 1.1032308340072632, + "learning_rate": 4.1358162956671295e-05, + "loss": 3.1781, + "step": 4717 + }, + { + "epoch": 6.91, + "grad_norm": 1.5497393608093262, + "learning_rate": 4.1320177228428646e-05, + "loss": 3.1825, + "step": 4718 + }, + { + "epoch": 6.91, + "grad_norm": 2.2472078800201416, + "learning_rate": 4.128220440935372e-05, + "loss": 3.1866, + "step": 4719 + }, + { + "epoch": 6.92, + "grad_norm": 92.44196319580078, + "learning_rate": 4.124424450780029e-05, + "loss": 3.2326, + "step": 4720 + }, + { + "epoch": 6.92, + "grad_norm": 1.094028115272522, + "learning_rate": 4.120629753211936e-05, + "loss": 3.2104, + "step": 4721 + }, + { + "epoch": 6.92, + "grad_norm": 0.9184134602546692, + "learning_rate": 4.116836349065892e-05, + "loss": 3.2233, + "step": 4722 + }, + { + "epoch": 6.92, + "grad_norm": 1.4922984838485718, + "learning_rate": 4.1130442391764214e-05, + "loss": 3.1623, + "step": 4723 + }, + { + "epoch": 6.92, + "grad_norm": 0.9635801911354065, + "learning_rate": 4.109253424377772e-05, + "loss": 3.1788, + "step": 4724 + }, + { + "epoch": 6.92, + "grad_norm": 1.1866681575775146, + "learning_rate": 4.105463905503888e-05, + "loss": 3.2056, + "step": 4725 + }, + { + "epoch": 6.93, + "grad_norm": 0.8471019268035889, + "learning_rate": 4.101675683388442e-05, + "loss": 3.1978, + "step": 4726 + }, + { + "epoch": 6.93, + "grad_norm": 0.5931624174118042, + "learning_rate": 4.0978887588648155e-05, + "loss": 3.2222, + "step": 4727 + }, + { + "epoch": 6.93, + "grad_norm": 0.9211606383323669, + "learning_rate": 4.094103132766109e-05, + "loss": 3.1993, + "step": 4728 + }, + { + "epoch": 6.93, + "grad_norm": 0.7005136013031006, + "learning_rate": 4.090318805925133e-05, + "loss": 3.2065, + "step": 4729 + }, + { + "epoch": 6.93, + "grad_norm": 0.9550567269325256, + "learning_rate": 4.086535779174415e-05, + "loss": 3.1617, + "step": 4730 + }, + { + "epoch": 6.93, + "grad_norm": 0.9668344259262085, + "learning_rate": 4.082754053346195e-05, + "loss": 3.2148, + "step": 4731 + }, + { + "epoch": 6.93, + "grad_norm": 0.6807712912559509, + "learning_rate": 4.078973629272428e-05, + "loss": 3.1848, + "step": 4732 + }, + { + "epoch": 6.93, + "eval_loss": 3.0342421531677246, + "eval_runtime": 1171.9875, + "eval_samples_per_second": 14.626, + "eval_steps_per_second": 0.915, + "step": 4732 + }, + { + "epoch": 6.94, + "grad_norm": 1.207279920578003, + "learning_rate": 4.075194507784782e-05, + "loss": 3.1595, + "step": 4733 + }, + { + "epoch": 6.94, + "grad_norm": 1.3139106035232544, + "learning_rate": 4.071416689714637e-05, + "loss": 3.1985, + "step": 4734 + }, + { + "epoch": 6.94, + "grad_norm": 1.053685188293457, + "learning_rate": 4.0676401758930894e-05, + "loss": 3.2048, + "step": 4735 + }, + { + "epoch": 6.94, + "grad_norm": 1.043835997581482, + "learning_rate": 4.063864967150945e-05, + "loss": 3.1763, + "step": 4736 + }, + { + "epoch": 6.94, + "grad_norm": 0.4672507047653198, + "learning_rate": 4.060091064318727e-05, + "loss": 3.1722, + "step": 4737 + }, + { + "epoch": 6.94, + "grad_norm": 1.6286187171936035, + "learning_rate": 4.0563184682266676e-05, + "loss": 3.2824, + "step": 4738 + }, + { + "epoch": 6.94, + "grad_norm": 0.8605678081512451, + "learning_rate": 4.052547179704711e-05, + "loss": 3.198, + "step": 4739 + }, + { + "epoch": 6.95, + "grad_norm": 1.1343073844909668, + "learning_rate": 4.048777199582518e-05, + "loss": 3.1726, + "step": 4740 + }, + { + "epoch": 6.95, + "grad_norm": 0.6983288526535034, + "learning_rate": 4.0450085286894565e-05, + "loss": 3.2139, + "step": 4741 + }, + { + "epoch": 6.95, + "grad_norm": 0.8509522080421448, + "learning_rate": 4.041241167854612e-05, + "loss": 3.176, + "step": 4742 + }, + { + "epoch": 6.95, + "grad_norm": 0.8853251934051514, + "learning_rate": 4.037475117906775e-05, + "loss": 3.1622, + "step": 4743 + }, + { + "epoch": 6.95, + "grad_norm": 1.2624071836471558, + "learning_rate": 4.033710379674458e-05, + "loss": 3.1633, + "step": 4744 + }, + { + "epoch": 6.95, + "grad_norm": 1.0235035419464111, + "learning_rate": 4.029946953985865e-05, + "loss": 3.1794, + "step": 4745 + }, + { + "epoch": 6.95, + "grad_norm": 1.0335735082626343, + "learning_rate": 4.0261848416689394e-05, + "loss": 3.2398, + "step": 4746 + }, + { + "epoch": 6.96, + "grad_norm": 1.2208091020584106, + "learning_rate": 4.022424043551312e-05, + "loss": 3.2424, + "step": 4747 + }, + { + "epoch": 6.96, + "grad_norm": 0.8216298818588257, + "learning_rate": 4.0186645604603304e-05, + "loss": 3.1948, + "step": 4748 + }, + { + "epoch": 6.96, + "grad_norm": 1.2754603624343872, + "learning_rate": 4.014906393223067e-05, + "loss": 3.1608, + "step": 4749 + }, + { + "epoch": 6.96, + "grad_norm": 1.283899188041687, + "learning_rate": 4.011149542666284e-05, + "loss": 3.2403, + "step": 4750 + }, + { + "epoch": 6.96, + "grad_norm": 0.569480836391449, + "learning_rate": 4.007394009616462e-05, + "loss": 3.2395, + "step": 4751 + }, + { + "epoch": 6.96, + "grad_norm": 0.7259752750396729, + "learning_rate": 4.003639794899804e-05, + "loss": 3.1733, + "step": 4752 + }, + { + "epoch": 6.97, + "grad_norm": 0.3225840628147125, + "learning_rate": 3.9998868993422026e-05, + "loss": 3.2326, + "step": 4753 + }, + { + "epoch": 6.97, + "grad_norm": 1.8098729848861694, + "learning_rate": 3.996135323769269e-05, + "loss": 3.2116, + "step": 4754 + }, + { + "epoch": 6.97, + "grad_norm": 0.893907904624939, + "learning_rate": 3.9923850690063345e-05, + "loss": 3.1768, + "step": 4755 + }, + { + "epoch": 6.97, + "grad_norm": 0.552627444267273, + "learning_rate": 3.988636135878421e-05, + "loss": 3.2011, + "step": 4756 + }, + { + "epoch": 6.97, + "grad_norm": 0.809914231300354, + "learning_rate": 3.984888525210266e-05, + "loss": 3.1777, + "step": 4757 + }, + { + "epoch": 6.97, + "grad_norm": 0.4486781060695648, + "learning_rate": 3.981142237826332e-05, + "loss": 3.193, + "step": 4758 + }, + { + "epoch": 6.97, + "grad_norm": 0.7444935441017151, + "learning_rate": 3.977397274550766e-05, + "loss": 3.1806, + "step": 4759 + }, + { + "epoch": 6.98, + "grad_norm": 0.49755245447158813, + "learning_rate": 3.973653636207437e-05, + "loss": 3.1676, + "step": 4760 + }, + { + "epoch": 6.98, + "grad_norm": 1.1105209589004517, + "learning_rate": 3.9699113236199226e-05, + "loss": 3.1885, + "step": 4761 + }, + { + "epoch": 6.98, + "grad_norm": 0.8656307458877563, + "learning_rate": 3.9661703376115034e-05, + "loss": 3.2385, + "step": 4762 + }, + { + "epoch": 6.98, + "grad_norm": 1.250411033630371, + "learning_rate": 3.962430679005175e-05, + "loss": 3.1652, + "step": 4763 + }, + { + "epoch": 6.98, + "grad_norm": 1.7875831127166748, + "learning_rate": 3.958692348623634e-05, + "loss": 3.1795, + "step": 4764 + }, + { + "epoch": 6.98, + "grad_norm": 0.575053870677948, + "learning_rate": 3.954955347289289e-05, + "loss": 3.2002, + "step": 4765 + }, + { + "epoch": 6.98, + "grad_norm": 1.1401422023773193, + "learning_rate": 3.9512196758242556e-05, + "loss": 3.1345, + "step": 4766 + }, + { + "epoch": 6.99, + "grad_norm": 0.6520401835441589, + "learning_rate": 3.947485335050357e-05, + "loss": 3.2271, + "step": 4767 + }, + { + "epoch": 6.99, + "grad_norm": 0.7057487964630127, + "learning_rate": 3.9437523257891215e-05, + "loss": 3.2535, + "step": 4768 + }, + { + "epoch": 6.99, + "grad_norm": 1.187171220779419, + "learning_rate": 3.940020648861792e-05, + "loss": 3.153, + "step": 4769 + }, + { + "epoch": 6.99, + "grad_norm": 0.44475510716438293, + "learning_rate": 3.9362903050892996e-05, + "loss": 3.1748, + "step": 4770 + }, + { + "epoch": 6.99, + "grad_norm": 2.0689454078674316, + "learning_rate": 3.9325612952923065e-05, + "loss": 3.17, + "step": 4771 + }, + { + "epoch": 6.99, + "grad_norm": 1.3041399717330933, + "learning_rate": 3.92883362029117e-05, + "loss": 3.185, + "step": 4772 + }, + { + "epoch": 6.99, + "grad_norm": 0.5959802865982056, + "learning_rate": 3.925107280905943e-05, + "loss": 3.1533, + "step": 4773 + }, + { + "epoch": 7.0, + "grad_norm": 1.501555323600769, + "learning_rate": 3.9213822779564056e-05, + "loss": 3.1333, + "step": 4774 + }, + { + "epoch": 7.0, + "grad_norm": 0.4824700951576233, + "learning_rate": 3.917658612262033e-05, + "loss": 3.173, + "step": 4775 + }, + { + "epoch": 7.0, + "grad_norm": 0.9984436631202698, + "learning_rate": 3.913936284641997e-05, + "loss": 3.1509, + "step": 4776 + }, + { + "epoch": 7.0, + "grad_norm": 0.47394973039627075, + "learning_rate": 3.910215295915195e-05, + "loss": 3.1744, + "step": 4777 + }, + { + "epoch": 7.0, + "grad_norm": 1.570947527885437, + "learning_rate": 3.906495646900219e-05, + "loss": 3.1831, + "step": 4778 + }, + { + "epoch": 7.0, + "grad_norm": 0.9516702890396118, + "learning_rate": 3.902777338415358e-05, + "loss": 3.1569, + "step": 4779 + }, + { + "epoch": 7.01, + "grad_norm": 2.1789183616638184, + "learning_rate": 3.899060371278627e-05, + "loss": 3.1952, + "step": 4780 + }, + { + "epoch": 7.01, + "grad_norm": 3.014054536819458, + "learning_rate": 3.895344746307723e-05, + "loss": 3.2197, + "step": 4781 + }, + { + "epoch": 7.01, + "grad_norm": 1.0081833600997925, + "learning_rate": 3.8916304643200604e-05, + "loss": 3.1867, + "step": 4782 + }, + { + "epoch": 7.01, + "grad_norm": 1.2014774084091187, + "learning_rate": 3.887917526132767e-05, + "loss": 3.2003, + "step": 4783 + }, + { + "epoch": 7.01, + "grad_norm": 2.214982748031616, + "learning_rate": 3.8842059325626536e-05, + "loss": 3.1835, + "step": 4784 + }, + { + "epoch": 7.01, + "grad_norm": 1.1680030822753906, + "learning_rate": 3.880495684426243e-05, + "loss": 3.2276, + "step": 4785 + }, + { + "epoch": 7.01, + "grad_norm": 1.6248522996902466, + "learning_rate": 3.8767867825397796e-05, + "loss": 3.1552, + "step": 4786 + }, + { + "epoch": 7.0, + "grad_norm": 1.5934488773345947, + "learning_rate": 3.873079227719184e-05, + "loss": 3.1989, + "step": 4787 + }, + { + "epoch": 7.0, + "grad_norm": 1.3276259899139404, + "learning_rate": 3.869373020780094e-05, + "loss": 3.2109, + "step": 4788 + }, + { + "epoch": 7.0, + "grad_norm": 1.9052053689956665, + "learning_rate": 3.865668162537861e-05, + "loss": 3.1952, + "step": 4789 + }, + { + "epoch": 7.0, + "grad_norm": 1.7343652248382568, + "learning_rate": 3.8619646538075184e-05, + "loss": 3.2074, + "step": 4790 + }, + { + "epoch": 7.01, + "grad_norm": 1.1672378778457642, + "learning_rate": 3.858262495403816e-05, + "loss": 3.1704, + "step": 4791 + }, + { + "epoch": 7.01, + "grad_norm": 1.6275311708450317, + "learning_rate": 3.854561688141205e-05, + "loss": 3.1723, + "step": 4792 + }, + { + "epoch": 7.01, + "grad_norm": 2.027491569519043, + "learning_rate": 3.8508622328338376e-05, + "loss": 3.2521, + "step": 4793 + }, + { + "epoch": 7.01, + "grad_norm": 1.3763444423675537, + "learning_rate": 3.8471641302955684e-05, + "loss": 3.1991, + "step": 4794 + }, + { + "epoch": 7.01, + "grad_norm": 1.7834948301315308, + "learning_rate": 3.843467381339957e-05, + "loss": 3.1685, + "step": 4795 + }, + { + "epoch": 7.01, + "grad_norm": 1.548448085784912, + "learning_rate": 3.839771986780263e-05, + "loss": 3.1715, + "step": 4796 + }, + { + "epoch": 7.02, + "grad_norm": 2.0860743522644043, + "learning_rate": 3.836077947429446e-05, + "loss": 3.2339, + "step": 4797 + }, + { + "epoch": 7.02, + "grad_norm": 2.655271053314209, + "learning_rate": 3.8323852641001733e-05, + "loss": 3.1881, + "step": 4798 + }, + { + "epoch": 7.02, + "grad_norm": 1.268930435180664, + "learning_rate": 3.828693937604809e-05, + "loss": 3.1635, + "step": 4799 + }, + { + "epoch": 7.02, + "grad_norm": 1.7928105592727661, + "learning_rate": 3.8250039687554195e-05, + "loss": 3.1764, + "step": 4800 + }, + { + "epoch": 7.02, + "grad_norm": 2.381535768508911, + "learning_rate": 3.821315358363775e-05, + "loss": 3.1892, + "step": 4801 + }, + { + "epoch": 7.02, + "grad_norm": 1.9187071323394775, + "learning_rate": 3.817628107241343e-05, + "loss": 3.1783, + "step": 4802 + }, + { + "epoch": 7.02, + "grad_norm": 0.999586284160614, + "learning_rate": 3.813942216199299e-05, + "loss": 3.1682, + "step": 4803 + }, + { + "epoch": 7.03, + "grad_norm": 1.993181824684143, + "learning_rate": 3.810257686048503e-05, + "loss": 3.2129, + "step": 4804 + }, + { + "epoch": 7.03, + "grad_norm": 1.1408189535140991, + "learning_rate": 3.806574517599537e-05, + "loss": 3.2541, + "step": 4805 + }, + { + "epoch": 7.03, + "grad_norm": 1.8819730281829834, + "learning_rate": 3.802892711662676e-05, + "loss": 3.1628, + "step": 4806 + }, + { + "epoch": 7.03, + "grad_norm": 2.9541916847229004, + "learning_rate": 3.799212269047878e-05, + "loss": 3.224, + "step": 4807 + }, + { + "epoch": 7.03, + "grad_norm": 1.8541984558105469, + "learning_rate": 3.795533190564831e-05, + "loss": 3.1792, + "step": 4808 + }, + { + "epoch": 7.03, + "grad_norm": 0.8892889022827148, + "learning_rate": 3.791855477022903e-05, + "loss": 3.2253, + "step": 4809 + }, + { + "epoch": 7.03, + "grad_norm": 1.7917122840881348, + "learning_rate": 3.788179129231159e-05, + "loss": 3.1897, + "step": 4810 + }, + { + "epoch": 7.04, + "grad_norm": 1.6101915836334229, + "learning_rate": 3.78450414799838e-05, + "loss": 3.1593, + "step": 4811 + }, + { + "epoch": 7.04, + "grad_norm": 0.8278429508209229, + "learning_rate": 3.780830534133037e-05, + "loss": 3.1712, + "step": 4812 + }, + { + "epoch": 7.04, + "grad_norm": 1.5922514200210571, + "learning_rate": 3.777158288443291e-05, + "loss": 3.1608, + "step": 4813 + }, + { + "epoch": 7.04, + "grad_norm": 0.9403831362724304, + "learning_rate": 3.7734874117370244e-05, + "loss": 3.2142, + "step": 4814 + }, + { + "epoch": 7.04, + "grad_norm": 1.2470545768737793, + "learning_rate": 3.769817904821796e-05, + "loss": 3.2069, + "step": 4815 + }, + { + "epoch": 7.04, + "grad_norm": 0.7299978733062744, + "learning_rate": 3.766149768504876e-05, + "loss": 3.1938, + "step": 4816 + }, + { + "epoch": 7.04, + "grad_norm": 1.2859150171279907, + "learning_rate": 3.762483003593229e-05, + "loss": 3.1506, + "step": 4817 + }, + { + "epoch": 7.05, + "grad_norm": 0.8429161906242371, + "learning_rate": 3.7588176108935194e-05, + "loss": 3.1859, + "step": 4818 + }, + { + "epoch": 7.05, + "grad_norm": 1.0546910762786865, + "learning_rate": 3.75515359121211e-05, + "loss": 3.1802, + "step": 4819 + }, + { + "epoch": 7.05, + "grad_norm": 1.477935791015625, + "learning_rate": 3.7514909453550575e-05, + "loss": 3.19, + "step": 4820 + }, + { + "epoch": 7.05, + "grad_norm": 0.5975956916809082, + "learning_rate": 3.7478296741281215e-05, + "loss": 3.192, + "step": 4821 + }, + { + "epoch": 7.05, + "grad_norm": 0.9366810321807861, + "learning_rate": 3.744169778336758e-05, + "loss": 3.1711, + "step": 4822 + }, + { + "epoch": 7.05, + "grad_norm": 0.5407786965370178, + "learning_rate": 3.740511258786118e-05, + "loss": 3.2555, + "step": 4823 + }, + { + "epoch": 7.06, + "grad_norm": 0.5257308483123779, + "learning_rate": 3.736854116281052e-05, + "loss": 3.1868, + "step": 4824 + }, + { + "epoch": 7.06, + "grad_norm": 0.5943123698234558, + "learning_rate": 3.7331983516261073e-05, + "loss": 3.1969, + "step": 4825 + }, + { + "epoch": 7.06, + "grad_norm": 0.31985798478126526, + "learning_rate": 3.729543965625526e-05, + "loss": 3.2147, + "step": 4826 + }, + { + "epoch": 7.06, + "grad_norm": 0.932587206363678, + "learning_rate": 3.7258909590832506e-05, + "loss": 3.1519, + "step": 4827 + }, + { + "epoch": 7.06, + "grad_norm": 1.0996321439743042, + "learning_rate": 3.722239332802916e-05, + "loss": 3.2083, + "step": 4828 + }, + { + "epoch": 7.06, + "grad_norm": 0.7307466864585876, + "learning_rate": 3.718589087587857e-05, + "loss": 3.1716, + "step": 4829 + }, + { + "epoch": 7.06, + "grad_norm": 0.39040249586105347, + "learning_rate": 3.7149402242411023e-05, + "loss": 3.1738, + "step": 4830 + }, + { + "epoch": 7.07, + "grad_norm": 0.8237361907958984, + "learning_rate": 3.711292743565378e-05, + "loss": 3.186, + "step": 4831 + }, + { + "epoch": 7.07, + "grad_norm": 0.6313247084617615, + "learning_rate": 3.707646646363104e-05, + "loss": 3.186, + "step": 4832 + }, + { + "epoch": 7.07, + "grad_norm": 1.1303566694259644, + "learning_rate": 3.7040019334363996e-05, + "loss": 3.2012, + "step": 4833 + }, + { + "epoch": 7.07, + "grad_norm": 1.1949049234390259, + "learning_rate": 3.7003586055870755e-05, + "loss": 3.2074, + "step": 4834 + }, + { + "epoch": 7.07, + "grad_norm": 0.9406141638755798, + "learning_rate": 3.696716663616641e-05, + "loss": 3.1975, + "step": 4835 + }, + { + "epoch": 7.07, + "grad_norm": 0.7907782793045044, + "learning_rate": 3.6930761083262955e-05, + "loss": 3.1521, + "step": 4836 + }, + { + "epoch": 7.07, + "grad_norm": 0.45947736501693726, + "learning_rate": 3.689436940516944e-05, + "loss": 3.1649, + "step": 4837 + }, + { + "epoch": 7.08, + "grad_norm": 1.1769413948059082, + "learning_rate": 3.6857991609891675e-05, + "loss": 3.1585, + "step": 4838 + }, + { + "epoch": 7.08, + "grad_norm": 0.3579048812389374, + "learning_rate": 3.6821627705432614e-05, + "loss": 3.2008, + "step": 4839 + }, + { + "epoch": 7.08, + "grad_norm": 1.2127845287322998, + "learning_rate": 3.678527769979211e-05, + "loss": 3.194, + "step": 4840 + }, + { + "epoch": 7.08, + "grad_norm": 0.7427309155464172, + "learning_rate": 3.6748941600966815e-05, + "loss": 3.171, + "step": 4841 + }, + { + "epoch": 7.08, + "grad_norm": 1.3657749891281128, + "learning_rate": 3.671261941695045e-05, + "loss": 3.2067, + "step": 4842 + }, + { + "epoch": 7.08, + "grad_norm": 0.7372667193412781, + "learning_rate": 3.6676311155733746e-05, + "loss": 3.1599, + "step": 4843 + }, + { + "epoch": 7.09, + "grad_norm": 1.225062370300293, + "learning_rate": 3.664001682530418e-05, + "loss": 3.1744, + "step": 4844 + }, + { + "epoch": 7.09, + "grad_norm": 1.3079476356506348, + "learning_rate": 3.6603736433646264e-05, + "loss": 3.149, + "step": 4845 + }, + { + "epoch": 7.09, + "grad_norm": 0.3697514533996582, + "learning_rate": 3.656746998874153e-05, + "loss": 3.1942, + "step": 4846 + }, + { + "epoch": 7.09, + "grad_norm": 2.1522505283355713, + "learning_rate": 3.653121749856826e-05, + "loss": 3.1853, + "step": 4847 + }, + { + "epoch": 7.09, + "grad_norm": 0.901470959186554, + "learning_rate": 3.649497897110179e-05, + "loss": 3.1828, + "step": 4848 + }, + { + "epoch": 7.09, + "grad_norm": 0.8554129004478455, + "learning_rate": 3.645875441431435e-05, + "loss": 3.224, + "step": 4849 + }, + { + "epoch": 7.09, + "grad_norm": 1.8987047672271729, + "learning_rate": 3.642254383617511e-05, + "loss": 3.1925, + "step": 4850 + }, + { + "epoch": 7.1, + "grad_norm": 0.45335274934768677, + "learning_rate": 3.6386347244650155e-05, + "loss": 3.1787, + "step": 4851 + }, + { + "epoch": 7.1, + "grad_norm": 0.8638046979904175, + "learning_rate": 3.635016464770249e-05, + "loss": 3.1878, + "step": 4852 + }, + { + "epoch": 7.1, + "grad_norm": 0.6001507639884949, + "learning_rate": 3.6313996053292034e-05, + "loss": 3.1502, + "step": 4853 + }, + { + "epoch": 7.1, + "grad_norm": 0.9555931687355042, + "learning_rate": 3.627784146937563e-05, + "loss": 3.2091, + "step": 4854 + }, + { + "epoch": 7.1, + "grad_norm": 0.5545978546142578, + "learning_rate": 3.624170090390706e-05, + "loss": 3.2264, + "step": 4855 + }, + { + "epoch": 7.1, + "grad_norm": 0.2554129362106323, + "learning_rate": 3.620557436483701e-05, + "loss": 3.1756, + "step": 4856 + }, + { + "epoch": 7.1, + "grad_norm": 0.648738443851471, + "learning_rate": 3.6169461860113074e-05, + "loss": 3.1885, + "step": 4857 + }, + { + "epoch": 7.11, + "grad_norm": 0.39746615290641785, + "learning_rate": 3.613336339767974e-05, + "loss": 3.1645, + "step": 4858 + }, + { + "epoch": 7.11, + "grad_norm": 0.6752750277519226, + "learning_rate": 3.6097278985478454e-05, + "loss": 3.1669, + "step": 4859 + }, + { + "epoch": 7.11, + "grad_norm": 0.4994068443775177, + "learning_rate": 3.606120863144753e-05, + "loss": 3.2155, + "step": 4860 + }, + { + "epoch": 7.11, + "grad_norm": 0.9360056519508362, + "learning_rate": 3.6025152343522205e-05, + "loss": 3.198, + "step": 4861 + }, + { + "epoch": 7.11, + "grad_norm": 0.3693661093711853, + "learning_rate": 3.598911012963463e-05, + "loss": 3.1723, + "step": 4862 + }, + { + "epoch": 7.11, + "grad_norm": 0.7912516593933105, + "learning_rate": 3.5953081997713876e-05, + "loss": 3.1946, + "step": 4863 + }, + { + "epoch": 7.11, + "grad_norm": 0.5820804834365845, + "learning_rate": 3.591706795568579e-05, + "loss": 3.1551, + "step": 4864 + }, + { + "epoch": 7.12, + "grad_norm": 0.41594260931015015, + "learning_rate": 3.588106801147333e-05, + "loss": 3.2372, + "step": 4865 + }, + { + "epoch": 7.12, + "grad_norm": 1.0947422981262207, + "learning_rate": 3.5845082172996224e-05, + "loss": 3.1926, + "step": 4866 + }, + { + "epoch": 7.12, + "grad_norm": 0.4444226324558258, + "learning_rate": 3.580911044817104e-05, + "loss": 3.2154, + "step": 4867 + }, + { + "epoch": 7.12, + "grad_norm": 0.8094576001167297, + "learning_rate": 3.577315284491138e-05, + "loss": 3.1802, + "step": 4868 + }, + { + "epoch": 7.12, + "grad_norm": 0.6356644034385681, + "learning_rate": 3.573720937112772e-05, + "loss": 3.1484, + "step": 4869 + }, + { + "epoch": 7.12, + "grad_norm": 0.9300285577774048, + "learning_rate": 3.570128003472725e-05, + "loss": 3.1834, + "step": 4870 + }, + { + "epoch": 7.13, + "grad_norm": 0.419990211725235, + "learning_rate": 3.5665364843614336e-05, + "loss": 3.1857, + "step": 4871 + }, + { + "epoch": 7.13, + "grad_norm": 1.047782063484192, + "learning_rate": 3.562946380568997e-05, + "loss": 3.2032, + "step": 4872 + }, + { + "epoch": 7.13, + "grad_norm": 0.564288318157196, + "learning_rate": 3.5593576928852134e-05, + "loss": 3.1501, + "step": 4873 + }, + { + "epoch": 7.13, + "grad_norm": 0.6687262654304504, + "learning_rate": 3.555770422099581e-05, + "loss": 3.1657, + "step": 4874 + }, + { + "epoch": 7.13, + "grad_norm": 0.4760758876800537, + "learning_rate": 3.5521845690012655e-05, + "loss": 3.1708, + "step": 4875 + }, + { + "epoch": 7.13, + "grad_norm": 0.6651167273521423, + "learning_rate": 3.5486001343791295e-05, + "loss": 3.1546, + "step": 4876 + }, + { + "epoch": 7.13, + "grad_norm": 0.554828941822052, + "learning_rate": 3.5450171190217364e-05, + "loss": 3.1451, + "step": 4877 + }, + { + "epoch": 7.14, + "grad_norm": 1.176149606704712, + "learning_rate": 3.5414355237173136e-05, + "loss": 3.2025, + "step": 4878 + }, + { + "epoch": 7.14, + "grad_norm": 0.4831768572330475, + "learning_rate": 3.537855349253788e-05, + "loss": 3.2084, + "step": 4879 + }, + { + "epoch": 7.14, + "grad_norm": 0.946924090385437, + "learning_rate": 3.5342765964187864e-05, + "loss": 3.2116, + "step": 4880 + }, + { + "epoch": 7.14, + "grad_norm": 0.4359959661960602, + "learning_rate": 3.530699265999599e-05, + "loss": 3.1784, + "step": 4881 + }, + { + "epoch": 7.14, + "grad_norm": 0.9432086944580078, + "learning_rate": 3.5271233587832164e-05, + "loss": 3.2123, + "step": 4882 + }, + { + "epoch": 7.14, + "grad_norm": 0.4890124797821045, + "learning_rate": 3.523548875556316e-05, + "loss": 3.2041, + "step": 4883 + }, + { + "epoch": 7.14, + "grad_norm": 0.6941928267478943, + "learning_rate": 3.51997581710526e-05, + "loss": 3.2032, + "step": 4884 + }, + { + "epoch": 7.15, + "grad_norm": 0.5665786862373352, + "learning_rate": 3.5164041842160966e-05, + "loss": 3.193, + "step": 4885 + }, + { + "epoch": 7.15, + "grad_norm": 0.7066811919212341, + "learning_rate": 3.5128339776745625e-05, + "loss": 3.1766, + "step": 4886 + }, + { + "epoch": 7.15, + "grad_norm": 0.6680645942687988, + "learning_rate": 3.5092651982660774e-05, + "loss": 3.2185, + "step": 4887 + }, + { + "epoch": 7.15, + "grad_norm": 0.7763155698776245, + "learning_rate": 3.5056978467757506e-05, + "loss": 3.18, + "step": 4888 + }, + { + "epoch": 7.15, + "grad_norm": 0.8374306559562683, + "learning_rate": 3.502131923988373e-05, + "loss": 3.2049, + "step": 4889 + }, + { + "epoch": 7.15, + "grad_norm": 1.1140564680099487, + "learning_rate": 3.4985674306884274e-05, + "loss": 3.195, + "step": 4890 + }, + { + "epoch": 7.15, + "grad_norm": 0.9668914675712585, + "learning_rate": 3.495004367660075e-05, + "loss": 3.1662, + "step": 4891 + }, + { + "epoch": 7.16, + "grad_norm": 0.8925195336341858, + "learning_rate": 3.491442735687167e-05, + "loss": 3.198, + "step": 4892 + }, + { + "epoch": 7.16, + "grad_norm": 1.2694224119186401, + "learning_rate": 3.4878825355532394e-05, + "loss": 3.1972, + "step": 4893 + }, + { + "epoch": 7.16, + "grad_norm": 0.7892391085624695, + "learning_rate": 3.4843237680415156e-05, + "loss": 3.2424, + "step": 4894 + }, + { + "epoch": 7.16, + "grad_norm": 0.8622798323631287, + "learning_rate": 3.480766433934889e-05, + "loss": 3.2112, + "step": 4895 + }, + { + "epoch": 7.16, + "grad_norm": 0.8695278167724609, + "learning_rate": 3.477210534015961e-05, + "loss": 3.2158, + "step": 4896 + }, + { + "epoch": 7.16, + "grad_norm": 0.8544620871543884, + "learning_rate": 3.473656069067005e-05, + "loss": 3.2501, + "step": 4897 + }, + { + "epoch": 7.17, + "grad_norm": 0.6005237102508545, + "learning_rate": 3.470103039869968e-05, + "loss": 3.205, + "step": 4898 + }, + { + "epoch": 7.17, + "grad_norm": 0.796326220035553, + "learning_rate": 3.466551447206506e-05, + "loss": 3.1567, + "step": 4899 + }, + { + "epoch": 7.17, + "grad_norm": 0.40793824195861816, + "learning_rate": 3.463001291857943e-05, + "loss": 3.1772, + "step": 4900 + }, + { + "epoch": 7.17, + "grad_norm": 0.804324209690094, + "learning_rate": 3.459452574605281e-05, + "loss": 3.181, + "step": 4901 + }, + { + "epoch": 7.17, + "eval_loss": 3.0279037952423096, + "eval_runtime": 1163.8402, + "eval_samples_per_second": 14.729, + "eval_steps_per_second": 0.921, + "step": 4901 + }, + { + "epoch": 7.17, + "grad_norm": 0.47186848521232605, + "learning_rate": 3.455905296229222e-05, + "loss": 3.1301, + "step": 4902 + }, + { + "epoch": 7.17, + "grad_norm": 0.7227790951728821, + "learning_rate": 3.452359457510145e-05, + "loss": 3.1657, + "step": 4903 + }, + { + "epoch": 7.17, + "grad_norm": 0.9041692614555359, + "learning_rate": 3.4488150592281e-05, + "loss": 3.2102, + "step": 4904 + }, + { + "epoch": 7.18, + "grad_norm": 1.273749589920044, + "learning_rate": 3.4452721021628445e-05, + "loss": 3.2073, + "step": 4905 + }, + { + "epoch": 7.18, + "grad_norm": 1.3366678953170776, + "learning_rate": 3.441730587093795e-05, + "loss": 3.1913, + "step": 4906 + }, + { + "epoch": 7.18, + "grad_norm": 0.45730334520339966, + "learning_rate": 3.438190514800063e-05, + "loss": 3.2091, + "step": 4907 + }, + { + "epoch": 7.18, + "grad_norm": 1.0588980913162231, + "learning_rate": 3.4346518860604485e-05, + "loss": 3.2037, + "step": 4908 + }, + { + "epoch": 7.18, + "grad_norm": 0.6257373094558716, + "learning_rate": 3.4311147016534164e-05, + "loss": 3.1525, + "step": 4909 + }, + { + "epoch": 7.18, + "grad_norm": 0.7281100749969482, + "learning_rate": 3.4275789623571234e-05, + "loss": 3.2051, + "step": 4910 + }, + { + "epoch": 7.18, + "grad_norm": 0.8004705905914307, + "learning_rate": 3.42404466894942e-05, + "loss": 3.201, + "step": 4911 + }, + { + "epoch": 7.19, + "grad_norm": 0.34791865944862366, + "learning_rate": 3.4205118222078145e-05, + "loss": 3.1746, + "step": 4912 + }, + { + "epoch": 7.19, + "grad_norm": 1.0412063598632812, + "learning_rate": 3.416980422909516e-05, + "loss": 3.2195, + "step": 4913 + }, + { + "epoch": 7.19, + "grad_norm": 0.489105224609375, + "learning_rate": 3.413450471831407e-05, + "loss": 3.1446, + "step": 4914 + }, + { + "epoch": 7.19, + "grad_norm": 0.49872738122940063, + "learning_rate": 3.4099219697500514e-05, + "loss": 3.128, + "step": 4915 + }, + { + "epoch": 7.19, + "grad_norm": 0.7999401092529297, + "learning_rate": 3.4063949174416984e-05, + "loss": 3.1953, + "step": 4916 + }, + { + "epoch": 7.19, + "grad_norm": 0.4927428364753723, + "learning_rate": 3.402869315682275e-05, + "loss": 3.1533, + "step": 4917 + }, + { + "epoch": 7.19, + "grad_norm": 0.29562923312187195, + "learning_rate": 3.3993451652473916e-05, + "loss": 3.1752, + "step": 4918 + }, + { + "epoch": 7.2, + "grad_norm": 0.9475629329681396, + "learning_rate": 3.395822466912335e-05, + "loss": 3.1526, + "step": 4919 + }, + { + "epoch": 7.2, + "grad_norm": 0.9323421120643616, + "learning_rate": 3.3923012214520763e-05, + "loss": 3.1457, + "step": 4920 + }, + { + "epoch": 7.2, + "grad_norm": 0.30518239736557007, + "learning_rate": 3.388781429641267e-05, + "loss": 3.1892, + "step": 4921 + }, + { + "epoch": 7.2, + "grad_norm": 0.4029679000377655, + "learning_rate": 3.385263092254236e-05, + "loss": 3.1918, + "step": 4922 + }, + { + "epoch": 7.2, + "grad_norm": 0.7984765768051147, + "learning_rate": 3.381746210064995e-05, + "loss": 3.2006, + "step": 4923 + }, + { + "epoch": 7.2, + "grad_norm": 0.9310452342033386, + "learning_rate": 3.378230783847234e-05, + "loss": 3.1526, + "step": 4924 + }, + { + "epoch": 7.21, + "grad_norm": 0.5400900840759277, + "learning_rate": 3.374716814374323e-05, + "loss": 3.2063, + "step": 4925 + }, + { + "epoch": 7.21, + "grad_norm": 0.7315930128097534, + "learning_rate": 3.371204302419313e-05, + "loss": 3.1709, + "step": 4926 + }, + { + "epoch": 7.21, + "grad_norm": 0.8185734152793884, + "learning_rate": 3.367693248754933e-05, + "loss": 3.1659, + "step": 4927 + }, + { + "epoch": 7.21, + "grad_norm": 0.4014480710029602, + "learning_rate": 3.364183654153592e-05, + "loss": 3.1291, + "step": 4928 + }, + { + "epoch": 7.21, + "grad_norm": 0.45573729276657104, + "learning_rate": 3.360675519387371e-05, + "loss": 3.1648, + "step": 4929 + }, + { + "epoch": 7.21, + "grad_norm": 0.49166086316108704, + "learning_rate": 3.357168845228043e-05, + "loss": 3.168, + "step": 4930 + }, + { + "epoch": 7.21, + "grad_norm": 0.37833172082901, + "learning_rate": 3.353663632447054e-05, + "loss": 3.1675, + "step": 4931 + }, + { + "epoch": 7.22, + "grad_norm": 0.9355205297470093, + "learning_rate": 3.350159881815518e-05, + "loss": 3.1798, + "step": 4932 + }, + { + "epoch": 7.22, + "grad_norm": 0.888450562953949, + "learning_rate": 3.346657594104246e-05, + "loss": 3.1609, + "step": 4933 + }, + { + "epoch": 7.22, + "grad_norm": 0.39355748891830444, + "learning_rate": 3.343156770083717e-05, + "loss": 3.1912, + "step": 4934 + }, + { + "epoch": 7.22, + "grad_norm": 0.4100874960422516, + "learning_rate": 3.339657410524081e-05, + "loss": 3.1741, + "step": 4935 + }, + { + "epoch": 7.22, + "grad_norm": 0.6284832954406738, + "learning_rate": 3.336159516195182e-05, + "loss": 3.1456, + "step": 4936 + }, + { + "epoch": 7.22, + "grad_norm": 9.06452465057373, + "learning_rate": 3.3326630878665334e-05, + "loss": 3.1432, + "step": 4937 + }, + { + "epoch": 7.22, + "grad_norm": 1.4369940757751465, + "learning_rate": 3.329168126307319e-05, + "loss": 3.1878, + "step": 4938 + }, + { + "epoch": 7.23, + "grad_norm": 0.6629584431648254, + "learning_rate": 3.3256746322864127e-05, + "loss": 3.1782, + "step": 4939 + }, + { + "epoch": 7.23, + "grad_norm": 1.5383516550064087, + "learning_rate": 3.322182606572357e-05, + "loss": 3.1567, + "step": 4940 + }, + { + "epoch": 7.23, + "grad_norm": 1.498374581336975, + "learning_rate": 3.318692049933375e-05, + "loss": 3.2173, + "step": 4941 + }, + { + "epoch": 7.23, + "grad_norm": 0.5786330699920654, + "learning_rate": 3.315202963137366e-05, + "loss": 3.1327, + "step": 4942 + }, + { + "epoch": 7.23, + "grad_norm": 1.2023847103118896, + "learning_rate": 3.311715346951906e-05, + "loss": 3.2942, + "step": 4943 + }, + { + "epoch": 7.23, + "grad_norm": 0.6449021697044373, + "learning_rate": 3.3082292021442454e-05, + "loss": 3.1564, + "step": 4944 + }, + { + "epoch": 7.24, + "grad_norm": 0.36229845881462097, + "learning_rate": 3.304744529481315e-05, + "loss": 3.1787, + "step": 4945 + }, + { + "epoch": 7.24, + "grad_norm": 0.9449754953384399, + "learning_rate": 3.301261329729719e-05, + "loss": 3.1928, + "step": 4946 + }, + { + "epoch": 7.24, + "grad_norm": 0.9095726609230042, + "learning_rate": 3.297779603655736e-05, + "loss": 3.2106, + "step": 4947 + }, + { + "epoch": 7.24, + "grad_norm": 0.346010684967041, + "learning_rate": 3.2942993520253254e-05, + "loss": 3.1908, + "step": 4948 + }, + { + "epoch": 7.24, + "grad_norm": 0.7208975553512573, + "learning_rate": 3.2908205756041176e-05, + "loss": 3.1851, + "step": 4949 + }, + { + "epoch": 7.24, + "grad_norm": 0.5966201424598694, + "learning_rate": 3.2873432751574194e-05, + "loss": 3.2213, + "step": 4950 + }, + { + "epoch": 7.24, + "grad_norm": 0.3596455156803131, + "learning_rate": 3.2838674514502154e-05, + "loss": 3.2082, + "step": 4951 + }, + { + "epoch": 7.25, + "grad_norm": 0.8812998533248901, + "learning_rate": 3.2803931052471615e-05, + "loss": 3.1496, + "step": 4952 + }, + { + "epoch": 7.25, + "grad_norm": 0.7055163979530334, + "learning_rate": 3.276920237312593e-05, + "loss": 3.1859, + "step": 4953 + }, + { + "epoch": 7.25, + "grad_norm": 0.7300843596458435, + "learning_rate": 3.273448848410515e-05, + "loss": 3.1804, + "step": 4954 + }, + { + "epoch": 7.25, + "grad_norm": 1.0114784240722656, + "learning_rate": 3.269978939304612e-05, + "loss": 3.1612, + "step": 4955 + }, + { + "epoch": 7.25, + "grad_norm": 0.46303462982177734, + "learning_rate": 3.266510510758238e-05, + "loss": 3.1675, + "step": 4956 + }, + { + "epoch": 7.25, + "grad_norm": 0.722417950630188, + "learning_rate": 3.263043563534428e-05, + "loss": 3.1732, + "step": 4957 + }, + { + "epoch": 7.25, + "grad_norm": 0.6245961785316467, + "learning_rate": 3.2595780983958834e-05, + "loss": 3.1481, + "step": 4958 + }, + { + "epoch": 7.26, + "grad_norm": 0.31019535660743713, + "learning_rate": 3.256114116104986e-05, + "loss": 3.1808, + "step": 4959 + }, + { + "epoch": 7.26, + "grad_norm": 1.2723808288574219, + "learning_rate": 3.25265161742379e-05, + "loss": 3.1502, + "step": 4960 + }, + { + "epoch": 7.26, + "grad_norm": 0.6151775121688843, + "learning_rate": 3.249190603114013e-05, + "loss": 3.1393, + "step": 4961 + }, + { + "epoch": 7.26, + "grad_norm": 1.0856354236602783, + "learning_rate": 3.245731073937068e-05, + "loss": 3.1964, + "step": 4962 + }, + { + "epoch": 7.26, + "grad_norm": 0.665896475315094, + "learning_rate": 3.2422730306540186e-05, + "loss": 3.1715, + "step": 4963 + }, + { + "epoch": 7.26, + "grad_norm": 1.0534874200820923, + "learning_rate": 3.23881647402561e-05, + "loss": 3.1952, + "step": 4964 + }, + { + "epoch": 7.26, + "grad_norm": 1.2023156881332397, + "learning_rate": 3.235361404812274e-05, + "loss": 3.1622, + "step": 4965 + }, + { + "epoch": 7.27, + "grad_norm": 0.9104660749435425, + "learning_rate": 3.23190782377409e-05, + "loss": 3.1731, + "step": 4966 + }, + { + "epoch": 7.27, + "grad_norm": 1.6818856000900269, + "learning_rate": 3.228455731670823e-05, + "loss": 3.2543, + "step": 4967 + }, + { + "epoch": 7.27, + "grad_norm": 0.40944576263427734, + "learning_rate": 3.225005129261922e-05, + "loss": 3.1609, + "step": 4968 + }, + { + "epoch": 7.27, + "grad_norm": 2.257155656814575, + "learning_rate": 3.221556017306484e-05, + "loss": 3.2051, + "step": 4969 + }, + { + "epoch": 7.27, + "grad_norm": 1.4875580072402954, + "learning_rate": 3.218108396563293e-05, + "loss": 3.1654, + "step": 4970 + }, + { + "epoch": 7.27, + "grad_norm": 0.7136067152023315, + "learning_rate": 3.21466226779081e-05, + "loss": 3.1851, + "step": 4971 + }, + { + "epoch": 7.28, + "grad_norm": 1.1615363359451294, + "learning_rate": 3.2112176317471535e-05, + "loss": 3.1378, + "step": 4972 + }, + { + "epoch": 7.28, + "grad_norm": 1.23421311378479, + "learning_rate": 3.20777448919012e-05, + "loss": 3.1976, + "step": 4973 + }, + { + "epoch": 7.28, + "grad_norm": 0.3943505883216858, + "learning_rate": 3.204332840877179e-05, + "loss": 3.1514, + "step": 4974 + }, + { + "epoch": 7.28, + "grad_norm": 1.666935682296753, + "learning_rate": 3.200892687565471e-05, + "loss": 3.1276, + "step": 4975 + }, + { + "epoch": 7.28, + "grad_norm": 0.877071738243103, + "learning_rate": 3.1974540300118064e-05, + "loss": 3.1572, + "step": 4976 + }, + { + "epoch": 7.28, + "grad_norm": 1.395138144493103, + "learning_rate": 3.1940168689726666e-05, + "loss": 3.1789, + "step": 4977 + }, + { + "epoch": 7.28, + "grad_norm": 1.2696112394332886, + "learning_rate": 3.190581205204204e-05, + "loss": 3.1909, + "step": 4978 + }, + { + "epoch": 7.29, + "grad_norm": 0.9251769185066223, + "learning_rate": 3.1871470394622404e-05, + "loss": 3.1486, + "step": 4979 + }, + { + "epoch": 7.29, + "grad_norm": 0.9306465983390808, + "learning_rate": 3.183714372502272e-05, + "loss": 3.1897, + "step": 4980 + }, + { + "epoch": 7.29, + "grad_norm": 0.6166656613349915, + "learning_rate": 3.1802832050794595e-05, + "loss": 3.1902, + "step": 4981 + }, + { + "epoch": 7.29, + "grad_norm": 1.0859153270721436, + "learning_rate": 3.1768535379486395e-05, + "loss": 3.2179, + "step": 4982 + }, + { + "epoch": 7.29, + "grad_norm": 0.5373363494873047, + "learning_rate": 3.173425371864314e-05, + "loss": 3.1732, + "step": 4983 + }, + { + "epoch": 7.29, + "grad_norm": 0.6277896761894226, + "learning_rate": 3.169998707580657e-05, + "loss": 3.1735, + "step": 4984 + }, + { + "epoch": 7.29, + "grad_norm": 0.8959996104240417, + "learning_rate": 3.166573545851514e-05, + "loss": 3.1425, + "step": 4985 + }, + { + "epoch": 7.3, + "grad_norm": 0.5353897213935852, + "learning_rate": 3.16314988743039e-05, + "loss": 3.1447, + "step": 4986 + }, + { + "epoch": 7.3, + "grad_norm": 0.8923834562301636, + "learning_rate": 3.1597277330704756e-05, + "loss": 3.1684, + "step": 4987 + }, + { + "epoch": 7.3, + "grad_norm": 0.456402063369751, + "learning_rate": 3.1563070835246213e-05, + "loss": 3.1917, + "step": 4988 + }, + { + "epoch": 7.3, + "grad_norm": 0.744957685470581, + "learning_rate": 3.152887939545337e-05, + "loss": 3.205, + "step": 4989 + }, + { + "epoch": 7.3, + "grad_norm": 0.7469781041145325, + "learning_rate": 3.149470301884823e-05, + "loss": 3.1544, + "step": 4990 + }, + { + "epoch": 7.3, + "grad_norm": 0.48788437247276306, + "learning_rate": 3.146054171294935e-05, + "loss": 3.1809, + "step": 4991 + }, + { + "epoch": 7.3, + "grad_norm": 1.1828556060791016, + "learning_rate": 3.14263954852719e-05, + "loss": 3.1552, + "step": 4992 + }, + { + "epoch": 7.31, + "grad_norm": 0.35930806398391724, + "learning_rate": 3.1392264343327914e-05, + "loss": 3.1549, + "step": 4993 + }, + { + "epoch": 7.31, + "grad_norm": 1.010923981666565, + "learning_rate": 3.135814829462602e-05, + "loss": 3.181, + "step": 4994 + }, + { + "epoch": 7.31, + "grad_norm": 1.8422837257385254, + "learning_rate": 3.132404734667142e-05, + "loss": 3.1774, + "step": 4995 + }, + { + "epoch": 7.31, + "grad_norm": 0.7072595357894897, + "learning_rate": 3.1289961506966214e-05, + "loss": 3.1996, + "step": 4996 + }, + { + "epoch": 7.31, + "grad_norm": 0.8350168466567993, + "learning_rate": 3.125589078300898e-05, + "loss": 3.2288, + "step": 4997 + }, + { + "epoch": 7.31, + "grad_norm": 0.7736479043960571, + "learning_rate": 3.122183518229503e-05, + "loss": 3.1702, + "step": 4998 + }, + { + "epoch": 7.32, + "grad_norm": 0.6932786107063293, + "learning_rate": 3.1187794712316474e-05, + "loss": 3.1662, + "step": 4999 + }, + { + "epoch": 7.32, + "grad_norm": 0.5773895382881165, + "learning_rate": 3.11537693805619e-05, + "loss": 3.1859, + "step": 5000 + }, + { + "epoch": 7.32, + "grad_norm": 0.8364612460136414, + "learning_rate": 3.111975919451664e-05, + "loss": 3.1769, + "step": 5001 + }, + { + "epoch": 7.32, + "grad_norm": 0.3345867097377777, + "learning_rate": 3.10857641616628e-05, + "loss": 3.1939, + "step": 5002 + }, + { + "epoch": 7.32, + "grad_norm": 0.6873171925544739, + "learning_rate": 3.105178428947898e-05, + "loss": 3.2176, + "step": 5003 + }, + { + "epoch": 7.32, + "grad_norm": 0.5719138383865356, + "learning_rate": 3.10178195854405e-05, + "loss": 3.174, + "step": 5004 + }, + { + "epoch": 7.32, + "grad_norm": 0.9275087714195251, + "learning_rate": 3.0983870057019484e-05, + "loss": 3.2174, + "step": 5005 + }, + { + "epoch": 7.33, + "grad_norm": 1.1784770488739014, + "learning_rate": 3.09499357116845e-05, + "loss": 3.1778, + "step": 5006 + }, + { + "epoch": 7.33, + "grad_norm": 0.7836307287216187, + "learning_rate": 3.0916016556900904e-05, + "loss": 3.1627, + "step": 5007 + }, + { + "epoch": 7.33, + "grad_norm": 0.482113242149353, + "learning_rate": 3.088211260013068e-05, + "loss": 3.1841, + "step": 5008 + }, + { + "epoch": 7.33, + "grad_norm": 0.48355069756507874, + "learning_rate": 3.084822384883249e-05, + "loss": 3.1691, + "step": 5009 + }, + { + "epoch": 7.33, + "grad_norm": 0.4077546000480652, + "learning_rate": 3.0814350310461615e-05, + "loss": 3.1581, + "step": 5010 + }, + { + "epoch": 7.33, + "grad_norm": 0.7942917943000793, + "learning_rate": 3.078049199247e-05, + "loss": 3.1964, + "step": 5011 + }, + { + "epoch": 7.33, + "grad_norm": 0.38686656951904297, + "learning_rate": 3.0746648902306264e-05, + "loss": 3.1606, + "step": 5012 + }, + { + "epoch": 7.34, + "grad_norm": 1.3776711225509644, + "learning_rate": 3.071282104741565e-05, + "loss": 3.2248, + "step": 5013 + }, + { + "epoch": 7.34, + "grad_norm": 0.2463265061378479, + "learning_rate": 3.067900843524008e-05, + "loss": 3.1848, + "step": 5014 + }, + { + "epoch": 7.34, + "grad_norm": 0.5669382214546204, + "learning_rate": 3.064521107321806e-05, + "loss": 3.1501, + "step": 5015 + }, + { + "epoch": 7.34, + "grad_norm": 0.40979495644569397, + "learning_rate": 3.061142896878486e-05, + "loss": 3.1762, + "step": 5016 + }, + { + "epoch": 7.34, + "grad_norm": 0.5121455192565918, + "learning_rate": 3.05776621293722e-05, + "loss": 3.1893, + "step": 5017 + }, + { + "epoch": 7.34, + "grad_norm": 0.3571344017982483, + "learning_rate": 3.054391056240865e-05, + "loss": 3.1726, + "step": 5018 + }, + { + "epoch": 7.34, + "grad_norm": 0.3279051184654236, + "learning_rate": 3.0510174275319325e-05, + "loss": 3.1238, + "step": 5019 + }, + { + "epoch": 7.35, + "grad_norm": 0.5023763179779053, + "learning_rate": 3.0476453275525906e-05, + "loss": 3.1893, + "step": 5020 + }, + { + "epoch": 7.35, + "grad_norm": 0.45298734307289124, + "learning_rate": 3.0442747570446872e-05, + "loss": 3.192, + "step": 5021 + }, + { + "epoch": 7.35, + "grad_norm": 0.580197274684906, + "learning_rate": 3.0409057167497258e-05, + "loss": 3.1934, + "step": 5022 + }, + { + "epoch": 7.35, + "grad_norm": 0.4648624360561371, + "learning_rate": 3.037538207408862e-05, + "loss": 3.1733, + "step": 5023 + }, + { + "epoch": 7.35, + "grad_norm": 0.8324884176254272, + "learning_rate": 3.034172229762935e-05, + "loss": 3.1584, + "step": 5024 + }, + { + "epoch": 7.35, + "grad_norm": 0.6418886184692383, + "learning_rate": 3.0308077845524384e-05, + "loss": 3.1809, + "step": 5025 + }, + { + "epoch": 7.36, + "grad_norm": 0.7795079946517944, + "learning_rate": 3.0274448725175176e-05, + "loss": 3.1515, + "step": 5026 + }, + { + "epoch": 7.36, + "grad_norm": 0.5563576221466064, + "learning_rate": 3.024083494397999e-05, + "loss": 3.1772, + "step": 5027 + }, + { + "epoch": 7.36, + "grad_norm": 0.42387935519218445, + "learning_rate": 3.0207236509333647e-05, + "loss": 3.1866, + "step": 5028 + }, + { + "epoch": 7.36, + "grad_norm": 0.5205991864204407, + "learning_rate": 3.017365342862748e-05, + "loss": 3.1517, + "step": 5029 + }, + { + "epoch": 7.36, + "grad_norm": 0.5210472941398621, + "learning_rate": 3.0140085709249667e-05, + "loss": 3.172, + "step": 5030 + }, + { + "epoch": 7.36, + "grad_norm": 0.5895211100578308, + "learning_rate": 3.010653335858479e-05, + "loss": 3.239, + "step": 5031 + }, + { + "epoch": 7.36, + "grad_norm": 0.38003700971603394, + "learning_rate": 3.007299638401416e-05, + "loss": 3.1581, + "step": 5032 + }, + { + "epoch": 7.37, + "grad_norm": 0.3639942705631256, + "learning_rate": 3.0039474792915702e-05, + "loss": 3.2157, + "step": 5033 + }, + { + "epoch": 7.37, + "grad_norm": 1.5683830976486206, + "learning_rate": 3.0005968592663925e-05, + "loss": 3.1839, + "step": 5034 + }, + { + "epoch": 7.37, + "grad_norm": 0.5200375914573669, + "learning_rate": 2.9972477790629994e-05, + "loss": 3.1836, + "step": 5035 + }, + { + "epoch": 7.37, + "grad_norm": 0.7715541124343872, + "learning_rate": 2.9939002394181637e-05, + "loss": 3.1778, + "step": 5036 + }, + { + "epoch": 7.37, + "grad_norm": 1.0058480501174927, + "learning_rate": 2.9905542410683218e-05, + "loss": 3.1782, + "step": 5037 + }, + { + "epoch": 7.37, + "grad_norm": 0.6472846269607544, + "learning_rate": 2.987209784749573e-05, + "loss": 3.1847, + "step": 5038 + }, + { + "epoch": 7.37, + "grad_norm": 1.0543522834777832, + "learning_rate": 2.9838668711976724e-05, + "loss": 3.1228, + "step": 5039 + }, + { + "epoch": 7.38, + "grad_norm": 0.9698706865310669, + "learning_rate": 2.9805255011480415e-05, + "loss": 3.163, + "step": 5040 + }, + { + "epoch": 7.38, + "grad_norm": 0.7419912219047546, + "learning_rate": 2.9771856753357573e-05, + "loss": 3.1658, + "step": 5041 + }, + { + "epoch": 7.38, + "grad_norm": 0.8493951559066772, + "learning_rate": 2.9738473944955603e-05, + "loss": 3.1264, + "step": 5042 + }, + { + "epoch": 7.38, + "grad_norm": 0.9127371311187744, + "learning_rate": 2.970510659361848e-05, + "loss": 3.2144, + "step": 5043 + }, + { + "epoch": 7.38, + "grad_norm": 0.5002891421318054, + "learning_rate": 2.967175470668683e-05, + "loss": 3.1733, + "step": 5044 + }, + { + "epoch": 7.38, + "grad_norm": 0.7260929346084595, + "learning_rate": 2.9638418291497817e-05, + "loss": 3.1292, + "step": 5045 + }, + { + "epoch": 7.39, + "grad_norm": 0.8466237783432007, + "learning_rate": 2.9605097355385227e-05, + "loss": 3.194, + "step": 5046 + }, + { + "epoch": 7.39, + "grad_norm": 0.29703596234321594, + "learning_rate": 2.9571791905679468e-05, + "loss": 3.1485, + "step": 5047 + }, + { + "epoch": 7.39, + "grad_norm": 0.8655198216438293, + "learning_rate": 2.9538501949707487e-05, + "loss": 3.1638, + "step": 5048 + }, + { + "epoch": 7.39, + "grad_norm": 0.5336217880249023, + "learning_rate": 2.9505227494792874e-05, + "loss": 3.1671, + "step": 5049 + }, + { + "epoch": 7.39, + "grad_norm": 0.2816140949726105, + "learning_rate": 2.947196854825581e-05, + "loss": 3.1982, + "step": 5050 + }, + { + "epoch": 7.39, + "grad_norm": 0.6799694299697876, + "learning_rate": 2.9438725117412947e-05, + "loss": 3.1824, + "step": 5051 + }, + { + "epoch": 7.39, + "grad_norm": 0.7424922585487366, + "learning_rate": 2.9405497209577703e-05, + "loss": 3.1787, + "step": 5052 + }, + { + "epoch": 7.4, + "grad_norm": 0.5237893462181091, + "learning_rate": 2.9372284832059993e-05, + "loss": 3.1705, + "step": 5053 + }, + { + "epoch": 7.4, + "grad_norm": 0.7378460764884949, + "learning_rate": 2.933908799216625e-05, + "loss": 3.1769, + "step": 5054 + }, + { + "epoch": 7.4, + "grad_norm": 0.9311485886573792, + "learning_rate": 2.9305906697199615e-05, + "loss": 3.1574, + "step": 5055 + }, + { + "epoch": 7.4, + "grad_norm": 0.41424453258514404, + "learning_rate": 2.9272740954459776e-05, + "loss": 3.132, + "step": 5056 + }, + { + "epoch": 7.4, + "grad_norm": 0.4336400330066681, + "learning_rate": 2.92395907712429e-05, + "loss": 3.1724, + "step": 5057 + }, + { + "epoch": 7.4, + "grad_norm": 0.902377188205719, + "learning_rate": 2.920645615484181e-05, + "loss": 3.1471, + "step": 5058 + }, + { + "epoch": 7.4, + "grad_norm": 0.7323518991470337, + "learning_rate": 2.9173337112545985e-05, + "loss": 3.2012, + "step": 5059 + }, + { + "epoch": 7.41, + "grad_norm": 0.6311755776405334, + "learning_rate": 2.9140233651641314e-05, + "loss": 3.1655, + "step": 5060 + }, + { + "epoch": 7.41, + "grad_norm": 1.4856607913970947, + "learning_rate": 2.9107145779410317e-05, + "loss": 3.1768, + "step": 5061 + }, + { + "epoch": 7.41, + "grad_norm": 0.5299568176269531, + "learning_rate": 2.9074073503132217e-05, + "loss": 3.1368, + "step": 5062 + }, + { + "epoch": 7.41, + "grad_norm": 0.5073440074920654, + "learning_rate": 2.904101683008259e-05, + "loss": 3.1784, + "step": 5063 + }, + { + "epoch": 7.41, + "grad_norm": 0.7405803799629211, + "learning_rate": 2.9007975767533714e-05, + "loss": 3.145, + "step": 5064 + }, + { + "epoch": 7.41, + "grad_norm": 0.39113080501556396, + "learning_rate": 2.897495032275439e-05, + "loss": 3.148, + "step": 5065 + }, + { + "epoch": 7.41, + "grad_norm": 0.8356419801712036, + "learning_rate": 2.8941940503010022e-05, + "loss": 3.1611, + "step": 5066 + }, + { + "epoch": 7.42, + "grad_norm": 0.6192629337310791, + "learning_rate": 2.8908946315562524e-05, + "loss": 3.1742, + "step": 5067 + }, + { + "epoch": 7.42, + "grad_norm": 1.1198054552078247, + "learning_rate": 2.8875967767670408e-05, + "loss": 3.1789, + "step": 5068 + }, + { + "epoch": 7.42, + "grad_norm": 0.4452805519104004, + "learning_rate": 2.884300486658873e-05, + "loss": 3.1533, + "step": 5069 + }, + { + "epoch": 7.42, + "grad_norm": 0.8051960468292236, + "learning_rate": 2.881005761956911e-05, + "loss": 3.1688, + "step": 5070 + }, + { + "epoch": 7.42, + "eval_loss": 3.0202746391296387, + "eval_runtime": 1162.3223, + "eval_samples_per_second": 14.748, + "eval_steps_per_second": 0.922, + "step": 5070 + }, + { + "epoch": 7.42, + "grad_norm": 1.1692814826965332, + "learning_rate": 2.8777126033859726e-05, + "loss": 3.1559, + "step": 5071 + }, + { + "epoch": 7.42, + "grad_norm": 0.3186073303222656, + "learning_rate": 2.8744210116705296e-05, + "loss": 3.1507, + "step": 5072 + }, + { + "epoch": 7.43, + "grad_norm": 0.6718719005584717, + "learning_rate": 2.8711309875347102e-05, + "loss": 3.1667, + "step": 5073 + }, + { + "epoch": 7.43, + "grad_norm": 0.7967109680175781, + "learning_rate": 2.8678425317022973e-05, + "loss": 3.1415, + "step": 5074 + }, + { + "epoch": 7.43, + "grad_norm": 0.36231231689453125, + "learning_rate": 2.8645556448967315e-05, + "loss": 3.1264, + "step": 5075 + }, + { + "epoch": 7.43, + "grad_norm": 0.5855808258056641, + "learning_rate": 2.8612703278411025e-05, + "loss": 3.1217, + "step": 5076 + }, + { + "epoch": 7.43, + "grad_norm": 0.6386885046958923, + "learning_rate": 2.8579865812581608e-05, + "loss": 3.1476, + "step": 5077 + }, + { + "epoch": 7.43, + "grad_norm": 0.4413103759288788, + "learning_rate": 2.8547044058703076e-05, + "loss": 3.2046, + "step": 5078 + }, + { + "epoch": 7.43, + "grad_norm": 0.7146703004837036, + "learning_rate": 2.851423802399599e-05, + "loss": 3.2516, + "step": 5079 + }, + { + "epoch": 7.44, + "grad_norm": 0.7587980031967163, + "learning_rate": 2.8481447715677466e-05, + "loss": 3.2025, + "step": 5080 + }, + { + "epoch": 7.44, + "grad_norm": 1.2787212133407593, + "learning_rate": 2.8448673140961147e-05, + "loss": 3.1572, + "step": 5081 + }, + { + "epoch": 7.44, + "grad_norm": 0.5361132621765137, + "learning_rate": 2.8415914307057245e-05, + "loss": 3.1564, + "step": 5082 + }, + { + "epoch": 7.44, + "grad_norm": 1.6003608703613281, + "learning_rate": 2.8383171221172412e-05, + "loss": 3.1482, + "step": 5083 + }, + { + "epoch": 7.44, + "grad_norm": 0.9171953201293945, + "learning_rate": 2.8350443890510003e-05, + "loss": 3.1886, + "step": 5084 + }, + { + "epoch": 7.44, + "grad_norm": 0.39613091945648193, + "learning_rate": 2.8317732322269754e-05, + "loss": 3.1506, + "step": 5085 + }, + { + "epoch": 7.44, + "grad_norm": 1.17970609664917, + "learning_rate": 2.8285036523647944e-05, + "loss": 3.1526, + "step": 5086 + }, + { + "epoch": 7.45, + "grad_norm": 0.781102180480957, + "learning_rate": 2.8252356501837562e-05, + "loss": 3.1663, + "step": 5087 + }, + { + "epoch": 7.45, + "grad_norm": 0.49025315046310425, + "learning_rate": 2.8219692264027886e-05, + "loss": 3.1647, + "step": 5088 + }, + { + "epoch": 7.45, + "grad_norm": 2.0356221199035645, + "learning_rate": 2.8187043817404833e-05, + "loss": 3.2115, + "step": 5089 + }, + { + "epoch": 7.45, + "grad_norm": 1.2565196752548218, + "learning_rate": 2.815441116915093e-05, + "loss": 3.2334, + "step": 5090 + }, + { + "epoch": 7.45, + "grad_norm": 0.6307234168052673, + "learning_rate": 2.812179432644506e-05, + "loss": 3.1614, + "step": 5091 + }, + { + "epoch": 7.45, + "grad_norm": 0.8804566264152527, + "learning_rate": 2.808919329646269e-05, + "loss": 3.1703, + "step": 5092 + }, + { + "epoch": 7.45, + "grad_norm": 0.5766822695732117, + "learning_rate": 2.8056608086375934e-05, + "loss": 3.1264, + "step": 5093 + }, + { + "epoch": 7.46, + "grad_norm": 0.9480114579200745, + "learning_rate": 2.8024038703353227e-05, + "loss": 3.1382, + "step": 5094 + }, + { + "epoch": 7.46, + "grad_norm": 0.5312039852142334, + "learning_rate": 2.7991485154559617e-05, + "loss": 3.1555, + "step": 5095 + }, + { + "epoch": 7.46, + "grad_norm": 0.5020516514778137, + "learning_rate": 2.7958947447156746e-05, + "loss": 3.1825, + "step": 5096 + }, + { + "epoch": 7.46, + "grad_norm": 0.8643108010292053, + "learning_rate": 2.7926425588302606e-05, + "loss": 3.1903, + "step": 5097 + }, + { + "epoch": 7.46, + "grad_norm": 0.38518649339675903, + "learning_rate": 2.789391958515183e-05, + "loss": 3.1832, + "step": 5098 + }, + { + "epoch": 7.46, + "grad_norm": 0.9875998497009277, + "learning_rate": 2.7861429444855514e-05, + "loss": 3.1695, + "step": 5099 + }, + { + "epoch": 7.47, + "grad_norm": 0.5301055312156677, + "learning_rate": 2.7828955174561257e-05, + "loss": 3.1645, + "step": 5100 + }, + { + "epoch": 7.47, + "grad_norm": 0.7682579755783081, + "learning_rate": 2.7796496781413206e-05, + "loss": 3.1384, + "step": 5101 + }, + { + "epoch": 7.47, + "grad_norm": 1.0809067487716675, + "learning_rate": 2.7764054272551965e-05, + "loss": 3.1427, + "step": 5102 + }, + { + "epoch": 7.47, + "grad_norm": 0.6575438380241394, + "learning_rate": 2.773162765511469e-05, + "loss": 3.1146, + "step": 5103 + }, + { + "epoch": 7.47, + "grad_norm": 1.6334797143936157, + "learning_rate": 2.7699216936234994e-05, + "loss": 3.1582, + "step": 5104 + }, + { + "epoch": 7.47, + "grad_norm": 1.1224489212036133, + "learning_rate": 2.7666822123043046e-05, + "loss": 3.1531, + "step": 5105 + }, + { + "epoch": 7.47, + "grad_norm": 0.6989713311195374, + "learning_rate": 2.7634443222665475e-05, + "loss": 3.1587, + "step": 5106 + }, + { + "epoch": 7.48, + "grad_norm": 1.6318511962890625, + "learning_rate": 2.760208024222546e-05, + "loss": 3.1676, + "step": 5107 + }, + { + "epoch": 7.48, + "grad_norm": 1.0403834581375122, + "learning_rate": 2.7569733188842528e-05, + "loss": 3.2076, + "step": 5108 + }, + { + "epoch": 7.48, + "grad_norm": 0.9188293814659119, + "learning_rate": 2.7537402069632922e-05, + "loss": 3.1641, + "step": 5109 + }, + { + "epoch": 7.48, + "grad_norm": 1.1109488010406494, + "learning_rate": 2.7505086891709277e-05, + "loss": 3.1396, + "step": 5110 + }, + { + "epoch": 7.48, + "grad_norm": 0.40556856989860535, + "learning_rate": 2.747278766218061e-05, + "loss": 3.1475, + "step": 5111 + }, + { + "epoch": 7.48, + "grad_norm": 1.6644717454910278, + "learning_rate": 2.744050438815263e-05, + "loss": 3.1237, + "step": 5112 + }, + { + "epoch": 7.48, + "grad_norm": 1.2431093454360962, + "learning_rate": 2.740823707672744e-05, + "loss": 3.1342, + "step": 5113 + }, + { + "epoch": 7.49, + "grad_norm": 0.5421004295349121, + "learning_rate": 2.7375985735003552e-05, + "loss": 3.153, + "step": 5114 + }, + { + "epoch": 7.49, + "grad_norm": 1.7382439374923706, + "learning_rate": 2.7343750370076125e-05, + "loss": 3.1935, + "step": 5115 + }, + { + "epoch": 7.49, + "grad_norm": 0.757526695728302, + "learning_rate": 2.7311530989036726e-05, + "loss": 3.173, + "step": 5116 + }, + { + "epoch": 7.49, + "grad_norm": 1.094909906387329, + "learning_rate": 2.7279327598973303e-05, + "loss": 3.1654, + "step": 5117 + }, + { + "epoch": 7.49, + "grad_norm": 1.3784464597702026, + "learning_rate": 2.7247140206970534e-05, + "loss": 3.1394, + "step": 5118 + }, + { + "epoch": 7.49, + "grad_norm": 0.33335521817207336, + "learning_rate": 2.721496882010931e-05, + "loss": 3.1698, + "step": 5119 + }, + { + "epoch": 7.49, + "grad_norm": 0.5880882740020752, + "learning_rate": 2.7182813445467138e-05, + "loss": 3.1336, + "step": 5120 + }, + { + "epoch": 7.5, + "grad_norm": 1.3899345397949219, + "learning_rate": 2.7150674090118054e-05, + "loss": 3.2018, + "step": 5121 + }, + { + "epoch": 7.5, + "grad_norm": 0.4972773790359497, + "learning_rate": 2.7118550761132444e-05, + "loss": 3.174, + "step": 5122 + }, + { + "epoch": 7.5, + "grad_norm": 1.0646159648895264, + "learning_rate": 2.7086443465577195e-05, + "loss": 3.16, + "step": 5123 + }, + { + "epoch": 7.5, + "grad_norm": 0.5726083517074585, + "learning_rate": 2.705435221051581e-05, + "loss": 3.1459, + "step": 5124 + }, + { + "epoch": 7.5, + "grad_norm": 0.5148174166679382, + "learning_rate": 2.7022277003008035e-05, + "loss": 3.1702, + "step": 5125 + }, + { + "epoch": 7.5, + "grad_norm": 0.5842214226722717, + "learning_rate": 2.6990217850110222e-05, + "loss": 3.189, + "step": 5126 + }, + { + "epoch": 7.51, + "grad_norm": 0.450296550989151, + "learning_rate": 2.6958174758875253e-05, + "loss": 3.1581, + "step": 5127 + }, + { + "epoch": 7.51, + "grad_norm": 0.6981111168861389, + "learning_rate": 2.6926147736352302e-05, + "loss": 3.1709, + "step": 5128 + }, + { + "epoch": 7.51, + "grad_norm": 0.6360636949539185, + "learning_rate": 2.689413678958713e-05, + "loss": 3.163, + "step": 5129 + }, + { + "epoch": 7.51, + "grad_norm": 0.6991342306137085, + "learning_rate": 2.686214192562193e-05, + "loss": 3.1673, + "step": 5130 + }, + { + "epoch": 7.51, + "grad_norm": 0.4301992952823639, + "learning_rate": 2.6830163151495358e-05, + "loss": 3.15, + "step": 5131 + }, + { + "epoch": 7.51, + "grad_norm": 0.6416985988616943, + "learning_rate": 2.679820047424253e-05, + "loss": 3.1364, + "step": 5132 + }, + { + "epoch": 7.51, + "grad_norm": 0.6713948845863342, + "learning_rate": 2.6766253900895023e-05, + "loss": 3.1668, + "step": 5133 + }, + { + "epoch": 7.52, + "grad_norm": 0.4859183728694916, + "learning_rate": 2.6734323438480858e-05, + "loss": 3.153, + "step": 5134 + }, + { + "epoch": 7.52, + "grad_norm": 0.5950589179992676, + "learning_rate": 2.6702409094024538e-05, + "loss": 3.164, + "step": 5135 + }, + { + "epoch": 7.52, + "grad_norm": 0.4921453297138214, + "learning_rate": 2.6670510874547006e-05, + "loss": 3.1513, + "step": 5136 + }, + { + "epoch": 7.52, + "grad_norm": 0.3888862133026123, + "learning_rate": 2.6638628787065633e-05, + "loss": 3.1621, + "step": 5137 + }, + { + "epoch": 7.52, + "grad_norm": 0.43584099411964417, + "learning_rate": 2.660676283859429e-05, + "loss": 3.1327, + "step": 5138 + }, + { + "epoch": 7.52, + "grad_norm": 0.659436047077179, + "learning_rate": 2.6574913036143266e-05, + "loss": 3.1782, + "step": 5139 + }, + { + "epoch": 7.52, + "grad_norm": 0.33366990089416504, + "learning_rate": 2.6543079386719295e-05, + "loss": 3.1657, + "step": 5140 + }, + { + "epoch": 7.53, + "grad_norm": 0.6816606521606445, + "learning_rate": 2.651126189732561e-05, + "loss": 3.145, + "step": 5141 + }, + { + "epoch": 7.53, + "grad_norm": 0.5122085213661194, + "learning_rate": 2.6479460574961746e-05, + "loss": 3.1146, + "step": 5142 + }, + { + "epoch": 7.53, + "grad_norm": 0.7051358819007874, + "learning_rate": 2.6447675426623864e-05, + "loss": 3.1565, + "step": 5143 + }, + { + "epoch": 7.53, + "grad_norm": 0.3721584379673004, + "learning_rate": 2.6415906459304495e-05, + "loss": 3.165, + "step": 5144 + }, + { + "epoch": 7.53, + "grad_norm": 0.30766353011131287, + "learning_rate": 2.638415367999252e-05, + "loss": 3.1973, + "step": 5145 + }, + { + "epoch": 7.53, + "grad_norm": 0.41517648100852966, + "learning_rate": 2.63524170956734e-05, + "loss": 3.1766, + "step": 5146 + }, + { + "epoch": 7.53, + "grad_norm": 0.433137983083725, + "learning_rate": 2.6320696713328986e-05, + "loss": 3.2217, + "step": 5147 + }, + { + "epoch": 7.54, + "grad_norm": 0.3852587044239044, + "learning_rate": 2.6288992539937463e-05, + "loss": 3.1605, + "step": 5148 + }, + { + "epoch": 7.54, + "grad_norm": 0.5037500858306885, + "learning_rate": 2.625730458247362e-05, + "loss": 3.1981, + "step": 5149 + }, + { + "epoch": 7.54, + "grad_norm": 0.4805777966976166, + "learning_rate": 2.62256328479086e-05, + "loss": 3.1315, + "step": 5150 + }, + { + "epoch": 7.54, + "grad_norm": 0.7443678975105286, + "learning_rate": 2.6193977343209874e-05, + "loss": 3.173, + "step": 5151 + }, + { + "epoch": 7.54, + "grad_norm": 0.6087154746055603, + "learning_rate": 2.6162338075341564e-05, + "loss": 3.1921, + "step": 5152 + }, + { + "epoch": 7.54, + "grad_norm": 2.0148138999938965, + "learning_rate": 2.613071505126402e-05, + "loss": 3.1314, + "step": 5153 + }, + { + "epoch": 7.55, + "grad_norm": 0.5105882883071899, + "learning_rate": 2.6099108277934103e-05, + "loss": 3.1305, + "step": 5154 + }, + { + "epoch": 7.55, + "grad_norm": 0.4445195496082306, + "learning_rate": 2.6067517762305116e-05, + "loss": 3.1308, + "step": 5155 + }, + { + "epoch": 7.55, + "grad_norm": 0.3402569890022278, + "learning_rate": 2.6035943511326733e-05, + "loss": 3.1384, + "step": 5156 + }, + { + "epoch": 7.55, + "grad_norm": 0.42875614762306213, + "learning_rate": 2.6004385531945107e-05, + "loss": 3.1476, + "step": 5157 + }, + { + "epoch": 7.55, + "grad_norm": 0.46508556604385376, + "learning_rate": 2.5972843831102755e-05, + "loss": 3.1542, + "step": 5158 + }, + { + "epoch": 7.55, + "grad_norm": 0.7432370185852051, + "learning_rate": 2.594131841573866e-05, + "loss": 3.1507, + "step": 5159 + }, + { + "epoch": 7.55, + "grad_norm": 0.5330306887626648, + "learning_rate": 2.590980929278819e-05, + "loss": 3.1661, + "step": 5160 + }, + { + "epoch": 7.56, + "grad_norm": 0.37541455030441284, + "learning_rate": 2.587831646918315e-05, + "loss": 3.1454, + "step": 5161 + }, + { + "epoch": 7.56, + "grad_norm": 0.6398892402648926, + "learning_rate": 2.584683995185174e-05, + "loss": 3.1901, + "step": 5162 + }, + { + "epoch": 7.56, + "grad_norm": 0.9277903437614441, + "learning_rate": 2.5815379747718594e-05, + "loss": 3.1594, + "step": 5163 + }, + { + "epoch": 7.56, + "grad_norm": 0.8719270825386047, + "learning_rate": 2.5783935863704732e-05, + "loss": 3.1646, + "step": 5164 + }, + { + "epoch": 7.56, + "grad_norm": 0.811159074306488, + "learning_rate": 2.5752508306727608e-05, + "loss": 3.2009, + "step": 5165 + }, + { + "epoch": 7.56, + "grad_norm": 0.834298849105835, + "learning_rate": 2.5721097083701084e-05, + "loss": 3.1747, + "step": 5166 + }, + { + "epoch": 7.56, + "grad_norm": 1.0362366437911987, + "learning_rate": 2.5689702201535394e-05, + "loss": 3.1428, + "step": 5167 + }, + { + "epoch": 7.57, + "grad_norm": 0.8834418058395386, + "learning_rate": 2.5658323667137218e-05, + "loss": 3.1559, + "step": 5168 + }, + { + "epoch": 7.57, + "grad_norm": 0.4252467155456543, + "learning_rate": 2.5626961487409616e-05, + "loss": 3.1528, + "step": 5169 + }, + { + "epoch": 7.57, + "grad_norm": 1.2788466215133667, + "learning_rate": 2.559561566925206e-05, + "loss": 3.1789, + "step": 5170 + }, + { + "epoch": 7.57, + "grad_norm": 0.736139714717865, + "learning_rate": 2.556428621956043e-05, + "loss": 3.1242, + "step": 5171 + }, + { + "epoch": 7.57, + "grad_norm": 0.38194677233695984, + "learning_rate": 2.553297314522698e-05, + "loss": 3.1737, + "step": 5172 + }, + { + "epoch": 7.57, + "grad_norm": 0.5845624804496765, + "learning_rate": 2.5501676453140378e-05, + "loss": 3.1726, + "step": 5173 + }, + { + "epoch": 7.58, + "grad_norm": 0.7914636731147766, + "learning_rate": 2.5470396150185692e-05, + "loss": 3.1579, + "step": 5174 + }, + { + "epoch": 7.58, + "grad_norm": 0.39859166741371155, + "learning_rate": 2.5439132243244412e-05, + "loss": 3.1319, + "step": 5175 + }, + { + "epoch": 7.58, + "grad_norm": 1.092436671257019, + "learning_rate": 2.540788473919433e-05, + "loss": 3.1286, + "step": 5176 + }, + { + "epoch": 7.58, + "grad_norm": 0.6279719471931458, + "learning_rate": 2.5376653644909675e-05, + "loss": 3.1328, + "step": 5177 + }, + { + "epoch": 7.58, + "grad_norm": 1.2582124471664429, + "learning_rate": 2.534543896726117e-05, + "loss": 3.1486, + "step": 5178 + }, + { + "epoch": 7.58, + "grad_norm": 0.6606663465499878, + "learning_rate": 2.5314240713115756e-05, + "loss": 3.1386, + "step": 5179 + }, + { + "epoch": 7.58, + "grad_norm": 0.5657997727394104, + "learning_rate": 2.528305888933683e-05, + "loss": 3.1366, + "step": 5180 + }, + { + "epoch": 7.59, + "grad_norm": 0.7459825873374939, + "learning_rate": 2.5251893502784275e-05, + "loss": 3.1625, + "step": 5181 + }, + { + "epoch": 7.59, + "grad_norm": 0.6678188443183899, + "learning_rate": 2.522074456031418e-05, + "loss": 3.1981, + "step": 5182 + }, + { + "epoch": 7.59, + "grad_norm": 0.7135474681854248, + "learning_rate": 2.5189612068779112e-05, + "loss": 3.1958, + "step": 5183 + }, + { + "epoch": 7.59, + "grad_norm": 0.5656193494796753, + "learning_rate": 2.5158496035028077e-05, + "loss": 3.1218, + "step": 5184 + }, + { + "epoch": 7.59, + "grad_norm": 0.5109965205192566, + "learning_rate": 2.5127396465906327e-05, + "loss": 3.1465, + "step": 5185 + }, + { + "epoch": 7.59, + "grad_norm": 0.5121713876724243, + "learning_rate": 2.5096313368255585e-05, + "loss": 3.1401, + "step": 5186 + }, + { + "epoch": 7.59, + "grad_norm": 0.38621145486831665, + "learning_rate": 2.506524674891392e-05, + "loss": 3.1519, + "step": 5187 + }, + { + "epoch": 7.6, + "grad_norm": 0.557099461555481, + "learning_rate": 2.5034196614715778e-05, + "loss": 3.1632, + "step": 5188 + }, + { + "epoch": 7.6, + "grad_norm": 0.42920613288879395, + "learning_rate": 2.500316297249198e-05, + "loss": 3.1272, + "step": 5189 + }, + { + "epoch": 7.6, + "grad_norm": 0.765871524810791, + "learning_rate": 2.4972145829069714e-05, + "loss": 3.1568, + "step": 5190 + }, + { + "epoch": 7.6, + "grad_norm": 0.4879997670650482, + "learning_rate": 2.4941145191272563e-05, + "loss": 3.174, + "step": 5191 + }, + { + "epoch": 7.6, + "grad_norm": 0.5009108781814575, + "learning_rate": 2.4910161065920445e-05, + "loss": 3.1467, + "step": 5192 + }, + { + "epoch": 7.6, + "grad_norm": 0.6382935643196106, + "learning_rate": 2.4879193459829665e-05, + "loss": 3.1552, + "step": 5193 + }, + { + "epoch": 7.6, + "grad_norm": 0.57361900806427, + "learning_rate": 2.484824237981288e-05, + "loss": 3.1511, + "step": 5194 + }, + { + "epoch": 7.61, + "grad_norm": 0.2623114287853241, + "learning_rate": 2.4817307832679126e-05, + "loss": 3.1664, + "step": 5195 + }, + { + "epoch": 7.61, + "grad_norm": 0.5569726228713989, + "learning_rate": 2.4786389825233802e-05, + "loss": 3.1431, + "step": 5196 + }, + { + "epoch": 7.61, + "grad_norm": 0.5350569486618042, + "learning_rate": 2.4755488364278656e-05, + "loss": 3.173, + "step": 5197 + }, + { + "epoch": 7.61, + "grad_norm": 0.38636651635169983, + "learning_rate": 2.4724603456611805e-05, + "loss": 3.1996, + "step": 5198 + }, + { + "epoch": 7.61, + "grad_norm": 0.6508245468139648, + "learning_rate": 2.4693735109027716e-05, + "loss": 3.1693, + "step": 5199 + }, + { + "epoch": 7.61, + "grad_norm": 0.7632821798324585, + "learning_rate": 2.4662883328317222e-05, + "loss": 3.1948, + "step": 5200 + }, + { + "epoch": 7.62, + "grad_norm": 0.5081233978271484, + "learning_rate": 2.4632048121267547e-05, + "loss": 3.1938, + "step": 5201 + }, + { + "epoch": 7.62, + "grad_norm": 0.7682433128356934, + "learning_rate": 2.4601229494662125e-05, + "loss": 3.1553, + "step": 5202 + }, + { + "epoch": 7.62, + "grad_norm": 1.0782290697097778, + "learning_rate": 2.4570427455280943e-05, + "loss": 3.1649, + "step": 5203 + }, + { + "epoch": 7.62, + "grad_norm": 0.6353168487548828, + "learning_rate": 2.4539642009900253e-05, + "loss": 3.1517, + "step": 5204 + }, + { + "epoch": 7.62, + "grad_norm": 0.7350844144821167, + "learning_rate": 2.4508873165292535e-05, + "loss": 3.1375, + "step": 5205 + }, + { + "epoch": 7.62, + "grad_norm": 0.9207132458686829, + "learning_rate": 2.4478120928226846e-05, + "loss": 3.1478, + "step": 5206 + }, + { + "epoch": 7.62, + "grad_norm": 0.7946307063102722, + "learning_rate": 2.444738530546845e-05, + "loss": 3.1392, + "step": 5207 + }, + { + "epoch": 7.63, + "grad_norm": 0.9071816802024841, + "learning_rate": 2.441666630377889e-05, + "loss": 3.1275, + "step": 5208 + }, + { + "epoch": 7.63, + "grad_norm": 1.2288057804107666, + "learning_rate": 2.438596392991628e-05, + "loss": 3.1333, + "step": 5209 + }, + { + "epoch": 7.63, + "grad_norm": 1.0466946363449097, + "learning_rate": 2.435527819063482e-05, + "loss": 3.143, + "step": 5210 + }, + { + "epoch": 7.63, + "grad_norm": 0.5097655057907104, + "learning_rate": 2.432460909268518e-05, + "loss": 3.2069, + "step": 5211 + }, + { + "epoch": 7.63, + "grad_norm": 0.8036677241325378, + "learning_rate": 2.4293956642814453e-05, + "loss": 3.1592, + "step": 5212 + }, + { + "epoch": 7.63, + "grad_norm": 3.4330554008483887, + "learning_rate": 2.4263320847765868e-05, + "loss": 3.1551, + "step": 5213 + }, + { + "epoch": 7.63, + "grad_norm": 0.9436863660812378, + "learning_rate": 2.4232701714279115e-05, + "loss": 3.1353, + "step": 5214 + }, + { + "epoch": 7.64, + "grad_norm": 0.30926942825317383, + "learning_rate": 2.420209924909026e-05, + "loss": 3.1577, + "step": 5215 + }, + { + "epoch": 7.64, + "grad_norm": 0.7689909338951111, + "learning_rate": 2.417151345893157e-05, + "loss": 3.1326, + "step": 5216 + }, + { + "epoch": 7.64, + "grad_norm": 0.8262006640434265, + "learning_rate": 2.4140944350531712e-05, + "loss": 3.1372, + "step": 5217 + }, + { + "epoch": 7.64, + "grad_norm": 0.35750913619995117, + "learning_rate": 2.411039193061577e-05, + "loss": 3.1526, + "step": 5218 + }, + { + "epoch": 7.64, + "grad_norm": 0.7210152745246887, + "learning_rate": 2.407985620590498e-05, + "loss": 3.1702, + "step": 5219 + }, + { + "epoch": 7.64, + "grad_norm": 0.37686386704444885, + "learning_rate": 2.404933718311704e-05, + "loss": 3.1609, + "step": 5220 + }, + { + "epoch": 7.64, + "grad_norm": 0.38531693816185, + "learning_rate": 2.401883486896591e-05, + "loss": 3.16, + "step": 5221 + }, + { + "epoch": 7.65, + "grad_norm": 3.7467172145843506, + "learning_rate": 2.3988349270161913e-05, + "loss": 3.1974, + "step": 5222 + }, + { + "epoch": 7.65, + "grad_norm": 0.4558284878730774, + "learning_rate": 2.3957880393411667e-05, + "loss": 3.1724, + "step": 5223 + }, + { + "epoch": 7.65, + "grad_norm": 0.5120177268981934, + "learning_rate": 2.3927428245418127e-05, + "loss": 3.1466, + "step": 5224 + }, + { + "epoch": 7.65, + "grad_norm": 0.47643178701400757, + "learning_rate": 2.3896992832880548e-05, + "loss": 3.1508, + "step": 5225 + }, + { + "epoch": 7.65, + "grad_norm": 0.333384245634079, + "learning_rate": 2.3866574162494538e-05, + "loss": 3.1346, + "step": 5226 + }, + { + "epoch": 7.65, + "grad_norm": 0.5989437699317932, + "learning_rate": 2.383617224095198e-05, + "loss": 3.1354, + "step": 5227 + }, + { + "epoch": 7.66, + "grad_norm": 0.35983359813690186, + "learning_rate": 2.38057870749411e-05, + "loss": 3.1599, + "step": 5228 + }, + { + "epoch": 7.66, + "grad_norm": 0.3712514042854309, + "learning_rate": 2.3775418671146443e-05, + "loss": 3.1584, + "step": 5229 + }, + { + "epoch": 7.66, + "grad_norm": 0.3716100752353668, + "learning_rate": 2.3745067036248836e-05, + "loss": 3.1766, + "step": 5230 + }, + { + "epoch": 7.66, + "grad_norm": 0.27831578254699707, + "learning_rate": 2.3714732176925446e-05, + "loss": 3.1285, + "step": 5231 + }, + { + "epoch": 7.66, + "grad_norm": 0.47703278064727783, + "learning_rate": 2.368441409984976e-05, + "loss": 3.1467, + "step": 5232 + }, + { + "epoch": 7.66, + "grad_norm": 0.577670156955719, + "learning_rate": 2.3654112811691477e-05, + "loss": 3.1222, + "step": 5233 + }, + { + "epoch": 7.66, + "grad_norm": 0.44678911566734314, + "learning_rate": 2.3623828319116748e-05, + "loss": 3.1783, + "step": 5234 + }, + { + "epoch": 7.67, + "grad_norm": 0.9070417881011963, + "learning_rate": 2.359356062878797e-05, + "loss": 3.1324, + "step": 5235 + }, + { + "epoch": 7.67, + "grad_norm": 0.5070209503173828, + "learning_rate": 2.356330974736375e-05, + "loss": 3.129, + "step": 5236 + }, + { + "epoch": 7.67, + "grad_norm": 0.6462606191635132, + "learning_rate": 2.353307568149915e-05, + "loss": 3.1474, + "step": 5237 + }, + { + "epoch": 7.67, + "grad_norm": 0.34412530064582825, + "learning_rate": 2.3502858437845476e-05, + "loss": 3.1597, + "step": 5238 + }, + { + "epoch": 7.67, + "grad_norm": 0.4765468239784241, + "learning_rate": 2.347265802305022e-05, + "loss": 3.1474, + "step": 5239 + }, + { + "epoch": 7.67, + "eval_loss": 3.013108015060425, + "eval_runtime": 1170.7492, + "eval_samples_per_second": 14.642, + "eval_steps_per_second": 0.916, + "step": 5239 + }, + { + "epoch": 7.67, + "grad_norm": 0.5959687232971191, + "learning_rate": 2.3442474443757366e-05, + "loss": 3.1172, + "step": 5240 + }, + { + "epoch": 7.67, + "grad_norm": 0.27279970049858093, + "learning_rate": 2.3412307706607108e-05, + "loss": 3.1737, + "step": 5241 + }, + { + "epoch": 7.68, + "grad_norm": 0.5665635466575623, + "learning_rate": 2.3382157818235816e-05, + "loss": 3.1308, + "step": 5242 + }, + { + "epoch": 7.68, + "grad_norm": 0.3132018446922302, + "learning_rate": 2.3352024785276404e-05, + "loss": 3.1657, + "step": 5243 + }, + { + "epoch": 7.68, + "grad_norm": 0.622239351272583, + "learning_rate": 2.3321908614357824e-05, + "loss": 3.1689, + "step": 5244 + }, + { + "epoch": 7.68, + "grad_norm": 0.5310043692588806, + "learning_rate": 2.3291809312105462e-05, + "loss": 3.2016, + "step": 5245 + }, + { + "epoch": 7.68, + "grad_norm": 0.3280445337295532, + "learning_rate": 2.3261726885141023e-05, + "loss": 3.1654, + "step": 5246 + }, + { + "epoch": 7.68, + "grad_norm": 0.4684401750564575, + "learning_rate": 2.3231661340082366e-05, + "loss": 3.1932, + "step": 5247 + }, + { + "epoch": 7.68, + "grad_norm": 0.5372167825698853, + "learning_rate": 2.3201612683543704e-05, + "loss": 3.2027, + "step": 5248 + }, + { + "epoch": 7.69, + "grad_norm": 0.865881621837616, + "learning_rate": 2.3171580922135627e-05, + "loss": 3.1653, + "step": 5249 + }, + { + "epoch": 7.69, + "grad_norm": 0.650731086730957, + "learning_rate": 2.3141566062464847e-05, + "loss": 3.1495, + "step": 5250 + }, + { + "epoch": 7.69, + "grad_norm": 1.0147230625152588, + "learning_rate": 2.311156811113444e-05, + "loss": 3.1422, + "step": 5251 + }, + { + "epoch": 7.69, + "grad_norm": 0.34079644083976746, + "learning_rate": 2.308158707474377e-05, + "loss": 3.125, + "step": 5252 + }, + { + "epoch": 7.69, + "grad_norm": 0.8186140656471252, + "learning_rate": 2.3051622959888464e-05, + "loss": 3.1991, + "step": 5253 + }, + { + "epoch": 7.69, + "grad_norm": 0.8478550314903259, + "learning_rate": 2.3021675773160424e-05, + "loss": 3.1535, + "step": 5254 + }, + { + "epoch": 7.7, + "grad_norm": 0.6345941424369812, + "learning_rate": 2.299174552114781e-05, + "loss": 3.1704, + "step": 5255 + }, + { + "epoch": 7.7, + "grad_norm": 0.9661188721656799, + "learning_rate": 2.296183221043511e-05, + "loss": 3.1492, + "step": 5256 + }, + { + "epoch": 7.7, + "grad_norm": 0.4356244206428528, + "learning_rate": 2.2931935847603037e-05, + "loss": 3.159, + "step": 5257 + }, + { + "epoch": 7.7, + "grad_norm": 0.6363608837127686, + "learning_rate": 2.290205643922859e-05, + "loss": 3.1579, + "step": 5258 + }, + { + "epoch": 7.7, + "grad_norm": 1.1210684776306152, + "learning_rate": 2.2872193991885026e-05, + "loss": 3.1412, + "step": 5259 + }, + { + "epoch": 7.7, + "grad_norm": 0.38606759905815125, + "learning_rate": 2.2842348512141908e-05, + "loss": 3.1999, + "step": 5260 + }, + { + "epoch": 7.7, + "grad_norm": 1.1856557130813599, + "learning_rate": 2.281252000656503e-05, + "loss": 3.1595, + "step": 5261 + }, + { + "epoch": 7.71, + "grad_norm": 1.063137412071228, + "learning_rate": 2.2782708481716463e-05, + "loss": 3.1713, + "step": 5262 + }, + { + "epoch": 7.71, + "grad_norm": 0.6546114087104797, + "learning_rate": 2.275291394415454e-05, + "loss": 3.1581, + "step": 5263 + }, + { + "epoch": 7.71, + "grad_norm": 1.1830710172653198, + "learning_rate": 2.2723136400433865e-05, + "loss": 3.1447, + "step": 5264 + }, + { + "epoch": 7.71, + "grad_norm": 0.6315742135047913, + "learning_rate": 2.269337585710529e-05, + "loss": 3.1487, + "step": 5265 + }, + { + "epoch": 7.71, + "grad_norm": 0.8535947799682617, + "learning_rate": 2.2663632320715976e-05, + "loss": 3.1631, + "step": 5266 + }, + { + "epoch": 7.71, + "grad_norm": 1.0756810903549194, + "learning_rate": 2.2633905797809197e-05, + "loss": 3.1389, + "step": 5267 + }, + { + "epoch": 7.71, + "grad_norm": 0.7916353344917297, + "learning_rate": 2.2604196294924694e-05, + "loss": 3.13, + "step": 5268 + }, + { + "epoch": 7.72, + "grad_norm": 0.46646764874458313, + "learning_rate": 2.257450381859836e-05, + "loss": 3.1633, + "step": 5269 + }, + { + "epoch": 7.72, + "grad_norm": 0.6788039803504944, + "learning_rate": 2.254482837536224e-05, + "loss": 3.1178, + "step": 5270 + }, + { + "epoch": 7.72, + "grad_norm": 0.5274389982223511, + "learning_rate": 2.251516997174482e-05, + "loss": 3.1824, + "step": 5271 + }, + { + "epoch": 7.72, + "grad_norm": 0.5712849497795105, + "learning_rate": 2.2485528614270757e-05, + "loss": 3.1324, + "step": 5272 + }, + { + "epoch": 7.72, + "grad_norm": 0.39241471886634827, + "learning_rate": 2.2455904309460886e-05, + "loss": 3.1527, + "step": 5273 + }, + { + "epoch": 7.72, + "grad_norm": 0.3917776346206665, + "learning_rate": 2.242629706383237e-05, + "loss": 3.1554, + "step": 5274 + }, + { + "epoch": 7.73, + "grad_norm": 0.44407832622528076, + "learning_rate": 2.2396706883898676e-05, + "loss": 3.1729, + "step": 5275 + }, + { + "epoch": 7.73, + "grad_norm": 0.6070168614387512, + "learning_rate": 2.236713377616936e-05, + "loss": 3.1218, + "step": 5276 + }, + { + "epoch": 7.73, + "grad_norm": 0.43469035625457764, + "learning_rate": 2.2337577747150353e-05, + "loss": 3.1484, + "step": 5277 + }, + { + "epoch": 7.73, + "grad_norm": 0.3631550967693329, + "learning_rate": 2.2308038803343755e-05, + "loss": 3.1192, + "step": 5278 + }, + { + "epoch": 7.73, + "grad_norm": 0.45660415291786194, + "learning_rate": 2.227851695124796e-05, + "loss": 3.1704, + "step": 5279 + }, + { + "epoch": 7.73, + "grad_norm": 0.3175928592681885, + "learning_rate": 2.224901219735758e-05, + "loss": 3.1534, + "step": 5280 + }, + { + "epoch": 7.73, + "grad_norm": 0.6207453608512878, + "learning_rate": 2.221952454816344e-05, + "loss": 3.191, + "step": 5281 + }, + { + "epoch": 7.74, + "grad_norm": 0.5500271320343018, + "learning_rate": 2.219005401015264e-05, + "loss": 3.1639, + "step": 5282 + }, + { + "epoch": 7.74, + "grad_norm": 0.5747382044792175, + "learning_rate": 2.21606005898085e-05, + "loss": 3.132, + "step": 5283 + }, + { + "epoch": 7.74, + "grad_norm": 0.26599442958831787, + "learning_rate": 2.2131164293610572e-05, + "loss": 3.1491, + "step": 5284 + }, + { + "epoch": 7.74, + "grad_norm": 0.46995365619659424, + "learning_rate": 2.2101745128034656e-05, + "loss": 3.1383, + "step": 5285 + }, + { + "epoch": 7.74, + "grad_norm": 0.23462985455989838, + "learning_rate": 2.207234309955275e-05, + "loss": 3.1604, + "step": 5286 + }, + { + "epoch": 7.74, + "grad_norm": 0.45813092589378357, + "learning_rate": 2.2042958214633115e-05, + "loss": 3.1589, + "step": 5287 + }, + { + "epoch": 7.74, + "grad_norm": 0.48195019364356995, + "learning_rate": 2.2013590479740232e-05, + "loss": 3.1074, + "step": 5288 + }, + { + "epoch": 7.75, + "grad_norm": 0.9091559052467346, + "learning_rate": 2.1984239901334802e-05, + "loss": 3.1518, + "step": 5289 + }, + { + "epoch": 7.75, + "grad_norm": 0.9648109078407288, + "learning_rate": 2.195490648587375e-05, + "loss": 3.2077, + "step": 5290 + }, + { + "epoch": 7.75, + "grad_norm": 0.6132333874702454, + "learning_rate": 2.1925590239810235e-05, + "loss": 3.1481, + "step": 5291 + }, + { + "epoch": 7.75, + "grad_norm": 0.4865446984767914, + "learning_rate": 2.1896291169593643e-05, + "loss": 3.1402, + "step": 5292 + }, + { + "epoch": 7.75, + "grad_norm": 0.561164140701294, + "learning_rate": 2.186700928166956e-05, + "loss": 3.1597, + "step": 5293 + }, + { + "epoch": 7.75, + "grad_norm": 0.46440842747688293, + "learning_rate": 2.1837744582479814e-05, + "loss": 3.1493, + "step": 5294 + }, + { + "epoch": 7.75, + "grad_norm": 0.5474674701690674, + "learning_rate": 2.1808497078462442e-05, + "loss": 3.1622, + "step": 5295 + }, + { + "epoch": 7.76, + "grad_norm": 0.304200679063797, + "learning_rate": 2.1779266776051688e-05, + "loss": 3.1756, + "step": 5296 + }, + { + "epoch": 7.76, + "grad_norm": 0.534535825252533, + "learning_rate": 2.175005368167804e-05, + "loss": 3.2078, + "step": 5297 + }, + { + "epoch": 7.76, + "grad_norm": 0.7717536687850952, + "learning_rate": 2.17208578017682e-05, + "loss": 3.1603, + "step": 5298 + }, + { + "epoch": 7.76, + "grad_norm": 0.8810925483703613, + "learning_rate": 2.169167914274498e-05, + "loss": 3.1335, + "step": 5299 + }, + { + "epoch": 7.76, + "grad_norm": 0.9111099243164062, + "learning_rate": 2.166251771102761e-05, + "loss": 3.1247, + "step": 5300 + }, + { + "epoch": 7.76, + "grad_norm": 0.5138559937477112, + "learning_rate": 2.163337351303132e-05, + "loss": 3.183, + "step": 5301 + }, + { + "epoch": 7.77, + "grad_norm": 0.5875454545021057, + "learning_rate": 2.1604246555167638e-05, + "loss": 3.1424, + "step": 5302 + }, + { + "epoch": 7.77, + "grad_norm": 0.35727164149284363, + "learning_rate": 2.157513684384439e-05, + "loss": 3.1641, + "step": 5303 + }, + { + "epoch": 7.77, + "grad_norm": 0.720236599445343, + "learning_rate": 2.1546044385465426e-05, + "loss": 3.1459, + "step": 5304 + }, + { + "epoch": 7.77, + "grad_norm": 0.49461638927459717, + "learning_rate": 2.151696918643089e-05, + "loss": 3.1613, + "step": 5305 + }, + { + "epoch": 7.77, + "grad_norm": 0.5594753623008728, + "learning_rate": 2.1487911253137217e-05, + "loss": 3.1251, + "step": 5306 + }, + { + "epoch": 7.77, + "grad_norm": 1.0345263481140137, + "learning_rate": 2.1458870591976877e-05, + "loss": 3.1507, + "step": 5307 + }, + { + "epoch": 7.77, + "grad_norm": 0.31530436873435974, + "learning_rate": 2.1429847209338628e-05, + "loss": 3.1615, + "step": 5308 + }, + { + "epoch": 7.78, + "grad_norm": 0.3019442558288574, + "learning_rate": 2.1400841111607483e-05, + "loss": 3.1626, + "step": 5309 + }, + { + "epoch": 7.78, + "grad_norm": 0.40825822949409485, + "learning_rate": 2.137185230516453e-05, + "loss": 3.1289, + "step": 5310 + }, + { + "epoch": 7.78, + "grad_norm": 0.4319787323474884, + "learning_rate": 2.134288079638712e-05, + "loss": 3.1428, + "step": 5311 + }, + { + "epoch": 7.78, + "grad_norm": 0.5881616473197937, + "learning_rate": 2.131392659164879e-05, + "loss": 3.1675, + "step": 5312 + }, + { + "epoch": 7.78, + "grad_norm": 0.4166703224182129, + "learning_rate": 2.128498969731929e-05, + "loss": 3.1611, + "step": 5313 + }, + { + "epoch": 7.78, + "grad_norm": 0.7921746373176575, + "learning_rate": 2.125607011976454e-05, + "loss": 3.1281, + "step": 5314 + }, + { + "epoch": 7.78, + "grad_norm": 0.454723060131073, + "learning_rate": 2.1227167865346632e-05, + "loss": 3.1568, + "step": 5315 + }, + { + "epoch": 7.79, + "grad_norm": 0.49115943908691406, + "learning_rate": 2.1198282940423887e-05, + "loss": 3.1556, + "step": 5316 + }, + { + "epoch": 7.79, + "grad_norm": 0.4243347644805908, + "learning_rate": 2.1169415351350786e-05, + "loss": 3.1701, + "step": 5317 + }, + { + "epoch": 7.79, + "grad_norm": 0.9867798089981079, + "learning_rate": 2.1140565104478004e-05, + "loss": 3.1729, + "step": 5318 + }, + { + "epoch": 7.79, + "grad_norm": 1.0699418783187866, + "learning_rate": 2.1111732206152424e-05, + "loss": 3.1711, + "step": 5319 + }, + { + "epoch": 7.79, + "grad_norm": 0.7602143287658691, + "learning_rate": 2.1082916662717055e-05, + "loss": 3.1824, + "step": 5320 + }, + { + "epoch": 7.79, + "grad_norm": 0.7598912715911865, + "learning_rate": 2.1054118480511154e-05, + "loss": 3.1232, + "step": 5321 + }, + { + "epoch": 7.79, + "grad_norm": 0.7849280834197998, + "learning_rate": 2.1025337665870104e-05, + "loss": 3.1169, + "step": 5322 + }, + { + "epoch": 7.8, + "grad_norm": 0.5915687084197998, + "learning_rate": 2.0996574225125532e-05, + "loss": 3.1269, + "step": 5323 + }, + { + "epoch": 7.8, + "grad_norm": 0.9598039984703064, + "learning_rate": 2.096782816460513e-05, + "loss": 3.1504, + "step": 5324 + }, + { + "epoch": 7.8, + "grad_norm": 0.6661960482597351, + "learning_rate": 2.0939099490632896e-05, + "loss": 3.1284, + "step": 5325 + }, + { + "epoch": 7.8, + "grad_norm": 0.4720488488674164, + "learning_rate": 2.0910388209528964e-05, + "loss": 3.1584, + "step": 5326 + }, + { + "epoch": 7.8, + "grad_norm": 0.6807810664176941, + "learning_rate": 2.0881694327609536e-05, + "loss": 3.1586, + "step": 5327 + }, + { + "epoch": 7.8, + "grad_norm": 0.7132989764213562, + "learning_rate": 2.085301785118716e-05, + "loss": 3.1526, + "step": 5328 + }, + { + "epoch": 7.81, + "grad_norm": 0.4347662031650543, + "learning_rate": 2.0824358786570453e-05, + "loss": 3.1645, + "step": 5329 + }, + { + "epoch": 7.81, + "grad_norm": 1.1308764219284058, + "learning_rate": 2.0795717140064163e-05, + "loss": 3.162, + "step": 5330 + }, + { + "epoch": 7.81, + "grad_norm": 0.759826123714447, + "learning_rate": 2.076709291796931e-05, + "loss": 3.184, + "step": 5331 + }, + { + "epoch": 7.81, + "grad_norm": 0.6993570923805237, + "learning_rate": 2.0738486126583056e-05, + "loss": 3.1636, + "step": 5332 + }, + { + "epoch": 7.81, + "grad_norm": 0.5368728637695312, + "learning_rate": 2.0709896772198612e-05, + "loss": 3.1551, + "step": 5333 + }, + { + "epoch": 7.81, + "grad_norm": 0.5664252042770386, + "learning_rate": 2.0681324861105556e-05, + "loss": 3.1349, + "step": 5334 + }, + { + "epoch": 7.81, + "grad_norm": 0.6612221598625183, + "learning_rate": 2.0652770399589427e-05, + "loss": 3.1495, + "step": 5335 + }, + { + "epoch": 7.82, + "grad_norm": 0.48035261034965515, + "learning_rate": 2.0624233393932024e-05, + "loss": 3.1624, + "step": 5336 + }, + { + "epoch": 7.82, + "grad_norm": 0.36596471071243286, + "learning_rate": 2.059571385041138e-05, + "loss": 3.1674, + "step": 5337 + }, + { + "epoch": 7.82, + "grad_norm": 0.6980510950088501, + "learning_rate": 2.056721177530151e-05, + "loss": 3.1126, + "step": 5338 + }, + { + "epoch": 7.82, + "grad_norm": 0.567153811454773, + "learning_rate": 2.0538727174872684e-05, + "loss": 3.1449, + "step": 5339 + }, + { + "epoch": 7.82, + "grad_norm": 0.6353252530097961, + "learning_rate": 2.0510260055391418e-05, + "loss": 3.1646, + "step": 5340 + }, + { + "epoch": 7.82, + "grad_norm": 0.6695141792297363, + "learning_rate": 2.0481810423120196e-05, + "loss": 3.1426, + "step": 5341 + }, + { + "epoch": 7.82, + "grad_norm": 1.0574151277542114, + "learning_rate": 2.045337828431775e-05, + "loss": 3.1435, + "step": 5342 + }, + { + "epoch": 7.83, + "grad_norm": 0.3314790725708008, + "learning_rate": 2.0424963645239027e-05, + "loss": 3.1263, + "step": 5343 + }, + { + "epoch": 7.83, + "grad_norm": 0.708838939666748, + "learning_rate": 2.0396566512134996e-05, + "loss": 3.178, + "step": 5344 + }, + { + "epoch": 7.83, + "grad_norm": 0.670356810092926, + "learning_rate": 2.036818689125284e-05, + "loss": 3.1153, + "step": 5345 + }, + { + "epoch": 7.83, + "grad_norm": 0.6451178193092346, + "learning_rate": 2.0339824788835916e-05, + "loss": 3.1816, + "step": 5346 + }, + { + "epoch": 7.83, + "grad_norm": 0.5722768306732178, + "learning_rate": 2.031148021112367e-05, + "loss": 3.1584, + "step": 5347 + }, + { + "epoch": 7.83, + "grad_norm": 0.8647962808609009, + "learning_rate": 2.028315316435173e-05, + "loss": 3.118, + "step": 5348 + }, + { + "epoch": 7.83, + "grad_norm": 0.5145653486251831, + "learning_rate": 2.0254843654751852e-05, + "loss": 3.1143, + "step": 5349 + }, + { + "epoch": 7.84, + "grad_norm": 0.7415019273757935, + "learning_rate": 2.0226551688551954e-05, + "loss": 3.1961, + "step": 5350 + }, + { + "epoch": 7.84, + "grad_norm": 0.6742720603942871, + "learning_rate": 2.0198277271976052e-05, + "loss": 3.107, + "step": 5351 + }, + { + "epoch": 7.84, + "grad_norm": 0.5243716835975647, + "learning_rate": 2.017002041124435e-05, + "loss": 3.1568, + "step": 5352 + }, + { + "epoch": 7.84, + "grad_norm": 0.6386037468910217, + "learning_rate": 2.0141781112573154e-05, + "loss": 3.1501, + "step": 5353 + }, + { + "epoch": 7.84, + "grad_norm": 0.9167563915252686, + "learning_rate": 2.011355938217496e-05, + "loss": 3.1591, + "step": 5354 + }, + { + "epoch": 7.84, + "grad_norm": 0.3091890811920166, + "learning_rate": 2.0085355226258285e-05, + "loss": 3.1714, + "step": 5355 + }, + { + "epoch": 7.85, + "grad_norm": 0.6145365834236145, + "learning_rate": 2.00571686510279e-05, + "loss": 3.1247, + "step": 5356 + }, + { + "epoch": 7.85, + "grad_norm": 0.3485010862350464, + "learning_rate": 2.00289996626847e-05, + "loss": 3.1696, + "step": 5357 + }, + { + "epoch": 7.85, + "grad_norm": 0.5586562156677246, + "learning_rate": 2.0000848267425575e-05, + "loss": 3.1249, + "step": 5358 + }, + { + "epoch": 7.85, + "grad_norm": 0.45884886384010315, + "learning_rate": 1.9972714471443733e-05, + "loss": 3.1264, + "step": 5359 + }, + { + "epoch": 7.85, + "grad_norm": 0.3801732361316681, + "learning_rate": 1.9944598280928408e-05, + "loss": 3.1699, + "step": 5360 + }, + { + "epoch": 7.85, + "grad_norm": 0.5741209983825684, + "learning_rate": 1.9916499702064894e-05, + "loss": 3.1625, + "step": 5361 + }, + { + "epoch": 7.85, + "grad_norm": 0.23692893981933594, + "learning_rate": 1.9888418741034787e-05, + "loss": 3.1282, + "step": 5362 + }, + { + "epoch": 7.86, + "grad_norm": 0.23494169116020203, + "learning_rate": 1.9860355404015696e-05, + "loss": 3.1757, + "step": 5363 + }, + { + "epoch": 7.86, + "grad_norm": 0.5047478675842285, + "learning_rate": 1.9832309697181294e-05, + "loss": 3.1371, + "step": 5364 + }, + { + "epoch": 7.86, + "grad_norm": 0.39867815375328064, + "learning_rate": 1.9804281626701525e-05, + "loss": 3.1522, + "step": 5365 + }, + { + "epoch": 7.86, + "grad_norm": 0.4054205119609833, + "learning_rate": 1.977627119874237e-05, + "loss": 3.1453, + "step": 5366 + }, + { + "epoch": 7.86, + "grad_norm": 0.37046530842781067, + "learning_rate": 1.9748278419465872e-05, + "loss": 3.1295, + "step": 5367 + }, + { + "epoch": 7.86, + "grad_norm": 0.4440865218639374, + "learning_rate": 1.972030329503035e-05, + "loss": 3.1436, + "step": 5368 + }, + { + "epoch": 7.86, + "grad_norm": 0.25685054063796997, + "learning_rate": 1.9692345831590064e-05, + "loss": 3.147, + "step": 5369 + }, + { + "epoch": 7.87, + "grad_norm": 0.423784077167511, + "learning_rate": 1.966440603529549e-05, + "loss": 3.161, + "step": 5370 + }, + { + "epoch": 7.87, + "grad_norm": 0.407570481300354, + "learning_rate": 1.963648391229321e-05, + "loss": 3.1711, + "step": 5371 + }, + { + "epoch": 7.87, + "grad_norm": 1.7502979040145874, + "learning_rate": 1.9608579468725895e-05, + "loss": 3.1717, + "step": 5372 + }, + { + "epoch": 7.87, + "grad_norm": 0.37750712037086487, + "learning_rate": 1.958069271073234e-05, + "loss": 3.1518, + "step": 5373 + }, + { + "epoch": 7.87, + "grad_norm": 0.44295188784599304, + "learning_rate": 1.9552823644447437e-05, + "loss": 3.1532, + "step": 5374 + }, + { + "epoch": 7.87, + "grad_norm": 0.6612734794616699, + "learning_rate": 1.9524972276002207e-05, + "loss": 3.1507, + "step": 5375 + }, + { + "epoch": 7.88, + "grad_norm": 0.5701128244400024, + "learning_rate": 1.9497138611523748e-05, + "loss": 3.1726, + "step": 5376 + }, + { + "epoch": 7.88, + "grad_norm": 0.48842257261276245, + "learning_rate": 1.9469322657135293e-05, + "loss": 3.1591, + "step": 5377 + }, + { + "epoch": 7.88, + "grad_norm": 0.40046364068984985, + "learning_rate": 1.944152441895616e-05, + "loss": 3.1994, + "step": 5378 + }, + { + "epoch": 7.88, + "grad_norm": 0.62181556224823, + "learning_rate": 1.9413743903101776e-05, + "loss": 3.1446, + "step": 5379 + }, + { + "epoch": 7.88, + "grad_norm": 0.7460499405860901, + "learning_rate": 1.9385981115683672e-05, + "loss": 3.1688, + "step": 5380 + }, + { + "epoch": 7.88, + "grad_norm": 0.2939889132976532, + "learning_rate": 1.9358236062809464e-05, + "loss": 3.1436, + "step": 5381 + }, + { + "epoch": 7.88, + "grad_norm": 0.6719468832015991, + "learning_rate": 1.9330508750582897e-05, + "loss": 3.1482, + "step": 5382 + }, + { + "epoch": 7.89, + "grad_norm": 0.41709572076797485, + "learning_rate": 1.930279918510377e-05, + "loss": 3.1564, + "step": 5383 + }, + { + "epoch": 7.89, + "grad_norm": 0.5399413704872131, + "learning_rate": 1.9275107372468014e-05, + "loss": 3.1684, + "step": 5384 + } + ], + "logging_steps": 1, + "max_steps": 6730, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "save_steps": 673, + "total_flos": 1.3297358716246216e+20, + "train_batch_size": 2, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-5384/training_args.bin b/checkpoint-5384/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..c9318904b092ceb1505e2e8f2b5300dd991baa7f --- /dev/null +++ b/checkpoint-5384/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a4ab9da40c5b7630b88beccc23069550f831ccb9b69b82772516407dd51150c +size 5688 diff --git a/checkpoint-6057/README.md b/checkpoint-6057/README.md new file mode 100644 index 0000000000000000000000000000000000000000..dbeb1c7761226042e1d72de6378a4fa69fdb3f06 --- /dev/null +++ b/checkpoint-6057/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: unsloth/gemma-7b +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.9.0 \ No newline at end of file diff --git a/checkpoint-6057/adapter_config.json b/checkpoint-6057/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..35971161ab187cbdc8af6c44079edb2d2b677921 --- /dev/null +++ b/checkpoint-6057/adapter_config.json @@ -0,0 +1,33 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "unsloth/gemma-7b", + "bias": "none", + "fan_in_fan_out": null, + "inference_mode": true, + "init_lora_weights": true, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 16, + "lora_dropout": 0.05, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 32, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "v_proj", + "o_proj", + "down_proj", + "k_proj", + "up_proj", + "gate_proj", + "q_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": false +} \ No newline at end of file diff --git a/checkpoint-6057/adapter_model.safetensors b/checkpoint-6057/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e3662b8cd825c88450f1a30942b00c1408a90b90 --- /dev/null +++ b/checkpoint-6057/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6f850e80a279531695cf2bb0073b9840bfe70aee595c14e4af88fe1c938a44b5 +size 400084608 diff --git a/checkpoint-6057/optimizer.pt b/checkpoint-6057/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..f581208e8d0eaa225400db9ad393fd243362830c --- /dev/null +++ b/checkpoint-6057/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e127e6f9dc3700bbcbc2fcbef7d836396e2d3cc9d2517caa08105229a7a1d4a +size 200866804 diff --git a/checkpoint-6057/rng_state_0.pth b/checkpoint-6057/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..e78935104d31337a2f1c9d9d7ef4e49d22691d2c --- /dev/null +++ b/checkpoint-6057/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3739cfeee2cd515bdc6eadb44e33550e1a754ff4108a1b5dd749f499a570f87f +size 15984 diff --git a/checkpoint-6057/rng_state_1.pth b/checkpoint-6057/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..c995f6bfbcacf6e143749e970f2619b0432dd484 --- /dev/null +++ b/checkpoint-6057/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:91c57c8c49005b92e268f85b4105d3cecfb85819e93f96eed5e56bc9495df5bc +size 15984 diff --git a/checkpoint-6057/rng_state_2.pth b/checkpoint-6057/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..f3be0e45cf374570999f466e3907a8624e5ceed5 --- /dev/null +++ b/checkpoint-6057/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9b35183dd7f8381694748975f1ac0ce580f7dc8ff555da6b121dec42115986cc +size 15984 diff --git a/checkpoint-6057/rng_state_3.pth b/checkpoint-6057/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..a2dcda5c3a469f8277053bb37a3ff3098c1c26d5 --- /dev/null +++ b/checkpoint-6057/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4efb6177d36229e8b4aad277dbe1bdf8bd75eb115e58ef43dc2608a9eb1ee60d +size 15984 diff --git a/checkpoint-6057/rng_state_4.pth b/checkpoint-6057/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..89fa830ba4173a06ef18cffaf5276fec28d8fca8 --- /dev/null +++ b/checkpoint-6057/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:75d8033314cd3e4ba9dc64d6e183751a48acfc3c5f56ac54b3cb6a2d4ed2b67f +size 15984 diff --git a/checkpoint-6057/rng_state_5.pth b/checkpoint-6057/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..69dc0ff9861f9ccd1b0b02d0ff343f680bf94b6b --- /dev/null +++ b/checkpoint-6057/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:630fb361c29cdcb3a863270eb9478e53d9131321d5cf98cfe6a5d0039962b9cd +size 15984 diff --git a/checkpoint-6057/rng_state_6.pth b/checkpoint-6057/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..58121dba3ef14a92c5b59107c0ca9fb73f174342 --- /dev/null +++ b/checkpoint-6057/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19243dd395d597ab405209a281bfad200b2a6ee1e3ca232f5f01401a43520f54 +size 15984 diff --git a/checkpoint-6057/rng_state_7.pth b/checkpoint-6057/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..3f9ffb9ca6f63bea4f0b4dffed99e2c860447d2a --- /dev/null +++ b/checkpoint-6057/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ae910e61c283a6ce3bff7fa670e43ffa38fdda0533b927878dbc33d67a638ca1 +size 15984 diff --git a/checkpoint-6057/scheduler.pt b/checkpoint-6057/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..afdeb9e6587217473a99bf4096f36202f9f3d6d6 --- /dev/null +++ b/checkpoint-6057/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:54f925ccaa4cbd6e900b2bd067596b5517028ecaba1c8138d9b3a8b31f9ee613 +size 1064 diff --git a/checkpoint-6057/special_tokens_map.json b/checkpoint-6057/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80 --- /dev/null +++ b/checkpoint-6057/special_tokens_map.json @@ -0,0 +1,34 @@ +{ + "additional_special_tokens": [ + "", + "" + ], + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-6057/tokenizer.json b/checkpoint-6057/tokenizer.json new file mode 100644 index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044 --- /dev/null +++ b/checkpoint-6057/tokenizer.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee +size 17477929 diff --git a/checkpoint-6057/tokenizer.model b/checkpoint-6057/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8 --- /dev/null +++ b/checkpoint-6057/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2 +size 4241003 diff --git a/checkpoint-6057/tokenizer_config.json b/checkpoint-6057/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..1451ca400a880c5056ab40cb960930cf827de2de --- /dev/null +++ b/checkpoint-6057/tokenizer_config.json @@ -0,0 +1,70 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "3": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "106": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "107": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "additional_special_tokens": [ + "", + "" + ], + "bos_token": "", + "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '' + role + '\n' + message['content'] | trim + '\n' }}{% endfor %}{% if add_generation_prompt %}{{'model\n'}}{% endif %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": null, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "tokenizer_class": "GemmaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-6057/trainer_state.json b/checkpoint-6057/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..1aa4b415c67635f43945841da0aa447d764018b9 --- /dev/null +++ b/checkpoint-6057/trainer_state.json @@ -0,0 +1,42708 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.870985706330053, + "eval_steps": 169, + "global_step": 6057, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.10426733642816544, + "learning_rate": 6.25e-06, + "loss": 39.5782, + "step": 1 + }, + { + "epoch": 0.0, + "eval_loss": 39.257938385009766, + "eval_runtime": 1143.223, + "eval_samples_per_second": 14.994, + "eval_steps_per_second": 0.938, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 0.10457848757505417, + "learning_rate": 1.25e-05, + "loss": 39.5213, + "step": 2 + }, + { + "epoch": 0.0, + "grad_norm": 0.10474710911512375, + "learning_rate": 1.8750000000000002e-05, + "loss": 39.4605, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.11379425972700119, + "learning_rate": 2.5e-05, + "loss": 39.4968, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.12108173966407776, + "learning_rate": 3.125e-05, + "loss": 39.4831, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.13807319104671478, + "learning_rate": 3.7500000000000003e-05, + "loss": 39.5751, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.1619432121515274, + "learning_rate": 4.375e-05, + "loss": 39.4458, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.1936173141002655, + "learning_rate": 5e-05, + "loss": 39.4696, + "step": 8 + }, + { + "epoch": 0.01, + "grad_norm": 0.23368136584758759, + "learning_rate": 5.6250000000000005e-05, + "loss": 39.5398, + "step": 9 + }, + { + "epoch": 0.01, + "grad_norm": 0.28147873282432556, + "learning_rate": 6.25e-05, + "loss": 39.5004, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.3371683359146118, + "learning_rate": 6.875e-05, + "loss": 39.4975, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.4020785391330719, + "learning_rate": 7.500000000000001e-05, + "loss": 39.5237, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.4757504463195801, + "learning_rate": 8.125000000000001e-05, + "loss": 39.5889, + "step": 13 + }, + { + "epoch": 0.02, + "grad_norm": 0.5553041100502014, + "learning_rate": 8.75e-05, + "loss": 39.5117, + "step": 14 + }, + { + "epoch": 0.02, + "grad_norm": 0.6490126848220825, + "learning_rate": 9.375e-05, + "loss": 39.5925, + "step": 15 + }, + { + "epoch": 0.02, + "grad_norm": 0.7478038668632507, + "learning_rate": 0.0001, + "loss": 39.5638, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.8497499227523804, + "learning_rate": 0.00010625000000000001, + "loss": 39.513, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.9426423907279968, + "learning_rate": 0.00011250000000000001, + "loss": 39.1585, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.9930742979049683, + "learning_rate": 0.00011875, + "loss": 38.4983, + "step": 19 + }, + { + "epoch": 0.03, + "grad_norm": 0.9656581878662109, + "learning_rate": 0.000125, + "loss": 37.7409, + "step": 20 + }, + { + "epoch": 0.03, + "grad_norm": 1.0122301578521729, + "learning_rate": 0.00013125000000000002, + "loss": 37.248, + "step": 21 + }, + { + "epoch": 0.03, + "grad_norm": 1.0719304084777832, + "learning_rate": 0.0001375, + "loss": 37.0875, + "step": 22 + }, + { + "epoch": 0.03, + "grad_norm": 1.1606364250183105, + "learning_rate": 0.00014375, + "loss": 36.8509, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 1.2129034996032715, + "learning_rate": 0.00015000000000000001, + "loss": 36.5097, + "step": 24 + }, + { + "epoch": 0.04, + "grad_norm": 1.3106859922409058, + "learning_rate": 0.00015625, + "loss": 36.101, + "step": 25 + }, + { + "epoch": 0.04, + "grad_norm": 1.395408272743225, + "learning_rate": 0.00016250000000000002, + "loss": 35.5219, + "step": 26 + }, + { + "epoch": 0.04, + "grad_norm": 1.5144782066345215, + "learning_rate": 0.00016875, + "loss": 35.1062, + "step": 27 + }, + { + "epoch": 0.04, + "grad_norm": 1.6357661485671997, + "learning_rate": 0.000175, + "loss": 34.562, + "step": 28 + }, + { + "epoch": 0.04, + "grad_norm": 1.7252875566482544, + "learning_rate": 0.00018125000000000001, + "loss": 33.7745, + "step": 29 + }, + { + "epoch": 0.04, + "grad_norm": 1.8169721364974976, + "learning_rate": 0.0001875, + "loss": 33.1849, + "step": 30 + }, + { + "epoch": 0.05, + "grad_norm": 1.8421262502670288, + "learning_rate": 0.00019375000000000002, + "loss": 32.4654, + "step": 31 + }, + { + "epoch": 0.05, + "grad_norm": 1.9798814058303833, + "learning_rate": 0.0002, + "loss": 31.5573, + "step": 32 + }, + { + "epoch": 0.05, + "grad_norm": 2.1344172954559326, + "learning_rate": 0.0001999999890003351, + "loss": 30.9011, + "step": 33 + }, + { + "epoch": 0.05, + "grad_norm": 2.1522722244262695, + "learning_rate": 0.00019999995600134275, + "loss": 30.0969, + "step": 34 + }, + { + "epoch": 0.05, + "grad_norm": 1.9861739873886108, + "learning_rate": 0.0001999999010030303, + "loss": 29.2185, + "step": 35 + }, + { + "epoch": 0.05, + "grad_norm": 1.967444896697998, + "learning_rate": 0.00019999982400540977, + "loss": 28.5804, + "step": 36 + }, + { + "epoch": 0.05, + "grad_norm": 1.9654927253723145, + "learning_rate": 0.00019999972500849814, + "loss": 27.8538, + "step": 37 + }, + { + "epoch": 0.06, + "grad_norm": 1.9123767614364624, + "learning_rate": 0.00019999960401231715, + "loss": 27.1902, + "step": 38 + }, + { + "epoch": 0.06, + "grad_norm": 1.7705096006393433, + "learning_rate": 0.00019999946101689346, + "loss": 26.4983, + "step": 39 + }, + { + "epoch": 0.06, + "grad_norm": 1.6448909044265747, + "learning_rate": 0.00019999929602225854, + "loss": 25.8354, + "step": 40 + }, + { + "epoch": 0.06, + "grad_norm": 1.5614784955978394, + "learning_rate": 0.00019999910902844862, + "loss": 25.2961, + "step": 41 + }, + { + "epoch": 0.06, + "grad_norm": 1.5540868043899536, + "learning_rate": 0.0001999989000355049, + "loss": 24.8097, + "step": 42 + }, + { + "epoch": 0.06, + "grad_norm": 1.4986692667007446, + "learning_rate": 0.00019999866904347332, + "loss": 24.3735, + "step": 43 + }, + { + "epoch": 0.07, + "grad_norm": 1.4384242296218872, + "learning_rate": 0.00019999841605240475, + "loss": 23.8929, + "step": 44 + }, + { + "epoch": 0.07, + "grad_norm": 1.290089726448059, + "learning_rate": 0.00019999814106235478, + "loss": 23.4627, + "step": 45 + }, + { + "epoch": 0.07, + "grad_norm": 1.3345509767532349, + "learning_rate": 0.0001999978440733839, + "loss": 23.1114, + "step": 46 + }, + { + "epoch": 0.07, + "grad_norm": 1.5627790689468384, + "learning_rate": 0.00019999752508555752, + "loss": 22.8306, + "step": 47 + }, + { + "epoch": 0.07, + "grad_norm": 1.67068612575531, + "learning_rate": 0.00019999718409894577, + "loss": 22.5982, + "step": 48 + }, + { + "epoch": 0.07, + "grad_norm": 1.6429119110107422, + "learning_rate": 0.00019999682111362368, + "loss": 22.2669, + "step": 49 + }, + { + "epoch": 0.07, + "grad_norm": 1.5682146549224854, + "learning_rate": 0.00019999643612967106, + "loss": 21.9562, + "step": 50 + }, + { + "epoch": 0.08, + "grad_norm": 1.3983449935913086, + "learning_rate": 0.00019999602914717266, + "loss": 21.6858, + "step": 51 + }, + { + "epoch": 0.08, + "grad_norm": 1.3296594619750977, + "learning_rate": 0.000199995600166218, + "loss": 21.4453, + "step": 52 + }, + { + "epoch": 0.08, + "grad_norm": 1.2462741136550903, + "learning_rate": 0.00019999514918690142, + "loss": 21.1379, + "step": 53 + }, + { + "epoch": 0.08, + "grad_norm": 1.2661601305007935, + "learning_rate": 0.0001999946762093222, + "loss": 20.8178, + "step": 54 + }, + { + "epoch": 0.08, + "grad_norm": 1.5054320096969604, + "learning_rate": 0.00019999418123358432, + "loss": 20.5357, + "step": 55 + }, + { + "epoch": 0.08, + "grad_norm": 1.8500558137893677, + "learning_rate": 0.0001999936642597967, + "loss": 20.2271, + "step": 56 + }, + { + "epoch": 0.08, + "grad_norm": 2.102200508117676, + "learning_rate": 0.00019999312528807308, + "loss": 19.9845, + "step": 57 + }, + { + "epoch": 0.09, + "grad_norm": 2.3885788917541504, + "learning_rate": 0.000199992564318532, + "loss": 19.6821, + "step": 58 + }, + { + "epoch": 0.09, + "grad_norm": 1.3269203901290894, + "learning_rate": 0.00019999198135129689, + "loss": 19.4197, + "step": 59 + }, + { + "epoch": 0.09, + "grad_norm": 1.0758070945739746, + "learning_rate": 0.000199991376386496, + "loss": 19.0985, + "step": 60 + }, + { + "epoch": 0.09, + "grad_norm": 1.1309177875518799, + "learning_rate": 0.00019999074942426246, + "loss": 18.8138, + "step": 61 + }, + { + "epoch": 0.09, + "grad_norm": 1.230851173400879, + "learning_rate": 0.00019999010046473412, + "loss": 18.5005, + "step": 62 + }, + { + "epoch": 0.09, + "grad_norm": 1.2155430316925049, + "learning_rate": 0.00019998942950805377, + "loss": 18.1152, + "step": 63 + }, + { + "epoch": 0.1, + "grad_norm": 1.2195416688919067, + "learning_rate": 0.00019998873655436905, + "loss": 17.7053, + "step": 64 + }, + { + "epoch": 0.1, + "grad_norm": 1.2259167432785034, + "learning_rate": 0.0001999880216038324, + "loss": 17.3123, + "step": 65 + }, + { + "epoch": 0.1, + "grad_norm": 1.2353639602661133, + "learning_rate": 0.00019998728465660105, + "loss": 16.9048, + "step": 66 + }, + { + "epoch": 0.1, + "grad_norm": 1.3028600215911865, + "learning_rate": 0.00019998652571283715, + "loss": 16.477, + "step": 67 + }, + { + "epoch": 0.1, + "grad_norm": 1.3070034980773926, + "learning_rate": 0.00019998574477270772, + "loss": 16.0892, + "step": 68 + }, + { + "epoch": 0.1, + "grad_norm": 1.231437087059021, + "learning_rate": 0.00019998494183638446, + "loss": 15.6727, + "step": 69 + }, + { + "epoch": 0.1, + "grad_norm": 1.143466591835022, + "learning_rate": 0.00019998411690404412, + "loss": 15.3314, + "step": 70 + }, + { + "epoch": 0.11, + "grad_norm": 1.4420065879821777, + "learning_rate": 0.0001999832699758681, + "loss": 15.1217, + "step": 71 + }, + { + "epoch": 0.11, + "grad_norm": 1.9184247255325317, + "learning_rate": 0.00019998240105204276, + "loss": 15.0641, + "step": 72 + }, + { + "epoch": 0.11, + "grad_norm": 2.11586332321167, + "learning_rate": 0.0001999815101327592, + "loss": 14.9107, + "step": 73 + }, + { + "epoch": 0.11, + "grad_norm": 2.1635212898254395, + "learning_rate": 0.0001999805972182135, + "loss": 14.6949, + "step": 74 + }, + { + "epoch": 0.11, + "grad_norm": 1.9647817611694336, + "learning_rate": 0.00019997966230860643, + "loss": 14.3436, + "step": 75 + }, + { + "epoch": 0.11, + "grad_norm": 1.6679894924163818, + "learning_rate": 0.00019997870540414367, + "loss": 13.917, + "step": 76 + }, + { + "epoch": 0.11, + "grad_norm": 1.407972812652588, + "learning_rate": 0.0001999777265050358, + "loss": 13.6019, + "step": 77 + }, + { + "epoch": 0.12, + "grad_norm": 1.442781686782837, + "learning_rate": 0.00019997672561149806, + "loss": 13.3002, + "step": 78 + }, + { + "epoch": 0.12, + "grad_norm": 1.4343981742858887, + "learning_rate": 0.00019997570272375074, + "loss": 13.111, + "step": 79 + }, + { + "epoch": 0.12, + "grad_norm": 1.4623301029205322, + "learning_rate": 0.00019997465784201882, + "loss": 12.9296, + "step": 80 + }, + { + "epoch": 0.12, + "grad_norm": 1.2028108835220337, + "learning_rate": 0.00019997359096653216, + "loss": 12.7946, + "step": 81 + }, + { + "epoch": 0.12, + "grad_norm": 0.7906871438026428, + "learning_rate": 0.0001999725020975255, + "loss": 12.6467, + "step": 82 + }, + { + "epoch": 0.12, + "grad_norm": 0.8495603799819946, + "learning_rate": 0.0001999713912352384, + "loss": 12.5962, + "step": 83 + }, + { + "epoch": 0.12, + "grad_norm": 1.952924370765686, + "learning_rate": 0.00019997025837991514, + "loss": 12.4475, + "step": 84 + }, + { + "epoch": 0.13, + "grad_norm": 2.176668405532837, + "learning_rate": 0.00019996910353180502, + "loss": 12.3597, + "step": 85 + }, + { + "epoch": 0.13, + "grad_norm": 1.4191577434539795, + "learning_rate": 0.0001999679266911621, + "loss": 12.2764, + "step": 86 + }, + { + "epoch": 0.13, + "grad_norm": 1.0766239166259766, + "learning_rate": 0.00019996672785824526, + "loss": 12.2163, + "step": 87 + }, + { + "epoch": 0.13, + "grad_norm": 0.8922137022018433, + "learning_rate": 0.00019996550703331825, + "loss": 12.004, + "step": 88 + }, + { + "epoch": 0.13, + "grad_norm": 0.8033492565155029, + "learning_rate": 0.00019996426421664963, + "loss": 11.8298, + "step": 89 + }, + { + "epoch": 0.13, + "grad_norm": 0.8409807682037354, + "learning_rate": 0.00019996299940851277, + "loss": 11.7005, + "step": 90 + }, + { + "epoch": 0.14, + "grad_norm": 1.0452065467834473, + "learning_rate": 0.000199961712609186, + "loss": 11.5059, + "step": 91 + }, + { + "epoch": 0.14, + "grad_norm": 1.1938868761062622, + "learning_rate": 0.00019996040381895237, + "loss": 11.3725, + "step": 92 + }, + { + "epoch": 0.14, + "grad_norm": 1.1088273525238037, + "learning_rate": 0.0001999590730380998, + "loss": 11.1496, + "step": 93 + }, + { + "epoch": 0.14, + "grad_norm": 0.9627100229263306, + "learning_rate": 0.00019995772026692107, + "loss": 10.9487, + "step": 94 + }, + { + "epoch": 0.14, + "grad_norm": 0.9412399530410767, + "learning_rate": 0.00019995634550571377, + "loss": 10.7797, + "step": 95 + }, + { + "epoch": 0.14, + "grad_norm": 0.962394654750824, + "learning_rate": 0.00019995494875478032, + "loss": 10.5443, + "step": 96 + }, + { + "epoch": 0.14, + "grad_norm": 0.7746237516403198, + "learning_rate": 0.000199953530014428, + "loss": 10.3263, + "step": 97 + }, + { + "epoch": 0.15, + "grad_norm": 0.821438729763031, + "learning_rate": 0.00019995208928496897, + "loss": 10.1812, + "step": 98 + }, + { + "epoch": 0.15, + "grad_norm": 1.4739282131195068, + "learning_rate": 0.00019995062656672012, + "loss": 9.9212, + "step": 99 + }, + { + "epoch": 0.15, + "grad_norm": 0.9201889038085938, + "learning_rate": 0.00019994914186000328, + "loss": 9.8514, + "step": 100 + }, + { + "epoch": 0.15, + "grad_norm": 0.909538745880127, + "learning_rate": 0.00019994763516514505, + "loss": 9.7503, + "step": 101 + }, + { + "epoch": 0.15, + "grad_norm": 0.8758969902992249, + "learning_rate": 0.0001999461064824769, + "loss": 9.6219, + "step": 102 + }, + { + "epoch": 0.15, + "grad_norm": 0.9529918432235718, + "learning_rate": 0.00019994455581233514, + "loss": 9.5321, + "step": 103 + }, + { + "epoch": 0.15, + "grad_norm": 1.0058927536010742, + "learning_rate": 0.0001999429831550609, + "loss": 9.3729, + "step": 104 + }, + { + "epoch": 0.16, + "grad_norm": 0.9739552140235901, + "learning_rate": 0.00019994138851100015, + "loss": 9.3018, + "step": 105 + }, + { + "epoch": 0.16, + "grad_norm": 0.8659080266952515, + "learning_rate": 0.0001999397718805037, + "loss": 9.1481, + "step": 106 + }, + { + "epoch": 0.16, + "grad_norm": 0.9334455132484436, + "learning_rate": 0.00019993813326392722, + "loss": 9.0994, + "step": 107 + }, + { + "epoch": 0.16, + "grad_norm": 0.9508799910545349, + "learning_rate": 0.00019993647266163117, + "loss": 9.0184, + "step": 108 + }, + { + "epoch": 0.16, + "grad_norm": 0.8480530381202698, + "learning_rate": 0.00019993479007398088, + "loss": 8.8884, + "step": 109 + }, + { + "epoch": 0.16, + "grad_norm": 0.7846106290817261, + "learning_rate": 0.0001999330855013465, + "loss": 8.7288, + "step": 110 + }, + { + "epoch": 0.16, + "grad_norm": 0.7611198425292969, + "learning_rate": 0.00019993135894410304, + "loss": 8.7147, + "step": 111 + }, + { + "epoch": 0.17, + "grad_norm": 0.7758539915084839, + "learning_rate": 0.00019992961040263033, + "loss": 8.7001, + "step": 112 + }, + { + "epoch": 0.17, + "grad_norm": 0.6993403434753418, + "learning_rate": 0.000199927839877313, + "loss": 8.5369, + "step": 113 + }, + { + "epoch": 0.17, + "grad_norm": 0.6209515929222107, + "learning_rate": 0.0001999260473685406, + "loss": 8.4196, + "step": 114 + }, + { + "epoch": 0.17, + "grad_norm": 0.6134201884269714, + "learning_rate": 0.00019992423287670748, + "loss": 8.3379, + "step": 115 + }, + { + "epoch": 0.17, + "grad_norm": 0.6407685875892639, + "learning_rate": 0.00019992239640221276, + "loss": 8.2981, + "step": 116 + }, + { + "epoch": 0.17, + "grad_norm": 0.5949248671531677, + "learning_rate": 0.00019992053794546047, + "loss": 8.266, + "step": 117 + }, + { + "epoch": 0.18, + "grad_norm": 0.5390576720237732, + "learning_rate": 0.0001999186575068595, + "loss": 8.2384, + "step": 118 + }, + { + "epoch": 0.18, + "grad_norm": 0.5707695484161377, + "learning_rate": 0.00019991675508682349, + "loss": 8.1337, + "step": 119 + }, + { + "epoch": 0.18, + "grad_norm": 0.6072248816490173, + "learning_rate": 0.00019991483068577097, + "loss": 8.1228, + "step": 120 + }, + { + "epoch": 0.18, + "grad_norm": 0.5436020493507385, + "learning_rate": 0.00019991288430412532, + "loss": 8.0083, + "step": 121 + }, + { + "epoch": 0.18, + "grad_norm": 0.43990185856819153, + "learning_rate": 0.00019991091594231464, + "loss": 7.943, + "step": 122 + }, + { + "epoch": 0.18, + "grad_norm": 0.7439638376235962, + "learning_rate": 0.0001999089256007721, + "loss": 7.9075, + "step": 123 + }, + { + "epoch": 0.18, + "grad_norm": 0.5559173822402954, + "learning_rate": 0.00019990691327993548, + "loss": 7.8031, + "step": 124 + }, + { + "epoch": 0.19, + "grad_norm": 0.5491620302200317, + "learning_rate": 0.0001999048789802475, + "loss": 7.7261, + "step": 125 + }, + { + "epoch": 0.19, + "grad_norm": 0.7270621657371521, + "learning_rate": 0.00019990282270215563, + "loss": 7.8013, + "step": 126 + }, + { + "epoch": 0.19, + "grad_norm": 0.5974888205528259, + "learning_rate": 0.0001999007444461123, + "loss": 7.6982, + "step": 127 + }, + { + "epoch": 0.19, + "grad_norm": 0.4842529892921448, + "learning_rate": 0.00019989864421257467, + "loss": 7.6482, + "step": 128 + }, + { + "epoch": 0.19, + "grad_norm": 0.6567961573600769, + "learning_rate": 0.00019989652200200485, + "loss": 7.6721, + "step": 129 + }, + { + "epoch": 0.19, + "grad_norm": 0.5840482711791992, + "learning_rate": 0.00019989437781486963, + "loss": 7.6075, + "step": 130 + }, + { + "epoch": 0.19, + "grad_norm": 0.41609910130500793, + "learning_rate": 0.0001998922116516408, + "loss": 7.5145, + "step": 131 + }, + { + "epoch": 0.2, + "grad_norm": 0.4465920329093933, + "learning_rate": 0.00019989002351279483, + "loss": 7.5013, + "step": 132 + }, + { + "epoch": 0.2, + "grad_norm": 0.6557760238647461, + "learning_rate": 0.0001998878133988131, + "loss": 7.5361, + "step": 133 + }, + { + "epoch": 0.2, + "grad_norm": 0.4710114598274231, + "learning_rate": 0.00019988558131018186, + "loss": 7.4989, + "step": 134 + }, + { + "epoch": 0.2, + "grad_norm": 0.67153000831604, + "learning_rate": 0.00019988332724739215, + "loss": 7.6135, + "step": 135 + }, + { + "epoch": 0.2, + "grad_norm": 1.6115596294403076, + "learning_rate": 0.00019988105121093982, + "loss": 7.8275, + "step": 136 + }, + { + "epoch": 0.2, + "grad_norm": 0.7458434700965881, + "learning_rate": 0.0001998787532013256, + "loss": 7.9861, + "step": 137 + }, + { + "epoch": 0.2, + "grad_norm": 0.6269705891609192, + "learning_rate": 0.000199876433219055, + "loss": 8.0249, + "step": 138 + }, + { + "epoch": 0.21, + "grad_norm": 0.7430456876754761, + "learning_rate": 0.00019987409126463846, + "loss": 8.0029, + "step": 139 + }, + { + "epoch": 0.21, + "grad_norm": 1.2338321208953857, + "learning_rate": 0.0001998717273385912, + "loss": 7.9936, + "step": 140 + }, + { + "epoch": 0.21, + "grad_norm": 0.7371585965156555, + "learning_rate": 0.00019986934144143316, + "loss": 7.8833, + "step": 141 + }, + { + "epoch": 0.21, + "grad_norm": 0.6107997894287109, + "learning_rate": 0.00019986693357368934, + "loss": 7.7912, + "step": 142 + }, + { + "epoch": 0.21, + "grad_norm": 1.231048583984375, + "learning_rate": 0.00019986450373588946, + "loss": 7.7342, + "step": 143 + }, + { + "epoch": 0.21, + "grad_norm": 0.6497113704681396, + "learning_rate": 0.00019986205192856795, + "loss": 7.7583, + "step": 144 + }, + { + "epoch": 0.22, + "grad_norm": 1.1496895551681519, + "learning_rate": 0.0001998595781522643, + "loss": 7.7372, + "step": 145 + }, + { + "epoch": 0.22, + "grad_norm": 0.949022114276886, + "learning_rate": 0.00019985708240752268, + "loss": 7.7063, + "step": 146 + }, + { + "epoch": 0.22, + "grad_norm": 3.4438350200653076, + "learning_rate": 0.00019985456469489213, + "loss": 7.6564, + "step": 147 + }, + { + "epoch": 0.22, + "grad_norm": 1.3692785501480103, + "learning_rate": 0.00019985202501492656, + "loss": 7.7306, + "step": 148 + }, + { + "epoch": 0.22, + "grad_norm": 1.341970443725586, + "learning_rate": 0.00019984946336818463, + "loss": 7.6558, + "step": 149 + }, + { + "epoch": 0.22, + "grad_norm": 0.8111562728881836, + "learning_rate": 0.00019984687975522993, + "loss": 7.5931, + "step": 150 + }, + { + "epoch": 0.22, + "grad_norm": 1.028735876083374, + "learning_rate": 0.00019984427417663085, + "loss": 7.5077, + "step": 151 + }, + { + "epoch": 0.23, + "grad_norm": 4.875123023986816, + "learning_rate": 0.00019984164663296058, + "loss": 7.4798, + "step": 152 + }, + { + "epoch": 0.23, + "grad_norm": 1.2443698644638062, + "learning_rate": 0.00019983899712479715, + "loss": 7.5151, + "step": 153 + }, + { + "epoch": 0.23, + "grad_norm": 38.03964614868164, + "learning_rate": 0.00019983632565272346, + "loss": 7.3952, + "step": 154 + }, + { + "epoch": 0.23, + "grad_norm": 2.3863110542297363, + "learning_rate": 0.00019983363221732716, + "loss": 7.37, + "step": 155 + }, + { + "epoch": 0.23, + "grad_norm": 1.0871200561523438, + "learning_rate": 0.00019983091681920084, + "loss": 7.4005, + "step": 156 + }, + { + "epoch": 0.23, + "grad_norm": 1.156063437461853, + "learning_rate": 0.0001998281794589419, + "loss": 7.3377, + "step": 157 + }, + { + "epoch": 0.23, + "grad_norm": 3.350292682647705, + "learning_rate": 0.00019982542013715244, + "loss": 7.0855, + "step": 158 + }, + { + "epoch": 0.24, + "grad_norm": 4.875136375427246, + "learning_rate": 0.00019982263885443956, + "loss": 7.0308, + "step": 159 + }, + { + "epoch": 0.24, + "grad_norm": 4.6525068283081055, + "learning_rate": 0.00019981983561141516, + "loss": 7.0757, + "step": 160 + }, + { + "epoch": 0.24, + "grad_norm": 2.9209208488464355, + "learning_rate": 0.00019981701040869585, + "loss": 7.1687, + "step": 161 + }, + { + "epoch": 0.24, + "grad_norm": 2.691115379333496, + "learning_rate": 0.00019981416324690317, + "loss": 7.2762, + "step": 162 + }, + { + "epoch": 0.24, + "grad_norm": 240.15269470214844, + "learning_rate": 0.00019981129412666358, + "loss": 7.3205, + "step": 163 + }, + { + "epoch": 0.24, + "grad_norm": 5.614213466644287, + "learning_rate": 0.00019980840304860813, + "loss": 7.2591, + "step": 164 + }, + { + "epoch": 0.25, + "grad_norm": 6.848481178283691, + "learning_rate": 0.0001998054900133729, + "loss": 7.3047, + "step": 165 + }, + { + "epoch": 0.25, + "grad_norm": 1.3473995923995972, + "learning_rate": 0.00019980255502159873, + "loss": 7.314, + "step": 166 + }, + { + "epoch": 0.25, + "grad_norm": 1.2133899927139282, + "learning_rate": 0.00019979959807393132, + "loss": 7.2487, + "step": 167 + }, + { + "epoch": 0.25, + "grad_norm": 46.43593215942383, + "learning_rate": 0.00019979661917102115, + "loss": 7.2762, + "step": 168 + }, + { + "epoch": 0.25, + "grad_norm": 1.915299892425537, + "learning_rate": 0.00019979361831352357, + "loss": 7.2511, + "step": 169 + }, + { + "epoch": 0.25, + "eval_loss": 7.077121257781982, + "eval_runtime": 1142.7781, + "eval_samples_per_second": 15.0, + "eval_steps_per_second": 0.938, + "step": 169 + }, + { + "epoch": 0.25, + "grad_norm": 2.1094720363616943, + "learning_rate": 0.00019979059550209876, + "loss": 7.2082, + "step": 170 + }, + { + "epoch": 0.25, + "grad_norm": 1.6243547201156616, + "learning_rate": 0.0001997875507374117, + "loss": 7.2193, + "step": 171 + }, + { + "epoch": 0.26, + "grad_norm": 1.493200659751892, + "learning_rate": 0.0001997844840201322, + "loss": 7.2206, + "step": 172 + }, + { + "epoch": 0.26, + "grad_norm": 5.889941215515137, + "learning_rate": 0.00019978139535093496, + "loss": 7.2216, + "step": 173 + }, + { + "epoch": 0.26, + "grad_norm": 16.500181198120117, + "learning_rate": 0.00019977828473049947, + "loss": 7.2762, + "step": 174 + }, + { + "epoch": 0.26, + "grad_norm": 1.3627996444702148, + "learning_rate": 0.00019977515215951, + "loss": 7.2056, + "step": 175 + }, + { + "epoch": 0.26, + "grad_norm": 3.392564535140991, + "learning_rate": 0.0001997719976386557, + "loss": 7.2832, + "step": 176 + }, + { + "epoch": 0.26, + "grad_norm": 4.66141939163208, + "learning_rate": 0.00019976882116863061, + "loss": 7.2547, + "step": 177 + }, + { + "epoch": 0.26, + "grad_norm": 1.6790283918380737, + "learning_rate": 0.00019976562275013344, + "loss": 7.1852, + "step": 178 + }, + { + "epoch": 0.27, + "grad_norm": 3.375530242919922, + "learning_rate": 0.00019976240238386787, + "loss": 7.2132, + "step": 179 + }, + { + "epoch": 0.27, + "grad_norm": 2.9105639457702637, + "learning_rate": 0.0001997591600705424, + "loss": 7.2096, + "step": 180 + }, + { + "epoch": 0.27, + "grad_norm": 3.507265329360962, + "learning_rate": 0.00019975589581087023, + "loss": 7.1184, + "step": 181 + }, + { + "epoch": 0.27, + "grad_norm": 3.2614657878875732, + "learning_rate": 0.0001997526096055695, + "loss": 7.174, + "step": 182 + }, + { + "epoch": 0.27, + "grad_norm": 20.52471923828125, + "learning_rate": 0.0001997493014553632, + "loss": 7.1458, + "step": 183 + }, + { + "epoch": 0.27, + "grad_norm": 7.462634563446045, + "learning_rate": 0.00019974597136097904, + "loss": 7.2018, + "step": 184 + }, + { + "epoch": 0.27, + "grad_norm": 3.244502544403076, + "learning_rate": 0.00019974261932314972, + "loss": 7.2068, + "step": 185 + }, + { + "epoch": 0.28, + "grad_norm": 3.8728437423706055, + "learning_rate": 0.0001997392453426125, + "loss": 7.2269, + "step": 186 + }, + { + "epoch": 0.28, + "grad_norm": 1.4866065979003906, + "learning_rate": 0.00019973584942010978, + "loss": 7.3578, + "step": 187 + }, + { + "epoch": 0.28, + "grad_norm": 11.506189346313477, + "learning_rate": 0.0001997324315563886, + "loss": 7.3122, + "step": 188 + }, + { + "epoch": 0.28, + "grad_norm": 30.93811798095703, + "learning_rate": 0.0001997289917522008, + "loss": 7.3178, + "step": 189 + }, + { + "epoch": 0.28, + "grad_norm": 27.9945125579834, + "learning_rate": 0.00019972553000830324, + "loss": 7.3128, + "step": 190 + }, + { + "epoch": 0.28, + "grad_norm": 144.04150390625, + "learning_rate": 0.00019972204632545736, + "loss": 7.2427, + "step": 191 + }, + { + "epoch": 0.29, + "grad_norm": 77.6709213256836, + "learning_rate": 0.00019971854070442962, + "loss": 7.1204, + "step": 192 + }, + { + "epoch": 0.29, + "grad_norm": 56.658504486083984, + "learning_rate": 0.00019971501314599118, + "loss": 7.1243, + "step": 193 + }, + { + "epoch": 0.29, + "grad_norm": 12.49243450164795, + "learning_rate": 0.00019971146365091813, + "loss": 7.0943, + "step": 194 + }, + { + "epoch": 0.29, + "grad_norm": 12.344218254089355, + "learning_rate": 0.0001997078922199913, + "loss": 7.0644, + "step": 195 + }, + { + "epoch": 0.29, + "grad_norm": 542.6148071289062, + "learning_rate": 0.0001997042988539964, + "loss": 7.1231, + "step": 196 + }, + { + "epoch": 0.29, + "grad_norm": 13.391535758972168, + "learning_rate": 0.00019970068355372393, + "loss": 7.0059, + "step": 197 + }, + { + "epoch": 0.29, + "grad_norm": 438.4042663574219, + "learning_rate": 0.00019969704631996927, + "loss": 7.0568, + "step": 198 + }, + { + "epoch": 0.3, + "grad_norm": 918.09326171875, + "learning_rate": 0.00019969338715353257, + "loss": 7.0436, + "step": 199 + }, + { + "epoch": 0.3, + "grad_norm": 8.38797664642334, + "learning_rate": 0.00019968970605521876, + "loss": 7.0259, + "step": 200 + }, + { + "epoch": 0.3, + "grad_norm": 99.3614501953125, + "learning_rate": 0.00019968600302583774, + "loss": 7.016, + "step": 201 + }, + { + "epoch": 0.3, + "grad_norm": 9.193567276000977, + "learning_rate": 0.0001996822780662041, + "loss": 7.0366, + "step": 202 + }, + { + "epoch": 0.3, + "grad_norm": 20.37701988220215, + "learning_rate": 0.00019967853117713739, + "loss": 7.0319, + "step": 203 + }, + { + "epoch": 0.3, + "grad_norm": 13.956972122192383, + "learning_rate": 0.00019967476235946175, + "loss": 7.1477, + "step": 204 + }, + { + "epoch": 0.3, + "grad_norm": 18.673934936523438, + "learning_rate": 0.00019967097161400642, + "loss": 7.0995, + "step": 205 + }, + { + "epoch": 0.31, + "grad_norm": 52.66343688964844, + "learning_rate": 0.0001996671589416053, + "loss": 7.1156, + "step": 206 + }, + { + "epoch": 0.31, + "grad_norm": 54.673892974853516, + "learning_rate": 0.00019966332434309716, + "loss": 7.0896, + "step": 207 + }, + { + "epoch": 0.31, + "grad_norm": 120.72140502929688, + "learning_rate": 0.00019965946781932558, + "loss": 7.0867, + "step": 208 + }, + { + "epoch": 0.31, + "grad_norm": 22.450502395629883, + "learning_rate": 0.00019965558937113894, + "loss": 7.1196, + "step": 209 + }, + { + "epoch": 0.31, + "grad_norm": 94.62016296386719, + "learning_rate": 0.00019965168899939052, + "loss": 7.1317, + "step": 210 + }, + { + "epoch": 0.31, + "grad_norm": 47.635772705078125, + "learning_rate": 0.00019964776670493837, + "loss": 7.1411, + "step": 211 + }, + { + "epoch": 0.31, + "grad_norm": 5.324208736419678, + "learning_rate": 0.00019964382248864536, + "loss": 7.1979, + "step": 212 + }, + { + "epoch": 0.32, + "grad_norm": 11.261038780212402, + "learning_rate": 0.00019963985635137915, + "loss": 7.1543, + "step": 213 + }, + { + "epoch": 0.32, + "grad_norm": 463.3558044433594, + "learning_rate": 0.0001996358682940123, + "loss": 7.1743, + "step": 214 + }, + { + "epoch": 0.32, + "grad_norm": 3.5599043369293213, + "learning_rate": 0.00019963185831742218, + "loss": 7.1325, + "step": 215 + }, + { + "epoch": 0.32, + "grad_norm": 3.592991828918457, + "learning_rate": 0.00019962782642249092, + "loss": 7.1094, + "step": 216 + }, + { + "epoch": 0.32, + "grad_norm": 59.75080108642578, + "learning_rate": 0.00019962377261010551, + "loss": 7.1571, + "step": 217 + }, + { + "epoch": 0.32, + "grad_norm": 3.108335494995117, + "learning_rate": 0.00019961969688115782, + "loss": 7.061, + "step": 218 + }, + { + "epoch": 0.33, + "grad_norm": 87.59527587890625, + "learning_rate": 0.0001996155992365444, + "loss": 7.0367, + "step": 219 + }, + { + "epoch": 0.33, + "grad_norm": 234.99960327148438, + "learning_rate": 0.00019961147967716673, + "loss": 7.1225, + "step": 220 + }, + { + "epoch": 0.33, + "grad_norm": 2.509389877319336, + "learning_rate": 0.00019960733820393114, + "loss": 7.0294, + "step": 221 + }, + { + "epoch": 0.33, + "grad_norm": 4.749375343322754, + "learning_rate": 0.00019960317481774865, + "loss": 6.9989, + "step": 222 + }, + { + "epoch": 0.33, + "grad_norm": 6.5035319328308105, + "learning_rate": 0.00019959898951953524, + "loss": 6.9879, + "step": 223 + }, + { + "epoch": 0.33, + "grad_norm": 15.936602592468262, + "learning_rate": 0.0001995947823102116, + "loss": 6.8966, + "step": 224 + }, + { + "epoch": 0.33, + "grad_norm": 27.570510864257812, + "learning_rate": 0.00019959055319070332, + "loss": 6.9171, + "step": 225 + }, + { + "epoch": 0.34, + "grad_norm": 31.393901824951172, + "learning_rate": 0.00019958630216194073, + "loss": 6.993, + "step": 226 + }, + { + "epoch": 0.34, + "grad_norm": 2.980260133743286, + "learning_rate": 0.0001995820292248591, + "loss": 7.0092, + "step": 227 + }, + { + "epoch": 0.34, + "grad_norm": 2.1385061740875244, + "learning_rate": 0.0001995777343803984, + "loss": 6.9541, + "step": 228 + }, + { + "epoch": 0.34, + "grad_norm": 4.7213053703308105, + "learning_rate": 0.00019957341762950344, + "loss": 6.8875, + "step": 229 + }, + { + "epoch": 0.34, + "grad_norm": 2.533954381942749, + "learning_rate": 0.000199569078973124, + "loss": 6.934, + "step": 230 + }, + { + "epoch": 0.34, + "grad_norm": 3.432656764984131, + "learning_rate": 0.00019956471841221438, + "loss": 6.8725, + "step": 231 + }, + { + "epoch": 0.34, + "grad_norm": 9.07863998413086, + "learning_rate": 0.000199560335947734, + "loss": 6.8456, + "step": 232 + }, + { + "epoch": 0.35, + "grad_norm": 2.4700138568878174, + "learning_rate": 0.0001995559315806469, + "loss": 6.8803, + "step": 233 + }, + { + "epoch": 0.35, + "grad_norm": 15.215128898620605, + "learning_rate": 0.0001995515053119221, + "loss": 6.7947, + "step": 234 + }, + { + "epoch": 0.35, + "grad_norm": 10.798171997070312, + "learning_rate": 0.00019954705714253328, + "loss": 6.8543, + "step": 235 + }, + { + "epoch": 0.35, + "grad_norm": 2.1929962635040283, + "learning_rate": 0.000199542587073459, + "loss": 6.7983, + "step": 236 + }, + { + "epoch": 0.35, + "grad_norm": 18.292057037353516, + "learning_rate": 0.00019953809510568271, + "loss": 6.704, + "step": 237 + }, + { + "epoch": 0.35, + "grad_norm": 0.7752164602279663, + "learning_rate": 0.00019953358124019253, + "loss": 6.7268, + "step": 238 + }, + { + "epoch": 0.35, + "grad_norm": 1.0236197710037231, + "learning_rate": 0.00019952904547798154, + "loss": 6.6618, + "step": 239 + }, + { + "epoch": 0.36, + "grad_norm": 1.3812897205352783, + "learning_rate": 0.00019952448782004756, + "loss": 6.6367, + "step": 240 + }, + { + "epoch": 0.36, + "grad_norm": 1.219442367553711, + "learning_rate": 0.00019951990826739323, + "loss": 6.6027, + "step": 241 + }, + { + "epoch": 0.36, + "grad_norm": 1.485183596611023, + "learning_rate": 0.00019951530682102604, + "loss": 6.5468, + "step": 242 + }, + { + "epoch": 0.36, + "grad_norm": 20.273746490478516, + "learning_rate": 0.00019951068348195828, + "loss": 6.5281, + "step": 243 + }, + { + "epoch": 0.36, + "grad_norm": 3.194899082183838, + "learning_rate": 0.00019950603825120706, + "loss": 6.5781, + "step": 244 + }, + { + "epoch": 0.36, + "grad_norm": 0.9288484454154968, + "learning_rate": 0.00019950137112979425, + "loss": 6.5043, + "step": 245 + }, + { + "epoch": 0.37, + "grad_norm": 0.9781804084777832, + "learning_rate": 0.00019949668211874665, + "loss": 6.4531, + "step": 246 + }, + { + "epoch": 0.37, + "grad_norm": 5.633649826049805, + "learning_rate": 0.00019949197121909575, + "loss": 6.4262, + "step": 247 + }, + { + "epoch": 0.37, + "grad_norm": 4.805322170257568, + "learning_rate": 0.00019948723843187796, + "loss": 6.3919, + "step": 248 + }, + { + "epoch": 0.37, + "grad_norm": 1.2492502927780151, + "learning_rate": 0.00019948248375813445, + "loss": 6.3459, + "step": 249 + }, + { + "epoch": 0.37, + "grad_norm": 1.5589463710784912, + "learning_rate": 0.0001994777071989112, + "loss": 6.2473, + "step": 250 + }, + { + "epoch": 0.37, + "grad_norm": 1.4179363250732422, + "learning_rate": 0.00019947290875525908, + "loss": 6.2753, + "step": 251 + }, + { + "epoch": 0.37, + "grad_norm": 1.064170479774475, + "learning_rate": 0.00019946808842823366, + "loss": 6.2785, + "step": 252 + }, + { + "epoch": 0.38, + "grad_norm": 1.806328296661377, + "learning_rate": 0.00019946324621889535, + "loss": 6.2752, + "step": 253 + }, + { + "epoch": 0.38, + "grad_norm": 1.2437814474105835, + "learning_rate": 0.0001994583821283095, + "loss": 6.1444, + "step": 254 + }, + { + "epoch": 0.38, + "grad_norm": 17.912715911865234, + "learning_rate": 0.00019945349615754608, + "loss": 6.1001, + "step": 255 + }, + { + "epoch": 0.38, + "grad_norm": 0.8929802775382996, + "learning_rate": 0.00019944858830768002, + "loss": 6.0709, + "step": 256 + }, + { + "epoch": 0.38, + "grad_norm": 2.8683338165283203, + "learning_rate": 0.000199443658579791, + "loss": 5.9837, + "step": 257 + }, + { + "epoch": 0.38, + "grad_norm": 1.3912129402160645, + "learning_rate": 0.00019943870697496357, + "loss": 5.9728, + "step": 258 + }, + { + "epoch": 0.38, + "grad_norm": 1.9068537950515747, + "learning_rate": 0.00019943373349428698, + "loss": 5.9657, + "step": 259 + }, + { + "epoch": 0.39, + "grad_norm": 2.0884199142456055, + "learning_rate": 0.0001994287381388554, + "loss": 5.963, + "step": 260 + }, + { + "epoch": 0.39, + "grad_norm": 0.8210564851760864, + "learning_rate": 0.00019942372090976779, + "loss": 5.8253, + "step": 261 + }, + { + "epoch": 0.39, + "grad_norm": 3.3612587451934814, + "learning_rate": 0.0001994186818081279, + "loss": 5.7743, + "step": 262 + }, + { + "epoch": 0.39, + "grad_norm": 2.1916985511779785, + "learning_rate": 0.00019941362083504425, + "loss": 5.7697, + "step": 263 + }, + { + "epoch": 0.39, + "grad_norm": 1.3054827451705933, + "learning_rate": 0.00019940853799163026, + "loss": 5.7867, + "step": 264 + }, + { + "epoch": 0.39, + "grad_norm": 12.717972755432129, + "learning_rate": 0.00019940343327900416, + "loss": 5.6708, + "step": 265 + }, + { + "epoch": 0.4, + "grad_norm": 3.4190762042999268, + "learning_rate": 0.00019939830669828888, + "loss": 5.6869, + "step": 266 + }, + { + "epoch": 0.4, + "grad_norm": 2.753885507583618, + "learning_rate": 0.00019939315825061223, + "loss": 5.5825, + "step": 267 + }, + { + "epoch": 0.4, + "grad_norm": 1.1797378063201904, + "learning_rate": 0.0001993879879371069, + "loss": 5.6165, + "step": 268 + }, + { + "epoch": 0.4, + "grad_norm": 2.9885246753692627, + "learning_rate": 0.00019938279575891033, + "loss": 5.5817, + "step": 269 + }, + { + "epoch": 0.4, + "grad_norm": 1.8179508447647095, + "learning_rate": 0.00019937758171716468, + "loss": 5.4549, + "step": 270 + }, + { + "epoch": 0.4, + "grad_norm": 7.133963584899902, + "learning_rate": 0.00019937234581301707, + "loss": 5.5333, + "step": 271 + }, + { + "epoch": 0.4, + "grad_norm": 7.718517780303955, + "learning_rate": 0.00019936708804761932, + "loss": 5.4891, + "step": 272 + }, + { + "epoch": 0.41, + "grad_norm": 1.2075421810150146, + "learning_rate": 0.00019936180842212818, + "loss": 5.3542, + "step": 273 + }, + { + "epoch": 0.41, + "grad_norm": 6.181577205657959, + "learning_rate": 0.000199356506937705, + "loss": 5.4501, + "step": 274 + }, + { + "epoch": 0.41, + "grad_norm": 6.531129837036133, + "learning_rate": 0.00019935118359551624, + "loss": 5.5179, + "step": 275 + }, + { + "epoch": 0.41, + "grad_norm": 4.549463272094727, + "learning_rate": 0.00019934583839673287, + "loss": 5.4167, + "step": 276 + }, + { + "epoch": 0.41, + "grad_norm": 1.8859062194824219, + "learning_rate": 0.00019934047134253081, + "loss": 5.2628, + "step": 277 + }, + { + "epoch": 0.41, + "grad_norm": 7.954891681671143, + "learning_rate": 0.00019933508243409086, + "loss": 5.3723, + "step": 278 + }, + { + "epoch": 0.41, + "grad_norm": 6.0001444816589355, + "learning_rate": 0.00019932967167259847, + "loss": 5.2979, + "step": 279 + }, + { + "epoch": 0.42, + "grad_norm": 6.450177192687988, + "learning_rate": 0.000199324239059244, + "loss": 5.2947, + "step": 280 + }, + { + "epoch": 0.42, + "grad_norm": 5.8015642166137695, + "learning_rate": 0.00019931878459522254, + "loss": 5.2849, + "step": 281 + }, + { + "epoch": 0.42, + "grad_norm": 8.846211433410645, + "learning_rate": 0.00019931330828173408, + "loss": 5.1841, + "step": 282 + }, + { + "epoch": 0.42, + "grad_norm": 4.6721086502075195, + "learning_rate": 0.00019930781011998338, + "loss": 5.2056, + "step": 283 + }, + { + "epoch": 0.42, + "grad_norm": 3.9085495471954346, + "learning_rate": 0.00019930229011118, + "loss": 5.1857, + "step": 284 + }, + { + "epoch": 0.42, + "grad_norm": 1.6466888189315796, + "learning_rate": 0.00019929674825653826, + "loss": 5.0631, + "step": 285 + }, + { + "epoch": 0.42, + "grad_norm": 2.1051926612854004, + "learning_rate": 0.0001992911845572774, + "loss": 5.033, + "step": 286 + }, + { + "epoch": 0.43, + "grad_norm": 3.247199058532715, + "learning_rate": 0.00019928559901462133, + "loss": 5.098, + "step": 287 + }, + { + "epoch": 0.43, + "grad_norm": 4.04020881652832, + "learning_rate": 0.00019927999162979885, + "loss": 5.0759, + "step": 288 + }, + { + "epoch": 0.43, + "grad_norm": 1.5520354509353638, + "learning_rate": 0.00019927436240404359, + "loss": 5.0395, + "step": 289 + }, + { + "epoch": 0.43, + "grad_norm": 1.2532438039779663, + "learning_rate": 0.00019926871133859388, + "loss": 4.9942, + "step": 290 + }, + { + "epoch": 0.43, + "grad_norm": 3.3151402473449707, + "learning_rate": 0.00019926303843469296, + "loss": 4.9396, + "step": 291 + }, + { + "epoch": 0.43, + "grad_norm": 1.311003565788269, + "learning_rate": 0.00019925734369358883, + "loss": 4.9452, + "step": 292 + }, + { + "epoch": 0.44, + "grad_norm": 3.930495500564575, + "learning_rate": 0.00019925162711653425, + "loss": 4.9488, + "step": 293 + }, + { + "epoch": 0.44, + "grad_norm": 1.0052605867385864, + "learning_rate": 0.00019924588870478687, + "loss": 4.8627, + "step": 294 + }, + { + "epoch": 0.44, + "grad_norm": 7.544490814208984, + "learning_rate": 0.00019924012845960908, + "loss": 4.9602, + "step": 295 + }, + { + "epoch": 0.44, + "grad_norm": 6.155104637145996, + "learning_rate": 0.00019923434638226812, + "loss": 4.9091, + "step": 296 + }, + { + "epoch": 0.44, + "grad_norm": 3.3449931144714355, + "learning_rate": 0.00019922854247403599, + "loss": 5.002, + "step": 297 + }, + { + "epoch": 0.44, + "grad_norm": 3.9289393424987793, + "learning_rate": 0.0001992227167361895, + "loss": 4.8124, + "step": 298 + }, + { + "epoch": 0.44, + "grad_norm": 6.716512680053711, + "learning_rate": 0.00019921686917001032, + "loss": 4.8421, + "step": 299 + }, + { + "epoch": 0.45, + "grad_norm": 4.2615132331848145, + "learning_rate": 0.00019921099977678484, + "loss": 4.8303, + "step": 300 + }, + { + "epoch": 0.45, + "grad_norm": 4.952175617218018, + "learning_rate": 0.00019920510855780428, + "loss": 4.8718, + "step": 301 + }, + { + "epoch": 0.45, + "grad_norm": 1.830321192741394, + "learning_rate": 0.00019919919551436468, + "loss": 4.8395, + "step": 302 + }, + { + "epoch": 0.45, + "grad_norm": 2.7058656215667725, + "learning_rate": 0.0001991932606477669, + "loss": 4.721, + "step": 303 + }, + { + "epoch": 0.45, + "grad_norm": 2.4603841304779053, + "learning_rate": 0.00019918730395931649, + "loss": 4.7365, + "step": 304 + }, + { + "epoch": 0.45, + "grad_norm": 1.5384434461593628, + "learning_rate": 0.00019918132545032396, + "loss": 4.6979, + "step": 305 + }, + { + "epoch": 0.45, + "grad_norm": 5.274871349334717, + "learning_rate": 0.0001991753251221045, + "loss": 4.7287, + "step": 306 + }, + { + "epoch": 0.46, + "grad_norm": 3.8402764797210693, + "learning_rate": 0.00019916930297597817, + "loss": 4.744, + "step": 307 + }, + { + "epoch": 0.46, + "grad_norm": 3.417302131652832, + "learning_rate": 0.0001991632590132698, + "loss": 4.6439, + "step": 308 + }, + { + "epoch": 0.46, + "grad_norm": 4.113287925720215, + "learning_rate": 0.00019915719323530898, + "loss": 4.6228, + "step": 309 + }, + { + "epoch": 0.46, + "grad_norm": 3.57785964012146, + "learning_rate": 0.0001991511056434302, + "loss": 4.6077, + "step": 310 + }, + { + "epoch": 0.46, + "grad_norm": 1.568590521812439, + "learning_rate": 0.00019914499623897264, + "loss": 4.6056, + "step": 311 + }, + { + "epoch": 0.46, + "grad_norm": 3.293492317199707, + "learning_rate": 0.0001991388650232804, + "loss": 4.5294, + "step": 312 + }, + { + "epoch": 0.46, + "grad_norm": 3.416661262512207, + "learning_rate": 0.00019913271199770217, + "loss": 4.6239, + "step": 313 + }, + { + "epoch": 0.47, + "grad_norm": 2.6851394176483154, + "learning_rate": 0.0001991265371635917, + "loss": 4.6183, + "step": 314 + }, + { + "epoch": 0.47, + "grad_norm": 3.812194585800171, + "learning_rate": 0.0001991203405223074, + "loss": 4.5595, + "step": 315 + }, + { + "epoch": 0.47, + "grad_norm": 1.8244884014129639, + "learning_rate": 0.0001991141220752124, + "loss": 4.5329, + "step": 316 + }, + { + "epoch": 0.47, + "grad_norm": 2.763340950012207, + "learning_rate": 0.0001991078818236748, + "loss": 4.5336, + "step": 317 + }, + { + "epoch": 0.47, + "grad_norm": 0.6997628808021545, + "learning_rate": 0.0001991016197690674, + "loss": 4.5623, + "step": 318 + }, + { + "epoch": 0.47, + "grad_norm": 6.273808479309082, + "learning_rate": 0.0001990953359127678, + "loss": 4.549, + "step": 319 + }, + { + "epoch": 0.48, + "grad_norm": 4.5749359130859375, + "learning_rate": 0.00019908903025615842, + "loss": 4.4894, + "step": 320 + }, + { + "epoch": 0.48, + "grad_norm": 4.788200378417969, + "learning_rate": 0.00019908270280062643, + "loss": 4.514, + "step": 321 + }, + { + "epoch": 0.48, + "grad_norm": 5.6347551345825195, + "learning_rate": 0.00019907635354756385, + "loss": 4.477, + "step": 322 + }, + { + "epoch": 0.48, + "grad_norm": 5.96523904800415, + "learning_rate": 0.00019906998249836745, + "loss": 4.4759, + "step": 323 + }, + { + "epoch": 0.48, + "grad_norm": 2.5423848628997803, + "learning_rate": 0.00019906358965443888, + "loss": 4.4077, + "step": 324 + }, + { + "epoch": 0.48, + "grad_norm": 1.0939940214157104, + "learning_rate": 0.00019905717501718444, + "loss": 4.372, + "step": 325 + }, + { + "epoch": 0.48, + "grad_norm": 2.4741616249084473, + "learning_rate": 0.00019905073858801536, + "loss": 4.4112, + "step": 326 + }, + { + "epoch": 0.49, + "grad_norm": 2.774264335632324, + "learning_rate": 0.00019904428036834763, + "loss": 4.4412, + "step": 327 + }, + { + "epoch": 0.49, + "grad_norm": 1.9465824365615845, + "learning_rate": 0.00019903780035960194, + "loss": 4.4442, + "step": 328 + }, + { + "epoch": 0.49, + "grad_norm": 1.6578731536865234, + "learning_rate": 0.0001990312985632039, + "loss": 4.4368, + "step": 329 + }, + { + "epoch": 0.49, + "grad_norm": 1.1936014890670776, + "learning_rate": 0.00019902477498058387, + "loss": 4.3445, + "step": 330 + }, + { + "epoch": 0.49, + "grad_norm": 1.1214628219604492, + "learning_rate": 0.00019901822961317697, + "loss": 4.4032, + "step": 331 + }, + { + "epoch": 0.49, + "grad_norm": 1.4763222932815552, + "learning_rate": 0.00019901166246242316, + "loss": 4.3713, + "step": 332 + }, + { + "epoch": 0.49, + "grad_norm": 3.219590425491333, + "learning_rate": 0.00019900507352976714, + "loss": 4.3248, + "step": 333 + }, + { + "epoch": 0.5, + "grad_norm": 1.5973373651504517, + "learning_rate": 0.00019899846281665843, + "loss": 4.3571, + "step": 334 + }, + { + "epoch": 0.5, + "grad_norm": 2.3610401153564453, + "learning_rate": 0.0001989918303245514, + "loss": 4.2991, + "step": 335 + }, + { + "epoch": 0.5, + "grad_norm": 2.8560168743133545, + "learning_rate": 0.0001989851760549051, + "loss": 4.3024, + "step": 336 + }, + { + "epoch": 0.5, + "grad_norm": 0.9701719880104065, + "learning_rate": 0.0001989785000091834, + "loss": 4.2602, + "step": 337 + }, + { + "epoch": 0.5, + "grad_norm": 1.6666287183761597, + "learning_rate": 0.00019897180218885507, + "loss": 4.2519, + "step": 338 + }, + { + "epoch": 0.5, + "eval_loss": 4.065431594848633, + "eval_runtime": 1149.6322, + "eval_samples_per_second": 14.911, + "eval_steps_per_second": 0.932, + "step": 338 + }, + { + "epoch": 0.5, + "grad_norm": 2.7144992351531982, + "learning_rate": 0.00019896508259539353, + "loss": 4.2504, + "step": 339 + }, + { + "epoch": 0.5, + "grad_norm": 1.088392734527588, + "learning_rate": 0.00019895834123027703, + "loss": 4.2467, + "step": 340 + }, + { + "epoch": 0.51, + "grad_norm": 1.7031919956207275, + "learning_rate": 0.00019895157809498868, + "loss": 4.2498, + "step": 341 + }, + { + "epoch": 0.51, + "grad_norm": 3.883507013320923, + "learning_rate": 0.00019894479319101626, + "loss": 4.289, + "step": 342 + }, + { + "epoch": 0.51, + "grad_norm": 1.1266521215438843, + "learning_rate": 0.00019893798651985246, + "loss": 4.2371, + "step": 343 + }, + { + "epoch": 0.51, + "grad_norm": 2.436180591583252, + "learning_rate": 0.00019893115808299468, + "loss": 4.213, + "step": 344 + }, + { + "epoch": 0.51, + "grad_norm": 6.882200241088867, + "learning_rate": 0.00019892430788194513, + "loss": 4.2289, + "step": 345 + }, + { + "epoch": 0.51, + "grad_norm": 4.444315433502197, + "learning_rate": 0.00019891743591821077, + "loss": 4.1858, + "step": 346 + }, + { + "epoch": 0.52, + "grad_norm": 2.7933340072631836, + "learning_rate": 0.00019891054219330344, + "loss": 4.1988, + "step": 347 + }, + { + "epoch": 0.52, + "grad_norm": 1.5705324411392212, + "learning_rate": 0.00019890362670873973, + "loss": 4.1639, + "step": 348 + }, + { + "epoch": 0.52, + "grad_norm": 4.15967321395874, + "learning_rate": 0.00019889668946604093, + "loss": 4.1391, + "step": 349 + }, + { + "epoch": 0.52, + "grad_norm": 3.6706178188323975, + "learning_rate": 0.00019888973046673324, + "loss": 4.1706, + "step": 350 + }, + { + "epoch": 0.52, + "grad_norm": 4.234200477600098, + "learning_rate": 0.00019888274971234757, + "loss": 4.3108, + "step": 351 + }, + { + "epoch": 0.52, + "grad_norm": 2.6971170902252197, + "learning_rate": 0.00019887574720441967, + "loss": 4.1871, + "step": 352 + }, + { + "epoch": 0.52, + "grad_norm": 1.4427449703216553, + "learning_rate": 0.00019886872294448997, + "loss": 4.1327, + "step": 353 + }, + { + "epoch": 0.53, + "grad_norm": 4.426568984985352, + "learning_rate": 0.00019886167693410384, + "loss": 4.195, + "step": 354 + }, + { + "epoch": 0.53, + "grad_norm": 4.152656555175781, + "learning_rate": 0.00019885460917481136, + "loss": 4.1858, + "step": 355 + }, + { + "epoch": 0.53, + "grad_norm": 5.731381893157959, + "learning_rate": 0.0001988475196681673, + "loss": 4.179, + "step": 356 + }, + { + "epoch": 0.53, + "grad_norm": 17.160917282104492, + "learning_rate": 0.0001988404084157314, + "loss": 4.0972, + "step": 357 + }, + { + "epoch": 0.53, + "grad_norm": 1.9398918151855469, + "learning_rate": 0.00019883327541906802, + "loss": 4.1238, + "step": 358 + }, + { + "epoch": 0.53, + "grad_norm": 2.0568935871124268, + "learning_rate": 0.00019882612067974643, + "loss": 4.0953, + "step": 359 + }, + { + "epoch": 0.53, + "grad_norm": 3.45890212059021, + "learning_rate": 0.00019881894419934056, + "loss": 4.1171, + "step": 360 + }, + { + "epoch": 0.54, + "grad_norm": 1.6760789155960083, + "learning_rate": 0.00019881174597942923, + "loss": 4.1128, + "step": 361 + }, + { + "epoch": 0.54, + "grad_norm": 307.13031005859375, + "learning_rate": 0.000198804526021596, + "loss": 4.0658, + "step": 362 + }, + { + "epoch": 0.54, + "grad_norm": 15.57320785522461, + "learning_rate": 0.00019879728432742918, + "loss": 4.2352, + "step": 363 + }, + { + "epoch": 0.54, + "grad_norm": 8.536359786987305, + "learning_rate": 0.00019879002089852197, + "loss": 4.2513, + "step": 364 + }, + { + "epoch": 0.54, + "grad_norm": 9.536478996276855, + "learning_rate": 0.00019878273573647218, + "loss": 4.2273, + "step": 365 + }, + { + "epoch": 0.54, + "grad_norm": 11.019887924194336, + "learning_rate": 0.00019877542884288256, + "loss": 4.2037, + "step": 366 + }, + { + "epoch": 0.55, + "grad_norm": 7.40153169631958, + "learning_rate": 0.00019876810021936053, + "loss": 4.1092, + "step": 367 + }, + { + "epoch": 0.55, + "grad_norm": 5.204662322998047, + "learning_rate": 0.0001987607498675184, + "loss": 4.0828, + "step": 368 + }, + { + "epoch": 0.55, + "grad_norm": 3.2306785583496094, + "learning_rate": 0.00019875337778897315, + "loss": 4.0828, + "step": 369 + }, + { + "epoch": 0.55, + "grad_norm": 9.242484092712402, + "learning_rate": 0.00019874598398534658, + "loss": 4.1478, + "step": 370 + }, + { + "epoch": 0.55, + "grad_norm": 10.423043251037598, + "learning_rate": 0.00019873856845826536, + "loss": 4.1709, + "step": 371 + }, + { + "epoch": 0.55, + "grad_norm": 2.9181222915649414, + "learning_rate": 0.00019873113120936074, + "loss": 4.0819, + "step": 372 + }, + { + "epoch": 0.55, + "grad_norm": 5.6168718338012695, + "learning_rate": 0.00019872367224026895, + "loss": 4.09, + "step": 373 + }, + { + "epoch": 0.56, + "grad_norm": 2.738548994064331, + "learning_rate": 0.00019871619155263087, + "loss": 4.1129, + "step": 374 + }, + { + "epoch": 0.56, + "grad_norm": 8.364383697509766, + "learning_rate": 0.0001987086891480922, + "loss": 4.0793, + "step": 375 + }, + { + "epoch": 0.56, + "grad_norm": 8.516968727111816, + "learning_rate": 0.00019870116502830346, + "loss": 4.1287, + "step": 376 + }, + { + "epoch": 0.56, + "grad_norm": 2.941425323486328, + "learning_rate": 0.00019869361919491987, + "loss": 4.1145, + "step": 377 + }, + { + "epoch": 0.56, + "grad_norm": 4.925361633300781, + "learning_rate": 0.00019868605164960145, + "loss": 4.1046, + "step": 378 + }, + { + "epoch": 0.56, + "grad_norm": 1.9746365547180176, + "learning_rate": 0.0001986784623940131, + "loss": 3.9944, + "step": 379 + }, + { + "epoch": 0.56, + "grad_norm": 8.789060592651367, + "learning_rate": 0.00019867085142982426, + "loss": 4.0482, + "step": 380 + }, + { + "epoch": 0.57, + "grad_norm": 7.4179301261901855, + "learning_rate": 0.0001986632187587094, + "loss": 4.0702, + "step": 381 + }, + { + "epoch": 0.57, + "grad_norm": 3.4922187328338623, + "learning_rate": 0.00019865556438234763, + "loss": 4.0579, + "step": 382 + }, + { + "epoch": 0.57, + "grad_norm": 5.763650417327881, + "learning_rate": 0.00019864788830242284, + "loss": 4.0776, + "step": 383 + }, + { + "epoch": 0.57, + "grad_norm": 3.3361189365386963, + "learning_rate": 0.00019864019052062375, + "loss": 4.0426, + "step": 384 + }, + { + "epoch": 0.57, + "grad_norm": 8.98303508758545, + "learning_rate": 0.0001986324710386438, + "loss": 4.0599, + "step": 385 + }, + { + "epoch": 0.57, + "grad_norm": 8.836548805236816, + "learning_rate": 0.00019862472985818124, + "loss": 4.0534, + "step": 386 + }, + { + "epoch": 0.57, + "grad_norm": 2.0397250652313232, + "learning_rate": 0.00019861696698093902, + "loss": 4.0105, + "step": 387 + }, + { + "epoch": 0.58, + "grad_norm": 3.091047763824463, + "learning_rate": 0.000198609182408625, + "loss": 3.9637, + "step": 388 + }, + { + "epoch": 0.58, + "grad_norm": 3.192955255508423, + "learning_rate": 0.00019860137614295168, + "loss": 3.9963, + "step": 389 + }, + { + "epoch": 0.58, + "grad_norm": 1.5237922668457031, + "learning_rate": 0.00019859354818563643, + "loss": 3.964, + "step": 390 + }, + { + "epoch": 0.58, + "grad_norm": 4.192271709442139, + "learning_rate": 0.00019858569853840131, + "loss": 3.9638, + "step": 391 + }, + { + "epoch": 0.58, + "grad_norm": 2.3870139122009277, + "learning_rate": 0.00019857782720297319, + "loss": 4.0272, + "step": 392 + }, + { + "epoch": 0.58, + "grad_norm": 6.654936790466309, + "learning_rate": 0.00019856993418108376, + "loss": 3.9616, + "step": 393 + }, + { + "epoch": 0.59, + "grad_norm": 4.117982387542725, + "learning_rate": 0.00019856201947446937, + "loss": 3.9983, + "step": 394 + }, + { + "epoch": 0.59, + "grad_norm": 5.394196033477783, + "learning_rate": 0.00019855408308487124, + "loss": 4.0334, + "step": 395 + }, + { + "epoch": 0.59, + "grad_norm": 5.410153865814209, + "learning_rate": 0.0001985461250140353, + "loss": 4.0342, + "step": 396 + }, + { + "epoch": 0.59, + "grad_norm": 1.220145583152771, + "learning_rate": 0.00019853814526371228, + "loss": 3.9964, + "step": 397 + }, + { + "epoch": 0.59, + "grad_norm": 1.3831374645233154, + "learning_rate": 0.00019853014383565772, + "loss": 3.8909, + "step": 398 + }, + { + "epoch": 0.59, + "grad_norm": 3.4708709716796875, + "learning_rate": 0.0001985221207316318, + "loss": 3.9548, + "step": 399 + }, + { + "epoch": 0.59, + "grad_norm": 1.6609090566635132, + "learning_rate": 0.00019851407595339958, + "loss": 3.9481, + "step": 400 + }, + { + "epoch": 0.6, + "grad_norm": 1.5527194738388062, + "learning_rate": 0.00019850600950273086, + "loss": 3.9567, + "step": 401 + }, + { + "epoch": 0.6, + "grad_norm": 6.021573066711426, + "learning_rate": 0.00019849792138140022, + "loss": 4.0133, + "step": 402 + }, + { + "epoch": 0.6, + "grad_norm": 4.150272846221924, + "learning_rate": 0.00019848981159118699, + "loss": 3.9614, + "step": 403 + }, + { + "epoch": 0.6, + "grad_norm": 7.311338901519775, + "learning_rate": 0.00019848168013387522, + "loss": 3.9789, + "step": 404 + }, + { + "epoch": 0.6, + "grad_norm": 7.797035217285156, + "learning_rate": 0.00019847352701125385, + "loss": 4.0036, + "step": 405 + }, + { + "epoch": 0.6, + "grad_norm": 2.067911386489868, + "learning_rate": 0.0001984653522251165, + "loss": 3.9421, + "step": 406 + }, + { + "epoch": 0.6, + "grad_norm": 5.364695072174072, + "learning_rate": 0.00019845715577726148, + "loss": 3.9846, + "step": 407 + }, + { + "epoch": 0.61, + "grad_norm": 3.663616895675659, + "learning_rate": 0.00019844893766949207, + "loss": 3.9255, + "step": 408 + }, + { + "epoch": 0.61, + "grad_norm": 8.939522743225098, + "learning_rate": 0.00019844069790361611, + "loss": 3.9433, + "step": 409 + }, + { + "epoch": 0.61, + "grad_norm": 9.791443824768066, + "learning_rate": 0.00019843243648144635, + "loss": 3.9879, + "step": 410 + }, + { + "epoch": 0.61, + "grad_norm": 1.5382161140441895, + "learning_rate": 0.00019842415340480025, + "loss": 3.9755, + "step": 411 + }, + { + "epoch": 0.61, + "grad_norm": 7.2470879554748535, + "learning_rate": 0.00019841584867549998, + "loss": 3.9603, + "step": 412 + }, + { + "epoch": 0.61, + "grad_norm": 7.930785655975342, + "learning_rate": 0.0001984075222953726, + "loss": 3.9416, + "step": 413 + }, + { + "epoch": 0.61, + "grad_norm": 3.421830654144287, + "learning_rate": 0.00019839917426624977, + "loss": 3.9023, + "step": 414 + }, + { + "epoch": 0.62, + "grad_norm": 13.186482429504395, + "learning_rate": 0.00019839080458996806, + "loss": 4.0143, + "step": 415 + }, + { + "epoch": 0.62, + "grad_norm": 18.97665023803711, + "learning_rate": 0.0001983824132683687, + "loss": 4.0361, + "step": 416 + }, + { + "epoch": 0.62, + "grad_norm": 12.884719848632812, + "learning_rate": 0.0001983740003032978, + "loss": 4.0086, + "step": 417 + }, + { + "epoch": 0.62, + "grad_norm": 2.3282957077026367, + "learning_rate": 0.00019836556569660608, + "loss": 3.8853, + "step": 418 + }, + { + "epoch": 0.62, + "grad_norm": 5.4361186027526855, + "learning_rate": 0.0001983571094501491, + "loss": 3.8995, + "step": 419 + }, + { + "epoch": 0.62, + "grad_norm": 2.480142116546631, + "learning_rate": 0.00019834863156578724, + "loss": 3.9204, + "step": 420 + }, + { + "epoch": 0.63, + "grad_norm": 8.683335304260254, + "learning_rate": 0.0001983401320453855, + "loss": 4.04, + "step": 421 + }, + { + "epoch": 0.63, + "grad_norm": 10.6290864944458, + "learning_rate": 0.00019833161089081381, + "loss": 3.9174, + "step": 422 + }, + { + "epoch": 0.63, + "grad_norm": 1.7693558931350708, + "learning_rate": 0.00019832306810394664, + "loss": 3.937, + "step": 423 + }, + { + "epoch": 0.63, + "grad_norm": 6.862966537475586, + "learning_rate": 0.00019831450368666348, + "loss": 3.9314, + "step": 424 + }, + { + "epoch": 0.63, + "grad_norm": 6.495865345001221, + "learning_rate": 0.0001983059176408484, + "loss": 3.8863, + "step": 425 + }, + { + "epoch": 0.63, + "grad_norm": 2.8406107425689697, + "learning_rate": 0.0001982973099683902, + "loss": 3.9525, + "step": 426 + }, + { + "epoch": 0.63, + "grad_norm": 10.725767135620117, + "learning_rate": 0.00019828868067118255, + "loss": 3.894, + "step": 427 + }, + { + "epoch": 0.64, + "grad_norm": 12.51372241973877, + "learning_rate": 0.0001982800297511239, + "loss": 4.0417, + "step": 428 + }, + { + "epoch": 0.64, + "grad_norm": 5.210300445556641, + "learning_rate": 0.00019827135721011736, + "loss": 3.9486, + "step": 429 + }, + { + "epoch": 0.64, + "grad_norm": 8.358055114746094, + "learning_rate": 0.0001982626630500708, + "loss": 4.0152, + "step": 430 + }, + { + "epoch": 0.64, + "grad_norm": 6.729689598083496, + "learning_rate": 0.0001982539472728969, + "loss": 3.8904, + "step": 431 + }, + { + "epoch": 0.64, + "grad_norm": 4.138473033905029, + "learning_rate": 0.00019824520988051306, + "loss": 3.8004, + "step": 432 + }, + { + "epoch": 0.64, + "grad_norm": 6.071408271789551, + "learning_rate": 0.00019823645087484147, + "loss": 4.0332, + "step": 433 + }, + { + "epoch": 0.64, + "grad_norm": 4.820131301879883, + "learning_rate": 0.000198227670257809, + "loss": 4.0076, + "step": 434 + }, + { + "epoch": 0.65, + "grad_norm": 3.6242270469665527, + "learning_rate": 0.00019821886803134738, + "loss": 3.9052, + "step": 435 + }, + { + "epoch": 0.65, + "grad_norm": 4.531324863433838, + "learning_rate": 0.00019821004419739307, + "loss": 3.8389, + "step": 436 + }, + { + "epoch": 0.65, + "grad_norm": 3.6086678504943848, + "learning_rate": 0.00019820119875788716, + "loss": 3.8785, + "step": 437 + }, + { + "epoch": 0.65, + "grad_norm": 3.600398063659668, + "learning_rate": 0.00019819233171477566, + "loss": 3.8425, + "step": 438 + }, + { + "epoch": 0.65, + "grad_norm": 3.8471834659576416, + "learning_rate": 0.00019818344307000925, + "loss": 3.8667, + "step": 439 + }, + { + "epoch": 0.65, + "grad_norm": 3.2653310298919678, + "learning_rate": 0.00019817453282554333, + "loss": 3.8744, + "step": 440 + }, + { + "epoch": 0.65, + "grad_norm": 4.489375591278076, + "learning_rate": 0.00019816560098333817, + "loss": 3.8705, + "step": 441 + }, + { + "epoch": 0.66, + "grad_norm": 3.816256284713745, + "learning_rate": 0.00019815664754535862, + "loss": 3.7886, + "step": 442 + }, + { + "epoch": 0.66, + "grad_norm": 6.624678134918213, + "learning_rate": 0.00019814767251357448, + "loss": 3.865, + "step": 443 + }, + { + "epoch": 0.66, + "grad_norm": 6.498514652252197, + "learning_rate": 0.00019813867588996012, + "loss": 3.8771, + "step": 444 + }, + { + "epoch": 0.66, + "grad_norm": 3.340604305267334, + "learning_rate": 0.00019812965767649474, + "loss": 3.8211, + "step": 445 + }, + { + "epoch": 0.66, + "grad_norm": 5.6404805183410645, + "learning_rate": 0.0001981206178751623, + "loss": 3.8503, + "step": 446 + }, + { + "epoch": 0.66, + "grad_norm": 2.282567024230957, + "learning_rate": 0.00019811155648795153, + "loss": 3.865, + "step": 447 + }, + { + "epoch": 0.67, + "grad_norm": 4.082645893096924, + "learning_rate": 0.00019810247351685587, + "loss": 3.7833, + "step": 448 + }, + { + "epoch": 0.67, + "grad_norm": 2.7963573932647705, + "learning_rate": 0.00019809336896387344, + "loss": 3.7991, + "step": 449 + }, + { + "epoch": 0.67, + "grad_norm": 8.172618865966797, + "learning_rate": 0.00019808424283100726, + "loss": 3.8369, + "step": 450 + }, + { + "epoch": 0.67, + "grad_norm": 8.013967514038086, + "learning_rate": 0.000198075095120265, + "loss": 3.8526, + "step": 451 + }, + { + "epoch": 0.67, + "grad_norm": 0.9244366884231567, + "learning_rate": 0.00019806592583365907, + "loss": 3.7882, + "step": 452 + }, + { + "epoch": 0.67, + "grad_norm": 4.72100305557251, + "learning_rate": 0.00019805673497320662, + "loss": 3.7314, + "step": 453 + }, + { + "epoch": 0.67, + "grad_norm": 2.556580066680908, + "learning_rate": 0.00019804752254092968, + "loss": 3.8743, + "step": 454 + }, + { + "epoch": 0.68, + "grad_norm": 10.014892578125, + "learning_rate": 0.00019803828853885485, + "loss": 3.8263, + "step": 455 + }, + { + "epoch": 0.68, + "grad_norm": 12.621180534362793, + "learning_rate": 0.00019802903296901357, + "loss": 3.8423, + "step": 456 + }, + { + "epoch": 0.68, + "grad_norm": 6.725605010986328, + "learning_rate": 0.000198019755833442, + "loss": 3.7592, + "step": 457 + }, + { + "epoch": 0.68, + "grad_norm": 5.9739603996276855, + "learning_rate": 0.00019801045713418102, + "loss": 3.8252, + "step": 458 + }, + { + "epoch": 0.68, + "grad_norm": 6.546073913574219, + "learning_rate": 0.00019800113687327633, + "loss": 3.8135, + "step": 459 + }, + { + "epoch": 0.68, + "grad_norm": 1.279313325881958, + "learning_rate": 0.0001979917950527783, + "loss": 3.8242, + "step": 460 + }, + { + "epoch": 0.68, + "grad_norm": 14.773170471191406, + "learning_rate": 0.00019798243167474207, + "loss": 3.9442, + "step": 461 + }, + { + "epoch": 0.69, + "grad_norm": 20.262290954589844, + "learning_rate": 0.00019797304674122753, + "loss": 3.9303, + "step": 462 + }, + { + "epoch": 0.69, + "grad_norm": 13.257652282714844, + "learning_rate": 0.00019796364025429926, + "loss": 3.8547, + "step": 463 + }, + { + "epoch": 0.69, + "grad_norm": 1.5865988731384277, + "learning_rate": 0.00019795421221602668, + "loss": 3.781, + "step": 464 + }, + { + "epoch": 0.69, + "grad_norm": 4.687791347503662, + "learning_rate": 0.00019794476262848388, + "loss": 3.7924, + "step": 465 + }, + { + "epoch": 0.69, + "grad_norm": 1.8210766315460205, + "learning_rate": 0.0001979352914937497, + "loss": 3.8706, + "step": 466 + }, + { + "epoch": 0.69, + "grad_norm": 10.269219398498535, + "learning_rate": 0.00019792579881390768, + "loss": 3.8314, + "step": 467 + }, + { + "epoch": 0.7, + "grad_norm": 11.758221626281738, + "learning_rate": 0.00019791628459104625, + "loss": 3.8405, + "step": 468 + }, + { + "epoch": 0.7, + "grad_norm": 3.8874833583831787, + "learning_rate": 0.0001979067488272584, + "loss": 3.7889, + "step": 469 + }, + { + "epoch": 0.7, + "grad_norm": 8.348686218261719, + "learning_rate": 0.00019789719152464192, + "loss": 3.815, + "step": 470 + }, + { + "epoch": 0.7, + "grad_norm": 11.831920623779297, + "learning_rate": 0.0001978876126852994, + "loss": 3.8381, + "step": 471 + }, + { + "epoch": 0.7, + "grad_norm": 7.59721040725708, + "learning_rate": 0.00019787801231133814, + "loss": 3.8394, + "step": 472 + }, + { + "epoch": 0.7, + "grad_norm": 4.788586616516113, + "learning_rate": 0.00019786839040487005, + "loss": 3.8047, + "step": 473 + }, + { + "epoch": 0.7, + "grad_norm": 9.041956901550293, + "learning_rate": 0.00019785874696801202, + "loss": 3.7394, + "step": 474 + }, + { + "epoch": 0.71, + "grad_norm": 3.841315746307373, + "learning_rate": 0.00019784908200288546, + "loss": 3.7839, + "step": 475 + }, + { + "epoch": 0.71, + "grad_norm": 6.382961750030518, + "learning_rate": 0.0001978393955116166, + "loss": 3.7502, + "step": 476 + }, + { + "epoch": 0.71, + "grad_norm": 8.265791893005371, + "learning_rate": 0.00019782968749633642, + "loss": 3.7888, + "step": 477 + }, + { + "epoch": 0.71, + "grad_norm": 4.210597038269043, + "learning_rate": 0.00019781995795918063, + "loss": 3.7096, + "step": 478 + }, + { + "epoch": 0.71, + "grad_norm": 10.053391456604004, + "learning_rate": 0.00019781020690228963, + "loss": 3.7276, + "step": 479 + }, + { + "epoch": 0.71, + "grad_norm": 14.174814224243164, + "learning_rate": 0.0001978004343278086, + "loss": 3.891, + "step": 480 + }, + { + "epoch": 0.71, + "grad_norm": 7.2850661277771, + "learning_rate": 0.00019779064023788746, + "loss": 3.8149, + "step": 481 + }, + { + "epoch": 0.72, + "grad_norm": 5.421473503112793, + "learning_rate": 0.00019778082463468085, + "loss": 3.8328, + "step": 482 + }, + { + "epoch": 0.72, + "grad_norm": 8.712848663330078, + "learning_rate": 0.00019777098752034807, + "loss": 3.826, + "step": 483 + }, + { + "epoch": 0.72, + "grad_norm": 4.683780670166016, + "learning_rate": 0.00019776112889705334, + "loss": 3.7366, + "step": 484 + }, + { + "epoch": 0.72, + "grad_norm": 5.190871715545654, + "learning_rate": 0.00019775124876696536, + "loss": 3.7527, + "step": 485 + }, + { + "epoch": 0.72, + "grad_norm": 8.225262641906738, + "learning_rate": 0.00019774134713225778, + "loss": 3.8112, + "step": 486 + }, + { + "epoch": 0.72, + "grad_norm": 2.200317859649658, + "learning_rate": 0.00019773142399510883, + "loss": 3.7121, + "step": 487 + }, + { + "epoch": 0.72, + "grad_norm": 9.15087604522705, + "learning_rate": 0.0001977214793577016, + "loss": 3.7827, + "step": 488 + }, + { + "epoch": 0.73, + "grad_norm": 11.307198524475098, + "learning_rate": 0.0001977115132222238, + "loss": 3.7558, + "step": 489 + }, + { + "epoch": 0.73, + "grad_norm": 8.0049409866333, + "learning_rate": 0.00019770152559086794, + "loss": 3.7133, + "step": 490 + }, + { + "epoch": 0.73, + "grad_norm": 4.207780361175537, + "learning_rate": 0.0001976915164658312, + "loss": 3.7517, + "step": 491 + }, + { + "epoch": 0.73, + "grad_norm": 9.044132232666016, + "learning_rate": 0.00019768148584931554, + "loss": 3.7111, + "step": 492 + }, + { + "epoch": 0.73, + "grad_norm": 6.856382846832275, + "learning_rate": 0.00019767143374352762, + "loss": 3.7271, + "step": 493 + }, + { + "epoch": 0.73, + "grad_norm": 3.427558183670044, + "learning_rate": 0.00019766136015067886, + "loss": 3.7357, + "step": 494 + }, + { + "epoch": 0.74, + "grad_norm": 4.354124069213867, + "learning_rate": 0.00019765126507298534, + "loss": 3.7107, + "step": 495 + }, + { + "epoch": 0.74, + "grad_norm": 3.53426194190979, + "learning_rate": 0.00019764114851266796, + "loss": 3.7585, + "step": 496 + }, + { + "epoch": 0.74, + "grad_norm": 2.1305716037750244, + "learning_rate": 0.00019763101047195225, + "loss": 3.688, + "step": 497 + }, + { + "epoch": 0.74, + "grad_norm": 108.82621002197266, + "learning_rate": 0.00019762085095306856, + "loss": 3.6435, + "step": 498 + }, + { + "epoch": 0.74, + "grad_norm": 8.900164604187012, + "learning_rate": 0.00019761066995825183, + "loss": 3.7853, + "step": 499 + }, + { + "epoch": 0.74, + "grad_norm": 8.347526550292969, + "learning_rate": 0.0001976004674897419, + "loss": 3.7845, + "step": 500 + }, + { + "epoch": 0.74, + "grad_norm": 3.418429374694824, + "learning_rate": 0.00019759024354978322, + "loss": 3.7529, + "step": 501 + }, + { + "epoch": 0.75, + "grad_norm": 14.40571403503418, + "learning_rate": 0.00019757999814062495, + "loss": 3.795, + "step": 502 + }, + { + "epoch": 0.75, + "grad_norm": 21.624300003051758, + "learning_rate": 0.00019756973126452108, + "loss": 4.0035, + "step": 503 + }, + { + "epoch": 0.75, + "grad_norm": 14.849161148071289, + "learning_rate": 0.00019755944292373017, + "loss": 3.9246, + "step": 504 + }, + { + "epoch": 0.75, + "grad_norm": 7.422677993774414, + "learning_rate": 0.00019754913312051568, + "loss": 3.9148, + "step": 505 + }, + { + "epoch": 0.75, + "grad_norm": 11.331315994262695, + "learning_rate": 0.00019753880185714563, + "loss": 3.7997, + "step": 506 + }, + { + "epoch": 0.75, + "grad_norm": 34.16909408569336, + "learning_rate": 0.00019752844913589285, + "loss": 3.7348, + "step": 507 + }, + { + "epoch": 0.75, + "eval_loss": 3.593658447265625, + "eval_runtime": 1156.954, + "eval_samples_per_second": 14.816, + "eval_steps_per_second": 0.927, + "step": 507 + }, + { + "epoch": 0.75, + "grad_norm": 1562.1485595703125, + "learning_rate": 0.00019751807495903484, + "loss": 3.8149, + "step": 508 + }, + { + "epoch": 0.76, + "grad_norm": 12.771462440490723, + "learning_rate": 0.0001975076793288539, + "loss": 3.9449, + "step": 509 + }, + { + "epoch": 0.76, + "grad_norm": 7.921811580657959, + "learning_rate": 0.00019749726224763697, + "loss": 3.9061, + "step": 510 + }, + { + "epoch": 0.76, + "grad_norm": 4.624162197113037, + "learning_rate": 0.00019748682371767575, + "loss": 3.8572, + "step": 511 + }, + { + "epoch": 0.76, + "grad_norm": 33.93694305419922, + "learning_rate": 0.00019747636374126663, + "loss": 3.8013, + "step": 512 + }, + { + "epoch": 0.76, + "grad_norm": 67.59894561767578, + "learning_rate": 0.00019746588232071076, + "loss": 3.9281, + "step": 513 + }, + { + "epoch": 0.76, + "grad_norm": 18.626140594482422, + "learning_rate": 0.00019745537945831397, + "loss": 4.0322, + "step": 514 + }, + { + "epoch": 0.76, + "grad_norm": 9.813076972961426, + "learning_rate": 0.00019744485515638677, + "loss": 4.0781, + "step": 515 + }, + { + "epoch": 0.77, + "grad_norm": 12.949821472167969, + "learning_rate": 0.00019743430941724453, + "loss": 4.0615, + "step": 516 + }, + { + "epoch": 0.77, + "grad_norm": 9.89267635345459, + "learning_rate": 0.00019742374224320717, + "loss": 3.9752, + "step": 517 + }, + { + "epoch": 0.77, + "grad_norm": 4.396897792816162, + "learning_rate": 0.00019741315363659942, + "loss": 3.768, + "step": 518 + }, + { + "epoch": 0.77, + "grad_norm": 3.6423070430755615, + "learning_rate": 0.00019740254359975072, + "loss": 3.7554, + "step": 519 + }, + { + "epoch": 0.77, + "grad_norm": 5.352052688598633, + "learning_rate": 0.0001973919121349952, + "loss": 3.8254, + "step": 520 + }, + { + "epoch": 0.77, + "grad_norm": 4.971579074859619, + "learning_rate": 0.00019738125924467168, + "loss": 3.9223, + "step": 521 + }, + { + "epoch": 0.78, + "grad_norm": 3.2794764041900635, + "learning_rate": 0.00019737058493112376, + "loss": 3.7504, + "step": 522 + }, + { + "epoch": 0.78, + "grad_norm": 3.714524269104004, + "learning_rate": 0.00019735988919669968, + "loss": 3.7884, + "step": 523 + }, + { + "epoch": 0.78, + "grad_norm": 5.291292190551758, + "learning_rate": 0.0001973491720437525, + "loss": 3.8398, + "step": 524 + }, + { + "epoch": 0.78, + "grad_norm": 4.59107780456543, + "learning_rate": 0.00019733843347463984, + "loss": 3.9075, + "step": 525 + }, + { + "epoch": 0.78, + "grad_norm": 6.6806793212890625, + "learning_rate": 0.00019732767349172413, + "loss": 3.7432, + "step": 526 + }, + { + "epoch": 0.78, + "grad_norm": 48.79610824584961, + "learning_rate": 0.00019731689209737257, + "loss": 3.7873, + "step": 527 + }, + { + "epoch": 0.78, + "grad_norm": 7.798208236694336, + "learning_rate": 0.0001973060892939569, + "loss": 3.8176, + "step": 528 + }, + { + "epoch": 0.79, + "grad_norm": 7.387636661529541, + "learning_rate": 0.00019729526508385369, + "loss": 3.892, + "step": 529 + }, + { + "epoch": 0.79, + "grad_norm": 3.599566698074341, + "learning_rate": 0.00019728441946944422, + "loss": 3.754, + "step": 530 + }, + { + "epoch": 0.79, + "grad_norm": 6.45220947265625, + "learning_rate": 0.00019727355245311446, + "loss": 3.7462, + "step": 531 + }, + { + "epoch": 0.79, + "grad_norm": 6.4421563148498535, + "learning_rate": 0.00019726266403725503, + "loss": 3.8014, + "step": 532 + }, + { + "epoch": 0.79, + "grad_norm": 9.638814926147461, + "learning_rate": 0.00019725175422426135, + "loss": 3.9211, + "step": 533 + }, + { + "epoch": 0.79, + "grad_norm": 4.907318592071533, + "learning_rate": 0.00019724082301653345, + "loss": 3.8431, + "step": 534 + }, + { + "epoch": 0.79, + "grad_norm": 3.3494908809661865, + "learning_rate": 0.00019722987041647622, + "loss": 3.8407, + "step": 535 + }, + { + "epoch": 0.8, + "grad_norm": 2.774167060852051, + "learning_rate": 0.00019721889642649912, + "loss": 3.7459, + "step": 536 + }, + { + "epoch": 0.8, + "grad_norm": 4.917426586151123, + "learning_rate": 0.0001972079010490163, + "loss": 3.8871, + "step": 537 + }, + { + "epoch": 0.8, + "grad_norm": 2.627586841583252, + "learning_rate": 0.0001971968842864467, + "loss": 3.7812, + "step": 538 + }, + { + "epoch": 0.8, + "grad_norm": 6.549022197723389, + "learning_rate": 0.00019718584614121396, + "loss": 3.7288, + "step": 539 + }, + { + "epoch": 0.8, + "grad_norm": 13.424997329711914, + "learning_rate": 0.00019717478661574638, + "loss": 3.7931, + "step": 540 + }, + { + "epoch": 0.8, + "grad_norm": 10.896014213562012, + "learning_rate": 0.00019716370571247696, + "loss": 3.8668, + "step": 541 + }, + { + "epoch": 0.8, + "grad_norm": 7.297813415527344, + "learning_rate": 0.00019715260343384347, + "loss": 3.8884, + "step": 542 + }, + { + "epoch": 0.81, + "grad_norm": 3.8043060302734375, + "learning_rate": 0.00019714147978228832, + "loss": 3.8604, + "step": 543 + }, + { + "epoch": 0.81, + "grad_norm": 25.301651000976562, + "learning_rate": 0.0001971303347602586, + "loss": 3.8794, + "step": 544 + }, + { + "epoch": 0.81, + "grad_norm": 215.01597595214844, + "learning_rate": 0.0001971191683702062, + "loss": 3.9304, + "step": 545 + }, + { + "epoch": 0.81, + "grad_norm": 253.02139282226562, + "learning_rate": 0.0001971079806145876, + "loss": 4.3177, + "step": 546 + }, + { + "epoch": 0.81, + "grad_norm": 41.9672966003418, + "learning_rate": 0.00019709677149586409, + "loss": 4.7431, + "step": 547 + }, + { + "epoch": 0.81, + "grad_norm": 34.384029388427734, + "learning_rate": 0.0001970855410165015, + "loss": 5.1589, + "step": 548 + }, + { + "epoch": 0.82, + "grad_norm": 18.074525833129883, + "learning_rate": 0.00019707428917897058, + "loss": 5.4518, + "step": 549 + }, + { + "epoch": 0.82, + "grad_norm": 10.624735832214355, + "learning_rate": 0.0001970630159857466, + "loss": 5.0911, + "step": 550 + }, + { + "epoch": 0.82, + "grad_norm": 15.860513687133789, + "learning_rate": 0.00019705172143930953, + "loss": 4.8221, + "step": 551 + }, + { + "epoch": 0.82, + "grad_norm": 9.580286979675293, + "learning_rate": 0.00019704040554214422, + "loss": 4.7721, + "step": 552 + }, + { + "epoch": 0.82, + "grad_norm": 16.33072280883789, + "learning_rate": 0.00019702906829674002, + "loss": 4.7985, + "step": 553 + }, + { + "epoch": 0.82, + "grad_norm": 15.135660171508789, + "learning_rate": 0.00019701770970559105, + "loss": 4.7531, + "step": 554 + }, + { + "epoch": 0.82, + "grad_norm": 15.646889686584473, + "learning_rate": 0.0001970063297711961, + "loss": 4.6453, + "step": 555 + }, + { + "epoch": 0.83, + "grad_norm": 13.652432441711426, + "learning_rate": 0.00019699492849605874, + "loss": 4.671, + "step": 556 + }, + { + "epoch": 0.83, + "grad_norm": 18.923864364624023, + "learning_rate": 0.00019698350588268713, + "loss": 4.7306, + "step": 557 + }, + { + "epoch": 0.83, + "grad_norm": 15.4611177444458, + "learning_rate": 0.0001969720619335942, + "loss": 4.6082, + "step": 558 + }, + { + "epoch": 0.83, + "grad_norm": 6.054503440856934, + "learning_rate": 0.0001969605966512975, + "loss": 4.4091, + "step": 559 + }, + { + "epoch": 0.83, + "grad_norm": 11.314775466918945, + "learning_rate": 0.00019694911003831933, + "loss": 4.1236, + "step": 560 + }, + { + "epoch": 0.83, + "grad_norm": 12.968482971191406, + "learning_rate": 0.0001969376020971867, + "loss": 4.023, + "step": 561 + }, + { + "epoch": 0.83, + "grad_norm": 8.118325233459473, + "learning_rate": 0.00019692607283043124, + "loss": 3.9331, + "step": 562 + }, + { + "epoch": 0.84, + "grad_norm": 4.455306529998779, + "learning_rate": 0.00019691452224058933, + "loss": 3.9195, + "step": 563 + }, + { + "epoch": 0.84, + "grad_norm": 5.133811950683594, + "learning_rate": 0.000196902950330202, + "loss": 3.8438, + "step": 564 + }, + { + "epoch": 0.84, + "grad_norm": 4.062127590179443, + "learning_rate": 0.00019689135710181503, + "loss": 3.7435, + "step": 565 + }, + { + "epoch": 0.84, + "grad_norm": 4.880388259887695, + "learning_rate": 0.00019687974255797885, + "loss": 3.7828, + "step": 566 + }, + { + "epoch": 0.84, + "grad_norm": 5.3774027824401855, + "learning_rate": 0.00019686810670124856, + "loss": 3.7977, + "step": 567 + }, + { + "epoch": 0.84, + "grad_norm": 5.431920528411865, + "learning_rate": 0.00019685644953418397, + "loss": 3.8113, + "step": 568 + }, + { + "epoch": 0.84, + "grad_norm": 3.2028257846832275, + "learning_rate": 0.00019684477105934956, + "loss": 3.7359, + "step": 569 + }, + { + "epoch": 0.85, + "grad_norm": 3.04168701171875, + "learning_rate": 0.00019683307127931458, + "loss": 3.6924, + "step": 570 + }, + { + "epoch": 0.85, + "grad_norm": 3.9302849769592285, + "learning_rate": 0.00019682135019665286, + "loss": 3.7543, + "step": 571 + }, + { + "epoch": 0.85, + "grad_norm": 3.5611188411712646, + "learning_rate": 0.00019680960781394297, + "loss": 3.751, + "step": 572 + }, + { + "epoch": 0.85, + "grad_norm": 2.1138036251068115, + "learning_rate": 0.00019679784413376811, + "loss": 3.7196, + "step": 573 + }, + { + "epoch": 0.85, + "grad_norm": 3.7343742847442627, + "learning_rate": 0.0001967860591587163, + "loss": 3.6888, + "step": 574 + }, + { + "epoch": 0.85, + "grad_norm": 3.8373682498931885, + "learning_rate": 0.00019677425289138007, + "loss": 3.6819, + "step": 575 + }, + { + "epoch": 0.86, + "grad_norm": 3.580359935760498, + "learning_rate": 0.00019676242533435678, + "loss": 3.662, + "step": 576 + }, + { + "epoch": 0.86, + "grad_norm": 3.4737274646759033, + "learning_rate": 0.00019675057649024837, + "loss": 3.6293, + "step": 577 + }, + { + "epoch": 0.86, + "grad_norm": 2.3330490589141846, + "learning_rate": 0.00019673870636166156, + "loss": 3.6554, + "step": 578 + }, + { + "epoch": 0.86, + "grad_norm": 4.6455512046813965, + "learning_rate": 0.00019672681495120763, + "loss": 3.6991, + "step": 579 + }, + { + "epoch": 0.86, + "grad_norm": 2.803642988204956, + "learning_rate": 0.00019671490226150264, + "loss": 3.6473, + "step": 580 + }, + { + "epoch": 0.86, + "grad_norm": 5.52310848236084, + "learning_rate": 0.00019670296829516732, + "loss": 3.6435, + "step": 581 + }, + { + "epoch": 0.86, + "grad_norm": 5.066474914550781, + "learning_rate": 0.00019669101305482705, + "loss": 3.6769, + "step": 582 + }, + { + "epoch": 0.87, + "grad_norm": 2.860041379928589, + "learning_rate": 0.00019667903654311194, + "loss": 3.6205, + "step": 583 + }, + { + "epoch": 0.87, + "grad_norm": 2.413363456726074, + "learning_rate": 0.00019666703876265666, + "loss": 3.6287, + "step": 584 + }, + { + "epoch": 0.87, + "grad_norm": 4.433047771453857, + "learning_rate": 0.0001966550197161007, + "loss": 3.6912, + "step": 585 + }, + { + "epoch": 0.87, + "grad_norm": 117.41354370117188, + "learning_rate": 0.00019664297940608812, + "loss": 3.6996, + "step": 586 + }, + { + "epoch": 0.87, + "grad_norm": 2.949551582336426, + "learning_rate": 0.0001966309178352678, + "loss": 3.6726, + "step": 587 + }, + { + "epoch": 0.87, + "grad_norm": 6.152908802032471, + "learning_rate": 0.0001966188350062931, + "loss": 3.6525, + "step": 588 + }, + { + "epoch": 0.87, + "grad_norm": 6.42512321472168, + "learning_rate": 0.00019660673092182225, + "loss": 3.6739, + "step": 589 + }, + { + "epoch": 0.88, + "grad_norm": 3.4023945331573486, + "learning_rate": 0.000196594605584518, + "loss": 3.6757, + "step": 590 + }, + { + "epoch": 0.88, + "grad_norm": 4.428458213806152, + "learning_rate": 0.00019658245899704789, + "loss": 3.6438, + "step": 591 + }, + { + "epoch": 0.88, + "grad_norm": 4.402431488037109, + "learning_rate": 0.00019657029116208405, + "loss": 3.6902, + "step": 592 + }, + { + "epoch": 0.88, + "grad_norm": 4.742648124694824, + "learning_rate": 0.00019655810208230333, + "loss": 3.5998, + "step": 593 + }, + { + "epoch": 0.88, + "grad_norm": 1.4105639457702637, + "learning_rate": 0.00019654589176038728, + "loss": 3.5836, + "step": 594 + }, + { + "epoch": 0.88, + "grad_norm": 5.09644079208374, + "learning_rate": 0.00019653366019902206, + "loss": 3.616, + "step": 595 + }, + { + "epoch": 0.89, + "grad_norm": 5.025322914123535, + "learning_rate": 0.00019652140740089854, + "loss": 3.6783, + "step": 596 + }, + { + "epoch": 0.89, + "grad_norm": 4.536464691162109, + "learning_rate": 0.00019650913336871222, + "loss": 3.6274, + "step": 597 + }, + { + "epoch": 0.89, + "grad_norm": 3.1868488788604736, + "learning_rate": 0.00019649683810516336, + "loss": 3.6113, + "step": 598 + }, + { + "epoch": 0.89, + "grad_norm": 4.735551834106445, + "learning_rate": 0.00019648452161295682, + "loss": 3.5986, + "step": 599 + }, + { + "epoch": 0.89, + "grad_norm": 4.411576747894287, + "learning_rate": 0.0001964721838948021, + "loss": 3.5865, + "step": 600 + }, + { + "epoch": 0.89, + "grad_norm": 2.665224075317383, + "learning_rate": 0.00019645982495341348, + "loss": 3.5546, + "step": 601 + }, + { + "epoch": 0.89, + "grad_norm": 4.546963691711426, + "learning_rate": 0.00019644744479150981, + "loss": 3.6332, + "step": 602 + }, + { + "epoch": 0.9, + "grad_norm": 1.664005160331726, + "learning_rate": 0.00019643504341181466, + "loss": 3.6299, + "step": 603 + }, + { + "epoch": 0.9, + "grad_norm": 2.6972901821136475, + "learning_rate": 0.0001964226208170562, + "loss": 3.6158, + "step": 604 + }, + { + "epoch": 0.9, + "grad_norm": 3.175882339477539, + "learning_rate": 0.00019641017700996736, + "loss": 3.5895, + "step": 605 + }, + { + "epoch": 0.9, + "grad_norm": 3.3170359134674072, + "learning_rate": 0.00019639771199328574, + "loss": 3.5989, + "step": 606 + }, + { + "epoch": 0.9, + "grad_norm": 3.245979070663452, + "learning_rate": 0.00019638522576975346, + "loss": 3.5763, + "step": 607 + }, + { + "epoch": 0.9, + "grad_norm": 1.0223109722137451, + "learning_rate": 0.00019637271834211748, + "loss": 3.5668, + "step": 608 + }, + { + "epoch": 0.9, + "grad_norm": 1.9072734117507935, + "learning_rate": 0.00019636018971312932, + "loss": 3.5738, + "step": 609 + }, + { + "epoch": 0.91, + "grad_norm": 2.925701141357422, + "learning_rate": 0.00019634763988554522, + "loss": 3.616, + "step": 610 + }, + { + "epoch": 0.91, + "grad_norm": 2.6939620971679688, + "learning_rate": 0.00019633506886212603, + "loss": 3.5764, + "step": 611 + }, + { + "epoch": 0.91, + "grad_norm": 2.020845413208008, + "learning_rate": 0.0001963224766456373, + "loss": 3.5868, + "step": 612 + }, + { + "epoch": 0.91, + "grad_norm": 3.805741548538208, + "learning_rate": 0.0001963098632388492, + "loss": 3.5979, + "step": 613 + }, + { + "epoch": 0.91, + "grad_norm": 1.6603795289993286, + "learning_rate": 0.00019629722864453663, + "loss": 3.5792, + "step": 614 + }, + { + "epoch": 0.91, + "grad_norm": 8.938972473144531, + "learning_rate": 0.00019628457286547914, + "loss": 3.5978, + "step": 615 + }, + { + "epoch": 0.91, + "grad_norm": 9.437355041503906, + "learning_rate": 0.0001962718959044609, + "loss": 3.6115, + "step": 616 + }, + { + "epoch": 0.92, + "grad_norm": 1.7519173622131348, + "learning_rate": 0.0001962591977642707, + "loss": 3.5745, + "step": 617 + }, + { + "epoch": 0.92, + "grad_norm": 8.598402976989746, + "learning_rate": 0.00019624647844770213, + "loss": 3.5939, + "step": 618 + }, + { + "epoch": 0.92, + "grad_norm": 11.709813117980957, + "learning_rate": 0.0001962337379575533, + "loss": 3.7499, + "step": 619 + }, + { + "epoch": 0.92, + "grad_norm": 8.366117477416992, + "learning_rate": 0.00019622097629662708, + "loss": 3.5727, + "step": 620 + }, + { + "epoch": 0.92, + "grad_norm": 2.4304378032684326, + "learning_rate": 0.0001962081934677309, + "loss": 3.6257, + "step": 621 + }, + { + "epoch": 0.92, + "grad_norm": 6.7245025634765625, + "learning_rate": 0.00019619538947367692, + "loss": 3.5817, + "step": 622 + }, + { + "epoch": 0.93, + "grad_norm": 2.964221239089966, + "learning_rate": 0.00019618256431728194, + "loss": 3.5903, + "step": 623 + }, + { + "epoch": 0.93, + "grad_norm": 5.846991539001465, + "learning_rate": 0.00019616971800136734, + "loss": 3.5198, + "step": 624 + }, + { + "epoch": 0.93, + "grad_norm": 6.497872829437256, + "learning_rate": 0.00019615685052875933, + "loss": 3.5556, + "step": 625 + }, + { + "epoch": 0.93, + "grad_norm": 1.711037039756775, + "learning_rate": 0.00019614396190228864, + "loss": 3.5108, + "step": 626 + }, + { + "epoch": 0.93, + "grad_norm": 9.816991806030273, + "learning_rate": 0.0001961310521247906, + "loss": 3.5854, + "step": 627 + }, + { + "epoch": 0.93, + "grad_norm": 12.794452667236328, + "learning_rate": 0.00019611812119910537, + "loss": 3.6549, + "step": 628 + }, + { + "epoch": 0.93, + "grad_norm": 6.7332916259765625, + "learning_rate": 0.0001961051691280776, + "loss": 3.6605, + "step": 629 + }, + { + "epoch": 0.94, + "grad_norm": 6.012147426605225, + "learning_rate": 0.00019609219591455675, + "loss": 3.6014, + "step": 630 + }, + { + "epoch": 0.94, + "grad_norm": 6.8346381187438965, + "learning_rate": 0.00019607920156139671, + "loss": 3.5713, + "step": 631 + }, + { + "epoch": 0.94, + "grad_norm": 3.088141679763794, + "learning_rate": 0.0001960661860714563, + "loss": 3.5767, + "step": 632 + }, + { + "epoch": 0.94, + "grad_norm": 8.472504615783691, + "learning_rate": 0.0001960531494475987, + "loss": 3.6236, + "step": 633 + }, + { + "epoch": 0.94, + "grad_norm": 9.718921661376953, + "learning_rate": 0.00019604009169269193, + "loss": 3.5673, + "step": 634 + }, + { + "epoch": 0.94, + "grad_norm": 2.083310127258301, + "learning_rate": 0.00019602701280960866, + "loss": 3.5786, + "step": 635 + }, + { + "epoch": 0.94, + "grad_norm": 8.658992767333984, + "learning_rate": 0.0001960139128012261, + "loss": 3.6109, + "step": 636 + }, + { + "epoch": 0.95, + "grad_norm": 10.503134727478027, + "learning_rate": 0.0001960007916704262, + "loss": 3.6035, + "step": 637 + }, + { + "epoch": 0.95, + "grad_norm": 7.250131130218506, + "learning_rate": 0.00019598764942009548, + "loss": 3.6211, + "step": 638 + }, + { + "epoch": 0.95, + "grad_norm": 4.671655178070068, + "learning_rate": 0.00019597448605312515, + "loss": 3.5175, + "step": 639 + }, + { + "epoch": 0.95, + "grad_norm": 9.631921768188477, + "learning_rate": 0.00019596130157241112, + "loss": 3.6144, + "step": 640 + }, + { + "epoch": 0.95, + "grad_norm": 5.835899353027344, + "learning_rate": 0.00019594809598085381, + "loss": 3.5662, + "step": 641 + }, + { + "epoch": 0.95, + "grad_norm": 5.0995893478393555, + "learning_rate": 0.00019593486928135841, + "loss": 3.5438, + "step": 642 + }, + { + "epoch": 0.95, + "grad_norm": 6.497475624084473, + "learning_rate": 0.00019592162147683473, + "loss": 3.536, + "step": 643 + }, + { + "epoch": 0.96, + "grad_norm": 2.704535722732544, + "learning_rate": 0.00019590835257019714, + "loss": 3.5268, + "step": 644 + }, + { + "epoch": 0.96, + "grad_norm": 8.847089767456055, + "learning_rate": 0.00019589506256436473, + "loss": 3.6095, + "step": 645 + }, + { + "epoch": 0.96, + "grad_norm": 10.879183769226074, + "learning_rate": 0.00019588175146226122, + "loss": 3.657, + "step": 646 + }, + { + "epoch": 0.96, + "grad_norm": 4.468491077423096, + "learning_rate": 0.00019586841926681495, + "loss": 3.5589, + "step": 647 + }, + { + "epoch": 0.96, + "grad_norm": 6.55201530456543, + "learning_rate": 0.00019585506598095897, + "loss": 3.5587, + "step": 648 + }, + { + "epoch": 0.96, + "grad_norm": 8.522590637207031, + "learning_rate": 0.0001958416916076308, + "loss": 3.5354, + "step": 649 + }, + { + "epoch": 0.97, + "grad_norm": 4.8020806312561035, + "learning_rate": 0.00019582829614977283, + "loss": 3.506, + "step": 650 + }, + { + "epoch": 0.97, + "grad_norm": 7.987406253814697, + "learning_rate": 0.00019581487961033188, + "loss": 3.5615, + "step": 651 + }, + { + "epoch": 0.97, + "grad_norm": 11.70871353149414, + "learning_rate": 0.00019580144199225953, + "loss": 3.5993, + "step": 652 + }, + { + "epoch": 0.97, + "grad_norm": 4.69012975692749, + "learning_rate": 0.00019578798329851203, + "loss": 3.542, + "step": 653 + }, + { + "epoch": 0.97, + "grad_norm": 6.301438331604004, + "learning_rate": 0.0001957745035320501, + "loss": 3.5313, + "step": 654 + }, + { + "epoch": 0.97, + "grad_norm": 8.368910789489746, + "learning_rate": 0.0001957610026958393, + "loss": 3.5855, + "step": 655 + }, + { + "epoch": 0.97, + "grad_norm": 5.793453693389893, + "learning_rate": 0.00019574748079284962, + "loss": 3.5403, + "step": 656 + }, + { + "epoch": 0.98, + "grad_norm": 4.990311622619629, + "learning_rate": 0.00019573393782605582, + "loss": 3.5196, + "step": 657 + }, + { + "epoch": 0.98, + "grad_norm": 9.033957481384277, + "learning_rate": 0.0001957203737984373, + "loss": 3.528, + "step": 658 + }, + { + "epoch": 0.98, + "grad_norm": 3.766148805618286, + "learning_rate": 0.00019570678871297804, + "loss": 3.5015, + "step": 659 + }, + { + "epoch": 0.98, + "grad_norm": 7.3363518714904785, + "learning_rate": 0.00019569318257266666, + "loss": 3.5311, + "step": 660 + }, + { + "epoch": 0.98, + "grad_norm": 9.119929313659668, + "learning_rate": 0.00019567955538049643, + "loss": 3.5767, + "step": 661 + }, + { + "epoch": 0.98, + "grad_norm": 5.856534957885742, + "learning_rate": 0.00019566590713946522, + "loss": 3.5836, + "step": 662 + }, + { + "epoch": 0.98, + "grad_norm": 4.941824436187744, + "learning_rate": 0.00019565223785257557, + "loss": 3.5366, + "step": 663 + }, + { + "epoch": 0.99, + "grad_norm": 8.7649507522583, + "learning_rate": 0.00019563854752283465, + "loss": 3.5562, + "step": 664 + }, + { + "epoch": 0.99, + "grad_norm": 3.0613012313842773, + "learning_rate": 0.00019562483615325418, + "loss": 3.5715, + "step": 665 + }, + { + "epoch": 0.99, + "grad_norm": 6.858389377593994, + "learning_rate": 0.00019561110374685062, + "loss": 3.4948, + "step": 666 + }, + { + "epoch": 0.99, + "grad_norm": 8.862404823303223, + "learning_rate": 0.00019559735030664496, + "loss": 3.5523, + "step": 667 + }, + { + "epoch": 0.99, + "grad_norm": 5.061213970184326, + "learning_rate": 0.00019558357583566297, + "loss": 3.4475, + "step": 668 + }, + { + "epoch": 0.99, + "grad_norm": 7.312166690826416, + "learning_rate": 0.0001955697803369348, + "loss": 3.5274, + "step": 669 + }, + { + "epoch": 0.99, + "grad_norm": 11.606597900390625, + "learning_rate": 0.00019555596381349547, + "loss": 3.5426, + "step": 670 + }, + { + "epoch": 1.0, + "grad_norm": 5.651201248168945, + "learning_rate": 0.0001955421262683845, + "loss": 3.533, + "step": 671 + }, + { + "epoch": 1.0, + "grad_norm": 5.392771244049072, + "learning_rate": 0.00019552826770464598, + "loss": 3.4873, + "step": 672 + }, + { + "epoch": 1.0, + "grad_norm": 7.194543361663818, + "learning_rate": 0.00019551438812532882, + "loss": 3.5313, + "step": 673 + }, + { + "epoch": 1.0, + "grad_norm": 3.8677308559417725, + "learning_rate": 0.00019550048753348638, + "loss": 3.5539, + "step": 674 + }, + { + "epoch": 1.0, + "grad_norm": 8.319040298461914, + "learning_rate": 0.0001954865659321767, + "loss": 3.5204, + "step": 675 + }, + { + "epoch": 1.0, + "grad_norm": 12.673453330993652, + "learning_rate": 0.00019547262332446241, + "loss": 3.4573, + "step": 676 + }, + { + "epoch": 1.0, + "eval_loss": 3.3125858306884766, + "eval_runtime": 1160.3441, + "eval_samples_per_second": 14.773, + "eval_steps_per_second": 0.924, + "step": 676 + }, + { + "epoch": 1.01, + "grad_norm": 6.508894920349121, + "learning_rate": 0.00019545865971341085, + "loss": 3.5108, + "step": 677 + }, + { + "epoch": 1.01, + "grad_norm": 5.700242519378662, + "learning_rate": 0.00019544467510209388, + "loss": 3.4928, + "step": 678 + }, + { + "epoch": 1.01, + "grad_norm": 7.889042854309082, + "learning_rate": 0.00019543066949358803, + "loss": 3.5309, + "step": 679 + }, + { + "epoch": 1.01, + "grad_norm": 4.5509724617004395, + "learning_rate": 0.0001954166428909744, + "loss": 3.5496, + "step": 680 + }, + { + "epoch": 1.01, + "grad_norm": 7.286065578460693, + "learning_rate": 0.00019540259529733883, + "loss": 3.5685, + "step": 681 + }, + { + "epoch": 1.01, + "grad_norm": 10.92095947265625, + "learning_rate": 0.00019538852671577165, + "loss": 3.5438, + "step": 682 + }, + { + "epoch": 1.01, + "grad_norm": 4.9438557624816895, + "learning_rate": 0.00019537443714936787, + "loss": 3.4908, + "step": 683 + }, + { + "epoch": 1.0, + "grad_norm": 6.148499011993408, + "learning_rate": 0.00019536032660122707, + "loss": 3.5438, + "step": 684 + }, + { + "epoch": 1.0, + "grad_norm": 8.24433708190918, + "learning_rate": 0.00019534619507445352, + "loss": 3.6016, + "step": 685 + }, + { + "epoch": 1.0, + "grad_norm": 5.07331657409668, + "learning_rate": 0.000195332042572156, + "loss": 3.6093, + "step": 686 + }, + { + "epoch": 1.01, + "grad_norm": 7.187658309936523, + "learning_rate": 0.00019531786909744803, + "loss": 3.5025, + "step": 687 + }, + { + "epoch": 1.01, + "grad_norm": 10.835110664367676, + "learning_rate": 0.0001953036746534476, + "loss": 3.5561, + "step": 688 + }, + { + "epoch": 1.01, + "grad_norm": 4.704497814178467, + "learning_rate": 0.00019528945924327751, + "loss": 3.5617, + "step": 689 + }, + { + "epoch": 1.01, + "grad_norm": 6.398228168487549, + "learning_rate": 0.00019527522287006495, + "loss": 3.5302, + "step": 690 + }, + { + "epoch": 1.01, + "grad_norm": 8.401050567626953, + "learning_rate": 0.00019526096553694187, + "loss": 3.5882, + "step": 691 + }, + { + "epoch": 1.01, + "grad_norm": 4.914839267730713, + "learning_rate": 0.00019524668724704478, + "loss": 3.4931, + "step": 692 + }, + { + "epoch": 1.01, + "grad_norm": 6.374270915985107, + "learning_rate": 0.00019523238800351483, + "loss": 3.4807, + "step": 693 + }, + { + "epoch": 1.02, + "grad_norm": 9.898938179016113, + "learning_rate": 0.0001952180678094977, + "loss": 3.5138, + "step": 694 + }, + { + "epoch": 1.02, + "grad_norm": 4.583561897277832, + "learning_rate": 0.00019520372666814378, + "loss": 3.4739, + "step": 695 + }, + { + "epoch": 1.02, + "grad_norm": 6.307168483734131, + "learning_rate": 0.00019518936458260807, + "loss": 3.5473, + "step": 696 + }, + { + "epoch": 1.02, + "grad_norm": 8.191679000854492, + "learning_rate": 0.00019517498155605003, + "loss": 3.5002, + "step": 697 + }, + { + "epoch": 1.02, + "grad_norm": 5.078609466552734, + "learning_rate": 0.0001951605775916339, + "loss": 3.4731, + "step": 698 + }, + { + "epoch": 1.02, + "grad_norm": 6.6793060302734375, + "learning_rate": 0.00019514615269252845, + "loss": 3.5543, + "step": 699 + }, + { + "epoch": 1.02, + "grad_norm": 10.510246276855469, + "learning_rate": 0.00019513170686190706, + "loss": 3.5589, + "step": 700 + }, + { + "epoch": 1.03, + "grad_norm": 4.769564151763916, + "learning_rate": 0.00019511724010294766, + "loss": 3.4587, + "step": 701 + }, + { + "epoch": 1.03, + "grad_norm": 6.157220363616943, + "learning_rate": 0.00019510275241883292, + "loss": 3.4658, + "step": 702 + }, + { + "epoch": 1.03, + "grad_norm": 8.317898750305176, + "learning_rate": 0.00019508824381274996, + "loss": 3.5147, + "step": 703 + }, + { + "epoch": 1.03, + "grad_norm": 4.906617164611816, + "learning_rate": 0.00019507371428789064, + "loss": 3.5319, + "step": 704 + }, + { + "epoch": 1.03, + "grad_norm": 6.44058084487915, + "learning_rate": 0.00019505916384745134, + "loss": 3.5268, + "step": 705 + }, + { + "epoch": 1.03, + "grad_norm": 10.473220825195312, + "learning_rate": 0.00019504459249463304, + "loss": 3.535, + "step": 706 + }, + { + "epoch": 1.04, + "grad_norm": 4.5666656494140625, + "learning_rate": 0.00019503000023264136, + "loss": 3.4976, + "step": 707 + }, + { + "epoch": 1.04, + "grad_norm": 6.2709221839904785, + "learning_rate": 0.0001950153870646865, + "loss": 3.5246, + "step": 708 + }, + { + "epoch": 1.04, + "grad_norm": 8.397104263305664, + "learning_rate": 0.00019500075299398324, + "loss": 3.5369, + "step": 709 + }, + { + "epoch": 1.04, + "grad_norm": 4.958198547363281, + "learning_rate": 0.00019498609802375097, + "loss": 3.4644, + "step": 710 + }, + { + "epoch": 1.04, + "grad_norm": 6.521581172943115, + "learning_rate": 0.00019497142215721373, + "loss": 3.4931, + "step": 711 + }, + { + "epoch": 1.04, + "grad_norm": 10.69038200378418, + "learning_rate": 0.00019495672539760007, + "loss": 3.4746, + "step": 712 + }, + { + "epoch": 1.04, + "grad_norm": 5.08820915222168, + "learning_rate": 0.0001949420077481432, + "loss": 3.4963, + "step": 713 + }, + { + "epoch": 1.05, + "grad_norm": 5.867018222808838, + "learning_rate": 0.00019492726921208092, + "loss": 3.4735, + "step": 714 + }, + { + "epoch": 1.05, + "grad_norm": 7.802560329437256, + "learning_rate": 0.00019491250979265554, + "loss": 3.514, + "step": 715 + }, + { + "epoch": 1.05, + "grad_norm": 4.548806190490723, + "learning_rate": 0.0001948977294931141, + "loss": 3.4695, + "step": 716 + }, + { + "epoch": 1.05, + "grad_norm": 7.293185710906982, + "learning_rate": 0.00019488292831670818, + "loss": 3.5301, + "step": 717 + }, + { + "epoch": 1.05, + "grad_norm": 10.711014747619629, + "learning_rate": 0.00019486810626669388, + "loss": 3.5166, + "step": 718 + }, + { + "epoch": 1.05, + "grad_norm": 5.207925796508789, + "learning_rate": 0.00019485326334633198, + "loss": 3.4741, + "step": 719 + }, + { + "epoch": 1.05, + "grad_norm": 5.915352821350098, + "learning_rate": 0.00019483839955888784, + "loss": 3.4837, + "step": 720 + }, + { + "epoch": 1.06, + "grad_norm": 8.018194198608398, + "learning_rate": 0.00019482351490763135, + "loss": 3.4991, + "step": 721 + }, + { + "epoch": 1.06, + "grad_norm": 4.433293342590332, + "learning_rate": 0.00019480860939583705, + "loss": 3.417, + "step": 722 + }, + { + "epoch": 1.06, + "grad_norm": 6.9495038986206055, + "learning_rate": 0.0001947936830267841, + "loss": 3.5061, + "step": 723 + }, + { + "epoch": 1.06, + "grad_norm": 10.787668228149414, + "learning_rate": 0.00019477873580375614, + "loss": 3.5036, + "step": 724 + }, + { + "epoch": 1.06, + "grad_norm": 5.106097221374512, + "learning_rate": 0.0001947637677300415, + "loss": 3.4732, + "step": 725 + }, + { + "epoch": 1.06, + "grad_norm": 5.890613555908203, + "learning_rate": 0.00019474877880893302, + "loss": 3.5222, + "step": 726 + }, + { + "epoch": 1.06, + "grad_norm": 7.86448335647583, + "learning_rate": 0.0001947337690437282, + "loss": 3.4886, + "step": 727 + }, + { + "epoch": 1.07, + "grad_norm": 4.345864772796631, + "learning_rate": 0.00019471873843772904, + "loss": 3.4902, + "step": 728 + }, + { + "epoch": 1.07, + "grad_norm": 7.271783351898193, + "learning_rate": 0.00019470368699424218, + "loss": 3.4832, + "step": 729 + }, + { + "epoch": 1.07, + "grad_norm": 11.227890968322754, + "learning_rate": 0.0001946886147165789, + "loss": 3.4815, + "step": 730 + }, + { + "epoch": 1.07, + "grad_norm": 5.0216264724731445, + "learning_rate": 0.00019467352160805498, + "loss": 3.4912, + "step": 731 + }, + { + "epoch": 1.07, + "grad_norm": 5.998202800750732, + "learning_rate": 0.00019465840767199073, + "loss": 3.4653, + "step": 732 + }, + { + "epoch": 1.07, + "grad_norm": 7.844666004180908, + "learning_rate": 0.00019464327291171117, + "loss": 3.5708, + "step": 733 + }, + { + "epoch": 1.08, + "grad_norm": 4.441153526306152, + "learning_rate": 0.00019462811733054588, + "loss": 3.479, + "step": 734 + }, + { + "epoch": 1.08, + "grad_norm": 7.58824348449707, + "learning_rate": 0.0001946129409318289, + "loss": 3.4787, + "step": 735 + }, + { + "epoch": 1.08, + "grad_norm": 11.293692588806152, + "learning_rate": 0.00019459774371889901, + "loss": 3.5108, + "step": 736 + }, + { + "epoch": 1.08, + "grad_norm": 6.113391399383545, + "learning_rate": 0.00019458252569509946, + "loss": 3.4456, + "step": 737 + }, + { + "epoch": 1.08, + "grad_norm": 4.897597312927246, + "learning_rate": 0.00019456728686377813, + "loss": 3.5001, + "step": 738 + }, + { + "epoch": 1.08, + "grad_norm": 6.808078765869141, + "learning_rate": 0.00019455202722828743, + "loss": 3.5318, + "step": 739 + }, + { + "epoch": 1.08, + "grad_norm": 2.7436025142669678, + "learning_rate": 0.0001945367467919844, + "loss": 3.452, + "step": 740 + }, + { + "epoch": 1.09, + "grad_norm": 10.119528770446777, + "learning_rate": 0.00019452144555823065, + "loss": 3.5012, + "step": 741 + }, + { + "epoch": 1.09, + "grad_norm": 15.063286781311035, + "learning_rate": 0.00019450612353039235, + "loss": 3.5615, + "step": 742 + }, + { + "epoch": 1.09, + "grad_norm": 9.542817115783691, + "learning_rate": 0.0001944907807118402, + "loss": 3.5215, + "step": 743 + }, + { + "epoch": 1.09, + "grad_norm": 3.414539098739624, + "learning_rate": 0.00019447541710594952, + "loss": 3.4854, + "step": 744 + }, + { + "epoch": 1.09, + "grad_norm": 5.853447437286377, + "learning_rate": 0.00019446003271610026, + "loss": 3.4727, + "step": 745 + }, + { + "epoch": 1.09, + "grad_norm": 2.5720930099487305, + "learning_rate": 0.00019444462754567682, + "loss": 3.4853, + "step": 746 + }, + { + "epoch": 1.09, + "grad_norm": 8.554189682006836, + "learning_rate": 0.0001944292015980683, + "loss": 3.4712, + "step": 747 + }, + { + "epoch": 1.1, + "grad_norm": 11.265769958496094, + "learning_rate": 0.00019441375487666822, + "loss": 3.5236, + "step": 748 + }, + { + "epoch": 1.1, + "grad_norm": 4.981095314025879, + "learning_rate": 0.00019439828738487483, + "loss": 3.5258, + "step": 749 + }, + { + "epoch": 1.1, + "grad_norm": 5.837092876434326, + "learning_rate": 0.00019438279912609082, + "loss": 3.5503, + "step": 750 + }, + { + "epoch": 1.1, + "grad_norm": 8.067007064819336, + "learning_rate": 0.00019436729010372355, + "loss": 3.5129, + "step": 751 + }, + { + "epoch": 1.1, + "grad_norm": 4.977865219116211, + "learning_rate": 0.00019435176032118486, + "loss": 3.4686, + "step": 752 + }, + { + "epoch": 1.1, + "grad_norm": 6.810112953186035, + "learning_rate": 0.00019433620978189122, + "loss": 3.4424, + "step": 753 + }, + { + "epoch": 1.1, + "grad_norm": 10.288273811340332, + "learning_rate": 0.00019432063848926365, + "loss": 3.4829, + "step": 754 + }, + { + "epoch": 1.11, + "grad_norm": 4.5541839599609375, + "learning_rate": 0.0001943050464467277, + "loss": 3.4592, + "step": 755 + }, + { + "epoch": 1.11, + "grad_norm": 5.883044719696045, + "learning_rate": 0.0001942894336577136, + "loss": 3.5044, + "step": 756 + }, + { + "epoch": 1.11, + "grad_norm": 7.786341667175293, + "learning_rate": 0.00019427380012565593, + "loss": 3.5129, + "step": 757 + }, + { + "epoch": 1.11, + "grad_norm": 4.628783226013184, + "learning_rate": 0.00019425814585399405, + "loss": 3.43, + "step": 758 + }, + { + "epoch": 1.11, + "grad_norm": 6.876189708709717, + "learning_rate": 0.00019424247084617177, + "loss": 3.4606, + "step": 759 + }, + { + "epoch": 1.11, + "grad_norm": 10.343745231628418, + "learning_rate": 0.00019422677510563748, + "loss": 3.5079, + "step": 760 + }, + { + "epoch": 1.12, + "grad_norm": 4.678600311279297, + "learning_rate": 0.00019421105863584413, + "loss": 3.5135, + "step": 761 + }, + { + "epoch": 1.12, + "grad_norm": 6.044957160949707, + "learning_rate": 0.0001941953214402493, + "loss": 3.5067, + "step": 762 + }, + { + "epoch": 1.12, + "grad_norm": 7.979968070983887, + "learning_rate": 0.00019417956352231502, + "loss": 3.4955, + "step": 763 + }, + { + "epoch": 1.12, + "grad_norm": 4.172381401062012, + "learning_rate": 0.00019416378488550787, + "loss": 3.4781, + "step": 764 + }, + { + "epoch": 1.12, + "grad_norm": 7.783565998077393, + "learning_rate": 0.00019414798553329916, + "loss": 3.4649, + "step": 765 + }, + { + "epoch": 1.12, + "grad_norm": 11.02284049987793, + "learning_rate": 0.00019413216546916457, + "loss": 3.5242, + "step": 766 + }, + { + "epoch": 1.12, + "grad_norm": 5.28484582901001, + "learning_rate": 0.00019411632469658443, + "loss": 3.4645, + "step": 767 + }, + { + "epoch": 1.13, + "grad_norm": 6.049994945526123, + "learning_rate": 0.00019410046321904356, + "loss": 3.4601, + "step": 768 + }, + { + "epoch": 1.13, + "grad_norm": 7.867702960968018, + "learning_rate": 0.00019408458104003143, + "loss": 3.4853, + "step": 769 + }, + { + "epoch": 1.13, + "grad_norm": 5.088557720184326, + "learning_rate": 0.00019406867816304204, + "loss": 3.5329, + "step": 770 + }, + { + "epoch": 1.13, + "grad_norm": 6.444199085235596, + "learning_rate": 0.00019405275459157383, + "loss": 3.4624, + "step": 771 + }, + { + "epoch": 1.13, + "grad_norm": 9.823208808898926, + "learning_rate": 0.00019403681032912994, + "loss": 3.4496, + "step": 772 + }, + { + "epoch": 1.13, + "grad_norm": 4.309892654418945, + "learning_rate": 0.00019402084537921798, + "loss": 3.4607, + "step": 773 + }, + { + "epoch": 1.13, + "grad_norm": 6.139223575592041, + "learning_rate": 0.00019400485974535014, + "loss": 3.4246, + "step": 774 + }, + { + "epoch": 1.14, + "grad_norm": 8.018882751464844, + "learning_rate": 0.00019398885343104315, + "loss": 3.4731, + "step": 775 + }, + { + "epoch": 1.14, + "grad_norm": 4.7042555809021, + "learning_rate": 0.00019397282643981826, + "loss": 3.4863, + "step": 776 + }, + { + "epoch": 1.14, + "grad_norm": 6.583994388580322, + "learning_rate": 0.00019395677877520137, + "loss": 3.4354, + "step": 777 + }, + { + "epoch": 1.14, + "grad_norm": 9.998130798339844, + "learning_rate": 0.00019394071044072283, + "loss": 3.4882, + "step": 778 + }, + { + "epoch": 1.14, + "grad_norm": 4.592165946960449, + "learning_rate": 0.00019392462143991753, + "loss": 3.4696, + "step": 779 + }, + { + "epoch": 1.14, + "grad_norm": 6.2591681480407715, + "learning_rate": 0.00019390851177632497, + "loss": 3.4393, + "step": 780 + }, + { + "epoch": 1.14, + "grad_norm": 8.170223236083984, + "learning_rate": 0.00019389238145348916, + "loss": 3.421, + "step": 781 + }, + { + "epoch": 1.15, + "grad_norm": 5.1918559074401855, + "learning_rate": 0.00019387623047495867, + "loss": 3.452, + "step": 782 + }, + { + "epoch": 1.15, + "grad_norm": 6.183322429656982, + "learning_rate": 0.0001938600588442866, + "loss": 3.5035, + "step": 783 + }, + { + "epoch": 1.15, + "grad_norm": 9.819025039672852, + "learning_rate": 0.00019384386656503064, + "loss": 3.4897, + "step": 784 + }, + { + "epoch": 1.15, + "grad_norm": 4.417552471160889, + "learning_rate": 0.00019382765364075292, + "loss": 3.5126, + "step": 785 + }, + { + "epoch": 1.15, + "grad_norm": 6.277004241943359, + "learning_rate": 0.0001938114200750202, + "loss": 3.4848, + "step": 786 + }, + { + "epoch": 1.15, + "grad_norm": 8.565668106079102, + "learning_rate": 0.0001937951658714038, + "loss": 3.5328, + "step": 787 + }, + { + "epoch": 1.16, + "grad_norm": 5.04703426361084, + "learning_rate": 0.00019377889103347945, + "loss": 3.4752, + "step": 788 + }, + { + "epoch": 1.16, + "grad_norm": 5.934195518493652, + "learning_rate": 0.0001937625955648276, + "loss": 3.5186, + "step": 789 + }, + { + "epoch": 1.16, + "grad_norm": 9.426629066467285, + "learning_rate": 0.00019374627946903305, + "loss": 3.4537, + "step": 790 + }, + { + "epoch": 1.16, + "grad_norm": 3.8387088775634766, + "learning_rate": 0.00019372994274968527, + "loss": 3.4172, + "step": 791 + }, + { + "epoch": 1.16, + "grad_norm": 6.488461971282959, + "learning_rate": 0.00019371358541037825, + "loss": 3.5177, + "step": 792 + }, + { + "epoch": 1.16, + "grad_norm": 8.451127052307129, + "learning_rate": 0.0001936972074547105, + "loss": 3.4701, + "step": 793 + }, + { + "epoch": 1.16, + "grad_norm": 5.298417568206787, + "learning_rate": 0.00019368080888628502, + "loss": 3.4579, + "step": 794 + }, + { + "epoch": 1.17, + "grad_norm": 6.457042694091797, + "learning_rate": 0.00019366438970870942, + "loss": 3.4666, + "step": 795 + }, + { + "epoch": 1.17, + "grad_norm": 9.55510139465332, + "learning_rate": 0.00019364794992559578, + "loss": 3.4449, + "step": 796 + }, + { + "epoch": 1.17, + "grad_norm": 3.862884998321533, + "learning_rate": 0.00019363148954056077, + "loss": 3.4185, + "step": 797 + }, + { + "epoch": 1.17, + "grad_norm": 6.591372966766357, + "learning_rate": 0.00019361500855722555, + "loss": 3.537, + "step": 798 + }, + { + "epoch": 1.17, + "grad_norm": 8.397138595581055, + "learning_rate": 0.0001935985069792158, + "loss": 3.4205, + "step": 799 + }, + { + "epoch": 1.17, + "grad_norm": 5.131923675537109, + "learning_rate": 0.00019358198481016182, + "loss": 3.4916, + "step": 800 + }, + { + "epoch": 1.17, + "grad_norm": 5.620177745819092, + "learning_rate": 0.00019356544205369834, + "loss": 3.4604, + "step": 801 + }, + { + "epoch": 1.18, + "grad_norm": 9.219866752624512, + "learning_rate": 0.00019354887871346463, + "loss": 3.467, + "step": 802 + }, + { + "epoch": 1.18, + "grad_norm": 4.237349033355713, + "learning_rate": 0.00019353229479310456, + "loss": 3.4847, + "step": 803 + }, + { + "epoch": 1.18, + "grad_norm": 6.673110485076904, + "learning_rate": 0.00019351569029626644, + "loss": 3.4412, + "step": 804 + }, + { + "epoch": 1.18, + "grad_norm": 8.31098747253418, + "learning_rate": 0.0001934990652266032, + "loss": 3.4838, + "step": 805 + }, + { + "epoch": 1.18, + "grad_norm": 5.009383201599121, + "learning_rate": 0.0001934824195877722, + "loss": 3.4408, + "step": 806 + }, + { + "epoch": 1.18, + "grad_norm": 5.483675956726074, + "learning_rate": 0.00019346575338343537, + "loss": 3.4663, + "step": 807 + }, + { + "epoch": 1.19, + "grad_norm": 9.371169090270996, + "learning_rate": 0.00019344906661725918, + "loss": 3.5116, + "step": 808 + }, + { + "epoch": 1.19, + "grad_norm": 3.6695902347564697, + "learning_rate": 0.00019343235929291462, + "loss": 3.4776, + "step": 809 + }, + { + "epoch": 1.19, + "grad_norm": 6.492842197418213, + "learning_rate": 0.00019341563141407714, + "loss": 3.4725, + "step": 810 + }, + { + "epoch": 1.19, + "grad_norm": 8.520849227905273, + "learning_rate": 0.00019339888298442682, + "loss": 3.4845, + "step": 811 + }, + { + "epoch": 1.19, + "grad_norm": 4.986138820648193, + "learning_rate": 0.00019338211400764816, + "loss": 3.4313, + "step": 812 + }, + { + "epoch": 1.19, + "grad_norm": 6.272872447967529, + "learning_rate": 0.00019336532448743023, + "loss": 3.4337, + "step": 813 + }, + { + "epoch": 1.19, + "grad_norm": 10.374552726745605, + "learning_rate": 0.00019334851442746664, + "loss": 3.4586, + "step": 814 + }, + { + "epoch": 1.2, + "grad_norm": 4.865425109863281, + "learning_rate": 0.00019333168383145547, + "loss": 3.5045, + "step": 815 + }, + { + "epoch": 1.2, + "grad_norm": 6.018640995025635, + "learning_rate": 0.0001933148327030993, + "loss": 3.4509, + "step": 816 + }, + { + "epoch": 1.2, + "grad_norm": 8.015512466430664, + "learning_rate": 0.00019329796104610537, + "loss": 3.4657, + "step": 817 + }, + { + "epoch": 1.2, + "grad_norm": 4.071451187133789, + "learning_rate": 0.00019328106886418522, + "loss": 3.4079, + "step": 818 + }, + { + "epoch": 1.2, + "grad_norm": 7.4237823486328125, + "learning_rate": 0.00019326415616105507, + "loss": 3.4906, + "step": 819 + }, + { + "epoch": 1.2, + "grad_norm": 10.776564598083496, + "learning_rate": 0.00019324722294043558, + "loss": 3.4639, + "step": 820 + }, + { + "epoch": 1.2, + "grad_norm": 5.143712997436523, + "learning_rate": 0.00019323026920605197, + "loss": 3.4356, + "step": 821 + }, + { + "epoch": 1.21, + "grad_norm": 5.7633891105651855, + "learning_rate": 0.00019321329496163397, + "loss": 3.428, + "step": 822 + }, + { + "epoch": 1.21, + "grad_norm": 7.827281475067139, + "learning_rate": 0.0001931963002109157, + "loss": 3.4913, + "step": 823 + }, + { + "epoch": 1.21, + "grad_norm": 4.470641136169434, + "learning_rate": 0.00019317928495763602, + "loss": 3.4441, + "step": 824 + }, + { + "epoch": 1.21, + "grad_norm": 7.5875091552734375, + "learning_rate": 0.00019316224920553807, + "loss": 3.468, + "step": 825 + }, + { + "epoch": 1.21, + "grad_norm": 10.94041919708252, + "learning_rate": 0.00019314519295836964, + "loss": 3.4932, + "step": 826 + }, + { + "epoch": 1.21, + "grad_norm": 5.624682426452637, + "learning_rate": 0.000193128116219883, + "loss": 3.4411, + "step": 827 + }, + { + "epoch": 1.21, + "grad_norm": 5.68918514251709, + "learning_rate": 0.00019311101899383492, + "loss": 3.4896, + "step": 828 + }, + { + "epoch": 1.22, + "grad_norm": 7.500881671905518, + "learning_rate": 0.00019309390128398665, + "loss": 3.4429, + "step": 829 + }, + { + "epoch": 1.22, + "grad_norm": 4.407763481140137, + "learning_rate": 0.000193076763094104, + "loss": 3.4694, + "step": 830 + }, + { + "epoch": 1.22, + "grad_norm": 7.296525955200195, + "learning_rate": 0.0001930596044279572, + "loss": 3.4065, + "step": 831 + }, + { + "epoch": 1.22, + "grad_norm": 11.3180513381958, + "learning_rate": 0.00019304242528932113, + "loss": 3.5025, + "step": 832 + }, + { + "epoch": 1.22, + "grad_norm": 5.404650688171387, + "learning_rate": 0.00019302522568197502, + "loss": 3.4129, + "step": 833 + }, + { + "epoch": 1.22, + "grad_norm": 5.595582485198975, + "learning_rate": 0.0001930080056097027, + "loss": 3.5212, + "step": 834 + }, + { + "epoch": 1.23, + "grad_norm": 7.934787273406982, + "learning_rate": 0.00019299076507629245, + "loss": 3.5551, + "step": 835 + }, + { + "epoch": 1.23, + "grad_norm": 4.486216068267822, + "learning_rate": 0.00019297350408553708, + "loss": 3.441, + "step": 836 + }, + { + "epoch": 1.23, + "grad_norm": 7.029922962188721, + "learning_rate": 0.0001929562226412339, + "loss": 3.4982, + "step": 837 + }, + { + "epoch": 1.23, + "grad_norm": 10.693680763244629, + "learning_rate": 0.0001929389207471847, + "loss": 3.4494, + "step": 838 + }, + { + "epoch": 1.23, + "grad_norm": 4.879317760467529, + "learning_rate": 0.00019292159840719576, + "loss": 3.4357, + "step": 839 + }, + { + "epoch": 1.23, + "grad_norm": 6.057565212249756, + "learning_rate": 0.00019290425562507792, + "loss": 3.4738, + "step": 840 + }, + { + "epoch": 1.23, + "grad_norm": 7.925189018249512, + "learning_rate": 0.0001928868924046465, + "loss": 3.4606, + "step": 841 + }, + { + "epoch": 1.24, + "grad_norm": 4.621508598327637, + "learning_rate": 0.0001928695087497212, + "loss": 3.4088, + "step": 842 + }, + { + "epoch": 1.24, + "grad_norm": 6.142604351043701, + "learning_rate": 0.00019285210466412635, + "loss": 3.4484, + "step": 843 + }, + { + "epoch": 1.24, + "grad_norm": 10.20028305053711, + "learning_rate": 0.00019283468015169076, + "loss": 3.5059, + "step": 844 + }, + { + "epoch": 1.24, + "grad_norm": 4.734007358551025, + "learning_rate": 0.0001928172352162477, + "loss": 3.4299, + "step": 845 + }, + { + "epoch": 1.24, + "eval_loss": 3.2429304122924805, + "eval_runtime": 1163.5436, + "eval_samples_per_second": 14.733, + "eval_steps_per_second": 0.921, + "step": 845 + }, + { + "epoch": 1.24, + "grad_norm": 6.036166191101074, + "learning_rate": 0.00019279976986163493, + "loss": 3.3913, + "step": 846 + }, + { + "epoch": 1.24, + "grad_norm": 8.014019012451172, + "learning_rate": 0.00019278228409169468, + "loss": 3.4981, + "step": 847 + }, + { + "epoch": 1.24, + "grad_norm": 4.856557369232178, + "learning_rate": 0.00019276477791027374, + "loss": 3.4562, + "step": 848 + }, + { + "epoch": 1.25, + "grad_norm": 6.650398254394531, + "learning_rate": 0.00019274725132122336, + "loss": 3.4407, + "step": 849 + }, + { + "epoch": 1.25, + "grad_norm": 10.177898406982422, + "learning_rate": 0.00019272970432839922, + "loss": 3.4437, + "step": 850 + }, + { + "epoch": 1.25, + "grad_norm": 4.559935569763184, + "learning_rate": 0.0001927121369356616, + "loss": 3.418, + "step": 851 + }, + { + "epoch": 1.25, + "grad_norm": 5.8679423332214355, + "learning_rate": 0.00019269454914687517, + "loss": 3.4338, + "step": 852 + }, + { + "epoch": 1.25, + "grad_norm": 8.044879913330078, + "learning_rate": 0.00019267694096590914, + "loss": 3.4146, + "step": 853 + }, + { + "epoch": 1.25, + "grad_norm": 4.654249668121338, + "learning_rate": 0.0001926593123966372, + "loss": 3.4353, + "step": 854 + }, + { + "epoch": 1.25, + "grad_norm": 6.326594352722168, + "learning_rate": 0.0001926416634429375, + "loss": 3.447, + "step": 855 + }, + { + "epoch": 1.26, + "grad_norm": 10.194587707519531, + "learning_rate": 0.0001926239941086927, + "loss": 3.4594, + "step": 856 + }, + { + "epoch": 1.26, + "grad_norm": 4.672416687011719, + "learning_rate": 0.00019260630439778994, + "loss": 3.4292, + "step": 857 + }, + { + "epoch": 1.26, + "grad_norm": 5.656294822692871, + "learning_rate": 0.00019258859431412083, + "loss": 3.4478, + "step": 858 + }, + { + "epoch": 1.26, + "grad_norm": 8.051904678344727, + "learning_rate": 0.00019257086386158148, + "loss": 3.4809, + "step": 859 + }, + { + "epoch": 1.26, + "grad_norm": 4.461113452911377, + "learning_rate": 0.00019255311304407243, + "loss": 3.4361, + "step": 860 + }, + { + "epoch": 1.26, + "grad_norm": 6.34873104095459, + "learning_rate": 0.00019253534186549883, + "loss": 3.424, + "step": 861 + }, + { + "epoch": 1.27, + "grad_norm": 10.255434036254883, + "learning_rate": 0.00019251755032977013, + "loss": 3.4578, + "step": 862 + }, + { + "epoch": 1.27, + "grad_norm": 4.895047187805176, + "learning_rate": 0.00019249973844080037, + "loss": 3.4606, + "step": 863 + }, + { + "epoch": 1.27, + "grad_norm": 6.042967319488525, + "learning_rate": 0.00019248190620250807, + "loss": 3.4398, + "step": 864 + }, + { + "epoch": 1.27, + "grad_norm": 7.916470050811768, + "learning_rate": 0.00019246405361881622, + "loss": 3.4921, + "step": 865 + }, + { + "epoch": 1.27, + "grad_norm": 4.985673904418945, + "learning_rate": 0.0001924461806936522, + "loss": 3.4425, + "step": 866 + }, + { + "epoch": 1.27, + "grad_norm": 6.128829002380371, + "learning_rate": 0.000192428287430948, + "loss": 3.4719, + "step": 867 + }, + { + "epoch": 1.27, + "grad_norm": 9.511234283447266, + "learning_rate": 0.00019241037383463998, + "loss": 3.4018, + "step": 868 + }, + { + "epoch": 1.28, + "grad_norm": 3.816689968109131, + "learning_rate": 0.000192392439908669, + "loss": 3.375, + "step": 869 + }, + { + "epoch": 1.28, + "grad_norm": 6.579503059387207, + "learning_rate": 0.00019237448565698046, + "loss": 3.4039, + "step": 870 + }, + { + "epoch": 1.28, + "grad_norm": 8.701712608337402, + "learning_rate": 0.00019235651108352416, + "loss": 3.3992, + "step": 871 + }, + { + "epoch": 1.28, + "grad_norm": 5.65900993347168, + "learning_rate": 0.00019233851619225432, + "loss": 3.4225, + "step": 872 + }, + { + "epoch": 1.28, + "grad_norm": 4.935523509979248, + "learning_rate": 0.00019232050098712974, + "loss": 3.4198, + "step": 873 + }, + { + "epoch": 1.28, + "grad_norm": 8.355915069580078, + "learning_rate": 0.00019230246547211367, + "loss": 3.4283, + "step": 874 + }, + { + "epoch": 1.28, + "grad_norm": 3.1533217430114746, + "learning_rate": 0.0001922844096511738, + "loss": 3.4058, + "step": 875 + }, + { + "epoch": 1.29, + "grad_norm": 7.010662078857422, + "learning_rate": 0.00019226633352828224, + "loss": 3.3914, + "step": 876 + }, + { + "epoch": 1.29, + "grad_norm": 8.548685073852539, + "learning_rate": 0.00019224823710741569, + "loss": 3.5008, + "step": 877 + }, + { + "epoch": 1.29, + "grad_norm": 5.439549446105957, + "learning_rate": 0.00019223012039255516, + "loss": 3.4421, + "step": 878 + }, + { + "epoch": 1.29, + "grad_norm": 5.446386814117432, + "learning_rate": 0.00019221198338768622, + "loss": 3.4071, + "step": 879 + }, + { + "epoch": 1.29, + "grad_norm": 9.378525733947754, + "learning_rate": 0.00019219382609679894, + "loss": 3.409, + "step": 880 + }, + { + "epoch": 1.29, + "grad_norm": 4.573349475860596, + "learning_rate": 0.0001921756485238878, + "loss": 3.4156, + "step": 881 + }, + { + "epoch": 1.29, + "grad_norm": 5.857920169830322, + "learning_rate": 0.00019215745067295169, + "loss": 3.433, + "step": 882 + }, + { + "epoch": 1.3, + "grad_norm": 7.654170036315918, + "learning_rate": 0.00019213923254799402, + "loss": 3.4011, + "step": 883 + }, + { + "epoch": 1.3, + "grad_norm": 4.262071132659912, + "learning_rate": 0.0001921209941530227, + "loss": 3.401, + "step": 884 + }, + { + "epoch": 1.3, + "grad_norm": 7.949652194976807, + "learning_rate": 0.00019210273549205003, + "loss": 3.5334, + "step": 885 + }, + { + "epoch": 1.3, + "grad_norm": 11.388256072998047, + "learning_rate": 0.0001920844565690928, + "loss": 3.4325, + "step": 886 + }, + { + "epoch": 1.3, + "grad_norm": 6.005974292755127, + "learning_rate": 0.00019206615738817225, + "loss": 3.4264, + "step": 887 + }, + { + "epoch": 1.3, + "grad_norm": 5.020950794219971, + "learning_rate": 0.0001920478379533141, + "loss": 3.426, + "step": 888 + }, + { + "epoch": 1.31, + "grad_norm": 6.8441386222839355, + "learning_rate": 0.00019202949826854844, + "loss": 3.4234, + "step": 889 + }, + { + "epoch": 1.31, + "grad_norm": 3.4975130558013916, + "learning_rate": 0.0001920111383379099, + "loss": 3.4667, + "step": 890 + }, + { + "epoch": 1.31, + "grad_norm": 7.94019889831543, + "learning_rate": 0.00019199275816543762, + "loss": 3.4208, + "step": 891 + }, + { + "epoch": 1.31, + "grad_norm": 11.041260719299316, + "learning_rate": 0.00019197435775517501, + "loss": 3.4392, + "step": 892 + }, + { + "epoch": 1.31, + "grad_norm": 5.074487209320068, + "learning_rate": 0.0001919559371111701, + "loss": 3.3906, + "step": 893 + }, + { + "epoch": 1.31, + "grad_norm": 5.909255027770996, + "learning_rate": 0.00019193749623747525, + "loss": 3.4508, + "step": 894 + }, + { + "epoch": 1.31, + "grad_norm": 7.794407367706299, + "learning_rate": 0.0001919190351381474, + "loss": 3.3957, + "step": 895 + }, + { + "epoch": 1.32, + "grad_norm": 4.833076000213623, + "learning_rate": 0.00019190055381724781, + "loss": 3.403, + "step": 896 + }, + { + "epoch": 1.32, + "grad_norm": 6.273806095123291, + "learning_rate": 0.0001918820522788423, + "loss": 3.4114, + "step": 897 + }, + { + "epoch": 1.32, + "grad_norm": 9.626479148864746, + "learning_rate": 0.00019186353052700104, + "loss": 3.4555, + "step": 898 + }, + { + "epoch": 1.32, + "grad_norm": 4.462099075317383, + "learning_rate": 0.00019184498856579868, + "loss": 3.4001, + "step": 899 + }, + { + "epoch": 1.32, + "grad_norm": 5.839742660522461, + "learning_rate": 0.00019182642639931438, + "loss": 3.4041, + "step": 900 + }, + { + "epoch": 1.32, + "grad_norm": 7.5265583992004395, + "learning_rate": 0.00019180784403163167, + "loss": 3.3898, + "step": 901 + }, + { + "epoch": 1.32, + "grad_norm": 4.089066028594971, + "learning_rate": 0.0001917892414668385, + "loss": 3.4162, + "step": 902 + }, + { + "epoch": 1.33, + "grad_norm": 7.329177379608154, + "learning_rate": 0.0001917706187090274, + "loss": 3.5439, + "step": 903 + }, + { + "epoch": 1.33, + "grad_norm": 10.022590637207031, + "learning_rate": 0.0001917519757622952, + "loss": 3.4009, + "step": 904 + }, + { + "epoch": 1.33, + "grad_norm": 5.030527591705322, + "learning_rate": 0.00019173331263074324, + "loss": 3.4314, + "step": 905 + }, + { + "epoch": 1.33, + "grad_norm": 5.504607677459717, + "learning_rate": 0.00019171462931847724, + "loss": 3.4463, + "step": 906 + }, + { + "epoch": 1.33, + "grad_norm": 7.219310283660889, + "learning_rate": 0.00019169592582960742, + "loss": 3.4878, + "step": 907 + }, + { + "epoch": 1.33, + "grad_norm": 4.046157360076904, + "learning_rate": 0.00019167720216824847, + "loss": 3.3643, + "step": 908 + }, + { + "epoch": 1.34, + "grad_norm": 7.463158130645752, + "learning_rate": 0.00019165845833851941, + "loss": 3.472, + "step": 909 + }, + { + "epoch": 1.34, + "grad_norm": 9.688533782958984, + "learning_rate": 0.00019163969434454382, + "loss": 3.4541, + "step": 910 + }, + { + "epoch": 1.34, + "grad_norm": 4.718085765838623, + "learning_rate": 0.00019162091019044958, + "loss": 3.4252, + "step": 911 + }, + { + "epoch": 1.34, + "grad_norm": 6.263822555541992, + "learning_rate": 0.00019160210588036911, + "loss": 3.4105, + "step": 912 + }, + { + "epoch": 1.34, + "grad_norm": 7.928195953369141, + "learning_rate": 0.00019158328141843928, + "loss": 3.4249, + "step": 913 + }, + { + "epoch": 1.34, + "grad_norm": 4.594407081604004, + "learning_rate": 0.00019156443680880126, + "loss": 3.453, + "step": 914 + }, + { + "epoch": 1.34, + "grad_norm": 6.126087188720703, + "learning_rate": 0.0001915455720556008, + "loss": 3.3944, + "step": 915 + }, + { + "epoch": 1.35, + "grad_norm": 9.558304786682129, + "learning_rate": 0.000191526687162988, + "loss": 3.3926, + "step": 916 + }, + { + "epoch": 1.35, + "grad_norm": 4.792506217956543, + "learning_rate": 0.0001915077821351174, + "loss": 3.4081, + "step": 917 + }, + { + "epoch": 1.35, + "grad_norm": 5.998870372772217, + "learning_rate": 0.00019148885697614797, + "loss": 3.4326, + "step": 918 + }, + { + "epoch": 1.35, + "grad_norm": 7.541780948638916, + "learning_rate": 0.00019146991169024315, + "loss": 3.4647, + "step": 919 + }, + { + "epoch": 1.35, + "grad_norm": 4.514527320861816, + "learning_rate": 0.00019145094628157078, + "loss": 3.3601, + "step": 920 + }, + { + "epoch": 1.35, + "grad_norm": 6.531849384307861, + "learning_rate": 0.0001914319607543031, + "loss": 3.4428, + "step": 921 + }, + { + "epoch": 1.35, + "grad_norm": 9.34991455078125, + "learning_rate": 0.00019141295511261678, + "loss": 3.393, + "step": 922 + }, + { + "epoch": 1.36, + "grad_norm": 4.035532474517822, + "learning_rate": 0.000191393929360693, + "loss": 3.3938, + "step": 923 + }, + { + "epoch": 1.36, + "grad_norm": 6.159792900085449, + "learning_rate": 0.0001913748835027172, + "loss": 3.4207, + "step": 924 + }, + { + "epoch": 1.36, + "grad_norm": 8.069183349609375, + "learning_rate": 0.00019135581754287944, + "loss": 3.4143, + "step": 925 + }, + { + "epoch": 1.36, + "grad_norm": 4.528735160827637, + "learning_rate": 0.00019133673148537405, + "loss": 3.4282, + "step": 926 + }, + { + "epoch": 1.36, + "grad_norm": 6.5554399490356445, + "learning_rate": 0.00019131762533439983, + "loss": 3.4149, + "step": 927 + }, + { + "epoch": 1.36, + "grad_norm": 9.77618408203125, + "learning_rate": 0.00019129849909416005, + "loss": 3.4509, + "step": 928 + }, + { + "epoch": 1.36, + "grad_norm": 4.406791687011719, + "learning_rate": 0.0001912793527688623, + "loss": 3.3955, + "step": 929 + }, + { + "epoch": 1.37, + "grad_norm": 6.1382622718811035, + "learning_rate": 0.00019126018636271867, + "loss": 3.4066, + "step": 930 + }, + { + "epoch": 1.37, + "grad_norm": 8.133618354797363, + "learning_rate": 0.00019124099987994565, + "loss": 3.4536, + "step": 931 + }, + { + "epoch": 1.37, + "grad_norm": 4.887189865112305, + "learning_rate": 0.0001912217933247641, + "loss": 3.4516, + "step": 932 + }, + { + "epoch": 1.37, + "grad_norm": 5.946772575378418, + "learning_rate": 0.0001912025667013994, + "loss": 3.3842, + "step": 933 + }, + { + "epoch": 1.37, + "grad_norm": 9.542755126953125, + "learning_rate": 0.00019118332001408122, + "loss": 3.3833, + "step": 934 + }, + { + "epoch": 1.37, + "grad_norm": 4.512613296508789, + "learning_rate": 0.00019116405326704373, + "loss": 3.3754, + "step": 935 + }, + { + "epoch": 1.38, + "grad_norm": 6.039595603942871, + "learning_rate": 0.00019114476646452547, + "loss": 3.4341, + "step": 936 + }, + { + "epoch": 1.38, + "grad_norm": 7.820281982421875, + "learning_rate": 0.0001911254596107694, + "loss": 3.3616, + "step": 937 + }, + { + "epoch": 1.38, + "grad_norm": 4.0570759773254395, + "learning_rate": 0.0001911061327100229, + "loss": 3.3999, + "step": 938 + }, + { + "epoch": 1.38, + "grad_norm": 6.659790992736816, + "learning_rate": 0.0001910867857665378, + "loss": 3.361, + "step": 939 + }, + { + "epoch": 1.38, + "grad_norm": 10.188064575195312, + "learning_rate": 0.00019106741878457026, + "loss": 3.3805, + "step": 940 + }, + { + "epoch": 1.38, + "grad_norm": 4.66610860824585, + "learning_rate": 0.0001910480317683809, + "loss": 3.3881, + "step": 941 + }, + { + "epoch": 1.38, + "grad_norm": 5.840310573577881, + "learning_rate": 0.0001910286247222347, + "loss": 3.4162, + "step": 942 + }, + { + "epoch": 1.39, + "grad_norm": 7.710467338562012, + "learning_rate": 0.00019100919765040113, + "loss": 3.396, + "step": 943 + }, + { + "epoch": 1.39, + "grad_norm": 4.53704833984375, + "learning_rate": 0.00019098975055715397, + "loss": 3.3605, + "step": 944 + }, + { + "epoch": 1.39, + "grad_norm": 6.515756607055664, + "learning_rate": 0.00019097028344677153, + "loss": 3.4119, + "step": 945 + }, + { + "epoch": 1.39, + "grad_norm": 10.256503105163574, + "learning_rate": 0.00019095079632353633, + "loss": 3.4173, + "step": 946 + }, + { + "epoch": 1.39, + "grad_norm": 4.8957672119140625, + "learning_rate": 0.0001909312891917355, + "loss": 3.3812, + "step": 947 + }, + { + "epoch": 1.39, + "grad_norm": 5.8118205070495605, + "learning_rate": 0.00019091176205566043, + "loss": 3.3734, + "step": 948 + }, + { + "epoch": 1.39, + "grad_norm": 7.7631025314331055, + "learning_rate": 0.00019089221491960695, + "loss": 3.4171, + "step": 949 + }, + { + "epoch": 1.4, + "grad_norm": 4.2992377281188965, + "learning_rate": 0.00019087264778787534, + "loss": 3.3724, + "step": 950 + }, + { + "epoch": 1.4, + "grad_norm": 6.633007049560547, + "learning_rate": 0.00019085306066477021, + "loss": 3.4198, + "step": 951 + }, + { + "epoch": 1.4, + "grad_norm": 9.867729187011719, + "learning_rate": 0.00019083345355460062, + "loss": 3.4247, + "step": 952 + }, + { + "epoch": 1.4, + "grad_norm": 4.476417541503906, + "learning_rate": 0.00019081382646167998, + "loss": 3.3617, + "step": 953 + }, + { + "epoch": 1.4, + "grad_norm": 5.963509559631348, + "learning_rate": 0.00019079417939032612, + "loss": 3.3945, + "step": 954 + }, + { + "epoch": 1.4, + "grad_norm": 7.921098232269287, + "learning_rate": 0.00019077451234486128, + "loss": 3.4305, + "step": 955 + }, + { + "epoch": 1.4, + "grad_norm": 4.772178649902344, + "learning_rate": 0.00019075482532961204, + "loss": 3.3836, + "step": 956 + }, + { + "epoch": 1.41, + "grad_norm": 6.6379265785217285, + "learning_rate": 0.00019073511834890949, + "loss": 3.3646, + "step": 957 + }, + { + "epoch": 1.41, + "grad_norm": 9.883600234985352, + "learning_rate": 0.00019071539140708894, + "loss": 3.398, + "step": 958 + }, + { + "epoch": 1.41, + "grad_norm": 4.0690178871154785, + "learning_rate": 0.00019069564450849025, + "loss": 3.3394, + "step": 959 + }, + { + "epoch": 1.41, + "grad_norm": 6.045716762542725, + "learning_rate": 0.00019067587765745757, + "loss": 3.4695, + "step": 960 + }, + { + "epoch": 1.41, + "grad_norm": 7.893038272857666, + "learning_rate": 0.00019065609085833947, + "loss": 3.4336, + "step": 961 + }, + { + "epoch": 1.41, + "grad_norm": 4.4083075523376465, + "learning_rate": 0.00019063628411548897, + "loss": 3.3471, + "step": 962 + }, + { + "epoch": 1.42, + "grad_norm": 6.27316951751709, + "learning_rate": 0.00019061645743326337, + "loss": 3.4311, + "step": 963 + }, + { + "epoch": 1.42, + "grad_norm": 9.92757797241211, + "learning_rate": 0.00019059661081602443, + "loss": 3.4114, + "step": 964 + }, + { + "epoch": 1.42, + "grad_norm": 5.099198341369629, + "learning_rate": 0.00019057674426813822, + "loss": 3.4386, + "step": 965 + }, + { + "epoch": 1.42, + "grad_norm": 5.424035549163818, + "learning_rate": 0.00019055685779397532, + "loss": 3.4275, + "step": 966 + }, + { + "epoch": 1.42, + "grad_norm": 7.513932704925537, + "learning_rate": 0.0001905369513979106, + "loss": 3.4135, + "step": 967 + }, + { + "epoch": 1.42, + "grad_norm": 4.388814449310303, + "learning_rate": 0.0001905170250843233, + "loss": 3.3889, + "step": 968 + }, + { + "epoch": 1.42, + "grad_norm": 6.554171562194824, + "learning_rate": 0.00019049707885759713, + "loss": 3.4184, + "step": 969 + }, + { + "epoch": 1.43, + "grad_norm": 9.21511173248291, + "learning_rate": 0.00019047711272212006, + "loss": 3.393, + "step": 970 + }, + { + "epoch": 1.43, + "grad_norm": 4.280778408050537, + "learning_rate": 0.00019045712668228456, + "loss": 3.3808, + "step": 971 + }, + { + "epoch": 1.43, + "grad_norm": 5.782205581665039, + "learning_rate": 0.0001904371207424874, + "loss": 3.3168, + "step": 972 + }, + { + "epoch": 1.43, + "grad_norm": 7.35162878036499, + "learning_rate": 0.00019041709490712978, + "loss": 3.4373, + "step": 973 + }, + { + "epoch": 1.43, + "grad_norm": 3.4769937992095947, + "learning_rate": 0.0001903970491806172, + "loss": 3.4039, + "step": 974 + }, + { + "epoch": 1.43, + "grad_norm": 8.196279525756836, + "learning_rate": 0.00019037698356735963, + "loss": 3.38, + "step": 975 + }, + { + "epoch": 1.43, + "grad_norm": 11.59374713897705, + "learning_rate": 0.00019035689807177136, + "loss": 3.4377, + "step": 976 + }, + { + "epoch": 1.44, + "grad_norm": 6.463580131530762, + "learning_rate": 0.00019033679269827105, + "loss": 3.3706, + "step": 977 + }, + { + "epoch": 1.44, + "grad_norm": 4.297844409942627, + "learning_rate": 0.00019031666745128177, + "loss": 3.4047, + "step": 978 + }, + { + "epoch": 1.44, + "grad_norm": 6.652123928070068, + "learning_rate": 0.00019029652233523093, + "loss": 3.3968, + "step": 979 + }, + { + "epoch": 1.44, + "grad_norm": 3.399991750717163, + "learning_rate": 0.00019027635735455027, + "loss": 3.3552, + "step": 980 + }, + { + "epoch": 1.44, + "grad_norm": 7.520792007446289, + "learning_rate": 0.00019025617251367604, + "loss": 3.3979, + "step": 981 + }, + { + "epoch": 1.44, + "grad_norm": 10.621402740478516, + "learning_rate": 0.00019023596781704872, + "loss": 3.4131, + "step": 982 + }, + { + "epoch": 1.44, + "grad_norm": 5.002892971038818, + "learning_rate": 0.0001902157432691132, + "loss": 3.3583, + "step": 983 + }, + { + "epoch": 1.45, + "grad_norm": 5.647467136383057, + "learning_rate": 0.00019019549887431877, + "loss": 3.384, + "step": 984 + }, + { + "epoch": 1.45, + "grad_norm": 7.81645393371582, + "learning_rate": 0.00019017523463711906, + "loss": 3.4198, + "step": 985 + }, + { + "epoch": 1.45, + "grad_norm": 4.508808135986328, + "learning_rate": 0.00019015495056197207, + "loss": 3.4205, + "step": 986 + }, + { + "epoch": 1.45, + "grad_norm": 6.004973411560059, + "learning_rate": 0.00019013464665334012, + "loss": 3.396, + "step": 987 + }, + { + "epoch": 1.45, + "grad_norm": 9.286575317382812, + "learning_rate": 0.00019011432291568995, + "loss": 3.4053, + "step": 988 + }, + { + "epoch": 1.45, + "grad_norm": 4.02089786529541, + "learning_rate": 0.00019009397935349268, + "loss": 3.3216, + "step": 989 + }, + { + "epoch": 1.46, + "grad_norm": 6.502218246459961, + "learning_rate": 0.00019007361597122373, + "loss": 3.4367, + "step": 990 + }, + { + "epoch": 1.46, + "grad_norm": 8.069449424743652, + "learning_rate": 0.0001900532327733629, + "loss": 3.3666, + "step": 991 + }, + { + "epoch": 1.46, + "grad_norm": 4.8657917976379395, + "learning_rate": 0.0001900328297643944, + "loss": 3.3401, + "step": 992 + }, + { + "epoch": 1.46, + "grad_norm": 5.57826042175293, + "learning_rate": 0.00019001240694880672, + "loss": 3.393, + "step": 993 + }, + { + "epoch": 1.46, + "grad_norm": 8.596120834350586, + "learning_rate": 0.00018999196433109274, + "loss": 3.416, + "step": 994 + }, + { + "epoch": 1.46, + "grad_norm": 4.3653130531311035, + "learning_rate": 0.0001899715019157497, + "loss": 3.3789, + "step": 995 + }, + { + "epoch": 1.46, + "grad_norm": 6.067572593688965, + "learning_rate": 0.0001899510197072792, + "loss": 3.3577, + "step": 996 + }, + { + "epoch": 1.47, + "grad_norm": 7.723760604858398, + "learning_rate": 0.00018993051771018719, + "loss": 3.4, + "step": 997 + }, + { + "epoch": 1.47, + "grad_norm": 4.619333744049072, + "learning_rate": 0.00018990999592898396, + "loss": 3.3979, + "step": 998 + }, + { + "epoch": 1.47, + "grad_norm": 6.598495006561279, + "learning_rate": 0.00018988945436818422, + "loss": 3.4331, + "step": 999 + }, + { + "epoch": 1.47, + "grad_norm": 9.450194358825684, + "learning_rate": 0.0001898688930323069, + "loss": 3.4156, + "step": 1000 + }, + { + "epoch": 1.47, + "grad_norm": 4.10836935043335, + "learning_rate": 0.00018984831192587538, + "loss": 3.3787, + "step": 1001 + }, + { + "epoch": 1.47, + "grad_norm": 6.087796211242676, + "learning_rate": 0.00018982771105341737, + "loss": 3.3969, + "step": 1002 + }, + { + "epoch": 1.47, + "grad_norm": 7.915310859680176, + "learning_rate": 0.00018980709041946497, + "loss": 3.3948, + "step": 1003 + }, + { + "epoch": 1.48, + "grad_norm": 4.227004528045654, + "learning_rate": 0.0001897864500285545, + "loss": 3.3238, + "step": 1004 + }, + { + "epoch": 1.48, + "grad_norm": 6.717720031738281, + "learning_rate": 0.00018976578988522675, + "loss": 3.3781, + "step": 1005 + }, + { + "epoch": 1.48, + "grad_norm": 10.050957679748535, + "learning_rate": 0.0001897451099940268, + "loss": 3.4204, + "step": 1006 + }, + { + "epoch": 1.48, + "grad_norm": 4.475175857543945, + "learning_rate": 0.0001897244103595041, + "loss": 3.4221, + "step": 1007 + }, + { + "epoch": 1.48, + "grad_norm": 5.719995975494385, + "learning_rate": 0.00018970369098621244, + "loss": 3.3555, + "step": 1008 + }, + { + "epoch": 1.48, + "grad_norm": 7.594791889190674, + "learning_rate": 0.00018968295187870996, + "loss": 3.4073, + "step": 1009 + }, + { + "epoch": 1.49, + "grad_norm": 4.333056926727295, + "learning_rate": 0.00018966219304155906, + "loss": 3.3556, + "step": 1010 + }, + { + "epoch": 1.49, + "grad_norm": 6.078528881072998, + "learning_rate": 0.00018964141447932656, + "loss": 3.3484, + "step": 1011 + }, + { + "epoch": 1.49, + "grad_norm": 8.950298309326172, + "learning_rate": 0.00018962061619658366, + "loss": 3.4055, + "step": 1012 + }, + { + "epoch": 1.49, + "grad_norm": 4.340825080871582, + "learning_rate": 0.0001895997981979058, + "loss": 3.334, + "step": 1013 + }, + { + "epoch": 1.49, + "grad_norm": 6.103879451751709, + "learning_rate": 0.00018957896048787278, + "loss": 3.4908, + "step": 1014 + }, + { + "epoch": 1.49, + "eval_loss": 3.206282138824463, + "eval_runtime": 1164.0205, + "eval_samples_per_second": 14.727, + "eval_steps_per_second": 0.921, + "step": 1014 + }, + { + "epoch": 1.49, + "grad_norm": 7.772497653961182, + "learning_rate": 0.00018955810307106881, + "loss": 3.3821, + "step": 1015 + }, + { + "epoch": 1.49, + "grad_norm": 4.639206886291504, + "learning_rate": 0.00018953722595208236, + "loss": 3.3776, + "step": 1016 + }, + { + "epoch": 1.5, + "grad_norm": 5.658569812774658, + "learning_rate": 0.00018951632913550626, + "loss": 3.339, + "step": 1017 + }, + { + "epoch": 1.5, + "grad_norm": 8.45776081085205, + "learning_rate": 0.00018949541262593762, + "loss": 3.4214, + "step": 1018 + }, + { + "epoch": 1.5, + "grad_norm": 3.2400307655334473, + "learning_rate": 0.000189474476427978, + "loss": 3.3666, + "step": 1019 + }, + { + "epoch": 1.5, + "grad_norm": 6.397870063781738, + "learning_rate": 0.0001894535205462332, + "loss": 3.4133, + "step": 1020 + }, + { + "epoch": 1.5, + "grad_norm": 8.270928382873535, + "learning_rate": 0.00018943254498531337, + "loss": 3.4133, + "step": 1021 + }, + { + "epoch": 1.5, + "grad_norm": 5.132183074951172, + "learning_rate": 0.00018941154974983298, + "loss": 3.3984, + "step": 1022 + }, + { + "epoch": 1.5, + "grad_norm": 5.842728137969971, + "learning_rate": 0.0001893905348444109, + "loss": 3.4019, + "step": 1023 + }, + { + "epoch": 1.51, + "grad_norm": 8.618907928466797, + "learning_rate": 0.00018936950027367017, + "loss": 3.4036, + "step": 1024 + }, + { + "epoch": 1.51, + "grad_norm": 3.155228853225708, + "learning_rate": 0.00018934844604223835, + "loss": 3.3512, + "step": 1025 + }, + { + "epoch": 1.51, + "grad_norm": 6.209192752838135, + "learning_rate": 0.00018932737215474714, + "loss": 3.388, + "step": 1026 + }, + { + "epoch": 1.51, + "grad_norm": 7.953395843505859, + "learning_rate": 0.00018930627861583275, + "loss": 3.402, + "step": 1027 + }, + { + "epoch": 1.51, + "grad_norm": 4.540157794952393, + "learning_rate": 0.00018928516543013552, + "loss": 3.3162, + "step": 1028 + }, + { + "epoch": 1.51, + "grad_norm": 6.285725116729736, + "learning_rate": 0.0001892640326023003, + "loss": 3.4329, + "step": 1029 + }, + { + "epoch": 1.51, + "grad_norm": 9.266477584838867, + "learning_rate": 0.0001892428801369761, + "loss": 3.381, + "step": 1030 + }, + { + "epoch": 1.52, + "grad_norm": 4.977963924407959, + "learning_rate": 0.00018922170803881635, + "loss": 3.3758, + "step": 1031 + }, + { + "epoch": 1.52, + "grad_norm": 5.400416374206543, + "learning_rate": 0.0001892005163124788, + "loss": 3.3279, + "step": 1032 + }, + { + "epoch": 1.52, + "grad_norm": 6.933363914489746, + "learning_rate": 0.00018917930496262542, + "loss": 3.3655, + "step": 1033 + }, + { + "epoch": 1.52, + "grad_norm": 3.6254875659942627, + "learning_rate": 0.00018915807399392262, + "loss": 3.3451, + "step": 1034 + }, + { + "epoch": 1.52, + "grad_norm": 7.3330841064453125, + "learning_rate": 0.00018913682341104105, + "loss": 3.4291, + "step": 1035 + }, + { + "epoch": 1.52, + "grad_norm": 10.801877975463867, + "learning_rate": 0.0001891155532186557, + "loss": 3.4253, + "step": 1036 + }, + { + "epoch": 1.53, + "grad_norm": 5.197689533233643, + "learning_rate": 0.00018909426342144587, + "loss": 3.3764, + "step": 1037 + }, + { + "epoch": 1.53, + "grad_norm": 5.4762797355651855, + "learning_rate": 0.00018907295402409514, + "loss": 3.3669, + "step": 1038 + }, + { + "epoch": 1.53, + "grad_norm": 9.155226707458496, + "learning_rate": 0.0001890516250312915, + "loss": 3.3748, + "step": 1039 + }, + { + "epoch": 1.53, + "grad_norm": 4.974143028259277, + "learning_rate": 0.00018903027644772715, + "loss": 3.3812, + "step": 1040 + }, + { + "epoch": 1.53, + "grad_norm": 3.935563802719116, + "learning_rate": 0.00018900890827809862, + "loss": 3.3748, + "step": 1041 + }, + { + "epoch": 1.53, + "grad_norm": 5.06904935836792, + "learning_rate": 0.0001889875205271068, + "loss": 3.4328, + "step": 1042 + }, + { + "epoch": 1.53, + "grad_norm": 1.7735739946365356, + "learning_rate": 0.0001889661131994568, + "loss": 3.3331, + "step": 1043 + }, + { + "epoch": 1.54, + "grad_norm": 3.266242504119873, + "learning_rate": 0.00018894468629985814, + "loss": 3.3797, + "step": 1044 + }, + { + "epoch": 1.54, + "grad_norm": 1.9417365789413452, + "learning_rate": 0.0001889232398330246, + "loss": 3.3556, + "step": 1045 + }, + { + "epoch": 1.54, + "grad_norm": 2.2055392265319824, + "learning_rate": 0.00018890177380367419, + "loss": 3.3695, + "step": 1046 + }, + { + "epoch": 1.54, + "grad_norm": 3.9984004497528076, + "learning_rate": 0.00018888028821652936, + "loss": 3.4158, + "step": 1047 + }, + { + "epoch": 1.54, + "grad_norm": 2.413911819458008, + "learning_rate": 0.0001888587830763168, + "loss": 3.4492, + "step": 1048 + }, + { + "epoch": 1.54, + "grad_norm": 4.597828388214111, + "learning_rate": 0.00018883725838776742, + "loss": 3.3518, + "step": 1049 + }, + { + "epoch": 1.54, + "grad_norm": 4.952722549438477, + "learning_rate": 0.00018881571415561656, + "loss": 3.371, + "step": 1050 + }, + { + "epoch": 1.55, + "grad_norm": 2.318856954574585, + "learning_rate": 0.00018879415038460383, + "loss": 3.3974, + "step": 1051 + }, + { + "epoch": 1.55, + "grad_norm": 2.227694034576416, + "learning_rate": 0.00018877256707947306, + "loss": 3.3784, + "step": 1052 + }, + { + "epoch": 1.55, + "grad_norm": 4.430450916290283, + "learning_rate": 0.00018875096424497248, + "loss": 3.3792, + "step": 1053 + }, + { + "epoch": 1.55, + "grad_norm": 2.9697399139404297, + "learning_rate": 0.00018872934188585454, + "loss": 3.4059, + "step": 1054 + }, + { + "epoch": 1.55, + "grad_norm": 4.346686840057373, + "learning_rate": 0.000188707700006876, + "loss": 3.4155, + "step": 1055 + }, + { + "epoch": 1.55, + "grad_norm": 4.2472381591796875, + "learning_rate": 0.000188686038612798, + "loss": 3.4242, + "step": 1056 + }, + { + "epoch": 1.55, + "grad_norm": 3.157696008682251, + "learning_rate": 0.0001886643577083858, + "loss": 3.4913, + "step": 1057 + }, + { + "epoch": 1.56, + "grad_norm": 2.940396308898926, + "learning_rate": 0.00018864265729840913, + "loss": 3.4734, + "step": 1058 + }, + { + "epoch": 1.56, + "grad_norm": 4.0289154052734375, + "learning_rate": 0.00018862093738764192, + "loss": 3.443, + "step": 1059 + }, + { + "epoch": 1.56, + "grad_norm": 3.4877893924713135, + "learning_rate": 0.00018859919798086238, + "loss": 3.5322, + "step": 1060 + }, + { + "epoch": 1.56, + "grad_norm": 4.22461462020874, + "learning_rate": 0.00018857743908285305, + "loss": 3.4562, + "step": 1061 + }, + { + "epoch": 1.56, + "grad_norm": 4.031467437744141, + "learning_rate": 0.00018855566069840077, + "loss": 3.4653, + "step": 1062 + }, + { + "epoch": 1.56, + "grad_norm": 3.602692127227783, + "learning_rate": 0.00018853386283229658, + "loss": 3.4555, + "step": 1063 + }, + { + "epoch": 1.57, + "grad_norm": 2.7948570251464844, + "learning_rate": 0.00018851204548933593, + "loss": 3.4255, + "step": 1064 + }, + { + "epoch": 1.57, + "grad_norm": 4.970945358276367, + "learning_rate": 0.00018849020867431842, + "loss": 3.4463, + "step": 1065 + }, + { + "epoch": 1.57, + "grad_norm": 4.409501552581787, + "learning_rate": 0.00018846835239204804, + "loss": 3.3992, + "step": 1066 + }, + { + "epoch": 1.57, + "grad_norm": 2.8853704929351807, + "learning_rate": 0.000188446476647333, + "loss": 3.4709, + "step": 1067 + }, + { + "epoch": 1.57, + "grad_norm": 2.3841402530670166, + "learning_rate": 0.00018842458144498588, + "loss": 3.4342, + "step": 1068 + }, + { + "epoch": 1.57, + "grad_norm": 4.642658710479736, + "learning_rate": 0.00018840266678982342, + "loss": 3.4102, + "step": 1069 + }, + { + "epoch": 1.57, + "grad_norm": 4.265832424163818, + "learning_rate": 0.00018838073268666672, + "loss": 3.4367, + "step": 1070 + }, + { + "epoch": 1.58, + "grad_norm": 2.900568723678589, + "learning_rate": 0.00018835877914034113, + "loss": 3.4052, + "step": 1071 + }, + { + "epoch": 1.58, + "grad_norm": 2.4887197017669678, + "learning_rate": 0.00018833680615567628, + "loss": 3.4175, + "step": 1072 + }, + { + "epoch": 1.58, + "grad_norm": 5.449963092803955, + "learning_rate": 0.0001883148137375061, + "loss": 3.4521, + "step": 1073 + }, + { + "epoch": 1.58, + "grad_norm": 5.6257643699646, + "learning_rate": 0.00018829280189066873, + "loss": 3.4741, + "step": 1074 + }, + { + "epoch": 1.58, + "grad_norm": 2.799276828765869, + "learning_rate": 0.00018827077062000668, + "loss": 3.4431, + "step": 1075 + }, + { + "epoch": 1.58, + "grad_norm": 2.863265037536621, + "learning_rate": 0.00018824871993036664, + "loss": 3.3978, + "step": 1076 + }, + { + "epoch": 1.58, + "grad_norm": 4.204624176025391, + "learning_rate": 0.00018822664982659964, + "loss": 3.4559, + "step": 1077 + }, + { + "epoch": 1.59, + "grad_norm": 3.0599143505096436, + "learning_rate": 0.00018820456031356093, + "loss": 3.3883, + "step": 1078 + }, + { + "epoch": 1.59, + "grad_norm": 4.109722137451172, + "learning_rate": 0.00018818245139611011, + "loss": 3.4014, + "step": 1079 + }, + { + "epoch": 1.59, + "grad_norm": 3.4017956256866455, + "learning_rate": 0.00018816032307911092, + "loss": 3.4348, + "step": 1080 + }, + { + "epoch": 1.59, + "grad_norm": 4.216831207275391, + "learning_rate": 0.00018813817536743148, + "loss": 3.452, + "step": 1081 + }, + { + "epoch": 1.59, + "grad_norm": 3.819659471511841, + "learning_rate": 0.00018811600826594416, + "loss": 3.3937, + "step": 1082 + }, + { + "epoch": 1.59, + "grad_norm": 3.5902295112609863, + "learning_rate": 0.00018809382177952552, + "loss": 3.3992, + "step": 1083 + }, + { + "epoch": 1.59, + "grad_norm": 2.6973037719726562, + "learning_rate": 0.0001880716159130565, + "loss": 3.4278, + "step": 1084 + }, + { + "epoch": 1.6, + "grad_norm": 5.055513858795166, + "learning_rate": 0.0001880493906714222, + "loss": 3.3913, + "step": 1085 + }, + { + "epoch": 1.6, + "grad_norm": 4.533970832824707, + "learning_rate": 0.00018802714605951199, + "loss": 3.4047, + "step": 1086 + }, + { + "epoch": 1.6, + "grad_norm": 2.5329983234405518, + "learning_rate": 0.00018800488208221963, + "loss": 3.4291, + "step": 1087 + }, + { + "epoch": 1.6, + "grad_norm": 1.993481993675232, + "learning_rate": 0.00018798259874444297, + "loss": 3.4334, + "step": 1088 + }, + { + "epoch": 1.6, + "grad_norm": 6.799105167388916, + "learning_rate": 0.00018796029605108424, + "loss": 3.4107, + "step": 1089 + }, + { + "epoch": 1.6, + "grad_norm": 6.37121057510376, + "learning_rate": 0.00018793797400704985, + "loss": 3.4237, + "step": 1090 + }, + { + "epoch": 1.61, + "grad_norm": 1.2064040899276733, + "learning_rate": 0.0001879156326172505, + "loss": 3.4487, + "step": 1091 + }, + { + "epoch": 1.61, + "grad_norm": 0.6396639943122864, + "learning_rate": 0.00018789327188660116, + "loss": 3.3509, + "step": 1092 + }, + { + "epoch": 1.61, + "grad_norm": 2.529014825820923, + "learning_rate": 0.00018787089182002105, + "loss": 3.4014, + "step": 1093 + }, + { + "epoch": 1.61, + "grad_norm": 2.6367127895355225, + "learning_rate": 0.00018784849242243362, + "loss": 3.3676, + "step": 1094 + }, + { + "epoch": 1.61, + "grad_norm": 0.968987226486206, + "learning_rate": 0.00018782607369876659, + "loss": 3.4156, + "step": 1095 + }, + { + "epoch": 1.61, + "grad_norm": 5.669402599334717, + "learning_rate": 0.00018780363565395192, + "loss": 3.4195, + "step": 1096 + }, + { + "epoch": 1.61, + "grad_norm": 4.468107223510742, + "learning_rate": 0.00018778117829292587, + "loss": 3.4453, + "step": 1097 + }, + { + "epoch": 1.62, + "grad_norm": 3.340261220932007, + "learning_rate": 0.00018775870162062886, + "loss": 3.4024, + "step": 1098 + }, + { + "epoch": 1.62, + "grad_norm": 2.798548698425293, + "learning_rate": 0.00018773620564200565, + "loss": 3.503, + "step": 1099 + }, + { + "epoch": 1.62, + "grad_norm": 5.848763465881348, + "learning_rate": 0.00018771369036200513, + "loss": 3.3962, + "step": 1100 + }, + { + "epoch": 1.62, + "grad_norm": 6.020981788635254, + "learning_rate": 0.00018769115578558062, + "loss": 3.4708, + "step": 1101 + }, + { + "epoch": 1.62, + "grad_norm": 2.3334991931915283, + "learning_rate": 0.00018766860191768946, + "loss": 3.3861, + "step": 1102 + }, + { + "epoch": 1.62, + "grad_norm": 3.1461942195892334, + "learning_rate": 0.00018764602876329345, + "loss": 3.3828, + "step": 1103 + }, + { + "epoch": 1.62, + "grad_norm": 2.833292007446289, + "learning_rate": 0.00018762343632735848, + "loss": 3.4628, + "step": 1104 + }, + { + "epoch": 1.63, + "grad_norm": 1.1478309631347656, + "learning_rate": 0.00018760082461485476, + "loss": 3.3512, + "step": 1105 + }, + { + "epoch": 1.63, + "grad_norm": 4.6011199951171875, + "learning_rate": 0.00018757819363075665, + "loss": 3.4301, + "step": 1106 + }, + { + "epoch": 1.63, + "grad_norm": 3.997016191482544, + "learning_rate": 0.0001875555433800429, + "loss": 3.4099, + "step": 1107 + }, + { + "epoch": 1.63, + "grad_norm": 4.506150722503662, + "learning_rate": 0.00018753287386769638, + "loss": 3.4169, + "step": 1108 + }, + { + "epoch": 1.63, + "grad_norm": 4.197047710418701, + "learning_rate": 0.0001875101850987042, + "loss": 3.3518, + "step": 1109 + }, + { + "epoch": 1.63, + "grad_norm": 3.504765033721924, + "learning_rate": 0.0001874874770780578, + "loss": 3.3838, + "step": 1110 + }, + { + "epoch": 1.64, + "grad_norm": 2.7405476570129395, + "learning_rate": 0.00018746474981075272, + "loss": 3.376, + "step": 1111 + }, + { + "epoch": 1.64, + "grad_norm": 4.3409037590026855, + "learning_rate": 0.00018744200330178887, + "loss": 3.4184, + "step": 1112 + }, + { + "epoch": 1.64, + "grad_norm": 4.318120002746582, + "learning_rate": 0.0001874192375561703, + "loss": 3.4744, + "step": 1113 + }, + { + "epoch": 1.64, + "grad_norm": 3.5253307819366455, + "learning_rate": 0.00018739645257890533, + "loss": 3.4058, + "step": 1114 + }, + { + "epoch": 1.64, + "grad_norm": 3.368002414703369, + "learning_rate": 0.0001873736483750065, + "loss": 3.4013, + "step": 1115 + }, + { + "epoch": 1.64, + "grad_norm": 3.804229736328125, + "learning_rate": 0.00018735082494949058, + "loss": 3.3625, + "step": 1116 + }, + { + "epoch": 1.64, + "grad_norm": 3.6312084197998047, + "learning_rate": 0.00018732798230737855, + "loss": 3.3568, + "step": 1117 + }, + { + "epoch": 1.65, + "grad_norm": 3.4278273582458496, + "learning_rate": 0.00018730512045369568, + "loss": 3.4077, + "step": 1118 + }, + { + "epoch": 1.65, + "grad_norm": 3.177262306213379, + "learning_rate": 0.0001872822393934714, + "loss": 3.41, + "step": 1119 + }, + { + "epoch": 1.65, + "grad_norm": 3.8335347175598145, + "learning_rate": 0.00018725933913173938, + "loss": 3.3968, + "step": 1120 + }, + { + "epoch": 1.65, + "grad_norm": 3.43869948387146, + "learning_rate": 0.00018723641967353756, + "loss": 3.4295, + "step": 1121 + }, + { + "epoch": 1.65, + "grad_norm": 4.323538780212402, + "learning_rate": 0.00018721348102390803, + "loss": 3.3734, + "step": 1122 + }, + { + "epoch": 1.65, + "grad_norm": 3.6614975929260254, + "learning_rate": 0.00018719052318789717, + "loss": 3.3522, + "step": 1123 + }, + { + "epoch": 1.65, + "grad_norm": 3.442986249923706, + "learning_rate": 0.0001871675461705555, + "loss": 3.3743, + "step": 1124 + }, + { + "epoch": 1.66, + "grad_norm": 3.7068862915039062, + "learning_rate": 0.00018714454997693783, + "loss": 3.3611, + "step": 1125 + }, + { + "epoch": 1.66, + "grad_norm": 3.6128504276275635, + "learning_rate": 0.00018712153461210326, + "loss": 3.3678, + "step": 1126 + }, + { + "epoch": 1.66, + "grad_norm": 3.2057278156280518, + "learning_rate": 0.00018709850008111488, + "loss": 3.4118, + "step": 1127 + }, + { + "epoch": 1.66, + "grad_norm": 4.726739406585693, + "learning_rate": 0.0001870754463890402, + "loss": 3.3669, + "step": 1128 + }, + { + "epoch": 1.66, + "grad_norm": 4.606038570404053, + "learning_rate": 0.00018705237354095085, + "loss": 3.3316, + "step": 1129 + }, + { + "epoch": 1.66, + "grad_norm": 2.9650778770446777, + "learning_rate": 0.00018702928154192274, + "loss": 3.3614, + "step": 1130 + }, + { + "epoch": 1.66, + "grad_norm": 2.7509312629699707, + "learning_rate": 0.00018700617039703595, + "loss": 3.3797, + "step": 1131 + }, + { + "epoch": 1.67, + "grad_norm": 5.575433731079102, + "learning_rate": 0.00018698304011137475, + "loss": 3.384, + "step": 1132 + }, + { + "epoch": 1.67, + "grad_norm": 4.589602947235107, + "learning_rate": 0.00018695989069002764, + "loss": 3.3729, + "step": 1133 + }, + { + "epoch": 1.67, + "grad_norm": 2.780677318572998, + "learning_rate": 0.00018693672213808738, + "loss": 3.3704, + "step": 1134 + }, + { + "epoch": 1.67, + "grad_norm": 1.9864367246627808, + "learning_rate": 0.00018691353446065087, + "loss": 3.3718, + "step": 1135 + }, + { + "epoch": 1.67, + "grad_norm": 5.260591983795166, + "learning_rate": 0.00018689032766281924, + "loss": 3.3856, + "step": 1136 + }, + { + "epoch": 1.67, + "grad_norm": 4.4715447425842285, + "learning_rate": 0.00018686710174969787, + "loss": 3.3433, + "step": 1137 + }, + { + "epoch": 1.68, + "grad_norm": 3.0349228382110596, + "learning_rate": 0.00018684385672639622, + "loss": 3.3361, + "step": 1138 + }, + { + "epoch": 1.68, + "grad_norm": 2.5944900512695312, + "learning_rate": 0.0001868205925980281, + "loss": 3.3379, + "step": 1139 + }, + { + "epoch": 1.68, + "grad_norm": 4.727088928222656, + "learning_rate": 0.00018679730936971148, + "loss": 3.3889, + "step": 1140 + }, + { + "epoch": 1.68, + "grad_norm": 5.072381019592285, + "learning_rate": 0.0001867740070465685, + "loss": 3.4185, + "step": 1141 + }, + { + "epoch": 1.68, + "grad_norm": 2.782468318939209, + "learning_rate": 0.00018675068563372545, + "loss": 3.3635, + "step": 1142 + }, + { + "epoch": 1.68, + "grad_norm": 2.7385001182556152, + "learning_rate": 0.000186727345136313, + "loss": 3.4115, + "step": 1143 + }, + { + "epoch": 1.68, + "grad_norm": 4.811634540557861, + "learning_rate": 0.0001867039855594658, + "loss": 3.4155, + "step": 1144 + }, + { + "epoch": 1.69, + "grad_norm": 4.225504398345947, + "learning_rate": 0.00018668060690832285, + "loss": 3.3729, + "step": 1145 + }, + { + "epoch": 1.69, + "grad_norm": 3.26381254196167, + "learning_rate": 0.0001866572091880273, + "loss": 3.3982, + "step": 1146 + }, + { + "epoch": 1.69, + "grad_norm": 2.842423915863037, + "learning_rate": 0.00018663379240372647, + "loss": 3.387, + "step": 1147 + }, + { + "epoch": 1.69, + "grad_norm": 4.7502217292785645, + "learning_rate": 0.0001866103565605719, + "loss": 3.377, + "step": 1148 + }, + { + "epoch": 1.69, + "grad_norm": 4.280509948730469, + "learning_rate": 0.00018658690166371933, + "loss": 3.3774, + "step": 1149 + }, + { + "epoch": 1.69, + "grad_norm": 3.5164103507995605, + "learning_rate": 0.00018656342771832866, + "loss": 3.357, + "step": 1150 + }, + { + "epoch": 1.69, + "grad_norm": 2.891474485397339, + "learning_rate": 0.00018653993472956404, + "loss": 3.3755, + "step": 1151 + }, + { + "epoch": 1.7, + "grad_norm": 4.281400203704834, + "learning_rate": 0.00018651642270259373, + "loss": 3.3511, + "step": 1152 + }, + { + "epoch": 1.7, + "grad_norm": 4.510589122772217, + "learning_rate": 0.00018649289164259024, + "loss": 3.4113, + "step": 1153 + }, + { + "epoch": 1.7, + "grad_norm": 2.6343142986297607, + "learning_rate": 0.00018646934155473022, + "loss": 3.3291, + "step": 1154 + }, + { + "epoch": 1.7, + "grad_norm": 2.3589959144592285, + "learning_rate": 0.00018644577244419455, + "loss": 3.3638, + "step": 1155 + }, + { + "epoch": 1.7, + "grad_norm": 5.584714889526367, + "learning_rate": 0.00018642218431616828, + "loss": 3.32, + "step": 1156 + }, + { + "epoch": 1.7, + "grad_norm": 4.909949779510498, + "learning_rate": 0.00018639857717584065, + "loss": 3.3947, + "step": 1157 + }, + { + "epoch": 1.7, + "grad_norm": 2.9206135272979736, + "learning_rate": 0.00018637495102840506, + "loss": 3.3826, + "step": 1158 + }, + { + "epoch": 1.71, + "grad_norm": 2.58848237991333, + "learning_rate": 0.0001863513058790591, + "loss": 3.3237, + "step": 1159 + }, + { + "epoch": 1.71, + "grad_norm": 5.167590141296387, + "learning_rate": 0.00018632764173300457, + "loss": 3.3972, + "step": 1160 + }, + { + "epoch": 1.71, + "grad_norm": 5.174654483795166, + "learning_rate": 0.00018630395859544735, + "loss": 3.357, + "step": 1161 + }, + { + "epoch": 1.71, + "grad_norm": 2.867471218109131, + "learning_rate": 0.00018628025647159766, + "loss": 3.3406, + "step": 1162 + }, + { + "epoch": 1.71, + "grad_norm": 2.780794620513916, + "learning_rate": 0.00018625653536666974, + "loss": 3.3527, + "step": 1163 + }, + { + "epoch": 1.71, + "grad_norm": 4.502468585968018, + "learning_rate": 0.00018623279528588214, + "loss": 3.3216, + "step": 1164 + }, + { + "epoch": 1.72, + "grad_norm": 3.549867868423462, + "learning_rate": 0.00018620903623445747, + "loss": 3.3949, + "step": 1165 + }, + { + "epoch": 1.72, + "grad_norm": 3.7050774097442627, + "learning_rate": 0.00018618525821762256, + "loss": 3.395, + "step": 1166 + }, + { + "epoch": 1.72, + "grad_norm": 3.520019769668579, + "learning_rate": 0.00018616146124060846, + "loss": 3.3757, + "step": 1167 + }, + { + "epoch": 1.72, + "grad_norm": 3.870093822479248, + "learning_rate": 0.00018613764530865028, + "loss": 3.3712, + "step": 1168 + }, + { + "epoch": 1.72, + "grad_norm": 3.196948766708374, + "learning_rate": 0.0001861138104269874, + "loss": 3.3337, + "step": 1169 + }, + { + "epoch": 1.72, + "grad_norm": 4.416697025299072, + "learning_rate": 0.00018608995660086337, + "loss": 3.4137, + "step": 1170 + }, + { + "epoch": 1.72, + "grad_norm": 3.9429569244384766, + "learning_rate": 0.00018606608383552582, + "loss": 3.389, + "step": 1171 + }, + { + "epoch": 1.73, + "grad_norm": 3.530494213104248, + "learning_rate": 0.0001860421921362266, + "loss": 3.3752, + "step": 1172 + }, + { + "epoch": 1.73, + "grad_norm": 3.766986608505249, + "learning_rate": 0.00018601828150822177, + "loss": 3.3386, + "step": 1173 + }, + { + "epoch": 1.73, + "grad_norm": 3.8910138607025146, + "learning_rate": 0.00018599435195677146, + "loss": 3.348, + "step": 1174 + }, + { + "epoch": 1.73, + "grad_norm": 3.5032382011413574, + "learning_rate": 0.00018597040348714003, + "loss": 3.408, + "step": 1175 + }, + { + "epoch": 1.73, + "grad_norm": 3.5396668910980225, + "learning_rate": 0.000185946436104596, + "loss": 3.317, + "step": 1176 + }, + { + "epoch": 1.73, + "grad_norm": 3.7018423080444336, + "learning_rate": 0.00018592244981441197, + "loss": 3.3456, + "step": 1177 + }, + { + "epoch": 1.73, + "grad_norm": 3.41996169090271, + "learning_rate": 0.00018589844462186484, + "loss": 3.4147, + "step": 1178 + }, + { + "epoch": 1.74, + "grad_norm": 2.907388210296631, + "learning_rate": 0.00018587442053223557, + "loss": 3.3377, + "step": 1179 + }, + { + "epoch": 1.74, + "grad_norm": 4.651727676391602, + "learning_rate": 0.00018585037755080926, + "loss": 3.3401, + "step": 1180 + }, + { + "epoch": 1.74, + "grad_norm": 4.285563945770264, + "learning_rate": 0.00018582631568287523, + "loss": 3.3671, + "step": 1181 + }, + { + "epoch": 1.74, + "grad_norm": 3.3015687465667725, + "learning_rate": 0.00018580223493372695, + "loss": 3.3049, + "step": 1182 + }, + { + "epoch": 1.74, + "grad_norm": 2.8395988941192627, + "learning_rate": 0.000185778135308662, + "loss": 3.3588, + "step": 1183 + }, + { + "epoch": 1.74, + "eval_loss": 3.1614086627960205, + "eval_runtime": 1168.1043, + "eval_samples_per_second": 14.675, + "eval_steps_per_second": 0.918, + "step": 1183 + }, + { + "epoch": 1.74, + "grad_norm": 4.551041603088379, + "learning_rate": 0.00018575401681298216, + "loss": 3.3579, + "step": 1184 + }, + { + "epoch": 1.74, + "grad_norm": 4.050691604614258, + "learning_rate": 0.00018572987945199328, + "loss": 3.3174, + "step": 1185 + }, + { + "epoch": 1.75, + "grad_norm": 3.483586072921753, + "learning_rate": 0.0001857057232310055, + "loss": 3.3689, + "step": 1186 + }, + { + "epoch": 1.75, + "grad_norm": 3.430351734161377, + "learning_rate": 0.00018568154815533293, + "loss": 3.3624, + "step": 1187 + }, + { + "epoch": 1.75, + "grad_norm": 4.206818103790283, + "learning_rate": 0.00018565735423029404, + "loss": 3.3365, + "step": 1188 + }, + { + "epoch": 1.75, + "grad_norm": 3.6076526641845703, + "learning_rate": 0.00018563314146121122, + "loss": 3.3728, + "step": 1189 + }, + { + "epoch": 1.75, + "grad_norm": 3.844933271408081, + "learning_rate": 0.0001856089098534112, + "loss": 3.3577, + "step": 1190 + }, + { + "epoch": 1.75, + "grad_norm": 3.1116933822631836, + "learning_rate": 0.00018558465941222472, + "loss": 3.3242, + "step": 1191 + }, + { + "epoch": 1.76, + "grad_norm": 4.367002010345459, + "learning_rate": 0.00018556039014298672, + "loss": 3.3382, + "step": 1192 + }, + { + "epoch": 1.76, + "grad_norm": 4.229283332824707, + "learning_rate": 0.0001855361020510363, + "loss": 3.3318, + "step": 1193 + }, + { + "epoch": 1.76, + "grad_norm": 3.6189987659454346, + "learning_rate": 0.00018551179514171668, + "loss": 3.3879, + "step": 1194 + }, + { + "epoch": 1.76, + "grad_norm": 3.045738935470581, + "learning_rate": 0.0001854874694203752, + "loss": 3.3407, + "step": 1195 + }, + { + "epoch": 1.76, + "grad_norm": 4.931478977203369, + "learning_rate": 0.00018546312489236335, + "loss": 3.3428, + "step": 1196 + }, + { + "epoch": 1.76, + "grad_norm": 4.312923908233643, + "learning_rate": 0.00018543876156303676, + "loss": 3.3768, + "step": 1197 + }, + { + "epoch": 1.76, + "grad_norm": 3.236755132675171, + "learning_rate": 0.00018541437943775523, + "loss": 3.3637, + "step": 1198 + }, + { + "epoch": 1.77, + "grad_norm": 2.9794223308563232, + "learning_rate": 0.00018538997852188266, + "loss": 3.2961, + "step": 1199 + }, + { + "epoch": 1.77, + "grad_norm": 4.777314186096191, + "learning_rate": 0.00018536555882078706, + "loss": 3.3838, + "step": 1200 + }, + { + "epoch": 1.77, + "grad_norm": 4.084963321685791, + "learning_rate": 0.0001853411203398406, + "loss": 3.3525, + "step": 1201 + }, + { + "epoch": 1.77, + "grad_norm": 3.598912477493286, + "learning_rate": 0.0001853166630844196, + "loss": 3.3362, + "step": 1202 + }, + { + "epoch": 1.77, + "grad_norm": 3.44163179397583, + "learning_rate": 0.0001852921870599045, + "loss": 3.4129, + "step": 1203 + }, + { + "epoch": 1.77, + "grad_norm": 3.8749027252197266, + "learning_rate": 0.00018526769227167984, + "loss": 3.3069, + "step": 1204 + }, + { + "epoch": 1.77, + "grad_norm": 3.8452203273773193, + "learning_rate": 0.00018524317872513433, + "loss": 3.3084, + "step": 1205 + }, + { + "epoch": 1.78, + "grad_norm": 3.8901963233947754, + "learning_rate": 0.00018521864642566076, + "loss": 3.322, + "step": 1206 + }, + { + "epoch": 1.78, + "grad_norm": 3.340991973876953, + "learning_rate": 0.00018519409537865608, + "loss": 3.3373, + "step": 1207 + }, + { + "epoch": 1.78, + "grad_norm": 4.287864685058594, + "learning_rate": 0.0001851695255895214, + "loss": 3.2985, + "step": 1208 + }, + { + "epoch": 1.78, + "grad_norm": 3.954977512359619, + "learning_rate": 0.00018514493706366184, + "loss": 3.3286, + "step": 1209 + }, + { + "epoch": 1.78, + "grad_norm": 3.5455312728881836, + "learning_rate": 0.00018512032980648674, + "loss": 3.3191, + "step": 1210 + }, + { + "epoch": 1.78, + "grad_norm": 3.2507898807525635, + "learning_rate": 0.00018509570382340955, + "loss": 3.3356, + "step": 1211 + }, + { + "epoch": 1.79, + "grad_norm": 4.234015464782715, + "learning_rate": 0.0001850710591198478, + "loss": 3.365, + "step": 1212 + }, + { + "epoch": 1.79, + "grad_norm": 3.4031906127929688, + "learning_rate": 0.00018504639570122316, + "loss": 3.3405, + "step": 1213 + }, + { + "epoch": 1.79, + "grad_norm": 3.9651551246643066, + "learning_rate": 0.00018502171357296144, + "loss": 3.3953, + "step": 1214 + }, + { + "epoch": 1.79, + "grad_norm": 3.6755764484405518, + "learning_rate": 0.00018499701274049248, + "loss": 3.3102, + "step": 1215 + }, + { + "epoch": 1.79, + "grad_norm": 3.6327264308929443, + "learning_rate": 0.00018497229320925038, + "loss": 3.318, + "step": 1216 + }, + { + "epoch": 1.79, + "grad_norm": 3.4605491161346436, + "learning_rate": 0.0001849475549846732, + "loss": 3.3059, + "step": 1217 + }, + { + "epoch": 1.79, + "grad_norm": 3.7450873851776123, + "learning_rate": 0.00018492279807220326, + "loss": 3.3303, + "step": 1218 + }, + { + "epoch": 1.8, + "grad_norm": 3.5053107738494873, + "learning_rate": 0.00018489802247728684, + "loss": 3.3211, + "step": 1219 + }, + { + "epoch": 1.8, + "grad_norm": 3.7314960956573486, + "learning_rate": 0.00018487322820537444, + "loss": 3.3359, + "step": 1220 + }, + { + "epoch": 1.8, + "grad_norm": 3.397026777267456, + "learning_rate": 0.00018484841526192065, + "loss": 3.3199, + "step": 1221 + }, + { + "epoch": 1.8, + "grad_norm": 3.9689042568206787, + "learning_rate": 0.00018482358365238413, + "loss": 3.3249, + "step": 1222 + }, + { + "epoch": 1.8, + "grad_norm": 3.4028615951538086, + "learning_rate": 0.00018479873338222763, + "loss": 3.3234, + "step": 1223 + }, + { + "epoch": 1.8, + "grad_norm": 4.0719780921936035, + "learning_rate": 0.00018477386445691813, + "loss": 3.3755, + "step": 1224 + }, + { + "epoch": 1.8, + "grad_norm": 3.8761250972747803, + "learning_rate": 0.00018474897688192655, + "loss": 3.3313, + "step": 1225 + }, + { + "epoch": 1.81, + "grad_norm": 3.763307809829712, + "learning_rate": 0.00018472407066272804, + "loss": 3.3112, + "step": 1226 + }, + { + "epoch": 1.81, + "grad_norm": 3.3675754070281982, + "learning_rate": 0.00018469914580480177, + "loss": 3.373, + "step": 1227 + }, + { + "epoch": 1.81, + "grad_norm": 3.7994701862335205, + "learning_rate": 0.00018467420231363103, + "loss": 3.3034, + "step": 1228 + }, + { + "epoch": 1.81, + "grad_norm": 3.433627128601074, + "learning_rate": 0.00018464924019470327, + "loss": 3.2695, + "step": 1229 + }, + { + "epoch": 1.81, + "grad_norm": 3.7266974449157715, + "learning_rate": 0.00018462425945350996, + "loss": 3.3498, + "step": 1230 + }, + { + "epoch": 1.81, + "grad_norm": 3.44728422164917, + "learning_rate": 0.00018459926009554666, + "loss": 3.3682, + "step": 1231 + }, + { + "epoch": 1.81, + "grad_norm": 4.175528049468994, + "learning_rate": 0.00018457424212631312, + "loss": 3.3374, + "step": 1232 + }, + { + "epoch": 1.82, + "grad_norm": 4.267003059387207, + "learning_rate": 0.0001845492055513131, + "loss": 3.2985, + "step": 1233 + }, + { + "epoch": 1.82, + "grad_norm": 3.53112530708313, + "learning_rate": 0.0001845241503760545, + "loss": 3.3654, + "step": 1234 + }, + { + "epoch": 1.82, + "grad_norm": 3.1011736392974854, + "learning_rate": 0.00018449907660604927, + "loss": 3.3169, + "step": 1235 + }, + { + "epoch": 1.82, + "grad_norm": 4.862052917480469, + "learning_rate": 0.0001844739842468135, + "loss": 3.329, + "step": 1236 + }, + { + "epoch": 1.82, + "grad_norm": 4.767108917236328, + "learning_rate": 0.00018444887330386727, + "loss": 3.3546, + "step": 1237 + }, + { + "epoch": 1.82, + "grad_norm": 3.3005921840667725, + "learning_rate": 0.00018442374378273488, + "loss": 3.338, + "step": 1238 + }, + { + "epoch": 1.83, + "grad_norm": 3.2703089714050293, + "learning_rate": 0.00018439859568894463, + "loss": 3.281, + "step": 1239 + }, + { + "epoch": 1.83, + "grad_norm": 4.4580888748168945, + "learning_rate": 0.00018437342902802897, + "loss": 3.3748, + "step": 1240 + }, + { + "epoch": 1.83, + "grad_norm": 3.4954566955566406, + "learning_rate": 0.00018434824380552433, + "loss": 3.3291, + "step": 1241 + }, + { + "epoch": 1.83, + "grad_norm": 3.5984458923339844, + "learning_rate": 0.00018432304002697137, + "loss": 3.3834, + "step": 1242 + }, + { + "epoch": 1.83, + "grad_norm": 3.217467784881592, + "learning_rate": 0.00018429781769791468, + "loss": 3.3814, + "step": 1243 + }, + { + "epoch": 1.83, + "grad_norm": 4.149362087249756, + "learning_rate": 0.00018427257682390304, + "loss": 3.3808, + "step": 1244 + }, + { + "epoch": 1.83, + "grad_norm": 3.7322371006011963, + "learning_rate": 0.0001842473174104893, + "loss": 3.3546, + "step": 1245 + }, + { + "epoch": 1.84, + "grad_norm": 3.667970895767212, + "learning_rate": 0.0001842220394632303, + "loss": 3.3144, + "step": 1246 + }, + { + "epoch": 1.84, + "grad_norm": 3.353289842605591, + "learning_rate": 0.00018419674298768705, + "loss": 3.367, + "step": 1247 + }, + { + "epoch": 1.84, + "grad_norm": 4.078193664550781, + "learning_rate": 0.00018417142798942462, + "loss": 3.3284, + "step": 1248 + }, + { + "epoch": 1.84, + "grad_norm": 3.791130304336548, + "learning_rate": 0.0001841460944740121, + "loss": 3.3223, + "step": 1249 + }, + { + "epoch": 1.84, + "grad_norm": 3.9178853034973145, + "learning_rate": 0.00018412074244702274, + "loss": 3.4204, + "step": 1250 + }, + { + "epoch": 1.84, + "grad_norm": 3.1923162937164307, + "learning_rate": 0.00018409537191403378, + "loss": 3.2976, + "step": 1251 + }, + { + "epoch": 1.84, + "grad_norm": 4.827831745147705, + "learning_rate": 0.00018406998288062658, + "loss": 3.3697, + "step": 1252 + }, + { + "epoch": 1.85, + "grad_norm": 5.54097843170166, + "learning_rate": 0.00018404457535238657, + "loss": 3.3319, + "step": 1253 + }, + { + "epoch": 1.85, + "grad_norm": 2.249396562576294, + "learning_rate": 0.00018401914933490322, + "loss": 3.3106, + "step": 1254 + }, + { + "epoch": 1.85, + "grad_norm": 1.4117510318756104, + "learning_rate": 0.0001839937048337701, + "loss": 3.3664, + "step": 1255 + }, + { + "epoch": 1.85, + "grad_norm": 5.818370819091797, + "learning_rate": 0.0001839682418545848, + "loss": 3.3226, + "step": 1256 + }, + { + "epoch": 1.85, + "grad_norm": 4.127283573150635, + "learning_rate": 0.00018394276040294903, + "loss": 3.2937, + "step": 1257 + }, + { + "epoch": 1.85, + "grad_norm": 3.759086847305298, + "learning_rate": 0.00018391726048446852, + "loss": 3.3379, + "step": 1258 + }, + { + "epoch": 1.85, + "grad_norm": 4.152595520019531, + "learning_rate": 0.0001838917421047531, + "loss": 3.3433, + "step": 1259 + }, + { + "epoch": 1.86, + "grad_norm": 2.377365827560425, + "learning_rate": 0.00018386620526941662, + "loss": 3.3103, + "step": 1260 + }, + { + "epoch": 1.86, + "grad_norm": 1.4664210081100464, + "learning_rate": 0.00018384064998407706, + "loss": 3.3038, + "step": 1261 + }, + { + "epoch": 1.86, + "grad_norm": 5.424790859222412, + "learning_rate": 0.00018381507625435633, + "loss": 3.3518, + "step": 1262 + }, + { + "epoch": 1.86, + "grad_norm": 4.665937900543213, + "learning_rate": 0.00018378948408588058, + "loss": 3.3437, + "step": 1263 + }, + { + "epoch": 1.86, + "grad_norm": 2.893569231033325, + "learning_rate": 0.00018376387348427984, + "loss": 3.2693, + "step": 1264 + }, + { + "epoch": 1.86, + "grad_norm": 3.674450159072876, + "learning_rate": 0.0001837382444551883, + "loss": 3.3324, + "step": 1265 + }, + { + "epoch": 1.87, + "grad_norm": 2.98838210105896, + "learning_rate": 0.00018371259700424416, + "loss": 3.3242, + "step": 1266 + }, + { + "epoch": 1.87, + "grad_norm": 2.417462110519409, + "learning_rate": 0.00018368693113708968, + "loss": 3.3771, + "step": 1267 + }, + { + "epoch": 1.87, + "grad_norm": 4.9014458656311035, + "learning_rate": 0.00018366124685937122, + "loss": 3.3235, + "step": 1268 + }, + { + "epoch": 1.87, + "grad_norm": 4.8541693687438965, + "learning_rate": 0.00018363554417673915, + "loss": 3.3149, + "step": 1269 + }, + { + "epoch": 1.87, + "grad_norm": 3.2948014736175537, + "learning_rate": 0.0001836098230948478, + "loss": 3.3211, + "step": 1270 + }, + { + "epoch": 1.87, + "grad_norm": 3.0249054431915283, + "learning_rate": 0.00018358408361935573, + "loss": 3.3073, + "step": 1271 + }, + { + "epoch": 1.87, + "grad_norm": 4.5272722244262695, + "learning_rate": 0.00018355832575592541, + "loss": 3.2809, + "step": 1272 + }, + { + "epoch": 1.88, + "grad_norm": 3.790753126144409, + "learning_rate": 0.0001835325495102234, + "loss": 3.3502, + "step": 1273 + }, + { + "epoch": 1.88, + "grad_norm": 3.7797255516052246, + "learning_rate": 0.00018350675488792035, + "loss": 3.2818, + "step": 1274 + }, + { + "epoch": 1.88, + "grad_norm": 3.5615875720977783, + "learning_rate": 0.00018348094189469082, + "loss": 3.3617, + "step": 1275 + }, + { + "epoch": 1.88, + "grad_norm": 3.750861167907715, + "learning_rate": 0.00018345511053621355, + "loss": 3.3649, + "step": 1276 + }, + { + "epoch": 1.88, + "grad_norm": 3.202655076980591, + "learning_rate": 0.00018342926081817127, + "loss": 3.3013, + "step": 1277 + }, + { + "epoch": 1.88, + "grad_norm": 3.6577436923980713, + "learning_rate": 0.0001834033927462507, + "loss": 3.3074, + "step": 1278 + }, + { + "epoch": 1.88, + "grad_norm": 3.164114236831665, + "learning_rate": 0.0001833775063261427, + "loss": 3.3211, + "step": 1279 + }, + { + "epoch": 1.89, + "grad_norm": 3.7693588733673096, + "learning_rate": 0.00018335160156354206, + "loss": 3.3229, + "step": 1280 + }, + { + "epoch": 1.89, + "grad_norm": 3.5220518112182617, + "learning_rate": 0.00018332567846414768, + "loss": 3.304, + "step": 1281 + }, + { + "epoch": 1.89, + "grad_norm": 3.884568691253662, + "learning_rate": 0.00018329973703366242, + "loss": 3.2905, + "step": 1282 + }, + { + "epoch": 1.89, + "grad_norm": 3.471947431564331, + "learning_rate": 0.0001832737772777933, + "loss": 3.3356, + "step": 1283 + }, + { + "epoch": 1.89, + "grad_norm": 3.8152236938476562, + "learning_rate": 0.00018324779920225127, + "loss": 3.3036, + "step": 1284 + }, + { + "epoch": 1.89, + "grad_norm": 3.813997507095337, + "learning_rate": 0.0001832218028127513, + "loss": 3.3465, + "step": 1285 + }, + { + "epoch": 1.89, + "grad_norm": 3.5465102195739746, + "learning_rate": 0.00018319578811501244, + "loss": 3.3241, + "step": 1286 + }, + { + "epoch": 1.9, + "grad_norm": 3.25008487701416, + "learning_rate": 0.00018316975511475776, + "loss": 3.3111, + "step": 1287 + }, + { + "epoch": 1.9, + "grad_norm": 4.218858242034912, + "learning_rate": 0.00018314370381771432, + "loss": 3.2914, + "step": 1288 + }, + { + "epoch": 1.9, + "grad_norm": 3.7324070930480957, + "learning_rate": 0.00018311763422961325, + "loss": 3.3188, + "step": 1289 + }, + { + "epoch": 1.9, + "grad_norm": 3.4570472240448, + "learning_rate": 0.00018309154635618965, + "loss": 3.3034, + "step": 1290 + }, + { + "epoch": 1.9, + "grad_norm": 2.9619879722595215, + "learning_rate": 0.00018306544020318274, + "loss": 3.3231, + "step": 1291 + }, + { + "epoch": 1.9, + "grad_norm": 4.235932350158691, + "learning_rate": 0.00018303931577633566, + "loss": 3.3226, + "step": 1292 + }, + { + "epoch": 1.91, + "grad_norm": 3.5511438846588135, + "learning_rate": 0.00018301317308139561, + "loss": 3.3305, + "step": 1293 + }, + { + "epoch": 1.91, + "grad_norm": 3.656703233718872, + "learning_rate": 0.00018298701212411382, + "loss": 3.3261, + "step": 1294 + }, + { + "epoch": 1.91, + "grad_norm": 3.338941812515259, + "learning_rate": 0.00018296083291024552, + "loss": 3.3564, + "step": 1295 + }, + { + "epoch": 1.91, + "grad_norm": 3.9150142669677734, + "learning_rate": 0.00018293463544554993, + "loss": 3.3292, + "step": 1296 + }, + { + "epoch": 1.91, + "grad_norm": 10.699624061584473, + "learning_rate": 0.00018290841973579039, + "loss": 3.3042, + "step": 1297 + }, + { + "epoch": 1.91, + "grad_norm": 1.4264583587646484, + "learning_rate": 0.0001828821857867341, + "loss": 3.3268, + "step": 1298 + }, + { + "epoch": 1.91, + "grad_norm": 6.050600528717041, + "learning_rate": 0.0001828559336041524, + "loss": 3.3629, + "step": 1299 + }, + { + "epoch": 1.92, + "grad_norm": 6.324507236480713, + "learning_rate": 0.00018282966319382057, + "loss": 3.3402, + "step": 1300 + }, + { + "epoch": 1.92, + "grad_norm": 1.820309042930603, + "learning_rate": 0.00018280337456151797, + "loss": 3.2955, + "step": 1301 + }, + { + "epoch": 1.92, + "grad_norm": 7.433422565460205, + "learning_rate": 0.00018277706771302788, + "loss": 3.3309, + "step": 1302 + }, + { + "epoch": 1.92, + "grad_norm": 9.409677505493164, + "learning_rate": 0.00018275074265413762, + "loss": 3.4307, + "step": 1303 + }, + { + "epoch": 1.92, + "grad_norm": 5.268000602722168, + "learning_rate": 0.00018272439939063856, + "loss": 3.4344, + "step": 1304 + }, + { + "epoch": 1.92, + "grad_norm": 5.51171875, + "learning_rate": 0.00018269803792832603, + "loss": 3.3385, + "step": 1305 + }, + { + "epoch": 1.92, + "grad_norm": 5.903563976287842, + "learning_rate": 0.00018267165827299934, + "loss": 3.3243, + "step": 1306 + }, + { + "epoch": 1.93, + "grad_norm": 3.7264065742492676, + "learning_rate": 0.00018264526043046192, + "loss": 3.3759, + "step": 1307 + }, + { + "epoch": 1.93, + "grad_norm": 5.053271293640137, + "learning_rate": 0.00018261884440652106, + "loss": 3.3202, + "step": 1308 + }, + { + "epoch": 1.93, + "grad_norm": 2.33587646484375, + "learning_rate": 0.0001825924102069881, + "loss": 3.2899, + "step": 1309 + }, + { + "epoch": 1.93, + "grad_norm": 4.696453094482422, + "learning_rate": 0.00018256595783767843, + "loss": 3.359, + "step": 1310 + }, + { + "epoch": 1.93, + "grad_norm": 4.715218544006348, + "learning_rate": 0.00018253948730441132, + "loss": 3.3466, + "step": 1311 + }, + { + "epoch": 1.93, + "grad_norm": 2.377490997314453, + "learning_rate": 0.0001825129986130102, + "loss": 3.3496, + "step": 1312 + }, + { + "epoch": 1.94, + "grad_norm": 2.269477367401123, + "learning_rate": 0.00018248649176930235, + "loss": 3.3404, + "step": 1313 + }, + { + "epoch": 1.94, + "grad_norm": 4.577181816101074, + "learning_rate": 0.0001824599667791191, + "loss": 3.3864, + "step": 1314 + }, + { + "epoch": 1.94, + "grad_norm": 4.408148288726807, + "learning_rate": 0.0001824334236482958, + "loss": 3.3405, + "step": 1315 + }, + { + "epoch": 1.94, + "grad_norm": 2.9572770595550537, + "learning_rate": 0.00018240686238267173, + "loss": 3.323, + "step": 1316 + }, + { + "epoch": 1.94, + "grad_norm": 4.4048380851745605, + "learning_rate": 0.0001823802829880902, + "loss": 3.2676, + "step": 1317 + }, + { + "epoch": 1.94, + "grad_norm": 1.4889674186706543, + "learning_rate": 0.0001823536854703985, + "loss": 3.2902, + "step": 1318 + }, + { + "epoch": 1.94, + "grad_norm": 1.6742371320724487, + "learning_rate": 0.00018232706983544792, + "loss": 3.2945, + "step": 1319 + }, + { + "epoch": 1.95, + "grad_norm": 3.9357426166534424, + "learning_rate": 0.0001823004360890937, + "loss": 3.3486, + "step": 1320 + }, + { + "epoch": 1.95, + "grad_norm": 2.596331834793091, + "learning_rate": 0.0001822737842371951, + "loss": 3.3539, + "step": 1321 + }, + { + "epoch": 1.95, + "grad_norm": 4.866861820220947, + "learning_rate": 0.00018224711428561533, + "loss": 3.3449, + "step": 1322 + }, + { + "epoch": 1.95, + "grad_norm": 5.0658955574035645, + "learning_rate": 0.0001822204262402216, + "loss": 3.3107, + "step": 1323 + }, + { + "epoch": 1.95, + "grad_norm": 1.9593414068222046, + "learning_rate": 0.00018219372010688515, + "loss": 3.348, + "step": 1324 + }, + { + "epoch": 1.95, + "grad_norm": 2.256303310394287, + "learning_rate": 0.0001821669958914811, + "loss": 3.3012, + "step": 1325 + }, + { + "epoch": 1.95, + "grad_norm": 3.2346420288085938, + "learning_rate": 0.0001821402535998886, + "loss": 3.3372, + "step": 1326 + }, + { + "epoch": 1.96, + "grad_norm": 1.7795019149780273, + "learning_rate": 0.00018211349323799078, + "loss": 3.3459, + "step": 1327 + }, + { + "epoch": 1.96, + "grad_norm": 7.262651443481445, + "learning_rate": 0.00018208671481167478, + "loss": 3.3454, + "step": 1328 + }, + { + "epoch": 1.96, + "grad_norm": 8.501165390014648, + "learning_rate": 0.00018205991832683164, + "loss": 3.3952, + "step": 1329 + }, + { + "epoch": 1.96, + "grad_norm": 2.0047447681427, + "learning_rate": 0.00018203310378935638, + "loss": 3.3304, + "step": 1330 + }, + { + "epoch": 1.96, + "grad_norm": 6.586772918701172, + "learning_rate": 0.00018200627120514807, + "loss": 3.3235, + "step": 1331 + }, + { + "epoch": 1.96, + "grad_norm": 6.3678083419799805, + "learning_rate": 0.00018197942058010967, + "loss": 3.4105, + "step": 1332 + }, + { + "epoch": 1.96, + "grad_norm": 3.969449520111084, + "learning_rate": 0.00018195255192014816, + "loss": 3.3689, + "step": 1333 + }, + { + "epoch": 1.97, + "grad_norm": 5.26508092880249, + "learning_rate": 0.0001819256652311744, + "loss": 3.3417, + "step": 1334 + }, + { + "epoch": 1.97, + "grad_norm": 5.3484673500061035, + "learning_rate": 0.00018189876051910337, + "loss": 3.32, + "step": 1335 + }, + { + "epoch": 1.97, + "grad_norm": 3.664024829864502, + "learning_rate": 0.00018187183778985388, + "loss": 3.4004, + "step": 1336 + }, + { + "epoch": 1.97, + "grad_norm": 5.414148807525635, + "learning_rate": 0.00018184489704934877, + "loss": 3.2964, + "step": 1337 + }, + { + "epoch": 1.97, + "grad_norm": 1.5016615390777588, + "learning_rate": 0.0001818179383035148, + "loss": 3.3444, + "step": 1338 + }, + { + "epoch": 1.97, + "grad_norm": 7.925120830535889, + "learning_rate": 0.00018179096155828274, + "loss": 3.3881, + "step": 1339 + }, + { + "epoch": 1.98, + "grad_norm": 9.725903511047363, + "learning_rate": 0.00018176396681958724, + "loss": 3.3698, + "step": 1340 + }, + { + "epoch": 1.98, + "grad_norm": 5.562051773071289, + "learning_rate": 0.00018173695409336702, + "loss": 3.3285, + "step": 1341 + }, + { + "epoch": 1.98, + "grad_norm": 4.413424491882324, + "learning_rate": 0.00018170992338556468, + "loss": 3.3209, + "step": 1342 + }, + { + "epoch": 1.98, + "grad_norm": 5.979745864868164, + "learning_rate": 0.00018168287470212677, + "loss": 3.3173, + "step": 1343 + }, + { + "epoch": 1.98, + "grad_norm": 2.968111753463745, + "learning_rate": 0.00018165580804900384, + "loss": 3.3304, + "step": 1344 + }, + { + "epoch": 1.98, + "grad_norm": 6.121416091918945, + "learning_rate": 0.00018162872343215037, + "loss": 3.325, + "step": 1345 + }, + { + "epoch": 1.98, + "grad_norm": 5.006643772125244, + "learning_rate": 0.0001816016208575248, + "loss": 3.3311, + "step": 1346 + }, + { + "epoch": 1.99, + "grad_norm": 2.833677291870117, + "learning_rate": 0.00018157450033108954, + "loss": 3.3484, + "step": 1347 + }, + { + "epoch": 1.99, + "grad_norm": 3.661172389984131, + "learning_rate": 0.00018154736185881088, + "loss": 3.3119, + "step": 1348 + }, + { + "epoch": 1.99, + "grad_norm": 1.4361642599105835, + "learning_rate": 0.0001815202054466591, + "loss": 3.3744, + "step": 1349 + }, + { + "epoch": 1.99, + "grad_norm": 2.236370801925659, + "learning_rate": 0.00018149303110060846, + "loss": 3.285, + "step": 1350 + }, + { + "epoch": 1.99, + "grad_norm": 3.9948387145996094, + "learning_rate": 0.00018146583882663712, + "loss": 3.3107, + "step": 1351 + }, + { + "epoch": 1.99, + "grad_norm": 2.2754180431365967, + "learning_rate": 0.0001814386286307272, + "loss": 3.3646, + "step": 1352 + }, + { + "epoch": 1.99, + "eval_loss": 3.131267786026001, + "eval_runtime": 1163.1, + "eval_samples_per_second": 14.738, + "eval_steps_per_second": 0.922, + "step": 1352 + }, + { + "epoch": 1.99, + "grad_norm": 4.385812282562256, + "learning_rate": 0.00018141140051886474, + "loss": 3.3364, + "step": 1353 + }, + { + "epoch": 2.0, + "grad_norm": 2.7103238105773926, + "learning_rate": 0.0001813841544970398, + "loss": 3.3096, + "step": 1354 + }, + { + "epoch": 2.0, + "grad_norm": 4.108714580535889, + "learning_rate": 0.00018135689057124625, + "loss": 3.3506, + "step": 1355 + }, + { + "epoch": 2.0, + "grad_norm": 3.6100189685821533, + "learning_rate": 0.000181329608747482, + "loss": 3.2999, + "step": 1356 + }, + { + "epoch": 2.0, + "grad_norm": 3.728175640106201, + "learning_rate": 0.0001813023090317489, + "loss": 3.2945, + "step": 1357 + }, + { + "epoch": 2.0, + "grad_norm": 3.231055974960327, + "learning_rate": 0.00018127499143005268, + "loss": 3.3229, + "step": 1358 + }, + { + "epoch": 2.0, + "grad_norm": 3.6905510425567627, + "learning_rate": 0.00018124765594840298, + "loss": 3.3186, + "step": 1359 + }, + { + "epoch": 2.0, + "grad_norm": 3.330751657485962, + "learning_rate": 0.0001812203025928135, + "loss": 3.3526, + "step": 1360 + }, + { + "epoch": 2.01, + "grad_norm": 3.546685218811035, + "learning_rate": 0.00018119293136930175, + "loss": 3.2914, + "step": 1361 + }, + { + "epoch": 2.01, + "grad_norm": 3.202545404434204, + "learning_rate": 0.00018116554228388924, + "loss": 3.3155, + "step": 1362 + }, + { + "epoch": 2.01, + "grad_norm": 3.762091875076294, + "learning_rate": 0.00018113813534260137, + "loss": 3.2653, + "step": 1363 + }, + { + "epoch": 2.01, + "grad_norm": 3.2883477210998535, + "learning_rate": 0.00018111071055146752, + "loss": 3.2802, + "step": 1364 + }, + { + "epoch": 2.01, + "grad_norm": 4.180018424987793, + "learning_rate": 0.0001810832679165209, + "loss": 3.2743, + "step": 1365 + }, + { + "epoch": 2.01, + "grad_norm": 3.009850025177002, + "learning_rate": 0.00018105580744379873, + "loss": 3.3396, + "step": 1366 + }, + { + "epoch": 2.02, + "grad_norm": 3.3239614963531494, + "learning_rate": 0.00018102832913934215, + "loss": 3.302, + "step": 1367 + }, + { + "epoch": 2.0, + "grad_norm": 2.7158830165863037, + "learning_rate": 0.00018100083300919616, + "loss": 3.2927, + "step": 1368 + }, + { + "epoch": 2.0, + "grad_norm": 4.113391399383545, + "learning_rate": 0.00018097331905940977, + "loss": 3.258, + "step": 1369 + }, + { + "epoch": 2.0, + "grad_norm": 4.452334880828857, + "learning_rate": 0.00018094578729603583, + "loss": 3.2988, + "step": 1370 + }, + { + "epoch": 2.0, + "grad_norm": 2.8618204593658447, + "learning_rate": 0.00018091823772513118, + "loss": 3.2938, + "step": 1371 + }, + { + "epoch": 2.01, + "grad_norm": 2.671171188354492, + "learning_rate": 0.00018089067035275652, + "loss": 3.2756, + "step": 1372 + }, + { + "epoch": 2.01, + "grad_norm": 4.292674541473389, + "learning_rate": 0.0001808630851849765, + "loss": 3.2655, + "step": 1373 + }, + { + "epoch": 2.01, + "grad_norm": 4.163233757019043, + "learning_rate": 0.00018083548222785965, + "loss": 3.2825, + "step": 1374 + }, + { + "epoch": 2.01, + "grad_norm": 2.8590521812438965, + "learning_rate": 0.00018080786148747842, + "loss": 3.3127, + "step": 1375 + }, + { + "epoch": 2.01, + "grad_norm": 2.4040653705596924, + "learning_rate": 0.00018078022296990923, + "loss": 3.348, + "step": 1376 + }, + { + "epoch": 2.01, + "grad_norm": 5.037558555603027, + "learning_rate": 0.00018075256668123238, + "loss": 3.275, + "step": 1377 + }, + { + "epoch": 2.02, + "grad_norm": 4.536971569061279, + "learning_rate": 0.000180724892627532, + "loss": 3.2533, + "step": 1378 + }, + { + "epoch": 2.02, + "grad_norm": 2.9320995807647705, + "learning_rate": 0.00018069720081489624, + "loss": 3.3223, + "step": 1379 + }, + { + "epoch": 2.02, + "grad_norm": 2.6839616298675537, + "learning_rate": 0.00018066949124941712, + "loss": 3.3313, + "step": 1380 + }, + { + "epoch": 2.02, + "grad_norm": 3.9813764095306396, + "learning_rate": 0.00018064176393719055, + "loss": 3.2821, + "step": 1381 + }, + { + "epoch": 2.02, + "grad_norm": 3.4426040649414062, + "learning_rate": 0.00018061401888431634, + "loss": 3.3005, + "step": 1382 + }, + { + "epoch": 2.02, + "grad_norm": 3.2250571250915527, + "learning_rate": 0.00018058625609689824, + "loss": 3.2879, + "step": 1383 + }, + { + "epoch": 2.02, + "grad_norm": 2.6666173934936523, + "learning_rate": 0.00018055847558104383, + "loss": 3.2522, + "step": 1384 + }, + { + "epoch": 2.03, + "grad_norm": 4.8121843338012695, + "learning_rate": 0.00018053067734286472, + "loss": 3.3785, + "step": 1385 + }, + { + "epoch": 2.03, + "grad_norm": 4.593588352203369, + "learning_rate": 0.00018050286138847628, + "loss": 3.2831, + "step": 1386 + }, + { + "epoch": 2.03, + "grad_norm": 3.0059878826141357, + "learning_rate": 0.0001804750277239978, + "loss": 3.2606, + "step": 1387 + }, + { + "epoch": 2.03, + "grad_norm": 2.9211039543151855, + "learning_rate": 0.00018044717635555259, + "loss": 3.2735, + "step": 1388 + }, + { + "epoch": 2.03, + "grad_norm": 3.4243338108062744, + "learning_rate": 0.00018041930728926768, + "loss": 3.2794, + "step": 1389 + }, + { + "epoch": 2.03, + "grad_norm": 3.398374557495117, + "learning_rate": 0.00018039142053127412, + "loss": 3.3066, + "step": 1390 + }, + { + "epoch": 2.03, + "grad_norm": 3.260343074798584, + "learning_rate": 0.0001803635160877068, + "loss": 3.3, + "step": 1391 + }, + { + "epoch": 2.04, + "grad_norm": 2.542720079421997, + "learning_rate": 0.00018033559396470454, + "loss": 3.2654, + "step": 1392 + }, + { + "epoch": 2.04, + "grad_norm": 4.721008777618408, + "learning_rate": 0.00018030765416840997, + "loss": 3.3286, + "step": 1393 + }, + { + "epoch": 2.04, + "grad_norm": 4.219196319580078, + "learning_rate": 0.00018027969670496966, + "loss": 3.2769, + "step": 1394 + }, + { + "epoch": 2.04, + "grad_norm": 2.8509461879730225, + "learning_rate": 0.00018025172158053413, + "loss": 3.2763, + "step": 1395 + }, + { + "epoch": 2.04, + "grad_norm": 3.1871495246887207, + "learning_rate": 0.00018022372880125765, + "loss": 3.3038, + "step": 1396 + }, + { + "epoch": 2.04, + "grad_norm": 4.717142105102539, + "learning_rate": 0.00018019571837329847, + "loss": 3.2649, + "step": 1397 + }, + { + "epoch": 2.04, + "grad_norm": 3.9490749835968018, + "learning_rate": 0.00018016769030281872, + "loss": 3.2709, + "step": 1398 + }, + { + "epoch": 2.05, + "grad_norm": 3.04848051071167, + "learning_rate": 0.00018013964459598434, + "loss": 3.2638, + "step": 1399 + }, + { + "epoch": 2.05, + "grad_norm": 2.971536636352539, + "learning_rate": 0.00018011158125896522, + "loss": 3.3796, + "step": 1400 + }, + { + "epoch": 2.05, + "grad_norm": 3.643239736557007, + "learning_rate": 0.0001800835002979351, + "loss": 3.2766, + "step": 1401 + }, + { + "epoch": 2.05, + "grad_norm": 3.0714406967163086, + "learning_rate": 0.00018005540171907162, + "loss": 3.2691, + "step": 1402 + }, + { + "epoch": 2.05, + "grad_norm": 3.928359031677246, + "learning_rate": 0.00018002728552855629, + "loss": 3.2842, + "step": 1403 + }, + { + "epoch": 2.05, + "grad_norm": 3.7628214359283447, + "learning_rate": 0.00017999915173257444, + "loss": 3.3371, + "step": 1404 + }, + { + "epoch": 2.06, + "grad_norm": 3.1093904972076416, + "learning_rate": 0.00017997100033731534, + "loss": 3.2866, + "step": 1405 + }, + { + "epoch": 2.06, + "grad_norm": 2.735902786254883, + "learning_rate": 0.0001799428313489721, + "loss": 3.2598, + "step": 1406 + }, + { + "epoch": 2.06, + "grad_norm": 3.890085458755493, + "learning_rate": 0.00017991464477374173, + "loss": 3.2565, + "step": 1407 + }, + { + "epoch": 2.06, + "grad_norm": 3.5979950428009033, + "learning_rate": 0.00017988644061782508, + "loss": 3.3153, + "step": 1408 + }, + { + "epoch": 2.06, + "grad_norm": 3.224681854248047, + "learning_rate": 0.00017985821888742685, + "loss": 3.2969, + "step": 1409 + }, + { + "epoch": 2.06, + "grad_norm": 3.645235776901245, + "learning_rate": 0.00017982997958875565, + "loss": 3.2823, + "step": 1410 + }, + { + "epoch": 2.06, + "grad_norm": 3.3547754287719727, + "learning_rate": 0.000179801722728024, + "loss": 3.3546, + "step": 1411 + }, + { + "epoch": 2.07, + "grad_norm": 2.8428351879119873, + "learning_rate": 0.00017977344831144807, + "loss": 3.3096, + "step": 1412 + }, + { + "epoch": 2.07, + "grad_norm": 4.488086223602295, + "learning_rate": 0.00017974515634524817, + "loss": 3.2783, + "step": 1413 + }, + { + "epoch": 2.07, + "grad_norm": 4.082129955291748, + "learning_rate": 0.0001797168468356483, + "loss": 3.3131, + "step": 1414 + }, + { + "epoch": 2.07, + "grad_norm": 3.037198305130005, + "learning_rate": 0.00017968851978887636, + "loss": 3.3435, + "step": 1415 + }, + { + "epoch": 2.07, + "grad_norm": 2.8374056816101074, + "learning_rate": 0.0001796601752111641, + "loss": 3.3026, + "step": 1416 + }, + { + "epoch": 2.07, + "grad_norm": 4.270615577697754, + "learning_rate": 0.00017963181310874716, + "loss": 3.3384, + "step": 1417 + }, + { + "epoch": 2.07, + "grad_norm": 4.1699323654174805, + "learning_rate": 0.00017960343348786504, + "loss": 3.2869, + "step": 1418 + }, + { + "epoch": 2.08, + "grad_norm": 2.7383670806884766, + "learning_rate": 0.00017957503635476099, + "loss": 3.2882, + "step": 1419 + }, + { + "epoch": 2.08, + "grad_norm": 2.707303762435913, + "learning_rate": 0.00017954662171568226, + "loss": 3.3616, + "step": 1420 + }, + { + "epoch": 2.08, + "grad_norm": 5.056338787078857, + "learning_rate": 0.00017951818957687981, + "loss": 3.381, + "step": 1421 + }, + { + "epoch": 2.08, + "grad_norm": 4.49586820602417, + "learning_rate": 0.0001794897399446086, + "loss": 3.2837, + "step": 1422 + }, + { + "epoch": 2.08, + "grad_norm": 2.5468156337738037, + "learning_rate": 0.0001794612728251273, + "loss": 3.2547, + "step": 1423 + }, + { + "epoch": 2.08, + "grad_norm": 2.2053098678588867, + "learning_rate": 0.00017943278822469852, + "loss": 3.3216, + "step": 1424 + }, + { + "epoch": 2.09, + "grad_norm": 4.875980854034424, + "learning_rate": 0.00017940428614958865, + "loss": 3.3319, + "step": 1425 + }, + { + "epoch": 2.09, + "grad_norm": 4.466098308563232, + "learning_rate": 0.000179375766606068, + "loss": 3.276, + "step": 1426 + }, + { + "epoch": 2.09, + "grad_norm": 2.8682076930999756, + "learning_rate": 0.0001793472296004106, + "loss": 3.2971, + "step": 1427 + }, + { + "epoch": 2.09, + "grad_norm": 2.2296552658081055, + "learning_rate": 0.00017931867513889448, + "loss": 3.2946, + "step": 1428 + }, + { + "epoch": 2.09, + "grad_norm": 4.954530239105225, + "learning_rate": 0.0001792901032278014, + "loss": 3.269, + "step": 1429 + }, + { + "epoch": 2.09, + "grad_norm": 4.960486888885498, + "learning_rate": 0.00017926151387341697, + "loss": 3.2616, + "step": 1430 + }, + { + "epoch": 2.09, + "grad_norm": 2.2952184677124023, + "learning_rate": 0.0001792329070820307, + "loss": 3.3348, + "step": 1431 + }, + { + "epoch": 2.1, + "grad_norm": 1.9920496940612793, + "learning_rate": 0.00017920428285993583, + "loss": 3.3137, + "step": 1432 + }, + { + "epoch": 2.1, + "grad_norm": 4.211342811584473, + "learning_rate": 0.00017917564121342958, + "loss": 3.3246, + "step": 1433 + }, + { + "epoch": 2.1, + "grad_norm": 3.4940345287323, + "learning_rate": 0.00017914698214881284, + "loss": 3.3421, + "step": 1434 + }, + { + "epoch": 2.1, + "grad_norm": 3.7058966159820557, + "learning_rate": 0.00017911830567239048, + "loss": 3.3377, + "step": 1435 + }, + { + "epoch": 2.1, + "grad_norm": 3.2200000286102295, + "learning_rate": 0.00017908961179047105, + "loss": 3.3125, + "step": 1436 + }, + { + "epoch": 2.1, + "grad_norm": 3.529256582260132, + "learning_rate": 0.00017906090050936712, + "loss": 3.2833, + "step": 1437 + }, + { + "epoch": 2.1, + "grad_norm": 3.189732313156128, + "learning_rate": 0.0001790321718353949, + "loss": 3.2662, + "step": 1438 + }, + { + "epoch": 2.11, + "grad_norm": 3.6067655086517334, + "learning_rate": 0.0001790034257748745, + "loss": 3.2795, + "step": 1439 + }, + { + "epoch": 2.11, + "grad_norm": 3.4113893508911133, + "learning_rate": 0.0001789746623341299, + "loss": 3.2666, + "step": 1440 + }, + { + "epoch": 2.11, + "grad_norm": 3.466991662979126, + "learning_rate": 0.00017894588151948885, + "loss": 3.325, + "step": 1441 + }, + { + "epoch": 2.11, + "grad_norm": 3.057206153869629, + "learning_rate": 0.00017891708333728296, + "loss": 3.2636, + "step": 1442 + }, + { + "epoch": 2.11, + "grad_norm": 3.8873162269592285, + "learning_rate": 0.0001788882677938476, + "loss": 3.3139, + "step": 1443 + }, + { + "epoch": 2.11, + "grad_norm": 3.4515161514282227, + "learning_rate": 0.000178859434895522, + "loss": 3.2884, + "step": 1444 + }, + { + "epoch": 2.11, + "grad_norm": 14.146556854248047, + "learning_rate": 0.00017883058464864923, + "loss": 3.2571, + "step": 1445 + }, + { + "epoch": 2.12, + "grad_norm": 9.11373519897461, + "learning_rate": 0.00017880171705957613, + "loss": 3.295, + "step": 1446 + }, + { + "epoch": 2.12, + "grad_norm": 5.7905049324035645, + "learning_rate": 0.0001787728321346534, + "loss": 3.2532, + "step": 1447 + }, + { + "epoch": 2.12, + "grad_norm": 3.7080447673797607, + "learning_rate": 0.0001787439298802355, + "loss": 3.3201, + "step": 1448 + }, + { + "epoch": 2.12, + "grad_norm": 4.587584018707275, + "learning_rate": 0.00017871501030268072, + "loss": 3.2702, + "step": 1449 + }, + { + "epoch": 2.12, + "grad_norm": 0.9951494932174683, + "learning_rate": 0.0001786860734083512, + "loss": 3.2862, + "step": 1450 + }, + { + "epoch": 2.12, + "grad_norm": 0.8612810373306274, + "learning_rate": 0.00017865711920361292, + "loss": 3.2968, + "step": 1451 + }, + { + "epoch": 2.13, + "grad_norm": 2.446154832839966, + "learning_rate": 0.00017862814769483549, + "loss": 3.2937, + "step": 1452 + }, + { + "epoch": 2.13, + "grad_norm": 1.2420624494552612, + "learning_rate": 0.00017859915888839253, + "loss": 3.2922, + "step": 1453 + }, + { + "epoch": 2.13, + "grad_norm": 1.9885892868041992, + "learning_rate": 0.00017857015279066135, + "loss": 3.3043, + "step": 1454 + }, + { + "epoch": 2.13, + "grad_norm": 2.127197504043579, + "learning_rate": 0.00017854112940802314, + "loss": 3.3503, + "step": 1455 + }, + { + "epoch": 2.13, + "grad_norm": 1.1510310173034668, + "learning_rate": 0.0001785120887468628, + "loss": 3.2314, + "step": 1456 + }, + { + "epoch": 2.13, + "grad_norm": 2.6665754318237305, + "learning_rate": 0.00017848303081356912, + "loss": 3.2788, + "step": 1457 + }, + { + "epoch": 2.13, + "grad_norm": 1.0793408155441284, + "learning_rate": 0.0001784539556145346, + "loss": 3.2623, + "step": 1458 + }, + { + "epoch": 2.14, + "grad_norm": 2.212766647338867, + "learning_rate": 0.00017842486315615563, + "loss": 3.3542, + "step": 1459 + }, + { + "epoch": 2.14, + "grad_norm": 1.8812615871429443, + "learning_rate": 0.00017839575344483238, + "loss": 3.2481, + "step": 1460 + }, + { + "epoch": 2.14, + "grad_norm": 1.537009358406067, + "learning_rate": 0.0001783666264869687, + "loss": 3.2492, + "step": 1461 + }, + { + "epoch": 2.14, + "grad_norm": 1.0844135284423828, + "learning_rate": 0.00017833748228897241, + "loss": 3.3093, + "step": 1462 + }, + { + "epoch": 2.14, + "grad_norm": 1.4660224914550781, + "learning_rate": 0.000178308320857255, + "loss": 3.2598, + "step": 1463 + }, + { + "epoch": 2.14, + "grad_norm": 0.9054508209228516, + "learning_rate": 0.00017827914219823183, + "loss": 3.3132, + "step": 1464 + }, + { + "epoch": 2.14, + "grad_norm": 5.584422588348389, + "learning_rate": 0.00017824994631832197, + "loss": 3.2961, + "step": 1465 + }, + { + "epoch": 2.15, + "grad_norm": 1.9765104055404663, + "learning_rate": 0.0001782207332239483, + "loss": 3.2963, + "step": 1466 + }, + { + "epoch": 2.15, + "grad_norm": 1.4504390954971313, + "learning_rate": 0.00017819150292153757, + "loss": 3.3335, + "step": 1467 + }, + { + "epoch": 2.15, + "grad_norm": 1.2761657238006592, + "learning_rate": 0.0001781622554175202, + "loss": 3.2779, + "step": 1468 + }, + { + "epoch": 2.15, + "grad_norm": 2.1837310791015625, + "learning_rate": 0.00017813299071833046, + "loss": 3.2791, + "step": 1469 + }, + { + "epoch": 2.15, + "grad_norm": 2.7105600833892822, + "learning_rate": 0.00017810370883040636, + "loss": 3.2983, + "step": 1470 + }, + { + "epoch": 2.15, + "grad_norm": 1.1672013998031616, + "learning_rate": 0.00017807440976018977, + "loss": 3.2989, + "step": 1471 + }, + { + "epoch": 2.15, + "grad_norm": 4.486381530761719, + "learning_rate": 0.00017804509351412629, + "loss": 3.3132, + "step": 1472 + }, + { + "epoch": 2.16, + "grad_norm": 3.2585389614105225, + "learning_rate": 0.0001780157600986652, + "loss": 3.3249, + "step": 1473 + }, + { + "epoch": 2.16, + "grad_norm": 3.1671297550201416, + "learning_rate": 0.00017798640952025978, + "loss": 3.31, + "step": 1474 + }, + { + "epoch": 2.16, + "grad_norm": 2.345313310623169, + "learning_rate": 0.0001779570417853669, + "loss": 3.2382, + "step": 1475 + }, + { + "epoch": 2.16, + "grad_norm": 5.773190498352051, + "learning_rate": 0.0001779276569004473, + "loss": 3.2903, + "step": 1476 + }, + { + "epoch": 2.16, + "grad_norm": 4.879696846008301, + "learning_rate": 0.00017789825487196537, + "loss": 3.3107, + "step": 1477 + }, + { + "epoch": 2.16, + "grad_norm": 4.470544815063477, + "learning_rate": 0.00017786883570638942, + "loss": 3.3, + "step": 1478 + }, + { + "epoch": 2.17, + "grad_norm": 5.051728248596191, + "learning_rate": 0.00017783939941019153, + "loss": 3.2964, + "step": 1479 + }, + { + "epoch": 2.17, + "grad_norm": 1.2672648429870605, + "learning_rate": 0.00017780994598984737, + "loss": 3.297, + "step": 1480 + }, + { + "epoch": 2.17, + "grad_norm": 5.323600769042969, + "learning_rate": 0.0001777804754518366, + "loss": 3.2878, + "step": 1481 + }, + { + "epoch": 2.17, + "grad_norm": 4.052299976348877, + "learning_rate": 0.00017775098780264243, + "loss": 3.3135, + "step": 1482 + }, + { + "epoch": 2.17, + "grad_norm": 3.538057565689087, + "learning_rate": 0.00017772148304875205, + "loss": 3.3209, + "step": 1483 + }, + { + "epoch": 2.17, + "grad_norm": 2.5743610858917236, + "learning_rate": 0.0001776919611966563, + "loss": 3.2468, + "step": 1484 + }, + { + "epoch": 2.17, + "grad_norm": 6.518936634063721, + "learning_rate": 0.00017766242225284966, + "loss": 3.2914, + "step": 1485 + }, + { + "epoch": 2.18, + "grad_norm": 5.448997497558594, + "learning_rate": 0.00017763286622383067, + "loss": 3.3127, + "step": 1486 + }, + { + "epoch": 2.18, + "grad_norm": 2.5418717861175537, + "learning_rate": 0.00017760329311610135, + "loss": 3.2913, + "step": 1487 + }, + { + "epoch": 2.18, + "grad_norm": 4.122663974761963, + "learning_rate": 0.00017757370293616762, + "loss": 3.3257, + "step": 1488 + }, + { + "epoch": 2.18, + "grad_norm": 3.3550655841827393, + "learning_rate": 0.00017754409569053917, + "loss": 3.271, + "step": 1489 + }, + { + "epoch": 2.18, + "grad_norm": 2.8279335498809814, + "learning_rate": 0.0001775144713857293, + "loss": 3.2989, + "step": 1490 + }, + { + "epoch": 2.18, + "grad_norm": 2.4097659587860107, + "learning_rate": 0.0001774848300282552, + "loss": 3.2557, + "step": 1491 + }, + { + "epoch": 2.18, + "grad_norm": 2.702836275100708, + "learning_rate": 0.00017745517162463776, + "loss": 3.2719, + "step": 1492 + }, + { + "epoch": 2.19, + "grad_norm": 1.937203288078308, + "learning_rate": 0.00017742549618140167, + "loss": 3.2801, + "step": 1493 + }, + { + "epoch": 2.19, + "grad_norm": 2.2853002548217773, + "learning_rate": 0.00017739580370507532, + "loss": 3.4001, + "step": 1494 + }, + { + "epoch": 2.19, + "grad_norm": 3.39026141166687, + "learning_rate": 0.00017736609420219082, + "loss": 3.2562, + "step": 1495 + }, + { + "epoch": 2.19, + "grad_norm": 2.192948341369629, + "learning_rate": 0.00017733636767928407, + "loss": 3.2548, + "step": 1496 + }, + { + "epoch": 2.19, + "grad_norm": 3.966799259185791, + "learning_rate": 0.00017730662414289472, + "loss": 3.2744, + "step": 1497 + }, + { + "epoch": 2.19, + "grad_norm": 2.6247634887695312, + "learning_rate": 0.00017727686359956615, + "loss": 3.287, + "step": 1498 + }, + { + "epoch": 2.19, + "grad_norm": 2.959963083267212, + "learning_rate": 0.00017724708605584545, + "loss": 3.2864, + "step": 1499 + }, + { + "epoch": 2.2, + "grad_norm": 1.7757418155670166, + "learning_rate": 0.00017721729151828355, + "loss": 3.3003, + "step": 1500 + }, + { + "epoch": 2.2, + "grad_norm": 3.2801480293273926, + "learning_rate": 0.00017718747999343498, + "loss": 3.2662, + "step": 1501 + }, + { + "epoch": 2.2, + "grad_norm": 2.6395089626312256, + "learning_rate": 0.0001771576514878581, + "loss": 3.2991, + "step": 1502 + }, + { + "epoch": 2.2, + "grad_norm": 2.687575340270996, + "learning_rate": 0.00017712780600811497, + "loss": 3.2676, + "step": 1503 + }, + { + "epoch": 2.2, + "grad_norm": 2.702707052230835, + "learning_rate": 0.00017709794356077143, + "loss": 3.2862, + "step": 1504 + }, + { + "epoch": 2.2, + "grad_norm": 2.10098934173584, + "learning_rate": 0.000177068064152397, + "loss": 3.2425, + "step": 1505 + }, + { + "epoch": 2.21, + "grad_norm": 1.3301888704299927, + "learning_rate": 0.0001770381677895649, + "loss": 3.2325, + "step": 1506 + }, + { + "epoch": 2.21, + "grad_norm": 3.560911178588867, + "learning_rate": 0.0001770082544788522, + "loss": 3.2914, + "step": 1507 + }, + { + "epoch": 2.21, + "grad_norm": 2.01891827583313, + "learning_rate": 0.0001769783242268396, + "loss": 3.2213, + "step": 1508 + }, + { + "epoch": 2.21, + "grad_norm": 3.6540088653564453, + "learning_rate": 0.00017694837704011155, + "loss": 3.3066, + "step": 1509 + }, + { + "epoch": 2.21, + "grad_norm": 3.1314940452575684, + "learning_rate": 0.00017691841292525625, + "loss": 3.2576, + "step": 1510 + }, + { + "epoch": 2.21, + "grad_norm": 3.816842555999756, + "learning_rate": 0.0001768884318888656, + "loss": 3.2784, + "step": 1511 + }, + { + "epoch": 2.21, + "grad_norm": 3.5961854457855225, + "learning_rate": 0.0001768584339375352, + "loss": 3.2562, + "step": 1512 + }, + { + "epoch": 2.22, + "grad_norm": 3.9628171920776367, + "learning_rate": 0.0001768284190778644, + "loss": 3.2561, + "step": 1513 + }, + { + "epoch": 2.22, + "grad_norm": 4.071855068206787, + "learning_rate": 0.00017679838731645632, + "loss": 3.2924, + "step": 1514 + }, + { + "epoch": 2.22, + "grad_norm": 2.176236152648926, + "learning_rate": 0.00017676833865991765, + "loss": 3.309, + "step": 1515 + }, + { + "epoch": 2.22, + "grad_norm": 1.6997188329696655, + "learning_rate": 0.000176738273114859, + "loss": 3.2835, + "step": 1516 + }, + { + "epoch": 2.22, + "grad_norm": 3.275249719619751, + "learning_rate": 0.00017670819068789454, + "loss": 3.2506, + "step": 1517 + }, + { + "epoch": 2.22, + "grad_norm": 1.3623771667480469, + "learning_rate": 0.0001766780913856422, + "loss": 3.2054, + "step": 1518 + }, + { + "epoch": 2.22, + "grad_norm": 5.944447994232178, + "learning_rate": 0.00017664797521472364, + "loss": 3.2794, + "step": 1519 + }, + { + "epoch": 2.23, + "grad_norm": 5.1751484870910645, + "learning_rate": 0.0001766178421817642, + "loss": 3.2498, + "step": 1520 + }, + { + "epoch": 2.23, + "grad_norm": 2.410102128982544, + "learning_rate": 0.00017658769229339294, + "loss": 3.2672, + "step": 1521 + }, + { + "epoch": 2.23, + "eval_loss": 3.0885374546051025, + "eval_runtime": 1164.3656, + "eval_samples_per_second": 14.722, + "eval_steps_per_second": 0.921, + "step": 1521 + }, + { + "epoch": 2.23, + "grad_norm": 2.7669081687927246, + "learning_rate": 0.00017655752555624263, + "loss": 3.2525, + "step": 1522 + }, + { + "epoch": 2.23, + "grad_norm": 4.010193347930908, + "learning_rate": 0.00017652734197694977, + "loss": 3.2827, + "step": 1523 + }, + { + "epoch": 2.23, + "grad_norm": 3.233734130859375, + "learning_rate": 0.00017649714156215456, + "loss": 3.2372, + "step": 1524 + }, + { + "epoch": 2.23, + "grad_norm": 3.2297024726867676, + "learning_rate": 0.00017646692431850085, + "loss": 3.2763, + "step": 1525 + }, + { + "epoch": 2.24, + "grad_norm": 2.7911462783813477, + "learning_rate": 0.00017643669025263625, + "loss": 3.2702, + "step": 1526 + }, + { + "epoch": 2.24, + "grad_norm": 2.8759915828704834, + "learning_rate": 0.00017640643937121207, + "loss": 3.2338, + "step": 1527 + }, + { + "epoch": 2.24, + "grad_norm": 2.3804805278778076, + "learning_rate": 0.00017637617168088325, + "loss": 3.2646, + "step": 1528 + }, + { + "epoch": 2.24, + "grad_norm": 4.322963237762451, + "learning_rate": 0.00017634588718830853, + "loss": 3.2743, + "step": 1529 + }, + { + "epoch": 2.24, + "grad_norm": 3.6001713275909424, + "learning_rate": 0.00017631558590015028, + "loss": 3.3172, + "step": 1530 + }, + { + "epoch": 2.24, + "grad_norm": 3.2861557006835938, + "learning_rate": 0.00017628526782307455, + "loss": 3.2723, + "step": 1531 + }, + { + "epoch": 2.24, + "grad_norm": 3.6503138542175293, + "learning_rate": 0.0001762549329637512, + "loss": 3.234, + "step": 1532 + }, + { + "epoch": 2.25, + "grad_norm": 3.1032047271728516, + "learning_rate": 0.00017622458132885357, + "loss": 3.2845, + "step": 1533 + }, + { + "epoch": 2.25, + "grad_norm": 2.5854177474975586, + "learning_rate": 0.00017619421292505892, + "loss": 3.2251, + "step": 1534 + }, + { + "epoch": 2.25, + "grad_norm": 4.162723064422607, + "learning_rate": 0.00017616382775904806, + "loss": 3.2519, + "step": 1535 + }, + { + "epoch": 2.25, + "grad_norm": 3.7609925270080566, + "learning_rate": 0.0001761334258375055, + "loss": 3.334, + "step": 1536 + }, + { + "epoch": 2.25, + "grad_norm": 3.426297664642334, + "learning_rate": 0.00017610300716711948, + "loss": 3.2644, + "step": 1537 + }, + { + "epoch": 2.25, + "grad_norm": 3.2877280712127686, + "learning_rate": 0.0001760725717545819, + "loss": 3.2466, + "step": 1538 + }, + { + "epoch": 2.25, + "grad_norm": 3.2899694442749023, + "learning_rate": 0.00017604211960658837, + "loss": 3.2245, + "step": 1539 + }, + { + "epoch": 2.26, + "grad_norm": 3.202381134033203, + "learning_rate": 0.0001760116507298381, + "loss": 3.2512, + "step": 1540 + }, + { + "epoch": 2.26, + "grad_norm": 9.299320220947266, + "learning_rate": 0.0001759811651310341, + "loss": 3.2295, + "step": 1541 + }, + { + "epoch": 2.26, + "grad_norm": 6.585042476654053, + "learning_rate": 0.000175950662816883, + "loss": 3.2845, + "step": 1542 + }, + { + "epoch": 2.26, + "grad_norm": 3.786369562149048, + "learning_rate": 0.00017592014379409502, + "loss": 3.2796, + "step": 1543 + }, + { + "epoch": 2.26, + "grad_norm": 5.25980281829834, + "learning_rate": 0.00017588960806938424, + "loss": 3.2427, + "step": 1544 + }, + { + "epoch": 2.26, + "grad_norm": 7.450433254241943, + "learning_rate": 0.00017585905564946827, + "loss": 3.3147, + "step": 1545 + }, + { + "epoch": 2.26, + "grad_norm": 3.1463863849639893, + "learning_rate": 0.00017582848654106845, + "loss": 3.2311, + "step": 1546 + }, + { + "epoch": 2.27, + "grad_norm": 5.310190200805664, + "learning_rate": 0.00017579790075090976, + "loss": 3.3092, + "step": 1547 + }, + { + "epoch": 2.27, + "grad_norm": 5.930779933929443, + "learning_rate": 0.0001757672982857209, + "loss": 3.2788, + "step": 1548 + }, + { + "epoch": 2.27, + "grad_norm": 2.841097831726074, + "learning_rate": 0.00017573667915223414, + "loss": 3.1998, + "step": 1549 + }, + { + "epoch": 2.27, + "grad_norm": 7.447371959686279, + "learning_rate": 0.0001757060433571856, + "loss": 3.2633, + "step": 1550 + }, + { + "epoch": 2.27, + "grad_norm": 7.628240585327148, + "learning_rate": 0.00017567539090731482, + "loss": 3.2626, + "step": 1551 + }, + { + "epoch": 2.27, + "grad_norm": 1.3023099899291992, + "learning_rate": 0.0001756447218093652, + "loss": 3.2525, + "step": 1552 + }, + { + "epoch": 2.28, + "grad_norm": 8.924667358398438, + "learning_rate": 0.00017561403607008376, + "loss": 3.2892, + "step": 1553 + }, + { + "epoch": 2.28, + "grad_norm": 11.807168006896973, + "learning_rate": 0.00017558333369622112, + "loss": 3.3097, + "step": 1554 + }, + { + "epoch": 2.28, + "grad_norm": 9.348532676696777, + "learning_rate": 0.0001755526146945316, + "loss": 3.3024, + "step": 1555 + }, + { + "epoch": 2.28, + "grad_norm": 1.3859421014785767, + "learning_rate": 0.00017552187907177315, + "loss": 3.2629, + "step": 1556 + }, + { + "epoch": 2.28, + "grad_norm": 12.93152141571045, + "learning_rate": 0.00017549112683470748, + "loss": 3.3065, + "step": 1557 + }, + { + "epoch": 2.28, + "grad_norm": 18.00539779663086, + "learning_rate": 0.00017546035799009978, + "loss": 3.3714, + "step": 1558 + }, + { + "epoch": 2.28, + "grad_norm": 13.751264572143555, + "learning_rate": 0.00017542957254471908, + "loss": 3.3417, + "step": 1559 + }, + { + "epoch": 2.29, + "grad_norm": 4.594606876373291, + "learning_rate": 0.0001753987705053379, + "loss": 3.343, + "step": 1560 + }, + { + "epoch": 2.29, + "grad_norm": 7.840448379516602, + "learning_rate": 0.0001753679518787325, + "loss": 3.3199, + "step": 1561 + }, + { + "epoch": 2.29, + "grad_norm": 11.789816856384277, + "learning_rate": 0.0001753371166716828, + "loss": 3.3301, + "step": 1562 + }, + { + "epoch": 2.29, + "grad_norm": 11.082632064819336, + "learning_rate": 0.0001753062648909723, + "loss": 3.3078, + "step": 1563 + }, + { + "epoch": 2.29, + "grad_norm": 5.516722679138184, + "learning_rate": 0.0001752753965433882, + "loss": 3.2917, + "step": 1564 + }, + { + "epoch": 2.29, + "grad_norm": 8.504383087158203, + "learning_rate": 0.00017524451163572136, + "loss": 3.3195, + "step": 1565 + }, + { + "epoch": 2.29, + "grad_norm": 14.364538192749023, + "learning_rate": 0.0001752136101747662, + "loss": 3.3471, + "step": 1566 + }, + { + "epoch": 2.3, + "grad_norm": 12.66904067993164, + "learning_rate": 0.00017518269216732087, + "loss": 3.3574, + "step": 1567 + }, + { + "epoch": 2.3, + "grad_norm": 4.23660945892334, + "learning_rate": 0.00017515175762018714, + "loss": 3.3283, + "step": 1568 + }, + { + "epoch": 2.3, + "grad_norm": 6.301276206970215, + "learning_rate": 0.00017512080654017036, + "loss": 3.3064, + "step": 1569 + }, + { + "epoch": 2.3, + "grad_norm": 8.942106246948242, + "learning_rate": 0.00017508983893407958, + "loss": 3.2761, + "step": 1570 + }, + { + "epoch": 2.3, + "grad_norm": 6.445720672607422, + "learning_rate": 0.00017505885480872745, + "loss": 3.3033, + "step": 1571 + }, + { + "epoch": 2.3, + "grad_norm": 3.0577292442321777, + "learning_rate": 0.0001750278541709303, + "loss": 3.2916, + "step": 1572 + }, + { + "epoch": 2.3, + "grad_norm": 5.473748683929443, + "learning_rate": 0.00017499683702750803, + "loss": 3.2609, + "step": 1573 + }, + { + "epoch": 2.31, + "grad_norm": 2.3021624088287354, + "learning_rate": 0.00017496580338528426, + "loss": 3.2478, + "step": 1574 + }, + { + "epoch": 2.31, + "grad_norm": 6.796987056732178, + "learning_rate": 0.00017493475325108611, + "loss": 3.3177, + "step": 1575 + }, + { + "epoch": 2.31, + "grad_norm": 8.425848007202148, + "learning_rate": 0.00017490368663174443, + "loss": 3.2999, + "step": 1576 + }, + { + "epoch": 2.31, + "grad_norm": 4.728596210479736, + "learning_rate": 0.0001748726035340937, + "loss": 3.2693, + "step": 1577 + }, + { + "epoch": 2.31, + "grad_norm": 6.588244915008545, + "learning_rate": 0.00017484150396497193, + "loss": 3.2474, + "step": 1578 + }, + { + "epoch": 2.31, + "grad_norm": 10.9650239944458, + "learning_rate": 0.00017481038793122088, + "loss": 3.3038, + "step": 1579 + }, + { + "epoch": 2.32, + "grad_norm": 5.750149726867676, + "learning_rate": 0.00017477925543968586, + "loss": 3.315, + "step": 1580 + }, + { + "epoch": 2.32, + "grad_norm": 4.661263465881348, + "learning_rate": 0.00017474810649721576, + "loss": 3.2819, + "step": 1581 + }, + { + "epoch": 2.32, + "grad_norm": 6.578607082366943, + "learning_rate": 0.0001747169411106632, + "loss": 3.268, + "step": 1582 + }, + { + "epoch": 2.32, + "grad_norm": 3.776803970336914, + "learning_rate": 0.0001746857592868843, + "loss": 3.2414, + "step": 1583 + }, + { + "epoch": 2.32, + "grad_norm": 5.648257732391357, + "learning_rate": 0.00017465456103273885, + "loss": 3.2601, + "step": 1584 + }, + { + "epoch": 2.32, + "grad_norm": 7.4345011711120605, + "learning_rate": 0.00017462334635509033, + "loss": 3.3107, + "step": 1585 + }, + { + "epoch": 2.32, + "grad_norm": 2.6171629428863525, + "learning_rate": 0.0001745921152608057, + "loss": 3.226, + "step": 1586 + }, + { + "epoch": 2.33, + "grad_norm": 6.692351341247559, + "learning_rate": 0.00017456086775675562, + "loss": 3.283, + "step": 1587 + }, + { + "epoch": 2.33, + "grad_norm": 7.625064373016357, + "learning_rate": 0.00017452960384981432, + "loss": 3.2796, + "step": 1588 + }, + { + "epoch": 2.33, + "grad_norm": 4.043941974639893, + "learning_rate": 0.00017449832354685961, + "loss": 3.2948, + "step": 1589 + }, + { + "epoch": 2.33, + "grad_norm": 6.564191818237305, + "learning_rate": 0.00017446702685477303, + "loss": 3.2824, + "step": 1590 + }, + { + "epoch": 2.33, + "grad_norm": 10.199933052062988, + "learning_rate": 0.00017443571378043959, + "loss": 3.3447, + "step": 1591 + }, + { + "epoch": 2.33, + "grad_norm": 6.0920729637146, + "learning_rate": 0.00017440438433074794, + "loss": 3.2682, + "step": 1592 + }, + { + "epoch": 2.33, + "grad_norm": 3.95531964302063, + "learning_rate": 0.00017437303851259038, + "loss": 3.2439, + "step": 1593 + }, + { + "epoch": 2.34, + "grad_norm": 6.185577869415283, + "learning_rate": 0.0001743416763328628, + "loss": 3.3002, + "step": 1594 + }, + { + "epoch": 2.34, + "grad_norm": 2.757133960723877, + "learning_rate": 0.00017431029779846463, + "loss": 3.2716, + "step": 1595 + }, + { + "epoch": 2.34, + "grad_norm": 8.002574920654297, + "learning_rate": 0.00017427890291629893, + "loss": 3.2844, + "step": 1596 + }, + { + "epoch": 2.34, + "grad_norm": 10.051774024963379, + "learning_rate": 0.0001742474916932724, + "loss": 3.2388, + "step": 1597 + }, + { + "epoch": 2.34, + "grad_norm": 5.608651638031006, + "learning_rate": 0.00017421606413629527, + "loss": 3.226, + "step": 1598 + }, + { + "epoch": 2.34, + "grad_norm": 4.5539231300354, + "learning_rate": 0.00017418462025228145, + "loss": 3.2424, + "step": 1599 + }, + { + "epoch": 2.34, + "grad_norm": 6.649083614349365, + "learning_rate": 0.0001741531600481483, + "loss": 3.2537, + "step": 1600 + }, + { + "epoch": 2.35, + "grad_norm": 3.6664934158325195, + "learning_rate": 0.00017412168353081686, + "loss": 3.243, + "step": 1601 + }, + { + "epoch": 2.35, + "grad_norm": 6.799222469329834, + "learning_rate": 0.00017409019070721183, + "loss": 3.2853, + "step": 1602 + }, + { + "epoch": 2.35, + "grad_norm": 9.875252723693848, + "learning_rate": 0.00017405868158426135, + "loss": 3.2505, + "step": 1603 + }, + { + "epoch": 2.35, + "grad_norm": 4.99692440032959, + "learning_rate": 0.00017402715616889728, + "loss": 3.3176, + "step": 1604 + }, + { + "epoch": 2.35, + "grad_norm": 6.484687328338623, + "learning_rate": 0.00017399561446805492, + "loss": 3.2547, + "step": 1605 + }, + { + "epoch": 2.35, + "grad_norm": 8.336297988891602, + "learning_rate": 0.00017396405648867328, + "loss": 3.2543, + "step": 1606 + }, + { + "epoch": 2.36, + "grad_norm": 5.791290283203125, + "learning_rate": 0.00017393248223769491, + "loss": 3.2538, + "step": 1607 + }, + { + "epoch": 2.36, + "grad_norm": 2.4937000274658203, + "learning_rate": 0.00017390089172206592, + "loss": 3.2665, + "step": 1608 + }, + { + "epoch": 2.36, + "grad_norm": 5.263038635253906, + "learning_rate": 0.00017386928494873602, + "loss": 3.2899, + "step": 1609 + }, + { + "epoch": 2.36, + "grad_norm": 0.9264150857925415, + "learning_rate": 0.00017383766192465846, + "loss": 3.2284, + "step": 1610 + }, + { + "epoch": 2.36, + "grad_norm": 7.343024730682373, + "learning_rate": 0.00017380602265679014, + "loss": 3.3191, + "step": 1611 + }, + { + "epoch": 2.36, + "grad_norm": 8.518827438354492, + "learning_rate": 0.00017377436715209144, + "loss": 3.3296, + "step": 1612 + }, + { + "epoch": 2.36, + "grad_norm": 5.042328357696533, + "learning_rate": 0.00017374269541752638, + "loss": 3.2711, + "step": 1613 + }, + { + "epoch": 2.37, + "grad_norm": 5.329464435577393, + "learning_rate": 0.00017371100746006252, + "loss": 3.2886, + "step": 1614 + }, + { + "epoch": 2.37, + "grad_norm": 8.302837371826172, + "learning_rate": 0.00017367930328667105, + "loss": 3.2565, + "step": 1615 + }, + { + "epoch": 2.37, + "grad_norm": 4.5046539306640625, + "learning_rate": 0.0001736475829043266, + "loss": 3.2309, + "step": 1616 + }, + { + "epoch": 2.37, + "grad_norm": 5.1940789222717285, + "learning_rate": 0.0001736158463200075, + "loss": 3.2687, + "step": 1617 + }, + { + "epoch": 2.37, + "grad_norm": 7.008042335510254, + "learning_rate": 0.00017358409354069554, + "loss": 3.2845, + "step": 1618 + }, + { + "epoch": 2.37, + "grad_norm": 3.637481689453125, + "learning_rate": 0.00017355232457337615, + "loss": 3.2221, + "step": 1619 + }, + { + "epoch": 2.37, + "grad_norm": 6.158731937408447, + "learning_rate": 0.00017352053942503828, + "loss": 3.2727, + "step": 1620 + }, + { + "epoch": 2.38, + "grad_norm": 9.95582103729248, + "learning_rate": 0.00017348873810267443, + "loss": 3.2826, + "step": 1621 + }, + { + "epoch": 2.38, + "grad_norm": 5.530641555786133, + "learning_rate": 0.0001734569206132807, + "loss": 3.291, + "step": 1622 + }, + { + "epoch": 2.38, + "grad_norm": 3.868178367614746, + "learning_rate": 0.00017342508696385674, + "loss": 3.2794, + "step": 1623 + }, + { + "epoch": 2.38, + "grad_norm": 6.0412516593933105, + "learning_rate": 0.00017339323716140573, + "loss": 3.2546, + "step": 1624 + }, + { + "epoch": 2.38, + "grad_norm": 2.217486619949341, + "learning_rate": 0.00017336137121293436, + "loss": 3.2526, + "step": 1625 + }, + { + "epoch": 2.38, + "grad_norm": 8.725723266601562, + "learning_rate": 0.000173329489125453, + "loss": 3.253, + "step": 1626 + }, + { + "epoch": 2.39, + "grad_norm": 12.251547813415527, + "learning_rate": 0.00017329759090597548, + "loss": 3.2593, + "step": 1627 + }, + { + "epoch": 2.39, + "grad_norm": 7.2429070472717285, + "learning_rate": 0.00017326567656151916, + "loss": 3.2555, + "step": 1628 + }, + { + "epoch": 2.39, + "grad_norm": 2.993010997772217, + "learning_rate": 0.000173233746099105, + "loss": 3.2165, + "step": 1629 + }, + { + "epoch": 2.39, + "grad_norm": 5.157336711883545, + "learning_rate": 0.0001732017995257575, + "loss": 3.3013, + "step": 1630 + }, + { + "epoch": 2.39, + "grad_norm": 2.0820674896240234, + "learning_rate": 0.00017316983684850467, + "loss": 3.2088, + "step": 1631 + }, + { + "epoch": 2.39, + "grad_norm": 8.305849075317383, + "learning_rate": 0.0001731378580743781, + "loss": 3.2902, + "step": 1632 + }, + { + "epoch": 2.39, + "grad_norm": 10.2841157913208, + "learning_rate": 0.00017310586321041288, + "loss": 3.2437, + "step": 1633 + }, + { + "epoch": 2.4, + "grad_norm": 4.831589698791504, + "learning_rate": 0.00017307385226364772, + "loss": 3.2177, + "step": 1634 + }, + { + "epoch": 2.4, + "grad_norm": 4.824088096618652, + "learning_rate": 0.00017304182524112478, + "loss": 3.2779, + "step": 1635 + }, + { + "epoch": 2.4, + "grad_norm": 7.442866802215576, + "learning_rate": 0.00017300978214988976, + "loss": 3.2716, + "step": 1636 + }, + { + "epoch": 2.4, + "grad_norm": 4.1812968254089355, + "learning_rate": 0.000172977722996992, + "loss": 3.2746, + "step": 1637 + }, + { + "epoch": 2.4, + "grad_norm": 5.722605228424072, + "learning_rate": 0.00017294564778948422, + "loss": 3.2238, + "step": 1638 + }, + { + "epoch": 2.4, + "grad_norm": 8.271087646484375, + "learning_rate": 0.00017291355653442278, + "loss": 3.3323, + "step": 1639 + }, + { + "epoch": 2.4, + "grad_norm": 3.2383551597595215, + "learning_rate": 0.00017288144923886757, + "loss": 3.2318, + "step": 1640 + }, + { + "epoch": 2.41, + "grad_norm": 6.346744537353516, + "learning_rate": 0.00017284932590988197, + "loss": 3.2569, + "step": 1641 + }, + { + "epoch": 2.41, + "grad_norm": 8.613646507263184, + "learning_rate": 0.00017281718655453286, + "loss": 3.2684, + "step": 1642 + }, + { + "epoch": 2.41, + "grad_norm": 5.7413129806518555, + "learning_rate": 0.00017278503117989073, + "loss": 3.2374, + "step": 1643 + }, + { + "epoch": 2.41, + "grad_norm": 3.2419638633728027, + "learning_rate": 0.0001727528597930295, + "loss": 3.2127, + "step": 1644 + }, + { + "epoch": 2.41, + "grad_norm": 6.177962779998779, + "learning_rate": 0.0001727206724010267, + "loss": 3.2859, + "step": 1645 + }, + { + "epoch": 2.41, + "grad_norm": 1.907907247543335, + "learning_rate": 0.0001726884690109633, + "loss": 3.2221, + "step": 1646 + }, + { + "epoch": 2.41, + "grad_norm": 6.6156415939331055, + "learning_rate": 0.0001726562496299239, + "loss": 3.2339, + "step": 1647 + }, + { + "epoch": 2.42, + "grad_norm": 8.058192253112793, + "learning_rate": 0.00017262401426499646, + "loss": 3.2718, + "step": 1648 + }, + { + "epoch": 2.42, + "grad_norm": 4.197020530700684, + "learning_rate": 0.00017259176292327259, + "loss": 3.3056, + "step": 1649 + }, + { + "epoch": 2.42, + "grad_norm": 5.8328166007995605, + "learning_rate": 0.00017255949561184738, + "loss": 3.2645, + "step": 1650 + }, + { + "epoch": 2.42, + "grad_norm": 9.331326484680176, + "learning_rate": 0.00017252721233781938, + "loss": 3.3117, + "step": 1651 + }, + { + "epoch": 2.42, + "grad_norm": 5.499820232391357, + "learning_rate": 0.00017249491310829078, + "loss": 3.278, + "step": 1652 + }, + { + "epoch": 2.42, + "grad_norm": 4.50565242767334, + "learning_rate": 0.00017246259793036707, + "loss": 3.2662, + "step": 1653 + }, + { + "epoch": 2.43, + "grad_norm": 6.363594055175781, + "learning_rate": 0.00017243026681115748, + "loss": 3.2784, + "step": 1654 + }, + { + "epoch": 2.43, + "grad_norm": 2.7347965240478516, + "learning_rate": 0.00017239791975777458, + "loss": 3.2907, + "step": 1655 + }, + { + "epoch": 2.43, + "grad_norm": 7.49398946762085, + "learning_rate": 0.00017236555677733453, + "loss": 3.3239, + "step": 1656 + }, + { + "epoch": 2.43, + "grad_norm": 10.45259952545166, + "learning_rate": 0.00017233317787695697, + "loss": 3.316, + "step": 1657 + }, + { + "epoch": 2.43, + "grad_norm": 5.672200679779053, + "learning_rate": 0.00017230078306376502, + "loss": 3.2043, + "step": 1658 + }, + { + "epoch": 2.43, + "grad_norm": 4.071951866149902, + "learning_rate": 0.00017226837234488534, + "loss": 3.2218, + "step": 1659 + }, + { + "epoch": 2.43, + "grad_norm": 5.800225257873535, + "learning_rate": 0.00017223594572744806, + "loss": 3.2586, + "step": 1660 + }, + { + "epoch": 2.44, + "grad_norm": 2.592320442199707, + "learning_rate": 0.0001722035032185868, + "loss": 3.244, + "step": 1661 + }, + { + "epoch": 2.44, + "grad_norm": 7.940516471862793, + "learning_rate": 0.00017217104482543876, + "loss": 3.2501, + "step": 1662 + }, + { + "epoch": 2.44, + "grad_norm": 10.571852684020996, + "learning_rate": 0.0001721385705551445, + "loss": 3.2631, + "step": 1663 + }, + { + "epoch": 2.44, + "grad_norm": 5.3643574714660645, + "learning_rate": 0.0001721060804148482, + "loss": 3.2116, + "step": 1664 + }, + { + "epoch": 2.44, + "grad_norm": 4.0979905128479, + "learning_rate": 0.00017207357441169742, + "loss": 3.1886, + "step": 1665 + }, + { + "epoch": 2.44, + "grad_norm": 6.099160194396973, + "learning_rate": 0.0001720410525528433, + "loss": 3.2502, + "step": 1666 + }, + { + "epoch": 2.44, + "grad_norm": 3.1805484294891357, + "learning_rate": 0.00017200851484544037, + "loss": 3.213, + "step": 1667 + }, + { + "epoch": 2.45, + "grad_norm": 7.588128089904785, + "learning_rate": 0.00017197596129664682, + "loss": 3.2935, + "step": 1668 + }, + { + "epoch": 2.45, + "grad_norm": 10.550485610961914, + "learning_rate": 0.0001719433919136241, + "loss": 3.2823, + "step": 1669 + }, + { + "epoch": 2.45, + "grad_norm": 5.52534818649292, + "learning_rate": 0.00017191080670353733, + "loss": 3.2452, + "step": 1670 + }, + { + "epoch": 2.45, + "grad_norm": 4.016864776611328, + "learning_rate": 0.00017187820567355497, + "loss": 3.2412, + "step": 1671 + }, + { + "epoch": 2.45, + "grad_norm": 6.307382583618164, + "learning_rate": 0.0001718455888308491, + "loss": 3.2699, + "step": 1672 + }, + { + "epoch": 2.45, + "grad_norm": 2.715496301651001, + "learning_rate": 0.00017181295618259515, + "loss": 3.2695, + "step": 1673 + }, + { + "epoch": 2.45, + "grad_norm": 7.303908348083496, + "learning_rate": 0.00017178030773597213, + "loss": 3.2392, + "step": 1674 + }, + { + "epoch": 2.46, + "grad_norm": 10.16519546508789, + "learning_rate": 0.00017174764349816245, + "loss": 3.2635, + "step": 1675 + }, + { + "epoch": 2.46, + "grad_norm": 5.080345630645752, + "learning_rate": 0.00017171496347635206, + "loss": 3.2302, + "step": 1676 + }, + { + "epoch": 2.46, + "grad_norm": 5.1449408531188965, + "learning_rate": 0.00017168226767773028, + "loss": 3.2135, + "step": 1677 + }, + { + "epoch": 2.46, + "grad_norm": 7.340461254119873, + "learning_rate": 0.00017164955610949004, + "loss": 3.2561, + "step": 1678 + }, + { + "epoch": 2.46, + "grad_norm": 3.821507453918457, + "learning_rate": 0.00017161682877882758, + "loss": 3.2111, + "step": 1679 + }, + { + "epoch": 2.46, + "grad_norm": 6.017604351043701, + "learning_rate": 0.0001715840856929428, + "loss": 3.3209, + "step": 1680 + }, + { + "epoch": 2.47, + "grad_norm": 8.912531852722168, + "learning_rate": 0.00017155132685903887, + "loss": 3.2441, + "step": 1681 + }, + { + "epoch": 2.47, + "grad_norm": 4.574288368225098, + "learning_rate": 0.00017151855228432253, + "loss": 3.2489, + "step": 1682 + }, + { + "epoch": 2.47, + "grad_norm": 4.964592456817627, + "learning_rate": 0.00017148576197600402, + "loss": 3.2607, + "step": 1683 + }, + { + "epoch": 2.47, + "grad_norm": 7.388319969177246, + "learning_rate": 0.00017145295594129693, + "loss": 3.2537, + "step": 1684 + }, + { + "epoch": 2.47, + "grad_norm": 4.345710277557373, + "learning_rate": 0.0001714201341874184, + "loss": 3.2435, + "step": 1685 + }, + { + "epoch": 2.47, + "grad_norm": 5.151455402374268, + "learning_rate": 0.000171387296721589, + "loss": 3.2198, + "step": 1686 + }, + { + "epoch": 2.47, + "grad_norm": 9.679162979125977, + "learning_rate": 0.0001713544435510327, + "loss": 3.3015, + "step": 1687 + }, + { + "epoch": 2.48, + "grad_norm": 4.5911383628845215, + "learning_rate": 0.00017132157468297704, + "loss": 3.2534, + "step": 1688 + }, + { + "epoch": 2.48, + "grad_norm": 3.5955042839050293, + "learning_rate": 0.00017128869012465292, + "loss": 3.2479, + "step": 1689 + }, + { + "epoch": 2.48, + "grad_norm": 4.769717693328857, + "learning_rate": 0.00017125578988329472, + "loss": 3.2706, + "step": 1690 + }, + { + "epoch": 2.48, + "eval_loss": 3.0677621364593506, + "eval_runtime": 1164.46, + "eval_samples_per_second": 14.721, + "eval_steps_per_second": 0.921, + "step": 1690 + }, + { + "epoch": 2.48, + "grad_norm": 0.9628227949142456, + "learning_rate": 0.0001712228739661403, + "loss": 3.2333, + "step": 1691 + }, + { + "epoch": 2.48, + "grad_norm": 9.899592399597168, + "learning_rate": 0.00017118994238043088, + "loss": 3.287, + "step": 1692 + }, + { + "epoch": 2.48, + "grad_norm": 13.610316276550293, + "learning_rate": 0.00017115699513341127, + "loss": 3.3172, + "step": 1693 + }, + { + "epoch": 2.48, + "grad_norm": 8.333012580871582, + "learning_rate": 0.00017112403223232964, + "loss": 3.2552, + "step": 1694 + }, + { + "epoch": 2.49, + "grad_norm": 1.985851764678955, + "learning_rate": 0.0001710910536844375, + "loss": 3.2494, + "step": 1695 + }, + { + "epoch": 2.49, + "grad_norm": 4.960620403289795, + "learning_rate": 0.00017105805949699, + "loss": 3.2411, + "step": 1696 + }, + { + "epoch": 2.49, + "grad_norm": 2.77993106842041, + "learning_rate": 0.00017102504967724563, + "loss": 3.264, + "step": 1697 + }, + { + "epoch": 2.49, + "grad_norm": 6.050425052642822, + "learning_rate": 0.0001709920242324663, + "loss": 3.2725, + "step": 1698 + }, + { + "epoch": 2.49, + "grad_norm": 6.0771307945251465, + "learning_rate": 0.00017095898316991744, + "loss": 3.199, + "step": 1699 + }, + { + "epoch": 2.49, + "grad_norm": 1.5405139923095703, + "learning_rate": 0.0001709259264968678, + "loss": 3.2523, + "step": 1700 + }, + { + "epoch": 2.49, + "grad_norm": 5.121715068817139, + "learning_rate": 0.00017089285422058968, + "loss": 3.2944, + "step": 1701 + }, + { + "epoch": 2.5, + "grad_norm": 4.806593894958496, + "learning_rate": 0.00017085976634835873, + "loss": 3.2461, + "step": 1702 + }, + { + "epoch": 2.5, + "grad_norm": 3.7719340324401855, + "learning_rate": 0.00017082666288745403, + "loss": 3.2067, + "step": 1703 + }, + { + "epoch": 2.5, + "grad_norm": 3.7723586559295654, + "learning_rate": 0.0001707935438451582, + "loss": 3.2715, + "step": 1704 + }, + { + "epoch": 2.5, + "grad_norm": 2.5075759887695312, + "learning_rate": 0.00017076040922875713, + "loss": 3.2053, + "step": 1705 + }, + { + "epoch": 2.5, + "grad_norm": 2.315262794494629, + "learning_rate": 0.00017072725904554025, + "loss": 3.2529, + "step": 1706 + }, + { + "epoch": 2.5, + "grad_norm": 3.849294900894165, + "learning_rate": 0.0001706940933028004, + "loss": 3.2501, + "step": 1707 + }, + { + "epoch": 2.51, + "grad_norm": 3.2091238498687744, + "learning_rate": 0.00017066091200783375, + "loss": 3.1893, + "step": 1708 + }, + { + "epoch": 2.51, + "grad_norm": 3.587019443511963, + "learning_rate": 0.00017062771516794002, + "loss": 3.2435, + "step": 1709 + }, + { + "epoch": 2.51, + "grad_norm": 2.367382526397705, + "learning_rate": 0.00017059450279042232, + "loss": 3.2453, + "step": 1710 + }, + { + "epoch": 2.51, + "grad_norm": 4.0962700843811035, + "learning_rate": 0.00017056127488258705, + "loss": 3.2043, + "step": 1711 + }, + { + "epoch": 2.51, + "grad_norm": 4.637650966644287, + "learning_rate": 0.0001705280314517442, + "loss": 3.2506, + "step": 1712 + }, + { + "epoch": 2.51, + "grad_norm": 1.244682788848877, + "learning_rate": 0.00017049477250520714, + "loss": 3.2169, + "step": 1713 + }, + { + "epoch": 2.51, + "grad_norm": 0.7056719660758972, + "learning_rate": 0.00017046149805029253, + "loss": 3.212, + "step": 1714 + }, + { + "epoch": 2.52, + "grad_norm": 2.6713907718658447, + "learning_rate": 0.00017042820809432056, + "loss": 3.2665, + "step": 1715 + }, + { + "epoch": 2.52, + "grad_norm": 1.15023934841156, + "learning_rate": 0.00017039490264461477, + "loss": 3.2735, + "step": 1716 + }, + { + "epoch": 2.52, + "grad_norm": 3.00900936126709, + "learning_rate": 0.0001703615817085022, + "loss": 3.2198, + "step": 1717 + }, + { + "epoch": 2.52, + "grad_norm": 3.382124423980713, + "learning_rate": 0.0001703282452933132, + "loss": 3.2279, + "step": 1718 + }, + { + "epoch": 2.52, + "grad_norm": 1.6238938570022583, + "learning_rate": 0.00017029489340638154, + "loss": 3.2521, + "step": 1719 + }, + { + "epoch": 2.52, + "grad_norm": 4.034934997558594, + "learning_rate": 0.00017026152605504442, + "loss": 3.2477, + "step": 1720 + }, + { + "epoch": 2.52, + "grad_norm": 4.463752269744873, + "learning_rate": 0.00017022814324664244, + "loss": 3.2478, + "step": 1721 + }, + { + "epoch": 2.53, + "grad_norm": 2.6497182846069336, + "learning_rate": 0.00017019474498851958, + "loss": 3.2774, + "step": 1722 + }, + { + "epoch": 2.53, + "grad_norm": 2.8634886741638184, + "learning_rate": 0.0001701613312880233, + "loss": 3.2704, + "step": 1723 + }, + { + "epoch": 2.53, + "grad_norm": 3.318505048751831, + "learning_rate": 0.00017012790215250432, + "loss": 3.2681, + "step": 1724 + }, + { + "epoch": 2.53, + "grad_norm": 2.9322783946990967, + "learning_rate": 0.0001700944575893168, + "loss": 3.2489, + "step": 1725 + }, + { + "epoch": 2.53, + "grad_norm": 1.9499863386154175, + "learning_rate": 0.00017006099760581838, + "loss": 3.2197, + "step": 1726 + }, + { + "epoch": 2.53, + "grad_norm": 2.5505218505859375, + "learning_rate": 0.00017002752220937004, + "loss": 3.2399, + "step": 1727 + }, + { + "epoch": 2.53, + "grad_norm": 2.46716570854187, + "learning_rate": 0.0001699940314073361, + "loss": 3.2078, + "step": 1728 + }, + { + "epoch": 2.54, + "grad_norm": 1.8553292751312256, + "learning_rate": 0.00016996052520708434, + "loss": 3.292, + "step": 1729 + }, + { + "epoch": 2.54, + "grad_norm": 2.041597366333008, + "learning_rate": 0.00016992700361598585, + "loss": 3.201, + "step": 1730 + }, + { + "epoch": 2.54, + "grad_norm": 2.173717737197876, + "learning_rate": 0.00016989346664141524, + "loss": 3.2707, + "step": 1731 + }, + { + "epoch": 2.54, + "grad_norm": 1.5214920043945312, + "learning_rate": 0.00016985991429075036, + "loss": 3.2218, + "step": 1732 + }, + { + "epoch": 2.54, + "grad_norm": 6.029373645782471, + "learning_rate": 0.00016982634657137252, + "loss": 3.3117, + "step": 1733 + }, + { + "epoch": 2.54, + "grad_norm": 5.91231107711792, + "learning_rate": 0.00016979276349066638, + "loss": 3.2202, + "step": 1734 + }, + { + "epoch": 2.55, + "grad_norm": 2.8163087368011475, + "learning_rate": 0.00016975916505602, + "loss": 3.2803, + "step": 1735 + }, + { + "epoch": 2.55, + "grad_norm": 3.7836880683898926, + "learning_rate": 0.00016972555127482483, + "loss": 3.2399, + "step": 1736 + }, + { + "epoch": 2.55, + "grad_norm": 3.339012622833252, + "learning_rate": 0.00016969192215447566, + "loss": 3.2619, + "step": 1737 + }, + { + "epoch": 2.55, + "grad_norm": 3.779021739959717, + "learning_rate": 0.00016965827770237067, + "loss": 3.2205, + "step": 1738 + }, + { + "epoch": 2.55, + "grad_norm": 0.7899141311645508, + "learning_rate": 0.00016962461792591137, + "loss": 3.2228, + "step": 1739 + }, + { + "epoch": 2.55, + "grad_norm": 2.307349443435669, + "learning_rate": 0.00016959094283250278, + "loss": 3.2412, + "step": 1740 + }, + { + "epoch": 2.55, + "grad_norm": 1.5662769079208374, + "learning_rate": 0.00016955725242955315, + "loss": 3.257, + "step": 1741 + }, + { + "epoch": 2.56, + "grad_norm": 1.4268426895141602, + "learning_rate": 0.0001695235467244741, + "loss": 3.2718, + "step": 1742 + }, + { + "epoch": 2.56, + "grad_norm": 1.7694865465164185, + "learning_rate": 0.0001694898257246807, + "loss": 3.2414, + "step": 1743 + }, + { + "epoch": 2.56, + "grad_norm": 2.555623769760132, + "learning_rate": 0.00016945608943759138, + "loss": 3.2384, + "step": 1744 + }, + { + "epoch": 2.56, + "grad_norm": 1.4633605480194092, + "learning_rate": 0.0001694223378706278, + "loss": 3.2164, + "step": 1745 + }, + { + "epoch": 2.56, + "grad_norm": 2.0434815883636475, + "learning_rate": 0.0001693885710312152, + "loss": 3.2547, + "step": 1746 + }, + { + "epoch": 2.56, + "grad_norm": 2.4210758209228516, + "learning_rate": 0.00016935478892678195, + "loss": 3.2367, + "step": 1747 + }, + { + "epoch": 2.56, + "grad_norm": 0.8179725408554077, + "learning_rate": 0.00016932099156475995, + "loss": 3.2619, + "step": 1748 + }, + { + "epoch": 2.57, + "grad_norm": 3.3028626441955566, + "learning_rate": 0.00016928717895258437, + "loss": 3.2256, + "step": 1749 + }, + { + "epoch": 2.57, + "grad_norm": 1.852142572402954, + "learning_rate": 0.00016925335109769374, + "loss": 3.2402, + "step": 1750 + }, + { + "epoch": 2.57, + "grad_norm": 2.416368007659912, + "learning_rate": 0.00016921950800753002, + "loss": 3.234, + "step": 1751 + }, + { + "epoch": 2.57, + "grad_norm": 1.5477232933044434, + "learning_rate": 0.00016918564968953842, + "loss": 3.2142, + "step": 1752 + }, + { + "epoch": 2.57, + "grad_norm": 1.5452466011047363, + "learning_rate": 0.00016915177615116754, + "loss": 3.1941, + "step": 1753 + }, + { + "epoch": 2.57, + "grad_norm": 3.213561773300171, + "learning_rate": 0.00016911788739986934, + "loss": 3.2154, + "step": 1754 + }, + { + "epoch": 2.58, + "grad_norm": 1.2881815433502197, + "learning_rate": 0.0001690839834430991, + "loss": 3.2909, + "step": 1755 + }, + { + "epoch": 2.58, + "grad_norm": 3.375774383544922, + "learning_rate": 0.00016905006428831552, + "loss": 3.2096, + "step": 1756 + }, + { + "epoch": 2.58, + "grad_norm": 2.6476569175720215, + "learning_rate": 0.00016901612994298053, + "loss": 3.2082, + "step": 1757 + }, + { + "epoch": 2.58, + "grad_norm": 3.5997047424316406, + "learning_rate": 0.0001689821804145595, + "loss": 3.1989, + "step": 1758 + }, + { + "epoch": 2.58, + "grad_norm": 2.435394763946533, + "learning_rate": 0.00016894821571052104, + "loss": 3.2109, + "step": 1759 + }, + { + "epoch": 2.58, + "grad_norm": 5.273232936859131, + "learning_rate": 0.00016891423583833724, + "loss": 3.2358, + "step": 1760 + }, + { + "epoch": 2.58, + "grad_norm": 4.958769798278809, + "learning_rate": 0.00016888024080548335, + "loss": 3.2315, + "step": 1761 + }, + { + "epoch": 2.59, + "grad_norm": 2.171224594116211, + "learning_rate": 0.00016884623061943813, + "loss": 3.2587, + "step": 1762 + }, + { + "epoch": 2.59, + "grad_norm": 1.9900020360946655, + "learning_rate": 0.00016881220528768356, + "loss": 3.2082, + "step": 1763 + }, + { + "epoch": 2.59, + "grad_norm": 2.3555827140808105, + "learning_rate": 0.00016877816481770496, + "loss": 3.2356, + "step": 1764 + }, + { + "epoch": 2.59, + "grad_norm": 1.4140695333480835, + "learning_rate": 0.00016874410921699105, + "loss": 3.1722, + "step": 1765 + }, + { + "epoch": 2.59, + "grad_norm": 4.323090076446533, + "learning_rate": 0.00016871003849303382, + "loss": 3.2533, + "step": 1766 + }, + { + "epoch": 2.59, + "grad_norm": 2.4270732402801514, + "learning_rate": 0.00016867595265332859, + "loss": 3.2092, + "step": 1767 + }, + { + "epoch": 2.59, + "grad_norm": 3.9084079265594482, + "learning_rate": 0.000168641851705374, + "loss": 3.2595, + "step": 1768 + }, + { + "epoch": 2.6, + "grad_norm": 3.986670970916748, + "learning_rate": 0.00016860773565667206, + "loss": 3.2302, + "step": 1769 + }, + { + "epoch": 2.6, + "grad_norm": 3.4050045013427734, + "learning_rate": 0.0001685736045147281, + "loss": 3.2589, + "step": 1770 + }, + { + "epoch": 2.6, + "grad_norm": 4.226673126220703, + "learning_rate": 0.0001685394582870507, + "loss": 3.235, + "step": 1771 + }, + { + "epoch": 2.6, + "grad_norm": 1.5907903909683228, + "learning_rate": 0.00016850529698115175, + "loss": 3.3035, + "step": 1772 + }, + { + "epoch": 2.6, + "grad_norm": 2.343108654022217, + "learning_rate": 0.00016847112060454665, + "loss": 3.1848, + "step": 1773 + }, + { + "epoch": 2.6, + "grad_norm": 3.3092055320739746, + "learning_rate": 0.00016843692916475384, + "loss": 3.2648, + "step": 1774 + }, + { + "epoch": 2.6, + "grad_norm": 1.7477455139160156, + "learning_rate": 0.00016840272266929523, + "loss": 3.206, + "step": 1775 + }, + { + "epoch": 2.61, + "grad_norm": 4.240764617919922, + "learning_rate": 0.00016836850112569613, + "loss": 3.2177, + "step": 1776 + }, + { + "epoch": 2.61, + "grad_norm": 4.2325758934021, + "learning_rate": 0.0001683342645414849, + "loss": 3.2154, + "step": 1777 + }, + { + "epoch": 2.61, + "grad_norm": 2.993154525756836, + "learning_rate": 0.00016830001292419345, + "loss": 3.1972, + "step": 1778 + }, + { + "epoch": 2.61, + "grad_norm": 3.2484753131866455, + "learning_rate": 0.00016826574628135688, + "loss": 3.232, + "step": 1779 + }, + { + "epoch": 2.61, + "grad_norm": 2.438793420791626, + "learning_rate": 0.00016823146462051364, + "loss": 3.2292, + "step": 1780 + }, + { + "epoch": 2.61, + "grad_norm": 2.0203206539154053, + "learning_rate": 0.0001681971679492054, + "loss": 3.1791, + "step": 1781 + }, + { + "epoch": 2.62, + "grad_norm": 4.8010125160217285, + "learning_rate": 0.0001681628562749773, + "loss": 3.2012, + "step": 1782 + }, + { + "epoch": 2.62, + "grad_norm": 4.596678256988525, + "learning_rate": 0.00016812852960537762, + "loss": 3.2204, + "step": 1783 + }, + { + "epoch": 2.62, + "grad_norm": 2.4350953102111816, + "learning_rate": 0.00016809418794795796, + "loss": 3.2175, + "step": 1784 + }, + { + "epoch": 2.62, + "grad_norm": 15.600502014160156, + "learning_rate": 0.00016805983131027333, + "loss": 3.256, + "step": 1785 + }, + { + "epoch": 2.62, + "grad_norm": 3.7827413082122803, + "learning_rate": 0.00016802545969988195, + "loss": 3.2165, + "step": 1786 + }, + { + "epoch": 2.62, + "grad_norm": 3.0212562084198, + "learning_rate": 0.0001679910731243453, + "loss": 3.2537, + "step": 1787 + }, + { + "epoch": 2.62, + "grad_norm": 2.496250629425049, + "learning_rate": 0.00016795667159122824, + "loss": 3.212, + "step": 1788 + }, + { + "epoch": 2.63, + "grad_norm": 3.222059488296509, + "learning_rate": 0.00016792225510809883, + "loss": 3.2665, + "step": 1789 + }, + { + "epoch": 2.63, + "grad_norm": 3.4902381896972656, + "learning_rate": 0.0001678878236825285, + "loss": 3.2326, + "step": 1790 + }, + { + "epoch": 2.63, + "grad_norm": 0.7913928627967834, + "learning_rate": 0.00016785337732209193, + "loss": 3.1662, + "step": 1791 + }, + { + "epoch": 2.63, + "grad_norm": 3.9567112922668457, + "learning_rate": 0.00016781891603436706, + "loss": 3.2217, + "step": 1792 + }, + { + "epoch": 2.63, + "grad_norm": 1.712324619293213, + "learning_rate": 0.00016778443982693519, + "loss": 3.1855, + "step": 1793 + }, + { + "epoch": 2.63, + "grad_norm": 1.2269927263259888, + "learning_rate": 0.0001677499487073808, + "loss": 3.2306, + "step": 1794 + }, + { + "epoch": 2.63, + "grad_norm": 3.1048622131347656, + "learning_rate": 0.00016771544268329177, + "loss": 3.1975, + "step": 1795 + }, + { + "epoch": 2.64, + "grad_norm": 1.18289315700531, + "learning_rate": 0.00016768092176225915, + "loss": 3.1904, + "step": 1796 + }, + { + "epoch": 2.64, + "grad_norm": 1.7925835847854614, + "learning_rate": 0.0001676463859518773, + "loss": 3.2143, + "step": 1797 + }, + { + "epoch": 2.64, + "grad_norm": 2.179051637649536, + "learning_rate": 0.0001676118352597439, + "loss": 3.2297, + "step": 1798 + }, + { + "epoch": 2.64, + "grad_norm": 1.2693228721618652, + "learning_rate": 0.00016757726969345984, + "loss": 3.2181, + "step": 1799 + }, + { + "epoch": 2.64, + "grad_norm": 5.075618743896484, + "learning_rate": 0.00016754268926062938, + "loss": 3.237, + "step": 1800 + }, + { + "epoch": 2.64, + "grad_norm": 5.278939723968506, + "learning_rate": 0.0001675080939688599, + "loss": 3.2266, + "step": 1801 + }, + { + "epoch": 2.64, + "grad_norm": 2.5977909564971924, + "learning_rate": 0.00016747348382576213, + "loss": 3.2409, + "step": 1802 + }, + { + "epoch": 2.65, + "grad_norm": 3.318488597869873, + "learning_rate": 0.00016743885883895013, + "loss": 3.231, + "step": 1803 + }, + { + "epoch": 2.65, + "grad_norm": 2.22575306892395, + "learning_rate": 0.00016740421901604118, + "loss": 3.2313, + "step": 1804 + }, + { + "epoch": 2.65, + "grad_norm": 2.436943769454956, + "learning_rate": 0.00016736956436465573, + "loss": 3.2268, + "step": 1805 + }, + { + "epoch": 2.65, + "grad_norm": 2.966697931289673, + "learning_rate": 0.0001673348948924176, + "loss": 3.2139, + "step": 1806 + }, + { + "epoch": 2.65, + "grad_norm": 2.458636999130249, + "learning_rate": 0.0001673002106069539, + "loss": 3.2228, + "step": 1807 + }, + { + "epoch": 2.65, + "grad_norm": 3.647672414779663, + "learning_rate": 0.00016726551151589485, + "loss": 3.256, + "step": 1808 + }, + { + "epoch": 2.66, + "grad_norm": 2.713827133178711, + "learning_rate": 0.0001672307976268741, + "loss": 3.2256, + "step": 1809 + }, + { + "epoch": 2.66, + "grad_norm": 3.452815294265747, + "learning_rate": 0.0001671960689475284, + "loss": 3.2875, + "step": 1810 + }, + { + "epoch": 2.66, + "grad_norm": 2.377556085586548, + "learning_rate": 0.00016716132548549786, + "loss": 3.2188, + "step": 1811 + }, + { + "epoch": 2.66, + "grad_norm": 3.791208028793335, + "learning_rate": 0.0001671265672484258, + "loss": 3.234, + "step": 1812 + }, + { + "epoch": 2.66, + "grad_norm": 3.7197184562683105, + "learning_rate": 0.00016709179424395884, + "loss": 3.1895, + "step": 1813 + }, + { + "epoch": 2.66, + "grad_norm": 2.4916012287139893, + "learning_rate": 0.00016705700647974675, + "loss": 3.2053, + "step": 1814 + }, + { + "epoch": 2.66, + "grad_norm": 2.569413185119629, + "learning_rate": 0.00016702220396344267, + "loss": 3.2358, + "step": 1815 + }, + { + "epoch": 2.67, + "grad_norm": 3.520432949066162, + "learning_rate": 0.00016698738670270285, + "loss": 3.2369, + "step": 1816 + }, + { + "epoch": 2.67, + "grad_norm": 2.880115032196045, + "learning_rate": 0.00016695255470518686, + "loss": 3.2232, + "step": 1817 + }, + { + "epoch": 2.67, + "grad_norm": 3.42354416847229, + "learning_rate": 0.00016691770797855758, + "loss": 3.1885, + "step": 1818 + }, + { + "epoch": 2.67, + "grad_norm": 3.31255841255188, + "learning_rate": 0.00016688284653048097, + "loss": 3.21, + "step": 1819 + }, + { + "epoch": 2.67, + "grad_norm": 3.157924175262451, + "learning_rate": 0.00016684797036862635, + "loss": 3.2055, + "step": 1820 + }, + { + "epoch": 2.67, + "grad_norm": 2.820952892303467, + "learning_rate": 0.00016681307950066628, + "loss": 3.1809, + "step": 1821 + }, + { + "epoch": 2.67, + "grad_norm": 3.921417236328125, + "learning_rate": 0.00016677817393427645, + "loss": 3.1902, + "step": 1822 + }, + { + "epoch": 2.68, + "grad_norm": 3.5899341106414795, + "learning_rate": 0.00016674325367713592, + "loss": 3.2222, + "step": 1823 + }, + { + "epoch": 2.68, + "grad_norm": 2.645939350128174, + "learning_rate": 0.00016670831873692683, + "loss": 3.1834, + "step": 1824 + }, + { + "epoch": 2.68, + "grad_norm": 2.4159257411956787, + "learning_rate": 0.00016667336912133468, + "loss": 3.2018, + "step": 1825 + }, + { + "epoch": 2.68, + "grad_norm": 3.56545090675354, + "learning_rate": 0.0001666384048380482, + "loss": 3.2096, + "step": 1826 + }, + { + "epoch": 2.68, + "grad_norm": 2.857130765914917, + "learning_rate": 0.0001666034258947592, + "loss": 3.2102, + "step": 1827 + }, + { + "epoch": 2.68, + "grad_norm": 3.4373629093170166, + "learning_rate": 0.00016656843229916285, + "loss": 3.2017, + "step": 1828 + }, + { + "epoch": 2.68, + "grad_norm": 2.5930848121643066, + "learning_rate": 0.00016653342405895756, + "loss": 3.2265, + "step": 1829 + }, + { + "epoch": 2.69, + "grad_norm": 4.415797710418701, + "learning_rate": 0.0001664984011818448, + "loss": 3.1589, + "step": 1830 + }, + { + "epoch": 2.69, + "grad_norm": 4.656808853149414, + "learning_rate": 0.0001664633636755295, + "loss": 3.2189, + "step": 1831 + }, + { + "epoch": 2.69, + "grad_norm": 1.3080215454101562, + "learning_rate": 0.00016642831154771957, + "loss": 3.2178, + "step": 1832 + }, + { + "epoch": 2.69, + "grad_norm": 1.491684913635254, + "learning_rate": 0.0001663932448061263, + "loss": 3.1922, + "step": 1833 + }, + { + "epoch": 2.69, + "grad_norm": 5.020033359527588, + "learning_rate": 0.0001663581634584641, + "loss": 3.2153, + "step": 1834 + }, + { + "epoch": 2.69, + "grad_norm": 4.078526973724365, + "learning_rate": 0.0001663230675124507, + "loss": 3.2165, + "step": 1835 + }, + { + "epoch": 2.7, + "grad_norm": 2.787774085998535, + "learning_rate": 0.00016628795697580686, + "loss": 3.1903, + "step": 1836 + }, + { + "epoch": 2.7, + "grad_norm": 2.4265925884246826, + "learning_rate": 0.0001662528318562568, + "loss": 3.1732, + "step": 1837 + }, + { + "epoch": 2.7, + "grad_norm": 3.579639434814453, + "learning_rate": 0.00016621769216152766, + "loss": 3.1891, + "step": 1838 + }, + { + "epoch": 2.7, + "grad_norm": 2.984905958175659, + "learning_rate": 0.00016618253789935005, + "loss": 3.2348, + "step": 1839 + }, + { + "epoch": 2.7, + "grad_norm": 4.036933898925781, + "learning_rate": 0.00016614736907745768, + "loss": 3.1677, + "step": 1840 + }, + { + "epoch": 2.7, + "grad_norm": 3.9433419704437256, + "learning_rate": 0.00016611218570358735, + "loss": 3.18, + "step": 1841 + }, + { + "epoch": 2.7, + "grad_norm": 2.285982370376587, + "learning_rate": 0.00016607698778547925, + "loss": 3.2298, + "step": 1842 + }, + { + "epoch": 2.71, + "grad_norm": 2.241262197494507, + "learning_rate": 0.00016604177533087667, + "loss": 3.1903, + "step": 1843 + }, + { + "epoch": 2.71, + "grad_norm": 3.7759506702423096, + "learning_rate": 0.00016600654834752611, + "loss": 3.2093, + "step": 1844 + }, + { + "epoch": 2.71, + "grad_norm": 3.2228505611419678, + "learning_rate": 0.00016597130684317725, + "loss": 3.1934, + "step": 1845 + }, + { + "epoch": 2.71, + "grad_norm": 3.3804237842559814, + "learning_rate": 0.00016593605082558305, + "loss": 3.2345, + "step": 1846 + }, + { + "epoch": 2.71, + "grad_norm": 2.6020681858062744, + "learning_rate": 0.0001659007803024995, + "loss": 3.1868, + "step": 1847 + }, + { + "epoch": 2.71, + "grad_norm": 3.3784382343292236, + "learning_rate": 0.00016586549528168596, + "loss": 3.1865, + "step": 1848 + }, + { + "epoch": 2.71, + "grad_norm": 2.881960391998291, + "learning_rate": 0.0001658301957709049, + "loss": 3.1906, + "step": 1849 + }, + { + "epoch": 2.72, + "grad_norm": 3.543879270553589, + "learning_rate": 0.00016579488177792187, + "loss": 3.2124, + "step": 1850 + }, + { + "epoch": 2.72, + "grad_norm": 3.8492488861083984, + "learning_rate": 0.00016575955331050584, + "loss": 3.3212, + "step": 1851 + }, + { + "epoch": 2.72, + "grad_norm": 2.2801854610443115, + "learning_rate": 0.00016572421037642875, + "loss": 3.2374, + "step": 1852 + }, + { + "epoch": 2.72, + "grad_norm": 2.0372071266174316, + "learning_rate": 0.00016568885298346586, + "loss": 3.2105, + "step": 1853 + }, + { + "epoch": 2.72, + "grad_norm": 4.137518405914307, + "learning_rate": 0.00016565348113939556, + "loss": 3.1718, + "step": 1854 + }, + { + "epoch": 2.72, + "grad_norm": 3.7536251544952393, + "learning_rate": 0.00016561809485199938, + "loss": 3.2118, + "step": 1855 + }, + { + "epoch": 2.73, + "grad_norm": 3.2339396476745605, + "learning_rate": 0.00016558269412906206, + "loss": 3.1746, + "step": 1856 + }, + { + "epoch": 2.73, + "grad_norm": 2.7833776473999023, + "learning_rate": 0.0001655472789783716, + "loss": 3.1876, + "step": 1857 + }, + { + "epoch": 2.73, + "grad_norm": 3.6448328495025635, + "learning_rate": 0.000165511849407719, + "loss": 3.2143, + "step": 1858 + }, + { + "epoch": 2.73, + "grad_norm": 3.4505040645599365, + "learning_rate": 0.0001654764054248986, + "loss": 3.173, + "step": 1859 + }, + { + "epoch": 2.73, + "eval_loss": 3.040992498397827, + "eval_runtime": 1164.9923, + "eval_samples_per_second": 14.714, + "eval_steps_per_second": 0.92, + "step": 1859 + }, + { + "epoch": 2.73, + "grad_norm": 2.8391804695129395, + "learning_rate": 0.0001654409470377078, + "loss": 3.2099, + "step": 1860 + }, + { + "epoch": 2.73, + "grad_norm": 2.7543411254882812, + "learning_rate": 0.0001654054742539472, + "loss": 3.1918, + "step": 1861 + }, + { + "epoch": 2.73, + "grad_norm": 3.5177597999572754, + "learning_rate": 0.00016536998708142062, + "loss": 3.245, + "step": 1862 + }, + { + "epoch": 2.74, + "grad_norm": 3.0664377212524414, + "learning_rate": 0.00016533448552793494, + "loss": 3.1616, + "step": 1863 + }, + { + "epoch": 2.74, + "grad_norm": 2.924795389175415, + "learning_rate": 0.00016529896960130032, + "loss": 3.229, + "step": 1864 + }, + { + "epoch": 2.74, + "grad_norm": 3.104602575302124, + "learning_rate": 0.00016526343930933, + "loss": 3.2294, + "step": 1865 + }, + { + "epoch": 2.74, + "grad_norm": 3.728653907775879, + "learning_rate": 0.0001652278946598404, + "loss": 3.2088, + "step": 1866 + }, + { + "epoch": 2.74, + "grad_norm": 3.854654312133789, + "learning_rate": 0.00016519233566065112, + "loss": 3.1829, + "step": 1867 + }, + { + "epoch": 2.74, + "grad_norm": 2.6493520736694336, + "learning_rate": 0.0001651567623195849, + "loss": 3.1807, + "step": 1868 + }, + { + "epoch": 2.74, + "grad_norm": 2.057978391647339, + "learning_rate": 0.00016512117464446762, + "loss": 3.141, + "step": 1869 + }, + { + "epoch": 2.75, + "grad_norm": 4.454994201660156, + "learning_rate": 0.00016508557264312835, + "loss": 3.1908, + "step": 1870 + }, + { + "epoch": 2.75, + "grad_norm": 4.243129253387451, + "learning_rate": 0.00016504995632339928, + "loss": 3.1819, + "step": 1871 + }, + { + "epoch": 2.75, + "grad_norm": 2.3739356994628906, + "learning_rate": 0.00016501432569311574, + "loss": 3.2002, + "step": 1872 + }, + { + "epoch": 2.75, + "grad_norm": 1.8129799365997314, + "learning_rate": 0.00016497868076011628, + "loss": 3.2084, + "step": 1873 + }, + { + "epoch": 2.75, + "grad_norm": 4.162952899932861, + "learning_rate": 0.0001649430215322425, + "loss": 3.1766, + "step": 1874 + }, + { + "epoch": 2.75, + "grad_norm": 3.205064535140991, + "learning_rate": 0.00016490734801733923, + "loss": 3.1601, + "step": 1875 + }, + { + "epoch": 2.75, + "grad_norm": 3.1646533012390137, + "learning_rate": 0.0001648716602232544, + "loss": 3.2347, + "step": 1876 + }, + { + "epoch": 2.76, + "grad_norm": 3.2153899669647217, + "learning_rate": 0.00016483595815783904, + "loss": 3.2166, + "step": 1877 + }, + { + "epoch": 2.76, + "grad_norm": 2.5029168128967285, + "learning_rate": 0.00016480024182894743, + "loss": 3.2366, + "step": 1878 + }, + { + "epoch": 2.76, + "grad_norm": 1.944775938987732, + "learning_rate": 0.0001647645112444369, + "loss": 3.1608, + "step": 1879 + }, + { + "epoch": 2.76, + "grad_norm": 4.4062700271606445, + "learning_rate": 0.00016472876641216786, + "loss": 3.226, + "step": 1880 + }, + { + "epoch": 2.76, + "grad_norm": 3.8181326389312744, + "learning_rate": 0.00016469300734000403, + "loss": 3.1934, + "step": 1881 + }, + { + "epoch": 2.76, + "grad_norm": 2.7566487789154053, + "learning_rate": 0.00016465723403581218, + "loss": 3.2031, + "step": 1882 + }, + { + "epoch": 2.77, + "grad_norm": 2.438339948654175, + "learning_rate": 0.00016462144650746213, + "loss": 3.2028, + "step": 1883 + }, + { + "epoch": 2.77, + "grad_norm": 3.6031992435455322, + "learning_rate": 0.0001645856447628269, + "loss": 3.2054, + "step": 1884 + }, + { + "epoch": 2.77, + "grad_norm": 2.86798095703125, + "learning_rate": 0.00016454982880978266, + "loss": 3.2189, + "step": 1885 + }, + { + "epoch": 2.77, + "grad_norm": 3.7704923152923584, + "learning_rate": 0.0001645139986562087, + "loss": 3.1854, + "step": 1886 + }, + { + "epoch": 2.77, + "grad_norm": 3.522306442260742, + "learning_rate": 0.0001644781543099874, + "loss": 3.2357, + "step": 1887 + }, + { + "epoch": 2.77, + "grad_norm": 2.8147850036621094, + "learning_rate": 0.00016444229577900423, + "loss": 3.2041, + "step": 1888 + }, + { + "epoch": 2.77, + "grad_norm": 4.882560729980469, + "learning_rate": 0.00016440642307114786, + "loss": 3.1773, + "step": 1889 + }, + { + "epoch": 2.78, + "grad_norm": 1.0316693782806396, + "learning_rate": 0.00016437053619431006, + "loss": 3.195, + "step": 1890 + }, + { + "epoch": 2.78, + "grad_norm": 1.5295097827911377, + "learning_rate": 0.0001643346351563857, + "loss": 3.2313, + "step": 1891 + }, + { + "epoch": 2.78, + "grad_norm": 1.8384897708892822, + "learning_rate": 0.00016429871996527276, + "loss": 3.1694, + "step": 1892 + }, + { + "epoch": 2.78, + "grad_norm": 1.1532273292541504, + "learning_rate": 0.00016426279062887233, + "loss": 3.1872, + "step": 1893 + }, + { + "epoch": 2.78, + "grad_norm": 0.97287917137146, + "learning_rate": 0.0001642268471550886, + "loss": 3.2283, + "step": 1894 + }, + { + "epoch": 2.78, + "grad_norm": 3.258812665939331, + "learning_rate": 0.00016419088955182897, + "loss": 3.2109, + "step": 1895 + }, + { + "epoch": 2.78, + "grad_norm": 3.9376001358032227, + "learning_rate": 0.00016415491782700384, + "loss": 3.213, + "step": 1896 + }, + { + "epoch": 2.79, + "grad_norm": 3.313781261444092, + "learning_rate": 0.00016411893198852668, + "loss": 3.2419, + "step": 1897 + }, + { + "epoch": 2.79, + "grad_norm": 1.2715352773666382, + "learning_rate": 0.00016408293204431425, + "loss": 3.1853, + "step": 1898 + }, + { + "epoch": 2.79, + "grad_norm": 4.407071113586426, + "learning_rate": 0.00016404691800228616, + "loss": 3.1633, + "step": 1899 + }, + { + "epoch": 2.79, + "grad_norm": 2.881059408187866, + "learning_rate": 0.00016401088987036538, + "loss": 3.1711, + "step": 1900 + }, + { + "epoch": 2.79, + "grad_norm": 213.681884765625, + "learning_rate": 0.0001639748476564778, + "loss": 3.219, + "step": 1901 + }, + { + "epoch": 2.79, + "grad_norm": 15.1211576461792, + "learning_rate": 0.00016393879136855248, + "loss": 3.3071, + "step": 1902 + }, + { + "epoch": 2.79, + "grad_norm": 227.103759765625, + "learning_rate": 0.00016390272101452156, + "loss": 3.3632, + "step": 1903 + }, + { + "epoch": 2.8, + "grad_norm": 94.93277740478516, + "learning_rate": 0.00016386663660232026, + "loss": 3.3738, + "step": 1904 + }, + { + "epoch": 2.8, + "grad_norm": 21.8365421295166, + "learning_rate": 0.00016383053813988694, + "loss": 3.4633, + "step": 1905 + }, + { + "epoch": 2.8, + "grad_norm": 461.8182373046875, + "learning_rate": 0.00016379442563516303, + "loss": 3.4446, + "step": 1906 + }, + { + "epoch": 2.8, + "grad_norm": 73.93539428710938, + "learning_rate": 0.00016375829909609296, + "loss": 3.511, + "step": 1907 + }, + { + "epoch": 2.8, + "grad_norm": 26.853099822998047, + "learning_rate": 0.00016372215853062438, + "loss": 3.4775, + "step": 1908 + }, + { + "epoch": 2.8, + "grad_norm": 449.6775207519531, + "learning_rate": 0.000163686003946708, + "loss": 3.5932, + "step": 1909 + }, + { + "epoch": 2.81, + "grad_norm": 1817.4749755859375, + "learning_rate": 0.00016364983535229754, + "loss": 3.6702, + "step": 1910 + }, + { + "epoch": 2.81, + "grad_norm": 431.15777587890625, + "learning_rate": 0.00016361365275534986, + "loss": 4.165, + "step": 1911 + }, + { + "epoch": 2.81, + "grad_norm": 1153.154296875, + "learning_rate": 0.0001635774561638249, + "loss": 4.4609, + "step": 1912 + }, + { + "epoch": 2.81, + "grad_norm": 378.61279296875, + "learning_rate": 0.00016354124558568567, + "loss": 4.9769, + "step": 1913 + }, + { + "epoch": 2.81, + "grad_norm": 228.8455352783203, + "learning_rate": 0.00016350502102889826, + "loss": 5.4494, + "step": 1914 + }, + { + "epoch": 2.81, + "grad_norm": 2911.531005859375, + "learning_rate": 0.0001634687825014318, + "loss": 5.2172, + "step": 1915 + }, + { + "epoch": 2.81, + "grad_norm": 8.003608703613281, + "learning_rate": 0.00016343253001125852, + "loss": 4.3254, + "step": 1916 + }, + { + "epoch": 2.82, + "grad_norm": 122.48676300048828, + "learning_rate": 0.00016339626356635372, + "loss": 3.9329, + "step": 1917 + }, + { + "epoch": 2.82, + "grad_norm": 165.21929931640625, + "learning_rate": 0.00016335998317469586, + "loss": 4.2273, + "step": 1918 + }, + { + "epoch": 2.82, + "grad_norm": 335.0215148925781, + "learning_rate": 0.00016332368884426626, + "loss": 4.668, + "step": 1919 + }, + { + "epoch": 2.82, + "grad_norm": 3772.19091796875, + "learning_rate": 0.00016328738058304956, + "loss": 5.2298, + "step": 1920 + }, + { + "epoch": 2.82, + "grad_norm": 34.06552505493164, + "learning_rate": 0.0001632510583990332, + "loss": 7.4421, + "step": 1921 + }, + { + "epoch": 2.82, + "grad_norm": 3216.72998046875, + "learning_rate": 0.00016321472230020794, + "loss": 7.2781, + "step": 1922 + }, + { + "epoch": 2.82, + "grad_norm": 9.185710906982422, + "learning_rate": 0.0001631783722945674, + "loss": 6.4889, + "step": 1923 + }, + { + "epoch": 2.83, + "grad_norm": 25.194772720336914, + "learning_rate": 0.00016314200839010834, + "loss": 6.3019, + "step": 1924 + }, + { + "epoch": 2.83, + "grad_norm": 35.877498626708984, + "learning_rate": 0.0001631056305948306, + "loss": 6.1576, + "step": 1925 + }, + { + "epoch": 2.83, + "grad_norm": 35.11639404296875, + "learning_rate": 0.00016306923891673707, + "loss": 5.7687, + "step": 1926 + }, + { + "epoch": 2.83, + "grad_norm": 7.56588077545166, + "learning_rate": 0.0001630328333638336, + "loss": 5.4341, + "step": 1927 + }, + { + "epoch": 2.83, + "grad_norm": 11.163580894470215, + "learning_rate": 0.00016299641394412927, + "loss": 5.2751, + "step": 1928 + }, + { + "epoch": 2.83, + "grad_norm": 9.635797500610352, + "learning_rate": 0.00016295998066563603, + "loss": 5.1351, + "step": 1929 + }, + { + "epoch": 2.83, + "grad_norm": 7.929530143737793, + "learning_rate": 0.000162923533536369, + "loss": 4.9651, + "step": 1930 + }, + { + "epoch": 2.84, + "grad_norm": 12.908093452453613, + "learning_rate": 0.00016288707256434623, + "loss": 4.9926, + "step": 1931 + }, + { + "epoch": 2.84, + "grad_norm": 17.120487213134766, + "learning_rate": 0.000162850597757589, + "loss": 4.9808, + "step": 1932 + }, + { + "epoch": 2.84, + "grad_norm": 12.659987449645996, + "learning_rate": 0.00016281410912412144, + "loss": 4.8239, + "step": 1933 + }, + { + "epoch": 2.84, + "grad_norm": 6.449459552764893, + "learning_rate": 0.00016277760667197087, + "loss": 4.7229, + "step": 1934 + }, + { + "epoch": 2.84, + "grad_norm": 11.50796127319336, + "learning_rate": 0.00016274109040916753, + "loss": 4.5966, + "step": 1935 + }, + { + "epoch": 2.84, + "grad_norm": 10.945295333862305, + "learning_rate": 0.00016270456034374474, + "loss": 4.5114, + "step": 1936 + }, + { + "epoch": 2.85, + "grad_norm": 6.565453052520752, + "learning_rate": 0.00016266801648373894, + "loss": 4.4361, + "step": 1937 + }, + { + "epoch": 2.85, + "grad_norm": 6.896590709686279, + "learning_rate": 0.0001626314588371895, + "loss": 4.3267, + "step": 1938 + }, + { + "epoch": 2.85, + "grad_norm": 6.761734485626221, + "learning_rate": 0.00016259488741213884, + "loss": 4.336, + "step": 1939 + }, + { + "epoch": 2.85, + "grad_norm": 7.6577301025390625, + "learning_rate": 0.00016255830221663243, + "loss": 4.2871, + "step": 1940 + }, + { + "epoch": 2.85, + "grad_norm": 5.5717620849609375, + "learning_rate": 0.00016252170325871878, + "loss": 4.2744, + "step": 1941 + }, + { + "epoch": 2.85, + "grad_norm": 6.425532341003418, + "learning_rate": 0.00016248509054644943, + "loss": 4.2368, + "step": 1942 + }, + { + "epoch": 2.85, + "grad_norm": 6.052181720733643, + "learning_rate": 0.00016244846408787895, + "loss": 4.1986, + "step": 1943 + }, + { + "epoch": 2.86, + "grad_norm": 4.308633804321289, + "learning_rate": 0.00016241182389106483, + "loss": 4.2131, + "step": 1944 + }, + { + "epoch": 2.86, + "grad_norm": 3.014502763748169, + "learning_rate": 0.00016237516996406775, + "loss": 4.1368, + "step": 1945 + }, + { + "epoch": 2.86, + "grad_norm": 3.824622869491577, + "learning_rate": 0.00016233850231495126, + "loss": 4.1542, + "step": 1946 + }, + { + "epoch": 2.86, + "grad_norm": 4.700507640838623, + "learning_rate": 0.00016230182095178206, + "loss": 4.032, + "step": 1947 + }, + { + "epoch": 2.86, + "grad_norm": 2.4007976055145264, + "learning_rate": 0.0001622651258826298, + "loss": 4.043, + "step": 1948 + }, + { + "epoch": 2.86, + "grad_norm": 2.202371835708618, + "learning_rate": 0.0001622284171155671, + "loss": 4.0561, + "step": 1949 + }, + { + "epoch": 2.86, + "grad_norm": 3.510922431945801, + "learning_rate": 0.00016219169465866965, + "loss": 4.0186, + "step": 1950 + }, + { + "epoch": 2.87, + "grad_norm": 3.223702907562256, + "learning_rate": 0.0001621549585200162, + "loss": 4.0432, + "step": 1951 + }, + { + "epoch": 2.87, + "grad_norm": 2.4226481914520264, + "learning_rate": 0.00016211820870768842, + "loss": 4.0018, + "step": 1952 + }, + { + "epoch": 2.87, + "grad_norm": 4.14686393737793, + "learning_rate": 0.000162081445229771, + "loss": 3.9202, + "step": 1953 + }, + { + "epoch": 2.87, + "grad_norm": 4.985760688781738, + "learning_rate": 0.0001620446680943517, + "loss": 4.0172, + "step": 1954 + }, + { + "epoch": 2.87, + "grad_norm": 4.1533427238464355, + "learning_rate": 0.00016200787730952124, + "loss": 3.9214, + "step": 1955 + }, + { + "epoch": 2.87, + "grad_norm": 2.108228921890259, + "learning_rate": 0.00016197107288337328, + "loss": 3.9857, + "step": 1956 + }, + { + "epoch": 2.88, + "grad_norm": 2.6513140201568604, + "learning_rate": 0.00016193425482400464, + "loss": 3.9326, + "step": 1957 + }, + { + "epoch": 2.88, + "grad_norm": 3.7441959381103516, + "learning_rate": 0.00016189742313951498, + "loss": 3.8733, + "step": 1958 + }, + { + "epoch": 2.88, + "grad_norm": 2.4456493854522705, + "learning_rate": 0.00016186057783800704, + "loss": 3.9193, + "step": 1959 + }, + { + "epoch": 2.88, + "grad_norm": 2.1966073513031006, + "learning_rate": 0.00016182371892758658, + "loss": 3.9049, + "step": 1960 + }, + { + "epoch": 2.88, + "grad_norm": 3.958984613418579, + "learning_rate": 0.00016178684641636226, + "loss": 3.8627, + "step": 1961 + }, + { + "epoch": 2.88, + "grad_norm": 2.8413755893707275, + "learning_rate": 0.00016174996031244581, + "loss": 3.9005, + "step": 1962 + }, + { + "epoch": 2.88, + "grad_norm": 2.8592469692230225, + "learning_rate": 0.00016171306062395192, + "loss": 3.8878, + "step": 1963 + }, + { + "epoch": 2.89, + "grad_norm": 5.321715831756592, + "learning_rate": 0.00016167614735899827, + "loss": 3.8257, + "step": 1964 + }, + { + "epoch": 2.89, + "grad_norm": 2.1376495361328125, + "learning_rate": 0.00016163922052570554, + "loss": 3.8189, + "step": 1965 + }, + { + "epoch": 2.89, + "grad_norm": 6.535236835479736, + "learning_rate": 0.0001616022801321974, + "loss": 3.8569, + "step": 1966 + }, + { + "epoch": 2.89, + "grad_norm": 4.882155895233154, + "learning_rate": 0.00016156532618660045, + "loss": 3.8699, + "step": 1967 + }, + { + "epoch": 2.89, + "grad_norm": 5.889296531677246, + "learning_rate": 0.00016152835869704432, + "loss": 3.8201, + "step": 1968 + }, + { + "epoch": 2.89, + "grad_norm": 5.7923102378845215, + "learning_rate": 0.00016149137767166165, + "loss": 3.8467, + "step": 1969 + }, + { + "epoch": 2.89, + "grad_norm": 4.793107509613037, + "learning_rate": 0.000161454383118588, + "loss": 3.789, + "step": 1970 + }, + { + "epoch": 2.9, + "grad_norm": 5.585588455200195, + "learning_rate": 0.00016141737504596186, + "loss": 3.8359, + "step": 1971 + }, + { + "epoch": 2.9, + "grad_norm": 1.328306794166565, + "learning_rate": 0.00016138035346192483, + "loss": 3.781, + "step": 1972 + }, + { + "epoch": 2.9, + "grad_norm": 1.4749982357025146, + "learning_rate": 0.00016134331837462142, + "loss": 3.7788, + "step": 1973 + }, + { + "epoch": 2.9, + "grad_norm": 4.906754016876221, + "learning_rate": 0.00016130626979219906, + "loss": 3.8094, + "step": 1974 + }, + { + "epoch": 2.9, + "grad_norm": 2.958904266357422, + "learning_rate": 0.00016126920772280818, + "loss": 3.8173, + "step": 1975 + }, + { + "epoch": 2.9, + "grad_norm": 8.751588821411133, + "learning_rate": 0.00016123213217460224, + "loss": 3.7913, + "step": 1976 + }, + { + "epoch": 2.9, + "grad_norm": 10.918651580810547, + "learning_rate": 0.00016119504315573755, + "loss": 3.7906, + "step": 1977 + }, + { + "epoch": 2.91, + "grad_norm": 2.1029868125915527, + "learning_rate": 0.0001611579406743735, + "loss": 3.8032, + "step": 1978 + }, + { + "epoch": 2.91, + "grad_norm": 9.132828712463379, + "learning_rate": 0.00016112082473867235, + "loss": 3.8242, + "step": 1979 + }, + { + "epoch": 2.91, + "grad_norm": 10.694255828857422, + "learning_rate": 0.0001610836953567994, + "loss": 3.8359, + "step": 1980 + }, + { + "epoch": 2.91, + "grad_norm": 6.331207752227783, + "learning_rate": 0.0001610465525369228, + "loss": 3.7975, + "step": 1981 + }, + { + "epoch": 2.91, + "grad_norm": 7.4574055671691895, + "learning_rate": 0.00016100939628721375, + "loss": 3.7936, + "step": 1982 + }, + { + "epoch": 2.91, + "grad_norm": 12.698988914489746, + "learning_rate": 0.00016097222661584643, + "loss": 3.8415, + "step": 1983 + }, + { + "epoch": 2.92, + "grad_norm": 7.2185750007629395, + "learning_rate": 0.00016093504353099783, + "loss": 3.8274, + "step": 1984 + }, + { + "epoch": 2.92, + "grad_norm": 5.854779243469238, + "learning_rate": 0.00016089784704084804, + "loss": 3.8396, + "step": 1985 + }, + { + "epoch": 2.92, + "grad_norm": 8.324865341186523, + "learning_rate": 0.00016086063715358003, + "loss": 3.8204, + "step": 1986 + }, + { + "epoch": 2.92, + "grad_norm": 3.311805009841919, + "learning_rate": 0.0001608234138773797, + "loss": 3.7652, + "step": 1987 + }, + { + "epoch": 2.92, + "grad_norm": 11.10246467590332, + "learning_rate": 0.00016078617722043595, + "loss": 3.8391, + "step": 1988 + }, + { + "epoch": 2.92, + "grad_norm": 15.776145935058594, + "learning_rate": 0.00016074892719094056, + "loss": 3.799, + "step": 1989 + }, + { + "epoch": 2.92, + "grad_norm": 7.907688617706299, + "learning_rate": 0.00016071166379708833, + "loss": 3.7895, + "step": 1990 + }, + { + "epoch": 2.93, + "grad_norm": 5.6883649826049805, + "learning_rate": 0.00016067438704707695, + "loss": 3.794, + "step": 1991 + }, + { + "epoch": 2.93, + "grad_norm": 8.36189079284668, + "learning_rate": 0.00016063709694910701, + "loss": 3.8036, + "step": 1992 + }, + { + "epoch": 2.93, + "grad_norm": 4.989264488220215, + "learning_rate": 0.00016059979351138213, + "loss": 3.7555, + "step": 1993 + }, + { + "epoch": 2.93, + "grad_norm": 7.9405670166015625, + "learning_rate": 0.0001605624767421088, + "loss": 3.7995, + "step": 1994 + }, + { + "epoch": 2.93, + "grad_norm": 11.919158935546875, + "learning_rate": 0.00016052514664949646, + "loss": 3.8064, + "step": 1995 + }, + { + "epoch": 2.93, + "grad_norm": 15.537467002868652, + "learning_rate": 0.00016048780324175746, + "loss": 3.8284, + "step": 1996 + }, + { + "epoch": 2.93, + "grad_norm": 3.631410837173462, + "learning_rate": 0.00016045044652710712, + "loss": 3.838, + "step": 1997 + }, + { + "epoch": 2.94, + "grad_norm": 2.047104597091675, + "learning_rate": 0.0001604130765137637, + "loss": 3.7354, + "step": 1998 + }, + { + "epoch": 2.94, + "grad_norm": 8.390944480895996, + "learning_rate": 0.00016037569320994827, + "loss": 3.8143, + "step": 1999 + }, + { + "epoch": 2.94, + "grad_norm": 8.165762901306152, + "learning_rate": 0.00016033829662388498, + "loss": 3.7676, + "step": 2000 + }, + { + "epoch": 2.94, + "grad_norm": 2.4152541160583496, + "learning_rate": 0.0001603008867638008, + "loss": 3.7495, + "step": 2001 + }, + { + "epoch": 2.94, + "grad_norm": 5.799648761749268, + "learning_rate": 0.00016026346363792567, + "loss": 3.7996, + "step": 2002 + }, + { + "epoch": 2.94, + "grad_norm": 3.751105785369873, + "learning_rate": 0.00016022602725449238, + "loss": 3.7672, + "step": 2003 + }, + { + "epoch": 2.94, + "grad_norm": 7.3117780685424805, + "learning_rate": 0.0001601885776217367, + "loss": 3.7477, + "step": 2004 + }, + { + "epoch": 2.95, + "grad_norm": 7.469902992248535, + "learning_rate": 0.00016015111474789733, + "loss": 3.7555, + "step": 2005 + }, + { + "epoch": 2.95, + "grad_norm": 2.65592622756958, + "learning_rate": 0.00016011363864121582, + "loss": 3.7347, + "step": 2006 + }, + { + "epoch": 2.95, + "grad_norm": 3.7348546981811523, + "learning_rate": 0.0001600761493099367, + "loss": 3.7731, + "step": 2007 + }, + { + "epoch": 2.95, + "grad_norm": 2.6096608638763428, + "learning_rate": 0.0001600386467623073, + "loss": 3.7667, + "step": 2008 + }, + { + "epoch": 2.95, + "grad_norm": 1.8336734771728516, + "learning_rate": 0.000160001131006578, + "loss": 3.7475, + "step": 2009 + }, + { + "epoch": 2.95, + "grad_norm": 9.613265991210938, + "learning_rate": 0.000159963602051002, + "loss": 3.7442, + "step": 2010 + }, + { + "epoch": 2.96, + "grad_norm": 4.934391021728516, + "learning_rate": 0.00015992605990383538, + "loss": 3.7025, + "step": 2011 + }, + { + "epoch": 2.96, + "grad_norm": 2.538320779800415, + "learning_rate": 0.0001598885045733372, + "loss": 3.7681, + "step": 2012 + }, + { + "epoch": 2.96, + "grad_norm": 2.606849193572998, + "learning_rate": 0.00015985093606776937, + "loss": 3.7237, + "step": 2013 + }, + { + "epoch": 2.96, + "grad_norm": 3.6826469898223877, + "learning_rate": 0.0001598133543953967, + "loss": 3.7382, + "step": 2014 + }, + { + "epoch": 2.96, + "grad_norm": 1.016801118850708, + "learning_rate": 0.0001597757595644869, + "loss": 3.7059, + "step": 2015 + }, + { + "epoch": 2.96, + "grad_norm": 6.855797290802002, + "learning_rate": 0.00015973815158331064, + "loss": 3.7315, + "step": 2016 + }, + { + "epoch": 2.96, + "grad_norm": 4.787756443023682, + "learning_rate": 0.00015970053046014137, + "loss": 3.7273, + "step": 2017 + }, + { + "epoch": 2.97, + "grad_norm": 5.884622573852539, + "learning_rate": 0.00015966289620325546, + "loss": 3.7079, + "step": 2018 + }, + { + "epoch": 2.97, + "grad_norm": 7.18374490737915, + "learning_rate": 0.00015962524882093227, + "loss": 3.7161, + "step": 2019 + }, + { + "epoch": 2.97, + "grad_norm": 1.6866484880447388, + "learning_rate": 0.00015958758832145388, + "loss": 3.6481, + "step": 2020 + }, + { + "epoch": 2.97, + "grad_norm": 10.191508293151855, + "learning_rate": 0.00015954991471310543, + "loss": 3.7034, + "step": 2021 + }, + { + "epoch": 2.97, + "grad_norm": 14.30430793762207, + "learning_rate": 0.0001595122280041748, + "loss": 3.7654, + "step": 2022 + }, + { + "epoch": 2.97, + "grad_norm": 5.003686904907227, + "learning_rate": 0.00015947452820295292, + "loss": 3.7148, + "step": 2023 + }, + { + "epoch": 2.97, + "grad_norm": 6.288287162780762, + "learning_rate": 0.00015943681531773333, + "loss": 3.7528, + "step": 2024 + }, + { + "epoch": 2.98, + "grad_norm": 8.045740127563477, + "learning_rate": 0.00015939908935681273, + "loss": 3.6763, + "step": 2025 + }, + { + "epoch": 2.98, + "grad_norm": 3.2493703365325928, + "learning_rate": 0.0001593613503284906, + "loss": 3.6939, + "step": 2026 + }, + { + "epoch": 2.98, + "grad_norm": 10.84262752532959, + "learning_rate": 0.0001593235982410691, + "loss": 3.7516, + "step": 2027 + }, + { + "epoch": 2.98, + "grad_norm": 15.299811363220215, + "learning_rate": 0.00015928583310285366, + "loss": 3.7319, + "step": 2028 + }, + { + "epoch": 2.98, + "eval_loss": 3.5392327308654785, + "eval_runtime": 1153.2794, + "eval_samples_per_second": 14.864, + "eval_steps_per_second": 0.93, + "step": 2028 + }, + { + "epoch": 2.98, + "grad_norm": 8.231132507324219, + "learning_rate": 0.0001592480549221522, + "loss": 3.7177, + "step": 2029 + }, + { + "epoch": 2.98, + "grad_norm": 5.159106254577637, + "learning_rate": 0.00015921026370727573, + "loss": 3.7017, + "step": 2030 + }, + { + "epoch": 2.98, + "grad_norm": 22.314449310302734, + "learning_rate": 0.00015917245946653807, + "loss": 3.6842, + "step": 2031 + }, + { + "epoch": 2.99, + "grad_norm": 8.433747291564941, + "learning_rate": 0.00015913464220825588, + "loss": 3.6786, + "step": 2032 + }, + { + "epoch": 2.99, + "grad_norm": 3.3623647689819336, + "learning_rate": 0.00015909681194074867, + "loss": 3.678, + "step": 2033 + }, + { + "epoch": 2.99, + "grad_norm": 10.62174129486084, + "learning_rate": 0.00015905896867233894, + "loss": 3.7098, + "step": 2034 + }, + { + "epoch": 2.99, + "grad_norm": 15.299424171447754, + "learning_rate": 0.00015902111241135187, + "loss": 3.6926, + "step": 2035 + }, + { + "epoch": 2.99, + "grad_norm": 9.536619186401367, + "learning_rate": 0.00015898324316611563, + "loss": 3.673, + "step": 2036 + }, + { + "epoch": 2.99, + "grad_norm": 5.059504508972168, + "learning_rate": 0.00015894536094496117, + "loss": 3.6625, + "step": 2037 + }, + { + "epoch": 3.0, + "grad_norm": 7.704652786254883, + "learning_rate": 0.00015890746575622231, + "loss": 3.6632, + "step": 2038 + }, + { + "epoch": 3.0, + "grad_norm": 4.4129557609558105, + "learning_rate": 0.00015886955760823578, + "loss": 3.6315, + "step": 2039 + }, + { + "epoch": 3.0, + "grad_norm": 7.341033935546875, + "learning_rate": 0.0001588316365093411, + "loss": 3.6763, + "step": 2040 + }, + { + "epoch": 3.0, + "grad_norm": 9.556234359741211, + "learning_rate": 0.00015879370246788067, + "loss": 3.6774, + "step": 2041 + }, + { + "epoch": 3.0, + "grad_norm": 3.8044204711914062, + "learning_rate": 0.00015875575549219973, + "loss": 3.6249, + "step": 2042 + }, + { + "epoch": 3.0, + "grad_norm": 7.746762275695801, + "learning_rate": 0.0001587177955906463, + "loss": 3.6366, + "step": 2043 + }, + { + "epoch": 3.0, + "grad_norm": 9.851776123046875, + "learning_rate": 0.0001586798227715714, + "loss": 3.6885, + "step": 2044 + }, + { + "epoch": 3.01, + "grad_norm": 5.628739833831787, + "learning_rate": 0.00015864183704332872, + "loss": 3.6248, + "step": 2045 + }, + { + "epoch": 3.01, + "grad_norm": 6.932154178619385, + "learning_rate": 0.00015860383841427485, + "loss": 3.6499, + "step": 2046 + }, + { + "epoch": 3.01, + "grad_norm": 10.925968170166016, + "learning_rate": 0.00015856582689276934, + "loss": 3.6793, + "step": 2047 + }, + { + "epoch": 3.01, + "grad_norm": 5.439330577850342, + "learning_rate": 0.00015852780248717442, + "loss": 3.654, + "step": 2048 + }, + { + "epoch": 3.01, + "grad_norm": 6.066278457641602, + "learning_rate": 0.0001584897652058551, + "loss": 3.6738, + "step": 2049 + }, + { + "epoch": 3.01, + "grad_norm": 8.065271377563477, + "learning_rate": 0.00015845171505717947, + "loss": 3.659, + "step": 2050 + }, + { + "epoch": 3.01, + "grad_norm": 3.8582706451416016, + "learning_rate": 0.00015841365204951826, + "loss": 3.6458, + "step": 2051 + }, + { + "epoch": 3.02, + "grad_norm": 9.716731071472168, + "learning_rate": 0.00015837557619124507, + "loss": 3.6313, + "step": 2052 + }, + { + "epoch": 3.0, + "grad_norm": 14.822004318237305, + "learning_rate": 0.00015833748749073632, + "loss": 3.6512, + "step": 2053 + }, + { + "epoch": 3.0, + "grad_norm": 8.674092292785645, + "learning_rate": 0.00015829938595637128, + "loss": 3.6251, + "step": 2054 + }, + { + "epoch": 3.0, + "grad_norm": 4.23799467086792, + "learning_rate": 0.00015826127159653203, + "loss": 3.6496, + "step": 2055 + }, + { + "epoch": 3.01, + "grad_norm": 9.582878112792969, + "learning_rate": 0.00015822314441960348, + "loss": 3.5982, + "step": 2056 + }, + { + "epoch": 3.01, + "grad_norm": 5.006598949432373, + "learning_rate": 0.00015818500443397337, + "loss": 3.6347, + "step": 2057 + }, + { + "epoch": 3.01, + "grad_norm": 4.559854984283447, + "learning_rate": 0.0001581468516480322, + "loss": 3.6414, + "step": 2058 + }, + { + "epoch": 3.01, + "grad_norm": 5.798408508300781, + "learning_rate": 0.00015810868607017337, + "loss": 3.6176, + "step": 2059 + }, + { + "epoch": 3.01, + "grad_norm": 1.9246889352798462, + "learning_rate": 0.00015807050770879304, + "loss": 3.6134, + "step": 2060 + }, + { + "epoch": 3.01, + "grad_norm": 1.124200701713562, + "learning_rate": 0.00015803231657229017, + "loss": 3.6295, + "step": 2061 + }, + { + "epoch": 3.01, + "grad_norm": 7.120904445648193, + "learning_rate": 0.00015799411266906657, + "loss": 3.6101, + "step": 2062 + }, + { + "epoch": 3.02, + "grad_norm": 5.709414482116699, + "learning_rate": 0.00015795589600752682, + "loss": 3.6255, + "step": 2063 + }, + { + "epoch": 3.02, + "grad_norm": 3.2987709045410156, + "learning_rate": 0.0001579176665960784, + "loss": 3.6194, + "step": 2064 + }, + { + "epoch": 3.02, + "grad_norm": 4.208385944366455, + "learning_rate": 0.00015787942444313145, + "loss": 3.6357, + "step": 2065 + }, + { + "epoch": 3.02, + "grad_norm": 2.485320568084717, + "learning_rate": 0.00015784116955709897, + "loss": 3.6233, + "step": 2066 + }, + { + "epoch": 3.02, + "grad_norm": 0.9779897332191467, + "learning_rate": 0.0001578029019463969, + "loss": 3.6382, + "step": 2067 + }, + { + "epoch": 3.02, + "grad_norm": 5.236642837524414, + "learning_rate": 0.0001577646216194437, + "loss": 3.5985, + "step": 2068 + }, + { + "epoch": 3.03, + "grad_norm": 4.276678562164307, + "learning_rate": 0.00015772632858466093, + "loss": 3.614, + "step": 2069 + }, + { + "epoch": 3.03, + "grad_norm": 4.390791893005371, + "learning_rate": 0.00015768802285047272, + "loss": 3.6166, + "step": 2070 + }, + { + "epoch": 3.03, + "grad_norm": 4.886763572692871, + "learning_rate": 0.00015764970442530609, + "loss": 3.5902, + "step": 2071 + }, + { + "epoch": 3.03, + "grad_norm": 3.569462776184082, + "learning_rate": 0.00015761137331759084, + "loss": 3.5779, + "step": 2072 + }, + { + "epoch": 3.03, + "grad_norm": 2.935269832611084, + "learning_rate": 0.00015757302953575954, + "loss": 3.6369, + "step": 2073 + }, + { + "epoch": 3.03, + "grad_norm": 4.911141872406006, + "learning_rate": 0.0001575346730882476, + "loss": 3.6034, + "step": 2074 + }, + { + "epoch": 3.03, + "grad_norm": 4.09962797164917, + "learning_rate": 0.00015749630398349316, + "loss": 3.6431, + "step": 2075 + }, + { + "epoch": 3.04, + "grad_norm": 4.247868537902832, + "learning_rate": 0.00015745792222993715, + "loss": 3.6258, + "step": 2076 + }, + { + "epoch": 3.04, + "grad_norm": 3.7722268104553223, + "learning_rate": 0.00015741952783602332, + "loss": 3.597, + "step": 2077 + }, + { + "epoch": 3.04, + "grad_norm": 5.005308151245117, + "learning_rate": 0.00015738112081019817, + "loss": 3.6146, + "step": 2078 + }, + { + "epoch": 3.04, + "grad_norm": 4.696826934814453, + "learning_rate": 0.000157342701160911, + "loss": 3.6477, + "step": 2079 + }, + { + "epoch": 3.04, + "grad_norm": 3.856837034225464, + "learning_rate": 0.00015730426889661386, + "loss": 3.6594, + "step": 2080 + }, + { + "epoch": 3.04, + "grad_norm": 3.4349594116210938, + "learning_rate": 0.0001572658240257616, + "loss": 3.5798, + "step": 2081 + }, + { + "epoch": 3.04, + "grad_norm": 4.7815842628479, + "learning_rate": 0.00015722736655681186, + "loss": 3.5797, + "step": 2082 + }, + { + "epoch": 3.05, + "grad_norm": 3.9778714179992676, + "learning_rate": 0.00015718889649822494, + "loss": 3.6173, + "step": 2083 + }, + { + "epoch": 3.05, + "grad_norm": 3.863435745239258, + "learning_rate": 0.0001571504138584641, + "loss": 3.5669, + "step": 2084 + }, + { + "epoch": 3.05, + "grad_norm": 3.9693548679351807, + "learning_rate": 0.0001571119186459952, + "loss": 3.6072, + "step": 2085 + }, + { + "epoch": 3.05, + "grad_norm": 4.673881530761719, + "learning_rate": 0.00015707341086928698, + "loss": 3.5794, + "step": 2086 + }, + { + "epoch": 3.05, + "grad_norm": 3.680610179901123, + "learning_rate": 0.0001570348905368108, + "loss": 3.5927, + "step": 2087 + }, + { + "epoch": 3.05, + "grad_norm": 4.484524726867676, + "learning_rate": 0.00015699635765704096, + "loss": 3.5588, + "step": 2088 + }, + { + "epoch": 3.05, + "grad_norm": 4.283873558044434, + "learning_rate": 0.00015695781223845441, + "loss": 3.6385, + "step": 2089 + }, + { + "epoch": 3.06, + "grad_norm": 3.3641796112060547, + "learning_rate": 0.00015691925428953089, + "loss": 3.6163, + "step": 2090 + }, + { + "epoch": 3.06, + "grad_norm": 2.774048089981079, + "learning_rate": 0.00015688068381875287, + "loss": 3.6696, + "step": 2091 + }, + { + "epoch": 3.06, + "grad_norm": 5.591458320617676, + "learning_rate": 0.0001568421008346056, + "loss": 3.54, + "step": 2092 + }, + { + "epoch": 3.06, + "grad_norm": 5.354968547821045, + "learning_rate": 0.0001568035053455771, + "loss": 3.5762, + "step": 2093 + }, + { + "epoch": 3.06, + "grad_norm": 2.219583749771118, + "learning_rate": 0.0001567648973601581, + "loss": 3.5527, + "step": 2094 + }, + { + "epoch": 3.06, + "grad_norm": 1.655458688735962, + "learning_rate": 0.0001567262768868421, + "loss": 3.5702, + "step": 2095 + }, + { + "epoch": 3.07, + "grad_norm": 5.801334381103516, + "learning_rate": 0.00015668764393412537, + "loss": 3.5317, + "step": 2096 + }, + { + "epoch": 3.07, + "grad_norm": 5.561006546020508, + "learning_rate": 0.00015664899851050684, + "loss": 3.5875, + "step": 2097 + }, + { + "epoch": 3.07, + "grad_norm": 2.8261959552764893, + "learning_rate": 0.0001566103406244883, + "loss": 3.5284, + "step": 2098 + }, + { + "epoch": 3.07, + "grad_norm": 2.3607285022735596, + "learning_rate": 0.00015657167028457423, + "loss": 3.5596, + "step": 2099 + }, + { + "epoch": 3.07, + "grad_norm": 6.529599666595459, + "learning_rate": 0.00015653298749927178, + "loss": 3.5603, + "step": 2100 + }, + { + "epoch": 3.07, + "grad_norm": 5.787454605102539, + "learning_rate": 0.00015649429227709098, + "loss": 3.5419, + "step": 2101 + }, + { + "epoch": 3.07, + "grad_norm": 1.2445347309112549, + "learning_rate": 0.00015645558462654447, + "loss": 3.5599, + "step": 2102 + }, + { + "epoch": 3.08, + "grad_norm": 1.686257243156433, + "learning_rate": 0.0001564168645561477, + "loss": 3.5051, + "step": 2103 + }, + { + "epoch": 3.08, + "grad_norm": 5.125936985015869, + "learning_rate": 0.00015637813207441882, + "loss": 3.4685, + "step": 2104 + }, + { + "epoch": 3.08, + "grad_norm": 4.658946514129639, + "learning_rate": 0.0001563393871898787, + "loss": 3.5291, + "step": 2105 + }, + { + "epoch": 3.08, + "grad_norm": 4.34171199798584, + "learning_rate": 0.00015630062991105098, + "loss": 3.5044, + "step": 2106 + }, + { + "epoch": 3.08, + "grad_norm": 5.2388458251953125, + "learning_rate": 0.00015626186024646197, + "loss": 3.4543, + "step": 2107 + }, + { + "epoch": 3.08, + "grad_norm": 3.39254093170166, + "learning_rate": 0.00015622307820464077, + "loss": 3.4678, + "step": 2108 + }, + { + "epoch": 3.08, + "grad_norm": 2.4865806102752686, + "learning_rate": 0.00015618428379411917, + "loss": 3.4326, + "step": 2109 + }, + { + "epoch": 3.09, + "grad_norm": 7.326157093048096, + "learning_rate": 0.00015614547702343166, + "loss": 3.4709, + "step": 2110 + }, + { + "epoch": 3.09, + "grad_norm": 7.048543930053711, + "learning_rate": 0.00015610665790111546, + "loss": 3.4498, + "step": 2111 + }, + { + "epoch": 3.09, + "grad_norm": 10.954419136047363, + "learning_rate": 0.00015606782643571055, + "loss": 3.3944, + "step": 2112 + }, + { + "epoch": 3.09, + "grad_norm": 11.594827651977539, + "learning_rate": 0.00015602898263575952, + "loss": 3.4224, + "step": 2113 + }, + { + "epoch": 3.09, + "grad_norm": 6.556940078735352, + "learning_rate": 0.0001559901265098079, + "loss": 3.4079, + "step": 2114 + }, + { + "epoch": 3.09, + "grad_norm": 3.0502734184265137, + "learning_rate": 0.00015595125806640357, + "loss": 3.409, + "step": 2115 + }, + { + "epoch": 3.09, + "grad_norm": 6.4218220710754395, + "learning_rate": 0.0001559123773140975, + "loss": 3.4382, + "step": 2116 + }, + { + "epoch": 3.1, + "grad_norm": 6.8841753005981445, + "learning_rate": 0.00015587348426144312, + "loss": 3.463, + "step": 2117 + }, + { + "epoch": 3.1, + "grad_norm": 35.60420227050781, + "learning_rate": 0.00015583457891699666, + "loss": 3.3878, + "step": 2118 + }, + { + "epoch": 3.1, + "grad_norm": 2.3054988384246826, + "learning_rate": 0.000155795661289317, + "loss": 3.382, + "step": 2119 + }, + { + "epoch": 3.1, + "grad_norm": 6.13869571685791, + "learning_rate": 0.00015575673138696578, + "loss": 3.4213, + "step": 2120 + }, + { + "epoch": 3.1, + "grad_norm": 5.998043537139893, + "learning_rate": 0.00015571778921850732, + "loss": 3.382, + "step": 2121 + }, + { + "epoch": 3.1, + "grad_norm": 1.6560511589050293, + "learning_rate": 0.00015567883479250863, + "loss": 3.4691, + "step": 2122 + }, + { + "epoch": 3.11, + "grad_norm": 2.120764970779419, + "learning_rate": 0.00015563986811753945, + "loss": 3.3711, + "step": 2123 + }, + { + "epoch": 3.11, + "grad_norm": 1.6065012216567993, + "learning_rate": 0.00015560088920217217, + "loss": 3.4219, + "step": 2124 + }, + { + "epoch": 3.11, + "grad_norm": 6.458485126495361, + "learning_rate": 0.00015556189805498188, + "loss": 3.4184, + "step": 2125 + }, + { + "epoch": 3.11, + "grad_norm": 5.5630202293396, + "learning_rate": 0.0001555228946845464, + "loss": 3.4185, + "step": 2126 + }, + { + "epoch": 3.11, + "grad_norm": 4.821037769317627, + "learning_rate": 0.00015548387909944614, + "loss": 3.4003, + "step": 2127 + }, + { + "epoch": 3.11, + "grad_norm": 13.251823425292969, + "learning_rate": 0.00015544485130826433, + "loss": 3.4081, + "step": 2128 + }, + { + "epoch": 3.11, + "grad_norm": 5.775404930114746, + "learning_rate": 0.00015540581131958682, + "loss": 3.3918, + "step": 2129 + }, + { + "epoch": 3.12, + "grad_norm": 6.066825866699219, + "learning_rate": 0.00015536675914200213, + "loss": 3.3863, + "step": 2130 + }, + { + "epoch": 3.12, + "grad_norm": 11.889984130859375, + "learning_rate": 0.0001553276947841015, + "loss": 3.4055, + "step": 2131 + }, + { + "epoch": 3.12, + "grad_norm": 3.267805576324463, + "learning_rate": 0.00015528861825447876, + "loss": 3.4031, + "step": 2132 + }, + { + "epoch": 3.12, + "grad_norm": 11.77675724029541, + "learning_rate": 0.00015524952956173056, + "loss": 3.4159, + "step": 2133 + }, + { + "epoch": 3.12, + "grad_norm": 16.67833137512207, + "learning_rate": 0.0001552104287144561, + "loss": 3.4706, + "step": 2134 + }, + { + "epoch": 3.12, + "grad_norm": 9.038000106811523, + "learning_rate": 0.00015517131572125736, + "loss": 3.4391, + "step": 2135 + }, + { + "epoch": 3.12, + "grad_norm": 7.349591255187988, + "learning_rate": 0.0001551321905907389, + "loss": 3.4181, + "step": 2136 + }, + { + "epoch": 3.13, + "grad_norm": 11.440853118896484, + "learning_rate": 0.00015509305333150797, + "loss": 3.4324, + "step": 2137 + }, + { + "epoch": 3.13, + "grad_norm": 9.080293655395508, + "learning_rate": 0.00015505390395217453, + "loss": 3.4841, + "step": 2138 + }, + { + "epoch": 3.13, + "grad_norm": 2.2852790355682373, + "learning_rate": 0.0001550147424613512, + "loss": 3.35, + "step": 2139 + }, + { + "epoch": 3.13, + "grad_norm": 17.961963653564453, + "learning_rate": 0.00015497556886765316, + "loss": 3.4265, + "step": 2140 + }, + { + "epoch": 3.13, + "grad_norm": 26.195865631103516, + "learning_rate": 0.00015493638317969844, + "loss": 3.6051, + "step": 2141 + }, + { + "epoch": 3.13, + "grad_norm": 19.95352554321289, + "learning_rate": 0.0001548971854061076, + "loss": 3.6381, + "step": 2142 + }, + { + "epoch": 3.13, + "grad_norm": 5.972691535949707, + "learning_rate": 0.00015485797555550386, + "loss": 3.5607, + "step": 2143 + }, + { + "epoch": 3.14, + "grad_norm": 13.078893661499023, + "learning_rate": 0.0001548187536365131, + "loss": 3.4778, + "step": 2144 + }, + { + "epoch": 3.14, + "grad_norm": 15.685088157653809, + "learning_rate": 0.000154779519657764, + "loss": 3.5011, + "step": 2145 + }, + { + "epoch": 3.14, + "grad_norm": 14.251028060913086, + "learning_rate": 0.00015474027362788763, + "loss": 3.6155, + "step": 2146 + }, + { + "epoch": 3.14, + "grad_norm": 10.396077156066895, + "learning_rate": 0.00015470101555551796, + "loss": 3.5646, + "step": 2147 + }, + { + "epoch": 3.14, + "grad_norm": 5.34552526473999, + "learning_rate": 0.0001546617454492914, + "loss": 3.4643, + "step": 2148 + }, + { + "epoch": 3.14, + "grad_norm": 18.76544189453125, + "learning_rate": 0.0001546224633178472, + "loss": 3.4686, + "step": 2149 + }, + { + "epoch": 3.15, + "grad_norm": 32.796051025390625, + "learning_rate": 0.00015458316916982715, + "loss": 3.6171, + "step": 2150 + }, + { + "epoch": 3.15, + "grad_norm": 31.727258682250977, + "learning_rate": 0.00015454386301387565, + "loss": 3.751, + "step": 2151 + }, + { + "epoch": 3.15, + "grad_norm": 18.45448112487793, + "learning_rate": 0.00015450454485863984, + "loss": 3.6405, + "step": 2152 + }, + { + "epoch": 3.15, + "grad_norm": 8.118375778198242, + "learning_rate": 0.00015446521471276946, + "loss": 3.6001, + "step": 2153 + }, + { + "epoch": 3.15, + "grad_norm": 16.14570426940918, + "learning_rate": 0.0001544258725849168, + "loss": 3.559, + "step": 2154 + }, + { + "epoch": 3.15, + "grad_norm": 16.983943939208984, + "learning_rate": 0.00015438651848373696, + "loss": 3.5108, + "step": 2155 + }, + { + "epoch": 3.15, + "grad_norm": 13.206223487854004, + "learning_rate": 0.00015434715241788752, + "loss": 3.4883, + "step": 2156 + }, + { + "epoch": 3.16, + "grad_norm": 8.664280891418457, + "learning_rate": 0.00015430777439602876, + "loss": 3.4346, + "step": 2157 + }, + { + "epoch": 3.16, + "grad_norm": 5.633321762084961, + "learning_rate": 0.00015426838442682362, + "loss": 3.4285, + "step": 2158 + }, + { + "epoch": 3.16, + "grad_norm": 14.557170867919922, + "learning_rate": 0.00015422898251893756, + "loss": 3.4706, + "step": 2159 + }, + { + "epoch": 3.16, + "grad_norm": 19.554941177368164, + "learning_rate": 0.00015418956868103878, + "loss": 3.4929, + "step": 2160 + }, + { + "epoch": 3.16, + "grad_norm": 15.579254150390625, + "learning_rate": 0.00015415014292179805, + "loss": 3.4144, + "step": 2161 + }, + { + "epoch": 3.16, + "grad_norm": 5.079093933105469, + "learning_rate": 0.0001541107052498888, + "loss": 3.3886, + "step": 2162 + }, + { + "epoch": 3.16, + "grad_norm": 10.48103141784668, + "learning_rate": 0.00015407125567398698, + "loss": 3.4392, + "step": 2163 + }, + { + "epoch": 3.17, + "grad_norm": 14.062845230102539, + "learning_rate": 0.00015403179420277131, + "loss": 3.4527, + "step": 2164 + }, + { + "epoch": 3.17, + "grad_norm": 13.237971305847168, + "learning_rate": 0.00015399232084492304, + "loss": 3.4518, + "step": 2165 + }, + { + "epoch": 3.17, + "grad_norm": 7.261989593505859, + "learning_rate": 0.000153952835609126, + "loss": 3.4716, + "step": 2166 + }, + { + "epoch": 3.17, + "grad_norm": 5.766734600067139, + "learning_rate": 0.0001539133385040667, + "loss": 3.3591, + "step": 2167 + }, + { + "epoch": 3.17, + "grad_norm": 10.160755157470703, + "learning_rate": 0.0001538738295384343, + "loss": 3.3638, + "step": 2168 + }, + { + "epoch": 3.17, + "grad_norm": 7.740986347198486, + "learning_rate": 0.00015383430872092037, + "loss": 3.4132, + "step": 2169 + }, + { + "epoch": 3.18, + "grad_norm": 2.3649628162384033, + "learning_rate": 0.00015379477606021933, + "loss": 3.4459, + "step": 2170 + }, + { + "epoch": 3.18, + "grad_norm": 4.513500213623047, + "learning_rate": 0.00015375523156502805, + "loss": 3.4318, + "step": 2171 + }, + { + "epoch": 3.18, + "grad_norm": 3.214402198791504, + "learning_rate": 0.00015371567524404613, + "loss": 3.3651, + "step": 2172 + }, + { + "epoch": 3.18, + "grad_norm": 3.8382275104522705, + "learning_rate": 0.0001536761071059756, + "loss": 3.4077, + "step": 2173 + }, + { + "epoch": 3.18, + "grad_norm": 1.4553401470184326, + "learning_rate": 0.0001536365271595212, + "loss": 3.3301, + "step": 2174 + }, + { + "epoch": 3.18, + "grad_norm": 4.466412544250488, + "learning_rate": 0.00015359693541339037, + "loss": 3.3988, + "step": 2175 + }, + { + "epoch": 3.18, + "grad_norm": 3.6478259563446045, + "learning_rate": 0.00015355733187629288, + "loss": 3.3779, + "step": 2176 + }, + { + "epoch": 3.19, + "grad_norm": 5.013825416564941, + "learning_rate": 0.0001535177165569413, + "loss": 3.413, + "step": 2177 + }, + { + "epoch": 3.19, + "grad_norm": 4.778167247772217, + "learning_rate": 0.00015347808946405075, + "loss": 3.3425, + "step": 2178 + }, + { + "epoch": 3.19, + "grad_norm": 5.024134159088135, + "learning_rate": 0.00015343845060633893, + "loss": 3.3676, + "step": 2179 + }, + { + "epoch": 3.19, + "grad_norm": 5.648217678070068, + "learning_rate": 0.00015339879999252606, + "loss": 3.3165, + "step": 2180 + }, + { + "epoch": 3.19, + "grad_norm": 2.347975015640259, + "learning_rate": 0.00015335913763133511, + "loss": 3.3588, + "step": 2181 + }, + { + "epoch": 3.19, + "grad_norm": 5.947681427001953, + "learning_rate": 0.00015331946353149144, + "loss": 3.3811, + "step": 2182 + }, + { + "epoch": 3.19, + "grad_norm": 5.794849395751953, + "learning_rate": 0.00015327977770172315, + "loss": 3.3751, + "step": 2183 + }, + { + "epoch": 3.2, + "grad_norm": 2.177245855331421, + "learning_rate": 0.0001532400801507608, + "loss": 3.3472, + "step": 2184 + }, + { + "epoch": 3.2, + "grad_norm": 4.221487045288086, + "learning_rate": 0.00015320037088733764, + "loss": 3.3766, + "step": 2185 + }, + { + "epoch": 3.2, + "grad_norm": 3.649068832397461, + "learning_rate": 0.00015316064992018945, + "loss": 3.3633, + "step": 2186 + }, + { + "epoch": 3.2, + "grad_norm": 3.780611038208008, + "learning_rate": 0.00015312091725805448, + "loss": 3.3371, + "step": 2187 + }, + { + "epoch": 3.2, + "grad_norm": 1.3155494928359985, + "learning_rate": 0.00015308117290967373, + "loss": 3.3083, + "step": 2188 + }, + { + "epoch": 3.2, + "grad_norm": 1.961712121963501, + "learning_rate": 0.0001530414168837907, + "loss": 3.3465, + "step": 2189 + }, + { + "epoch": 3.2, + "grad_norm": 2.569746494293213, + "learning_rate": 0.00015300164918915138, + "loss": 3.3853, + "step": 2190 + }, + { + "epoch": 3.21, + "grad_norm": 1.9574949741363525, + "learning_rate": 0.00015296186983450446, + "loss": 3.3444, + "step": 2191 + }, + { + "epoch": 3.21, + "grad_norm": 2.3811464309692383, + "learning_rate": 0.00015292207882860114, + "loss": 3.316, + "step": 2192 + }, + { + "epoch": 3.21, + "grad_norm": 1.6529579162597656, + "learning_rate": 0.00015288227618019513, + "loss": 3.3269, + "step": 2193 + }, + { + "epoch": 3.21, + "grad_norm": 1.4792758226394653, + "learning_rate": 0.00015284246189804272, + "loss": 3.3406, + "step": 2194 + }, + { + "epoch": 3.21, + "grad_norm": 1.3875869512557983, + "learning_rate": 0.0001528026359909029, + "loss": 3.3492, + "step": 2195 + }, + { + "epoch": 3.21, + "grad_norm": 3.4284157752990723, + "learning_rate": 0.00015276279846753696, + "loss": 3.3176, + "step": 2196 + }, + { + "epoch": 3.22, + "grad_norm": 1.68557608127594, + "learning_rate": 0.000152722949336709, + "loss": 3.3142, + "step": 2197 + }, + { + "epoch": 3.22, + "eval_loss": 3.1609973907470703, + "eval_runtime": 1163.3297, + "eval_samples_per_second": 14.735, + "eval_steps_per_second": 0.921, + "step": 2197 + }, + { + "epoch": 3.22, + "grad_norm": 4.416086673736572, + "learning_rate": 0.0001526830886071855, + "loss": 3.3143, + "step": 2198 + }, + { + "epoch": 3.22, + "grad_norm": 2.9269638061523438, + "learning_rate": 0.0001526432162877356, + "loss": 3.3417, + "step": 2199 + }, + { + "epoch": 3.22, + "grad_norm": 3.496257781982422, + "learning_rate": 0.00015260333238713088, + "loss": 3.3356, + "step": 2200 + }, + { + "epoch": 3.22, + "grad_norm": 3.9582598209381104, + "learning_rate": 0.00015256343691414558, + "loss": 3.3688, + "step": 2201 + }, + { + "epoch": 3.22, + "grad_norm": 3.0187695026397705, + "learning_rate": 0.0001525235298775564, + "loss": 3.3381, + "step": 2202 + }, + { + "epoch": 3.22, + "grad_norm": 1.4246793985366821, + "learning_rate": 0.0001524836112861427, + "loss": 3.3228, + "step": 2203 + }, + { + "epoch": 3.23, + "grad_norm": 2.6875085830688477, + "learning_rate": 0.00015244368114868625, + "loss": 3.2918, + "step": 2204 + }, + { + "epoch": 3.23, + "grad_norm": 2.1740024089813232, + "learning_rate": 0.00015240373947397133, + "loss": 3.2933, + "step": 2205 + }, + { + "epoch": 3.23, + "grad_norm": 5.561516284942627, + "learning_rate": 0.000152363786270785, + "loss": 3.3344, + "step": 2206 + }, + { + "epoch": 3.23, + "grad_norm": 2.453502655029297, + "learning_rate": 0.00015232382154791655, + "loss": 3.3204, + "step": 2207 + }, + { + "epoch": 3.23, + "grad_norm": 3.4560389518737793, + "learning_rate": 0.0001522838453141581, + "loss": 3.3054, + "step": 2208 + }, + { + "epoch": 3.23, + "grad_norm": 1.4101018905639648, + "learning_rate": 0.000152243857578304, + "loss": 3.3298, + "step": 2209 + }, + { + "epoch": 3.23, + "grad_norm": 1.3143606185913086, + "learning_rate": 0.0001522038583491514, + "loss": 3.3294, + "step": 2210 + }, + { + "epoch": 3.24, + "grad_norm": 2.1382877826690674, + "learning_rate": 0.00015216384763549977, + "loss": 3.2964, + "step": 2211 + }, + { + "epoch": 3.24, + "grad_norm": 3.5953447818756104, + "learning_rate": 0.00015212382544615126, + "loss": 3.3138, + "step": 2212 + }, + { + "epoch": 3.24, + "grad_norm": 2.013397216796875, + "learning_rate": 0.0001520837917899105, + "loss": 3.3318, + "step": 2213 + }, + { + "epoch": 3.24, + "grad_norm": 5.66317892074585, + "learning_rate": 0.0001520437466755846, + "loss": 3.3226, + "step": 2214 + }, + { + "epoch": 3.24, + "grad_norm": 5.973227024078369, + "learning_rate": 0.00015200369011198319, + "loss": 3.2917, + "step": 2215 + }, + { + "epoch": 3.24, + "grad_norm": 1.0823651552200317, + "learning_rate": 0.00015196362210791845, + "loss": 3.2837, + "step": 2216 + }, + { + "epoch": 3.24, + "grad_norm": 2.2828619480133057, + "learning_rate": 0.00015192354267220512, + "loss": 3.2994, + "step": 2217 + }, + { + "epoch": 3.25, + "grad_norm": 3.3102993965148926, + "learning_rate": 0.00015188345181366036, + "loss": 3.3674, + "step": 2218 + }, + { + "epoch": 3.25, + "grad_norm": 5.152593612670898, + "learning_rate": 0.00015184334954110394, + "loss": 3.3411, + "step": 2219 + }, + { + "epoch": 3.25, + "grad_norm": 6.140212535858154, + "learning_rate": 0.00015180323586335806, + "loss": 3.3188, + "step": 2220 + }, + { + "epoch": 3.25, + "grad_norm": 2.0337228775024414, + "learning_rate": 0.0001517631107892474, + "loss": 3.2678, + "step": 2221 + }, + { + "epoch": 3.25, + "grad_norm": 4.024622917175293, + "learning_rate": 0.0001517229743275993, + "loss": 3.2708, + "step": 2222 + }, + { + "epoch": 3.25, + "grad_norm": 2.378216505050659, + "learning_rate": 0.00015168282648724351, + "loss": 3.3462, + "step": 2223 + }, + { + "epoch": 3.26, + "grad_norm": 4.922878742218018, + "learning_rate": 0.0001516426672770122, + "loss": 3.291, + "step": 2224 + }, + { + "epoch": 3.26, + "grad_norm": 4.67563533782959, + "learning_rate": 0.00015160249670574026, + "loss": 3.2711, + "step": 2225 + }, + { + "epoch": 3.26, + "grad_norm": 4.26483154296875, + "learning_rate": 0.0001515623147822648, + "loss": 3.2862, + "step": 2226 + }, + { + "epoch": 3.26, + "grad_norm": 6.510219097137451, + "learning_rate": 0.00015152212151542567, + "loss": 3.3276, + "step": 2227 + }, + { + "epoch": 3.26, + "grad_norm": 4.469378471374512, + "learning_rate": 0.0001514819169140651, + "loss": 3.3276, + "step": 2228 + }, + { + "epoch": 3.26, + "grad_norm": 4.4898681640625, + "learning_rate": 0.0001514417009870278, + "loss": 3.2907, + "step": 2229 + }, + { + "epoch": 3.26, + "grad_norm": 5.1355485916137695, + "learning_rate": 0.00015140147374316106, + "loss": 3.3146, + "step": 2230 + }, + { + "epoch": 3.27, + "grad_norm": 3.954402208328247, + "learning_rate": 0.00015136123519131456, + "loss": 3.2888, + "step": 2231 + }, + { + "epoch": 3.27, + "grad_norm": 8.745506286621094, + "learning_rate": 0.00015132098534034052, + "loss": 3.3684, + "step": 2232 + }, + { + "epoch": 3.27, + "grad_norm": 7.295302391052246, + "learning_rate": 0.00015128072419909368, + "loss": 3.3378, + "step": 2233 + }, + { + "epoch": 3.27, + "grad_norm": 1.574733853340149, + "learning_rate": 0.00015124045177643112, + "loss": 3.3003, + "step": 2234 + }, + { + "epoch": 3.27, + "grad_norm": 6.663206100463867, + "learning_rate": 0.0001512001680812126, + "loss": 3.3466, + "step": 2235 + }, + { + "epoch": 3.27, + "grad_norm": 7.278547286987305, + "learning_rate": 0.00015115987312230026, + "loss": 3.2945, + "step": 2236 + }, + { + "epoch": 3.27, + "grad_norm": 3.385538101196289, + "learning_rate": 0.00015111956690855863, + "loss": 3.2958, + "step": 2237 + }, + { + "epoch": 3.28, + "grad_norm": 7.496275424957275, + "learning_rate": 0.00015107924944885486, + "loss": 3.3214, + "step": 2238 + }, + { + "epoch": 3.28, + "grad_norm": 25.01158332824707, + "learning_rate": 0.00015103892075205862, + "loss": 3.4204, + "step": 2239 + }, + { + "epoch": 3.28, + "grad_norm": 9.22580337524414, + "learning_rate": 0.00015099858082704182, + "loss": 3.2895, + "step": 2240 + }, + { + "epoch": 3.28, + "grad_norm": 1.5853315591812134, + "learning_rate": 0.00015095822968267896, + "loss": 3.3388, + "step": 2241 + }, + { + "epoch": 3.28, + "grad_norm": 8.119154930114746, + "learning_rate": 0.00015091786732784716, + "loss": 3.35, + "step": 2242 + }, + { + "epoch": 3.28, + "grad_norm": 10.604775428771973, + "learning_rate": 0.00015087749377142574, + "loss": 3.3851, + "step": 2243 + }, + { + "epoch": 3.28, + "grad_norm": 7.424951076507568, + "learning_rate": 0.00015083710902229664, + "loss": 3.3619, + "step": 2244 + }, + { + "epoch": 3.29, + "grad_norm": 5.7252373695373535, + "learning_rate": 0.0001507967130893443, + "loss": 3.342, + "step": 2245 + }, + { + "epoch": 3.29, + "grad_norm": 8.934366226196289, + "learning_rate": 0.00015075630598145547, + "loss": 3.3553, + "step": 2246 + }, + { + "epoch": 3.29, + "grad_norm": 6.990628719329834, + "learning_rate": 0.00015071588770751947, + "loss": 3.3318, + "step": 2247 + }, + { + "epoch": 3.29, + "grad_norm": 19.209877014160156, + "learning_rate": 0.00015067545827642808, + "loss": 3.3181, + "step": 2248 + }, + { + "epoch": 3.29, + "grad_norm": 6.441047191619873, + "learning_rate": 0.00015063501769707548, + "loss": 3.3515, + "step": 2249 + }, + { + "epoch": 3.29, + "grad_norm": 5.99086332321167, + "learning_rate": 0.00015059456597835835, + "loss": 3.334, + "step": 2250 + }, + { + "epoch": 3.3, + "grad_norm": 2.6446921825408936, + "learning_rate": 0.00015055410312917573, + "loss": 3.332, + "step": 2251 + }, + { + "epoch": 3.3, + "grad_norm": 266.819091796875, + "learning_rate": 0.00015051362915842923, + "loss": 3.3605, + "step": 2252 + }, + { + "epoch": 3.3, + "grad_norm": 16.00495719909668, + "learning_rate": 0.00015047314407502283, + "loss": 3.3521, + "step": 2253 + }, + { + "epoch": 3.3, + "grad_norm": 8.721566200256348, + "learning_rate": 0.00015043264788786305, + "loss": 3.3846, + "step": 2254 + }, + { + "epoch": 3.3, + "grad_norm": 3.7059578895568848, + "learning_rate": 0.00015039214060585864, + "loss": 3.39, + "step": 2255 + }, + { + "epoch": 3.3, + "grad_norm": 5.986879825592041, + "learning_rate": 0.00015035162223792106, + "loss": 3.3791, + "step": 2256 + }, + { + "epoch": 3.3, + "grad_norm": 4.337857723236084, + "learning_rate": 0.00015031109279296402, + "loss": 3.3309, + "step": 2257 + }, + { + "epoch": 3.31, + "grad_norm": 4.526264190673828, + "learning_rate": 0.00015027055227990373, + "loss": 3.3768, + "step": 2258 + }, + { + "epoch": 3.31, + "grad_norm": 7.14890718460083, + "learning_rate": 0.00015023000070765884, + "loss": 3.3677, + "step": 2259 + }, + { + "epoch": 3.31, + "grad_norm": 3.9721760749816895, + "learning_rate": 0.0001501894380851504, + "loss": 3.3136, + "step": 2260 + }, + { + "epoch": 3.31, + "grad_norm": 5.246433734893799, + "learning_rate": 0.00015014886442130197, + "loss": 3.3263, + "step": 2261 + }, + { + "epoch": 3.31, + "grad_norm": 7.421927452087402, + "learning_rate": 0.00015010827972503944, + "loss": 3.3386, + "step": 2262 + }, + { + "epoch": 3.31, + "grad_norm": 3.354271173477173, + "learning_rate": 0.00015006768400529117, + "loss": 3.3509, + "step": 2263 + }, + { + "epoch": 3.31, + "grad_norm": 5.732051849365234, + "learning_rate": 0.00015002707727098798, + "loss": 3.3885, + "step": 2264 + }, + { + "epoch": 3.32, + "grad_norm": 6.248003005981445, + "learning_rate": 0.00014998645953106303, + "loss": 3.3291, + "step": 2265 + }, + { + "epoch": 3.32, + "grad_norm": 2.3988747596740723, + "learning_rate": 0.000149945830794452, + "loss": 3.2951, + "step": 2266 + }, + { + "epoch": 3.32, + "grad_norm": 4.345824241638184, + "learning_rate": 0.00014990519107009292, + "loss": 3.3256, + "step": 2267 + }, + { + "epoch": 3.32, + "grad_norm": 4.299257278442383, + "learning_rate": 0.0001498645403669262, + "loss": 3.3154, + "step": 2268 + }, + { + "epoch": 3.32, + "grad_norm": 4.104548931121826, + "learning_rate": 0.00014982387869389485, + "loss": 3.3248, + "step": 2269 + }, + { + "epoch": 3.32, + "grad_norm": 3.4004967212677, + "learning_rate": 0.00014978320605994405, + "loss": 3.3147, + "step": 2270 + }, + { + "epoch": 3.33, + "grad_norm": 3.225266456604004, + "learning_rate": 0.00014974252247402156, + "loss": 3.2917, + "step": 2271 + }, + { + "epoch": 3.33, + "grad_norm": 3.6412038803100586, + "learning_rate": 0.00014970182794507748, + "loss": 3.3557, + "step": 2272 + }, + { + "epoch": 3.33, + "grad_norm": 3.6593449115753174, + "learning_rate": 0.00014966112248206436, + "loss": 3.3048, + "step": 2273 + }, + { + "epoch": 3.33, + "grad_norm": 3.3542096614837646, + "learning_rate": 0.0001496204060939371, + "loss": 3.2807, + "step": 2274 + }, + { + "epoch": 3.33, + "grad_norm": 3.2932636737823486, + "learning_rate": 0.000149579678789653, + "loss": 3.3457, + "step": 2275 + }, + { + "epoch": 3.33, + "grad_norm": 2.769178867340088, + "learning_rate": 0.00014953894057817188, + "loss": 3.3023, + "step": 2276 + }, + { + "epoch": 3.33, + "grad_norm": 3.352829694747925, + "learning_rate": 0.0001494981914684558, + "loss": 3.2836, + "step": 2277 + }, + { + "epoch": 3.34, + "grad_norm": 1.4692974090576172, + "learning_rate": 0.00014945743146946937, + "loss": 3.354, + "step": 2278 + }, + { + "epoch": 3.34, + "grad_norm": 5.136416912078857, + "learning_rate": 0.0001494166605901794, + "loss": 3.2915, + "step": 2279 + }, + { + "epoch": 3.34, + "grad_norm": 4.615678310394287, + "learning_rate": 0.00014937587883955534, + "loss": 3.2766, + "step": 2280 + }, + { + "epoch": 3.34, + "grad_norm": 1.0332159996032715, + "learning_rate": 0.00014933508622656878, + "loss": 3.3291, + "step": 2281 + }, + { + "epoch": 3.34, + "grad_norm": 1.1297160387039185, + "learning_rate": 0.00014929428276019388, + "loss": 3.3465, + "step": 2282 + }, + { + "epoch": 3.34, + "grad_norm": 3.5306038856506348, + "learning_rate": 0.00014925346844940718, + "loss": 3.2554, + "step": 2283 + }, + { + "epoch": 3.34, + "grad_norm": 9.56828498840332, + "learning_rate": 0.00014921264330318746, + "loss": 3.3203, + "step": 2284 + }, + { + "epoch": 3.35, + "grad_norm": 2.0608508586883545, + "learning_rate": 0.00014917180733051602, + "loss": 3.2899, + "step": 2285 + }, + { + "epoch": 3.35, + "grad_norm": 1.6151994466781616, + "learning_rate": 0.00014913096054037654, + "loss": 3.3209, + "step": 2286 + }, + { + "epoch": 3.35, + "grad_norm": 2.8889124393463135, + "learning_rate": 0.00014909010294175497, + "loss": 3.3144, + "step": 2287 + }, + { + "epoch": 3.35, + "grad_norm": 0.650810182094574, + "learning_rate": 0.00014904923454363975, + "loss": 3.2888, + "step": 2288 + }, + { + "epoch": 3.35, + "grad_norm": 2.751948833465576, + "learning_rate": 0.00014900835535502165, + "loss": 3.3317, + "step": 2289 + }, + { + "epoch": 3.35, + "grad_norm": 1.9524742364883423, + "learning_rate": 0.0001489674653848938, + "loss": 3.2798, + "step": 2290 + }, + { + "epoch": 3.35, + "grad_norm": 0.6142035126686096, + "learning_rate": 0.00014892656464225175, + "loss": 3.2485, + "step": 2291 + }, + { + "epoch": 3.36, + "grad_norm": 1.9580166339874268, + "learning_rate": 0.00014888565313609333, + "loss": 3.3353, + "step": 2292 + }, + { + "epoch": 3.36, + "grad_norm": 4.1013898849487305, + "learning_rate": 0.0001488447308754189, + "loss": 3.3178, + "step": 2293 + }, + { + "epoch": 3.36, + "grad_norm": 1.7551053762435913, + "learning_rate": 0.00014880379786923098, + "loss": 3.2874, + "step": 2294 + }, + { + "epoch": 3.36, + "grad_norm": 5.447327136993408, + "learning_rate": 0.00014876285412653456, + "loss": 3.2928, + "step": 2295 + }, + { + "epoch": 3.36, + "grad_norm": 5.519541263580322, + "learning_rate": 0.00014872189965633707, + "loss": 3.3418, + "step": 2296 + }, + { + "epoch": 3.36, + "grad_norm": 2.044977903366089, + "learning_rate": 0.0001486809344676482, + "loss": 3.3401, + "step": 2297 + }, + { + "epoch": 3.37, + "grad_norm": 5.334053993225098, + "learning_rate": 0.00014863995856947995, + "loss": 3.3156, + "step": 2298 + }, + { + "epoch": 3.37, + "grad_norm": 3.4951860904693604, + "learning_rate": 0.0001485989719708468, + "loss": 3.2812, + "step": 2299 + }, + { + "epoch": 3.37, + "grad_norm": 5.83620548248291, + "learning_rate": 0.00014855797468076552, + "loss": 3.3016, + "step": 2300 + }, + { + "epoch": 3.37, + "grad_norm": 6.08860969543457, + "learning_rate": 0.00014851696670825523, + "loss": 3.2941, + "step": 2301 + }, + { + "epoch": 3.37, + "grad_norm": 7.569385051727295, + "learning_rate": 0.00014847594806233737, + "loss": 3.3478, + "step": 2302 + }, + { + "epoch": 3.37, + "grad_norm": 5.372185230255127, + "learning_rate": 0.00014843491875203584, + "loss": 3.2692, + "step": 2303 + }, + { + "epoch": 3.37, + "grad_norm": 4.030975341796875, + "learning_rate": 0.0001483938787863768, + "loss": 3.2826, + "step": 2304 + }, + { + "epoch": 3.38, + "grad_norm": 6.85429048538208, + "learning_rate": 0.00014835282817438875, + "loss": 3.2932, + "step": 2305 + }, + { + "epoch": 3.38, + "grad_norm": 6.535613059997559, + "learning_rate": 0.0001483117669251025, + "loss": 3.3158, + "step": 2306 + }, + { + "epoch": 3.38, + "grad_norm": 3.962482452392578, + "learning_rate": 0.00014827069504755132, + "loss": 3.308, + "step": 2307 + }, + { + "epoch": 3.38, + "grad_norm": 6.860799312591553, + "learning_rate": 0.00014822961255077073, + "loss": 3.3145, + "step": 2308 + }, + { + "epoch": 3.38, + "grad_norm": 9.554657936096191, + "learning_rate": 0.0001481885194437986, + "loss": 3.3366, + "step": 2309 + }, + { + "epoch": 3.38, + "grad_norm": 4.180676460266113, + "learning_rate": 0.00014814741573567514, + "loss": 3.3096, + "step": 2310 + }, + { + "epoch": 3.38, + "grad_norm": 6.105428218841553, + "learning_rate": 0.0001481063014354429, + "loss": 3.2919, + "step": 2311 + }, + { + "epoch": 3.39, + "grad_norm": 7.584985733032227, + "learning_rate": 0.00014806517655214672, + "loss": 3.2756, + "step": 2312 + }, + { + "epoch": 3.39, + "grad_norm": 5.3676300048828125, + "learning_rate": 0.00014802404109483383, + "loss": 3.2969, + "step": 2313 + }, + { + "epoch": 3.39, + "grad_norm": 3.991481065750122, + "learning_rate": 0.00014798289507255373, + "loss": 3.2808, + "step": 2314 + }, + { + "epoch": 3.39, + "grad_norm": 6.212815761566162, + "learning_rate": 0.00014794173849435826, + "loss": 3.2809, + "step": 2315 + }, + { + "epoch": 3.39, + "grad_norm": 1.0110559463500977, + "learning_rate": 0.00014790057136930166, + "loss": 3.3016, + "step": 2316 + }, + { + "epoch": 3.39, + "grad_norm": 9.187312126159668, + "learning_rate": 0.00014785939370644036, + "loss": 3.275, + "step": 2317 + }, + { + "epoch": 3.39, + "grad_norm": 12.635149955749512, + "learning_rate": 0.00014781820551483315, + "loss": 3.3793, + "step": 2318 + }, + { + "epoch": 3.4, + "grad_norm": 9.69156551361084, + "learning_rate": 0.00014777700680354124, + "loss": 3.3063, + "step": 2319 + }, + { + "epoch": 3.4, + "grad_norm": 1.8201349973678589, + "learning_rate": 0.000147735797581628, + "loss": 3.3122, + "step": 2320 + }, + { + "epoch": 3.4, + "grad_norm": 11.608341217041016, + "learning_rate": 0.0001476945778581592, + "loss": 3.341, + "step": 2321 + }, + { + "epoch": 3.4, + "grad_norm": 13.871359825134277, + "learning_rate": 0.0001476533476422029, + "loss": 3.3459, + "step": 2322 + }, + { + "epoch": 3.4, + "grad_norm": 9.090968132019043, + "learning_rate": 0.00014761210694282947, + "loss": 3.369, + "step": 2323 + }, + { + "epoch": 3.4, + "grad_norm": 4.659329414367676, + "learning_rate": 0.0001475708557691116, + "loss": 3.417, + "step": 2324 + }, + { + "epoch": 3.41, + "grad_norm": 6.619305610656738, + "learning_rate": 0.00014752959413012428, + "loss": 3.3524, + "step": 2325 + }, + { + "epoch": 3.41, + "grad_norm": 9.944043159484863, + "learning_rate": 0.00014748832203494478, + "loss": 3.3326, + "step": 2326 + }, + { + "epoch": 3.41, + "grad_norm": 5.2585062980651855, + "learning_rate": 0.00014744703949265268, + "loss": 3.3292, + "step": 2327 + }, + { + "epoch": 3.41, + "grad_norm": 5.1725311279296875, + "learning_rate": 0.00014740574651232987, + "loss": 3.3172, + "step": 2328 + }, + { + "epoch": 3.41, + "grad_norm": 30.300024032592773, + "learning_rate": 0.00014736444310306052, + "loss": 3.3064, + "step": 2329 + }, + { + "epoch": 3.41, + "grad_norm": 9.257378578186035, + "learning_rate": 0.00014732312927393112, + "loss": 3.2872, + "step": 2330 + }, + { + "epoch": 3.41, + "grad_norm": 2.5596399307250977, + "learning_rate": 0.0001472818050340304, + "loss": 3.2914, + "step": 2331 + }, + { + "epoch": 3.42, + "grad_norm": 6.8813934326171875, + "learning_rate": 0.00014724047039244947, + "loss": 3.3698, + "step": 2332 + }, + { + "epoch": 3.42, + "grad_norm": 10.194779396057129, + "learning_rate": 0.00014719912535828166, + "loss": 3.2775, + "step": 2333 + }, + { + "epoch": 3.42, + "grad_norm": 5.2471489906311035, + "learning_rate": 0.00014715776994062254, + "loss": 3.3149, + "step": 2334 + }, + { + "epoch": 3.42, + "grad_norm": 3.753305196762085, + "learning_rate": 0.00014711640414857008, + "loss": 3.3024, + "step": 2335 + }, + { + "epoch": 3.42, + "grad_norm": 5.786775588989258, + "learning_rate": 0.00014707502799122447, + "loss": 3.2799, + "step": 2336 + }, + { + "epoch": 3.42, + "grad_norm": 1.7370816469192505, + "learning_rate": 0.00014703364147768822, + "loss": 3.2881, + "step": 2337 + }, + { + "epoch": 3.42, + "grad_norm": 7.590539932250977, + "learning_rate": 0.000146992244617066, + "loss": 3.3002, + "step": 2338 + }, + { + "epoch": 3.43, + "grad_norm": 9.67332935333252, + "learning_rate": 0.00014695083741846492, + "loss": 3.3037, + "step": 2339 + }, + { + "epoch": 3.43, + "grad_norm": 5.325073719024658, + "learning_rate": 0.00014690941989099426, + "loss": 3.3241, + "step": 2340 + }, + { + "epoch": 3.43, + "grad_norm": 7.251718044281006, + "learning_rate": 0.0001468679920437656, + "loss": 3.316, + "step": 2341 + }, + { + "epoch": 3.43, + "grad_norm": 8.890804290771484, + "learning_rate": 0.00014682655388589274, + "loss": 3.3177, + "step": 2342 + }, + { + "epoch": 3.43, + "grad_norm": 3.9834563732147217, + "learning_rate": 0.0001467851054264919, + "loss": 3.3002, + "step": 2343 + }, + { + "epoch": 3.43, + "grad_norm": 5.180391788482666, + "learning_rate": 0.0001467436466746814, + "loss": 3.3166, + "step": 2344 + }, + { + "epoch": 3.43, + "grad_norm": 9.531121253967285, + "learning_rate": 0.00014670217763958188, + "loss": 3.2482, + "step": 2345 + }, + { + "epoch": 3.44, + "grad_norm": 4.674380302429199, + "learning_rate": 0.00014666069833031624, + "loss": 3.2703, + "step": 2346 + }, + { + "epoch": 3.44, + "grad_norm": 3.9839441776275635, + "learning_rate": 0.00014661920875600973, + "loss": 3.2701, + "step": 2347 + }, + { + "epoch": 3.44, + "grad_norm": 4.408015251159668, + "learning_rate": 0.00014657770892578968, + "loss": 3.2791, + "step": 2348 + }, + { + "epoch": 3.44, + "grad_norm": 2.449519395828247, + "learning_rate": 0.0001465361988487858, + "loss": 3.2654, + "step": 2349 + }, + { + "epoch": 3.44, + "grad_norm": 1.4051802158355713, + "learning_rate": 0.00014649467853413008, + "loss": 3.3101, + "step": 2350 + }, + { + "epoch": 3.44, + "grad_norm": 6.514969825744629, + "learning_rate": 0.00014645314799095665, + "loss": 3.2282, + "step": 2351 + }, + { + "epoch": 3.45, + "grad_norm": 5.476304054260254, + "learning_rate": 0.000146411607228402, + "loss": 3.3043, + "step": 2352 + }, + { + "epoch": 3.45, + "grad_norm": 2.443798780441284, + "learning_rate": 0.00014637005625560477, + "loss": 3.3045, + "step": 2353 + }, + { + "epoch": 3.45, + "grad_norm": 6.676492214202881, + "learning_rate": 0.00014632849508170595, + "loss": 3.2794, + "step": 2354 + }, + { + "epoch": 3.45, + "grad_norm": 2.0837247371673584, + "learning_rate": 0.0001462869237158487, + "loss": 3.3134, + "step": 2355 + }, + { + "epoch": 3.45, + "grad_norm": 7.334787368774414, + "learning_rate": 0.00014624534216717841, + "loss": 3.2919, + "step": 2356 + }, + { + "epoch": 3.45, + "grad_norm": 8.694489479064941, + "learning_rate": 0.00014620375044484277, + "loss": 3.2639, + "step": 2357 + }, + { + "epoch": 3.45, + "grad_norm": 2.46984601020813, + "learning_rate": 0.00014616214855799167, + "loss": 3.2571, + "step": 2358 + }, + { + "epoch": 3.46, + "grad_norm": 7.231738567352295, + "learning_rate": 0.00014612053651577727, + "loss": 3.3468, + "step": 2359 + }, + { + "epoch": 3.46, + "grad_norm": 7.998284816741943, + "learning_rate": 0.0001460789143273539, + "loss": 3.3764, + "step": 2360 + }, + { + "epoch": 3.46, + "grad_norm": 4.452177047729492, + "learning_rate": 0.00014603728200187824, + "loss": 3.3293, + "step": 2361 + }, + { + "epoch": 3.46, + "grad_norm": 9.913199424743652, + "learning_rate": 0.00014599563954850901, + "loss": 3.2631, + "step": 2362 + }, + { + "epoch": 3.46, + "grad_norm": 15.90157413482666, + "learning_rate": 0.00014595398697640733, + "loss": 3.3812, + "step": 2363 + }, + { + "epoch": 3.46, + "grad_norm": 12.660530090332031, + "learning_rate": 0.0001459123242947365, + "loss": 3.371, + "step": 2364 + }, + { + "epoch": 3.46, + "grad_norm": 5.289348125457764, + "learning_rate": 0.00014587065151266203, + "loss": 3.3892, + "step": 2365 + }, + { + "epoch": 3.47, + "grad_norm": 9.648274421691895, + "learning_rate": 0.00014582896863935162, + "loss": 3.2931, + "step": 2366 + }, + { + "epoch": 3.47, + "eval_loss": 3.133899211883545, + "eval_runtime": 1162.2679, + "eval_samples_per_second": 14.749, + "eval_steps_per_second": 0.922, + "step": 2366 + }, + { + "epoch": 3.47, + "grad_norm": 9.562708854675293, + "learning_rate": 0.00014578727568397522, + "loss": 3.3485, + "step": 2367 + }, + { + "epoch": 3.47, + "grad_norm": 19.97218894958496, + "learning_rate": 0.00014574557265570505, + "loss": 3.3403, + "step": 2368 + }, + { + "epoch": 3.47, + "grad_norm": 6.222564220428467, + "learning_rate": 0.0001457038595637155, + "loss": 3.3743, + "step": 2369 + }, + { + "epoch": 3.47, + "grad_norm": 7.945571422576904, + "learning_rate": 0.00014566213641718307, + "loss": 3.2679, + "step": 2370 + }, + { + "epoch": 3.47, + "grad_norm": 8.883635520935059, + "learning_rate": 0.0001456204032252867, + "loss": 3.2295, + "step": 2371 + }, + { + "epoch": 3.48, + "grad_norm": 5.227753639221191, + "learning_rate": 0.00014557865999720735, + "loss": 3.2934, + "step": 2372 + }, + { + "epoch": 3.48, + "grad_norm": 8.865499496459961, + "learning_rate": 0.0001455369067421282, + "loss": 3.3526, + "step": 2373 + }, + { + "epoch": 3.48, + "grad_norm": 10.483139991760254, + "learning_rate": 0.00014549514346923477, + "loss": 3.3058, + "step": 2374 + }, + { + "epoch": 3.48, + "grad_norm": 12.922052383422852, + "learning_rate": 0.00014545337018771472, + "loss": 3.2366, + "step": 2375 + }, + { + "epoch": 3.48, + "grad_norm": 2.131599187850952, + "learning_rate": 0.00014541158690675778, + "loss": 3.2613, + "step": 2376 + }, + { + "epoch": 3.48, + "grad_norm": 10.010235786437988, + "learning_rate": 0.0001453697936355561, + "loss": 3.2698, + "step": 2377 + }, + { + "epoch": 3.48, + "grad_norm": 14.034173011779785, + "learning_rate": 0.00014532799038330385, + "loss": 3.3691, + "step": 2378 + }, + { + "epoch": 3.49, + "grad_norm": 9.930380821228027, + "learning_rate": 0.0001452861771591975, + "loss": 3.304, + "step": 2379 + }, + { + "epoch": 3.49, + "grad_norm": 7.6726555824279785, + "learning_rate": 0.00014524435397243567, + "loss": 3.2806, + "step": 2380 + }, + { + "epoch": 3.49, + "grad_norm": 8.018184661865234, + "learning_rate": 0.00014520252083221916, + "loss": 3.2673, + "step": 2381 + }, + { + "epoch": 3.49, + "grad_norm": 8.354979515075684, + "learning_rate": 0.00014516067774775101, + "loss": 3.295, + "step": 2382 + }, + { + "epoch": 3.49, + "grad_norm": 3.118138313293457, + "learning_rate": 0.00014511882472823643, + "loss": 3.2593, + "step": 2383 + }, + { + "epoch": 3.49, + "grad_norm": 9.12159538269043, + "learning_rate": 0.00014507696178288278, + "loss": 3.2893, + "step": 2384 + }, + { + "epoch": 3.49, + "grad_norm": 13.92984390258789, + "learning_rate": 0.0001450350889208996, + "loss": 3.3191, + "step": 2385 + }, + { + "epoch": 3.5, + "grad_norm": 9.545883178710938, + "learning_rate": 0.0001449932061514987, + "loss": 3.2753, + "step": 2386 + }, + { + "epoch": 3.5, + "grad_norm": 1.6055418252944946, + "learning_rate": 0.00014495131348389394, + "loss": 3.2373, + "step": 2387 + }, + { + "epoch": 3.5, + "grad_norm": 8.169912338256836, + "learning_rate": 0.00014490941092730152, + "loss": 3.2748, + "step": 2388 + }, + { + "epoch": 3.5, + "grad_norm": 8.912496566772461, + "learning_rate": 0.00014486749849093963, + "loss": 3.302, + "step": 2389 + }, + { + "epoch": 3.5, + "grad_norm": 6.068753242492676, + "learning_rate": 0.00014482557618402875, + "loss": 3.2805, + "step": 2390 + }, + { + "epoch": 3.5, + "grad_norm": 4.4562578201293945, + "learning_rate": 0.00014478364401579154, + "loss": 3.2488, + "step": 2391 + }, + { + "epoch": 3.5, + "grad_norm": 4.690038204193115, + "learning_rate": 0.00014474170199545275, + "loss": 3.2896, + "step": 2392 + }, + { + "epoch": 3.51, + "grad_norm": 3.260173797607422, + "learning_rate": 0.0001446997501322394, + "loss": 3.2497, + "step": 2393 + }, + { + "epoch": 3.51, + "grad_norm": 3.311274528503418, + "learning_rate": 0.0001446577884353806, + "loss": 3.2359, + "step": 2394 + }, + { + "epoch": 3.51, + "grad_norm": 2.5797505378723145, + "learning_rate": 0.00014461581691410758, + "loss": 3.3372, + "step": 2395 + }, + { + "epoch": 3.51, + "grad_norm": 1.9033737182617188, + "learning_rate": 0.00014457383557765386, + "loss": 3.2308, + "step": 2396 + }, + { + "epoch": 3.51, + "grad_norm": 3.9633514881134033, + "learning_rate": 0.000144531844435255, + "loss": 3.2787, + "step": 2397 + }, + { + "epoch": 3.51, + "grad_norm": 3.756439685821533, + "learning_rate": 0.00014448984349614886, + "loss": 3.3238, + "step": 2398 + }, + { + "epoch": 3.52, + "grad_norm": 3.5643348693847656, + "learning_rate": 0.00014444783276957524, + "loss": 3.2315, + "step": 2399 + }, + { + "epoch": 3.52, + "grad_norm": 4.421493053436279, + "learning_rate": 0.00014440581226477634, + "loss": 3.277, + "step": 2400 + }, + { + "epoch": 3.52, + "grad_norm": 4.980103969573975, + "learning_rate": 0.00014436378199099634, + "loss": 3.248, + "step": 2401 + }, + { + "epoch": 3.52, + "grad_norm": 7.962924003601074, + "learning_rate": 0.0001443217419574816, + "loss": 3.2628, + "step": 2402 + }, + { + "epoch": 3.52, + "grad_norm": 3.279404401779175, + "learning_rate": 0.00014427969217348065, + "loss": 3.2662, + "step": 2403 + }, + { + "epoch": 3.52, + "grad_norm": 6.002499580383301, + "learning_rate": 0.00014423763264824417, + "loss": 3.2498, + "step": 2404 + }, + { + "epoch": 3.52, + "grad_norm": 8.197328567504883, + "learning_rate": 0.00014419556339102503, + "loss": 3.3432, + "step": 2405 + }, + { + "epoch": 3.53, + "grad_norm": 3.395263910293579, + "learning_rate": 0.00014415348441107806, + "loss": 3.2308, + "step": 2406 + }, + { + "epoch": 3.53, + "grad_norm": 6.569970607757568, + "learning_rate": 0.00014411139571766042, + "loss": 3.2445, + "step": 2407 + }, + { + "epoch": 3.53, + "grad_norm": 9.0120849609375, + "learning_rate": 0.0001440692973200314, + "loss": 3.2689, + "step": 2408 + }, + { + "epoch": 3.53, + "grad_norm": 5.56130313873291, + "learning_rate": 0.00014402718922745228, + "loss": 3.2685, + "step": 2409 + }, + { + "epoch": 3.53, + "grad_norm": 4.367111682891846, + "learning_rate": 0.00014398507144918656, + "loss": 3.3052, + "step": 2410 + }, + { + "epoch": 3.53, + "grad_norm": 6.1577558517456055, + "learning_rate": 0.00014394294399449996, + "loss": 3.2687, + "step": 2411 + }, + { + "epoch": 3.53, + "grad_norm": 14.60132122039795, + "learning_rate": 0.00014390080687266013, + "loss": 3.2542, + "step": 2412 + }, + { + "epoch": 3.54, + "grad_norm": 19.244531631469727, + "learning_rate": 0.000143858660092937, + "loss": 3.2623, + "step": 2413 + }, + { + "epoch": 3.54, + "grad_norm": 5.000961780548096, + "learning_rate": 0.0001438165036646026, + "loss": 3.2887, + "step": 2414 + }, + { + "epoch": 3.54, + "grad_norm": 1.7813776731491089, + "learning_rate": 0.000143774337596931, + "loss": 3.2522, + "step": 2415 + }, + { + "epoch": 3.54, + "grad_norm": 11.567626953125, + "learning_rate": 0.00014373216189919851, + "loss": 3.3463, + "step": 2416 + }, + { + "epoch": 3.54, + "grad_norm": 17.07219696044922, + "learning_rate": 0.0001436899765806835, + "loss": 3.3653, + "step": 2417 + }, + { + "epoch": 3.54, + "grad_norm": 10.967621803283691, + "learning_rate": 0.00014364778165066642, + "loss": 3.339, + "step": 2418 + }, + { + "epoch": 3.54, + "grad_norm": 5.372880935668945, + "learning_rate": 0.00014360557711842991, + "loss": 3.3025, + "step": 2419 + }, + { + "epoch": 3.55, + "grad_norm": 8.31612777709961, + "learning_rate": 0.00014356336299325864, + "loss": 3.2886, + "step": 2420 + }, + { + "epoch": 3.55, + "grad_norm": 6.364083290100098, + "learning_rate": 0.00014352113928443947, + "loss": 3.3376, + "step": 2421 + }, + { + "epoch": 3.55, + "grad_norm": 39.3469123840332, + "learning_rate": 0.00014347890600126134, + "loss": 3.2798, + "step": 2422 + }, + { + "epoch": 3.55, + "grad_norm": 15.578664779663086, + "learning_rate": 0.00014343666315301526, + "loss": 3.3638, + "step": 2423 + }, + { + "epoch": 3.55, + "grad_norm": 15.445240020751953, + "learning_rate": 0.0001433944107489944, + "loss": 3.3672, + "step": 2424 + }, + { + "epoch": 3.55, + "grad_norm": 13.184659957885742, + "learning_rate": 0.00014335214879849394, + "loss": 3.3172, + "step": 2425 + }, + { + "epoch": 3.56, + "grad_norm": 4.225596904754639, + "learning_rate": 0.0001433098773108113, + "loss": 3.2598, + "step": 2426 + }, + { + "epoch": 3.56, + "grad_norm": 9.756654739379883, + "learning_rate": 0.00014326759629524594, + "loss": 3.3621, + "step": 2427 + }, + { + "epoch": 3.56, + "grad_norm": 14.042159080505371, + "learning_rate": 0.00014322530576109933, + "loss": 3.3233, + "step": 2428 + }, + { + "epoch": 3.56, + "grad_norm": 14.112271308898926, + "learning_rate": 0.00014318300571767513, + "loss": 3.3309, + "step": 2429 + }, + { + "epoch": 3.56, + "grad_norm": 9.754487037658691, + "learning_rate": 0.00014314069617427907, + "loss": 3.3182, + "step": 2430 + }, + { + "epoch": 3.56, + "grad_norm": 2.0633256435394287, + "learning_rate": 0.00014309837714021902, + "loss": 3.2493, + "step": 2431 + }, + { + "epoch": 3.56, + "grad_norm": 8.613134384155273, + "learning_rate": 0.00014305604862480478, + "loss": 3.3103, + "step": 2432 + }, + { + "epoch": 3.57, + "grad_norm": 9.88028335571289, + "learning_rate": 0.0001430137106373484, + "loss": 3.3556, + "step": 2433 + }, + { + "epoch": 3.57, + "grad_norm": 8.193672180175781, + "learning_rate": 0.00014297136318716395, + "loss": 3.3321, + "step": 2434 + }, + { + "epoch": 3.57, + "grad_norm": 4.98108434677124, + "learning_rate": 0.00014292900628356758, + "loss": 3.265, + "step": 2435 + }, + { + "epoch": 3.57, + "grad_norm": 5.972259521484375, + "learning_rate": 0.0001428866399358775, + "loss": 3.2759, + "step": 2436 + }, + { + "epoch": 3.57, + "grad_norm": 4.662635803222656, + "learning_rate": 0.00014284426415341408, + "loss": 3.3012, + "step": 2437 + }, + { + "epoch": 3.57, + "grad_norm": 4.476476192474365, + "learning_rate": 0.00014280187894549968, + "loss": 3.2919, + "step": 2438 + }, + { + "epoch": 3.57, + "grad_norm": 3.099634885787964, + "learning_rate": 0.00014275948432145876, + "loss": 3.3264, + "step": 2439 + }, + { + "epoch": 3.58, + "grad_norm": 6.200466632843018, + "learning_rate": 0.00014271708029061785, + "loss": 3.2982, + "step": 2440 + }, + { + "epoch": 3.58, + "grad_norm": 3.2733237743377686, + "learning_rate": 0.00014267466686230553, + "loss": 3.2932, + "step": 2441 + }, + { + "epoch": 3.58, + "grad_norm": 3.1728217601776123, + "learning_rate": 0.0001426322440458525, + "loss": 3.2532, + "step": 2442 + }, + { + "epoch": 3.58, + "grad_norm": 3.4622151851654053, + "learning_rate": 0.0001425898118505915, + "loss": 3.2794, + "step": 2443 + }, + { + "epoch": 3.58, + "grad_norm": 7.931596755981445, + "learning_rate": 0.0001425473702858573, + "loss": 3.2692, + "step": 2444 + }, + { + "epoch": 3.58, + "grad_norm": 3.586261749267578, + "learning_rate": 0.00014250491936098678, + "loss": 3.2432, + "step": 2445 + }, + { + "epoch": 3.58, + "grad_norm": 3.096931219100952, + "learning_rate": 0.00014246245908531882, + "loss": 3.2589, + "step": 2446 + }, + { + "epoch": 3.59, + "grad_norm": 3.957979202270508, + "learning_rate": 0.0001424199894681945, + "loss": 3.278, + "step": 2447 + }, + { + "epoch": 3.59, + "grad_norm": 0.8693408370018005, + "learning_rate": 0.00014237751051895676, + "loss": 3.2366, + "step": 2448 + }, + { + "epoch": 3.59, + "grad_norm": 2.3588359355926514, + "learning_rate": 0.0001423350222469507, + "loss": 3.2469, + "step": 2449 + }, + { + "epoch": 3.59, + "grad_norm": 1.2426791191101074, + "learning_rate": 0.00014229252466152345, + "loss": 3.2929, + "step": 2450 + }, + { + "epoch": 3.59, + "grad_norm": 2.7103374004364014, + "learning_rate": 0.00014225001777202424, + "loss": 3.248, + "step": 2451 + }, + { + "epoch": 3.59, + "grad_norm": 2.5343198776245117, + "learning_rate": 0.00014220750158780424, + "loss": 3.257, + "step": 2452 + }, + { + "epoch": 3.6, + "grad_norm": 1.3891563415527344, + "learning_rate": 0.00014216497611821676, + "loss": 3.2348, + "step": 2453 + }, + { + "epoch": 3.6, + "grad_norm": 3.7499053478240967, + "learning_rate": 0.0001421224413726171, + "loss": 3.2513, + "step": 2454 + }, + { + "epoch": 3.6, + "grad_norm": 1.5479190349578857, + "learning_rate": 0.00014207989736036267, + "loss": 3.2785, + "step": 2455 + }, + { + "epoch": 3.6, + "grad_norm": 2.783564567565918, + "learning_rate": 0.0001420373440908128, + "loss": 3.292, + "step": 2456 + }, + { + "epoch": 3.6, + "grad_norm": 3.4745991230010986, + "learning_rate": 0.00014199478157332895, + "loss": 3.2399, + "step": 2457 + }, + { + "epoch": 3.6, + "grad_norm": 2.5730254650115967, + "learning_rate": 0.0001419522098172746, + "loss": 3.257, + "step": 2458 + }, + { + "epoch": 3.6, + "grad_norm": 3.455461025238037, + "learning_rate": 0.00014190962883201523, + "loss": 3.225, + "step": 2459 + }, + { + "epoch": 3.61, + "grad_norm": 4.234341621398926, + "learning_rate": 0.0001418670386269184, + "loss": 3.2997, + "step": 2460 + }, + { + "epoch": 3.61, + "grad_norm": 9.0831937789917, + "learning_rate": 0.00014182443921135362, + "loss": 3.221, + "step": 2461 + }, + { + "epoch": 3.61, + "grad_norm": 3.933647871017456, + "learning_rate": 0.00014178183059469253, + "loss": 3.2355, + "step": 2462 + }, + { + "epoch": 3.61, + "grad_norm": 1.695810079574585, + "learning_rate": 0.00014173921278630872, + "loss": 3.2724, + "step": 2463 + }, + { + "epoch": 3.61, + "grad_norm": 2.3507513999938965, + "learning_rate": 0.0001416965857955778, + "loss": 3.273, + "step": 2464 + }, + { + "epoch": 3.61, + "grad_norm": 2.5299062728881836, + "learning_rate": 0.00014165394963187747, + "loss": 3.2366, + "step": 2465 + }, + { + "epoch": 3.61, + "grad_norm": 0.9136064052581787, + "learning_rate": 0.00014161130430458737, + "loss": 3.2318, + "step": 2466 + }, + { + "epoch": 3.62, + "grad_norm": 2.159097671508789, + "learning_rate": 0.00014156864982308917, + "loss": 3.2139, + "step": 2467 + }, + { + "epoch": 3.62, + "grad_norm": 2.3743743896484375, + "learning_rate": 0.0001415259861967666, + "loss": 3.2442, + "step": 2468 + }, + { + "epoch": 3.62, + "grad_norm": 1.7210527658462524, + "learning_rate": 0.00014148331343500538, + "loss": 3.2565, + "step": 2469 + }, + { + "epoch": 3.62, + "grad_norm": 4.262208938598633, + "learning_rate": 0.0001414406315471932, + "loss": 3.2542, + "step": 2470 + }, + { + "epoch": 3.62, + "grad_norm": 3.7595646381378174, + "learning_rate": 0.00014139794054271977, + "loss": 3.2048, + "step": 2471 + }, + { + "epoch": 3.62, + "grad_norm": 3.6839802265167236, + "learning_rate": 0.00014135524043097693, + "loss": 3.2873, + "step": 2472 + }, + { + "epoch": 3.63, + "grad_norm": 3.4619247913360596, + "learning_rate": 0.0001413125312213583, + "loss": 3.2562, + "step": 2473 + }, + { + "epoch": 3.63, + "grad_norm": 3.164567708969116, + "learning_rate": 0.00014126981292325965, + "loss": 3.2521, + "step": 2474 + }, + { + "epoch": 3.63, + "grad_norm": 3.066319227218628, + "learning_rate": 0.00014122708554607875, + "loss": 3.2671, + "step": 2475 + }, + { + "epoch": 3.63, + "grad_norm": 2.757352590560913, + "learning_rate": 0.00014118434909921534, + "loss": 3.2061, + "step": 2476 + }, + { + "epoch": 3.63, + "grad_norm": 1.9947091341018677, + "learning_rate": 0.0001411416035920711, + "loss": 3.2391, + "step": 2477 + }, + { + "epoch": 3.63, + "grad_norm": 4.6580986976623535, + "learning_rate": 0.0001410988490340498, + "loss": 3.252, + "step": 2478 + }, + { + "epoch": 3.63, + "grad_norm": 4.087027072906494, + "learning_rate": 0.00014105608543455716, + "loss": 3.2371, + "step": 2479 + }, + { + "epoch": 3.64, + "grad_norm": 3.322829008102417, + "learning_rate": 0.0001410133128030009, + "loss": 3.2261, + "step": 2480 + }, + { + "epoch": 3.64, + "grad_norm": 3.1107428073883057, + "learning_rate": 0.00014097053114879061, + "loss": 3.2418, + "step": 2481 + }, + { + "epoch": 3.64, + "grad_norm": 2.6654446125030518, + "learning_rate": 0.00014092774048133809, + "loss": 3.232, + "step": 2482 + }, + { + "epoch": 3.64, + "grad_norm": 1.8028963804244995, + "learning_rate": 0.00014088494081005695, + "loss": 3.2122, + "step": 2483 + }, + { + "epoch": 3.64, + "grad_norm": 9.199921607971191, + "learning_rate": 0.00014084213214436277, + "loss": 3.2409, + "step": 2484 + }, + { + "epoch": 3.64, + "grad_norm": 7.354410648345947, + "learning_rate": 0.00014079931449367328, + "loss": 3.2151, + "step": 2485 + }, + { + "epoch": 3.64, + "grad_norm": 2.6317131519317627, + "learning_rate": 0.00014075648786740806, + "loss": 3.2449, + "step": 2486 + }, + { + "epoch": 3.65, + "grad_norm": 6.4712042808532715, + "learning_rate": 0.00014071365227498857, + "loss": 3.2384, + "step": 2487 + }, + { + "epoch": 3.65, + "grad_norm": 7.566592216491699, + "learning_rate": 0.00014067080772583846, + "loss": 3.2208, + "step": 2488 + }, + { + "epoch": 3.65, + "grad_norm": 3.7115042209625244, + "learning_rate": 0.00014062795422938325, + "loss": 3.2403, + "step": 2489 + }, + { + "epoch": 3.65, + "grad_norm": 7.227043151855469, + "learning_rate": 0.00014058509179505036, + "loss": 3.2533, + "step": 2490 + }, + { + "epoch": 3.65, + "grad_norm": 10.836502075195312, + "learning_rate": 0.00014054222043226926, + "loss": 3.3034, + "step": 2491 + }, + { + "epoch": 3.65, + "grad_norm": 8.85554313659668, + "learning_rate": 0.00014049934015047136, + "loss": 3.2838, + "step": 2492 + }, + { + "epoch": 3.65, + "grad_norm": 7.335647106170654, + "learning_rate": 0.00014045645095909004, + "loss": 3.2506, + "step": 2493 + }, + { + "epoch": 3.66, + "grad_norm": 6.849621295928955, + "learning_rate": 0.00014041355286756066, + "loss": 3.266, + "step": 2494 + }, + { + "epoch": 3.66, + "grad_norm": 5.7214837074279785, + "learning_rate": 0.00014037064588532047, + "loss": 3.229, + "step": 2495 + }, + { + "epoch": 3.66, + "grad_norm": 1.1197165250778198, + "learning_rate": 0.00014032773002180875, + "loss": 3.189, + "step": 2496 + }, + { + "epoch": 3.66, + "grad_norm": 2.7683522701263428, + "learning_rate": 0.00014028480528646667, + "loss": 3.2357, + "step": 2497 + }, + { + "epoch": 3.66, + "grad_norm": 10.467253684997559, + "learning_rate": 0.0001402418716887374, + "loss": 3.2702, + "step": 2498 + }, + { + "epoch": 3.66, + "grad_norm": 5.283757209777832, + "learning_rate": 0.0001401989292380661, + "loss": 3.2404, + "step": 2499 + }, + { + "epoch": 3.67, + "grad_norm": 3.8525490760803223, + "learning_rate": 0.00014015597794389973, + "loss": 3.2352, + "step": 2500 + }, + { + "epoch": 3.67, + "grad_norm": 4.830957412719727, + "learning_rate": 0.0001401130178156873, + "loss": 3.2432, + "step": 2501 + }, + { + "epoch": 3.67, + "grad_norm": 28.55122184753418, + "learning_rate": 0.0001400700488628798, + "loss": 3.2176, + "step": 2502 + }, + { + "epoch": 3.67, + "grad_norm": 6.477649211883545, + "learning_rate": 0.0001400270710949301, + "loss": 3.1952, + "step": 2503 + }, + { + "epoch": 3.67, + "grad_norm": 4.75806188583374, + "learning_rate": 0.00013998408452129302, + "loss": 3.2384, + "step": 2504 + }, + { + "epoch": 3.67, + "grad_norm": 5.451016426086426, + "learning_rate": 0.00013994108915142526, + "loss": 3.2404, + "step": 2505 + }, + { + "epoch": 3.67, + "grad_norm": 2.510890245437622, + "learning_rate": 0.0001398980849947856, + "loss": 3.221, + "step": 2506 + }, + { + "epoch": 3.68, + "grad_norm": 6.08433198928833, + "learning_rate": 0.0001398550720608346, + "loss": 3.2386, + "step": 2507 + }, + { + "epoch": 3.68, + "grad_norm": 7.88702392578125, + "learning_rate": 0.00013981205035903485, + "loss": 3.2582, + "step": 2508 + }, + { + "epoch": 3.68, + "grad_norm": 3.3581769466400146, + "learning_rate": 0.00013976901989885086, + "loss": 3.2437, + "step": 2509 + }, + { + "epoch": 3.68, + "grad_norm": 5.539381980895996, + "learning_rate": 0.000139725980689749, + "loss": 3.2349, + "step": 2510 + }, + { + "epoch": 3.68, + "grad_norm": 6.059652328491211, + "learning_rate": 0.0001396829327411976, + "loss": 3.3016, + "step": 2511 + }, + { + "epoch": 3.68, + "grad_norm": 1.3593266010284424, + "learning_rate": 0.00013963987606266693, + "loss": 3.2587, + "step": 2512 + }, + { + "epoch": 3.68, + "grad_norm": 9.557854652404785, + "learning_rate": 0.0001395968106636292, + "loss": 3.2695, + "step": 2513 + }, + { + "epoch": 3.69, + "grad_norm": 26.98328971862793, + "learning_rate": 0.0001395537365535585, + "loss": 3.3668, + "step": 2514 + }, + { + "epoch": 3.69, + "grad_norm": 14.079690933227539, + "learning_rate": 0.0001395106537419308, + "loss": 3.2736, + "step": 2515 + }, + { + "epoch": 3.69, + "grad_norm": 5.151083946228027, + "learning_rate": 0.00013946756223822413, + "loss": 3.2461, + "step": 2516 + }, + { + "epoch": 3.69, + "grad_norm": 7.186163902282715, + "learning_rate": 0.00013942446205191824, + "loss": 3.2817, + "step": 2517 + }, + { + "epoch": 3.69, + "grad_norm": 9.277379035949707, + "learning_rate": 0.0001393813531924949, + "loss": 3.2008, + "step": 2518 + }, + { + "epoch": 3.69, + "grad_norm": 19.572731018066406, + "learning_rate": 0.00013933823566943777, + "loss": 3.2131, + "step": 2519 + }, + { + "epoch": 3.69, + "grad_norm": 4.6605329513549805, + "learning_rate": 0.00013929510949223245, + "loss": 3.2512, + "step": 2520 + }, + { + "epoch": 3.7, + "grad_norm": 151.2240753173828, + "learning_rate": 0.00013925197467036635, + "loss": 3.2621, + "step": 2521 + }, + { + "epoch": 3.7, + "grad_norm": 13.225665092468262, + "learning_rate": 0.0001392088312133289, + "loss": 3.2814, + "step": 2522 + }, + { + "epoch": 3.7, + "grad_norm": 14.404492378234863, + "learning_rate": 0.00013916567913061133, + "loss": 3.2903, + "step": 2523 + }, + { + "epoch": 3.7, + "grad_norm": 10.506006240844727, + "learning_rate": 0.00013912251843170687, + "loss": 3.226, + "step": 2524 + }, + { + "epoch": 3.7, + "grad_norm": 3.039416790008545, + "learning_rate": 0.0001390793491261105, + "loss": 3.26, + "step": 2525 + }, + { + "epoch": 3.7, + "grad_norm": 6.364565372467041, + "learning_rate": 0.00013903617122331923, + "loss": 3.1935, + "step": 2526 + }, + { + "epoch": 3.71, + "grad_norm": 3.801405668258667, + "learning_rate": 0.0001389929847328319, + "loss": 3.2204, + "step": 2527 + }, + { + "epoch": 3.71, + "grad_norm": 13.04414176940918, + "learning_rate": 0.00013894978966414926, + "loss": 3.2408, + "step": 2528 + }, + { + "epoch": 3.71, + "grad_norm": 14.216852188110352, + "learning_rate": 0.0001389065860267739, + "loss": 3.2739, + "step": 2529 + }, + { + "epoch": 3.71, + "grad_norm": 10.664838790893555, + "learning_rate": 0.0001388633738302104, + "loss": 3.2684, + "step": 2530 + }, + { + "epoch": 3.71, + "grad_norm": 6.671474456787109, + "learning_rate": 0.00013882015308396508, + "loss": 3.3327, + "step": 2531 + }, + { + "epoch": 3.71, + "grad_norm": 15.43130874633789, + "learning_rate": 0.00013877692379754626, + "loss": 3.2355, + "step": 2532 + }, + { + "epoch": 3.71, + "grad_norm": 44.657169342041016, + "learning_rate": 0.00013873368598046404, + "loss": 3.2309, + "step": 2533 + }, + { + "epoch": 3.72, + "grad_norm": 18.107580184936523, + "learning_rate": 0.00013869043964223051, + "loss": 3.2906, + "step": 2534 + }, + { + "epoch": 3.72, + "grad_norm": 11.789257049560547, + "learning_rate": 0.00013864718479235954, + "loss": 3.3045, + "step": 2535 + }, + { + "epoch": 3.72, + "eval_loss": 3.0709898471832275, + "eval_runtime": 1163.6634, + "eval_samples_per_second": 14.731, + "eval_steps_per_second": 0.921, + "step": 2535 + }, + { + "epoch": 3.72, + "grad_norm": 7.7072319984436035, + "learning_rate": 0.00013860392144036693, + "loss": 3.2131, + "step": 2536 + }, + { + "epoch": 3.72, + "grad_norm": 4.083733081817627, + "learning_rate": 0.0001385606495957703, + "loss": 3.2641, + "step": 2537 + }, + { + "epoch": 3.72, + "grad_norm": 11.785395622253418, + "learning_rate": 0.00013851736926808917, + "loss": 3.2106, + "step": 2538 + }, + { + "epoch": 3.72, + "grad_norm": 13.176704406738281, + "learning_rate": 0.00013847408046684497, + "loss": 3.2857, + "step": 2539 + }, + { + "epoch": 3.72, + "grad_norm": 5.579313278198242, + "learning_rate": 0.00013843078320156088, + "loss": 3.2861, + "step": 2540 + }, + { + "epoch": 3.73, + "grad_norm": 4.398528099060059, + "learning_rate": 0.0001383874774817621, + "loss": 3.225, + "step": 2541 + }, + { + "epoch": 3.73, + "grad_norm": 6.542925834655762, + "learning_rate": 0.00013834416331697544, + "loss": 3.204, + "step": 2542 + }, + { + "epoch": 3.73, + "grad_norm": 9.607292175292969, + "learning_rate": 0.00013830084071672985, + "loss": 3.2284, + "step": 2543 + }, + { + "epoch": 3.73, + "grad_norm": 2.1741347312927246, + "learning_rate": 0.00013825750969055604, + "loss": 3.2318, + "step": 2544 + }, + { + "epoch": 3.73, + "grad_norm": 2.4551427364349365, + "learning_rate": 0.00013821417024798642, + "loss": 3.2454, + "step": 2545 + }, + { + "epoch": 3.73, + "grad_norm": 1.8988637924194336, + "learning_rate": 0.0001381708223985555, + "loss": 3.2688, + "step": 2546 + }, + { + "epoch": 3.73, + "grad_norm": 1.2011829614639282, + "learning_rate": 0.00013812746615179944, + "loss": 3.2418, + "step": 2547 + }, + { + "epoch": 3.74, + "grad_norm": 2.4243390560150146, + "learning_rate": 0.0001380841015172563, + "loss": 3.2759, + "step": 2548 + }, + { + "epoch": 3.74, + "grad_norm": 0.9519268274307251, + "learning_rate": 0.00013804072850446612, + "loss": 3.2498, + "step": 2549 + }, + { + "epoch": 3.74, + "grad_norm": 5.196446418762207, + "learning_rate": 0.00013799734712297059, + "loss": 3.2358, + "step": 2550 + }, + { + "epoch": 3.74, + "grad_norm": 3.189202308654785, + "learning_rate": 0.00013795395738231333, + "loss": 3.2582, + "step": 2551 + }, + { + "epoch": 3.74, + "grad_norm": 5.616455078125, + "learning_rate": 0.0001379105592920398, + "loss": 3.2066, + "step": 2552 + }, + { + "epoch": 3.74, + "grad_norm": 9.704096794128418, + "learning_rate": 0.0001378671528616973, + "loss": 3.2245, + "step": 2553 + }, + { + "epoch": 3.75, + "grad_norm": 3.5445096492767334, + "learning_rate": 0.00013782373810083493, + "loss": 3.2097, + "step": 2554 + }, + { + "epoch": 3.75, + "grad_norm": 3.731013774871826, + "learning_rate": 0.00013778031501900367, + "loss": 3.2075, + "step": 2555 + }, + { + "epoch": 3.75, + "grad_norm": 8.907066345214844, + "learning_rate": 0.0001377368836257563, + "loss": 3.2249, + "step": 2556 + }, + { + "epoch": 3.75, + "grad_norm": 4.645308017730713, + "learning_rate": 0.00013769344393064743, + "loss": 3.2139, + "step": 2557 + }, + { + "epoch": 3.75, + "grad_norm": 4.715915679931641, + "learning_rate": 0.00013764999594323353, + "loss": 3.2613, + "step": 2558 + }, + { + "epoch": 3.75, + "grad_norm": 5.21357536315918, + "learning_rate": 0.0001376065396730728, + "loss": 3.2247, + "step": 2559 + }, + { + "epoch": 3.75, + "grad_norm": 1.811521053314209, + "learning_rate": 0.00013756307512972538, + "loss": 3.2714, + "step": 2560 + }, + { + "epoch": 3.76, + "grad_norm": 5.335707664489746, + "learning_rate": 0.0001375196023227532, + "loss": 3.2079, + "step": 2561 + }, + { + "epoch": 3.76, + "grad_norm": 3.844973087310791, + "learning_rate": 0.0001374761212617199, + "loss": 3.2411, + "step": 2562 + }, + { + "epoch": 3.76, + "grad_norm": 4.505429267883301, + "learning_rate": 0.00013743263195619115, + "loss": 3.2508, + "step": 2563 + }, + { + "epoch": 3.76, + "grad_norm": 5.144504547119141, + "learning_rate": 0.00013738913441573418, + "loss": 3.2112, + "step": 2564 + }, + { + "epoch": 3.76, + "grad_norm": 23.4456787109375, + "learning_rate": 0.00013734562864991823, + "loss": 3.2106, + "step": 2565 + }, + { + "epoch": 3.76, + "grad_norm": 4.499874591827393, + "learning_rate": 0.0001373021146683142, + "loss": 3.2324, + "step": 2566 + }, + { + "epoch": 3.76, + "grad_norm": 2.9865691661834717, + "learning_rate": 0.000137258592480495, + "loss": 3.2498, + "step": 2567 + }, + { + "epoch": 3.77, + "grad_norm": 3.308457374572754, + "learning_rate": 0.00013721506209603512, + "loss": 3.2132, + "step": 2568 + }, + { + "epoch": 3.77, + "grad_norm": 2.9479405879974365, + "learning_rate": 0.000137171523524511, + "loss": 3.2817, + "step": 2569 + }, + { + "epoch": 3.77, + "grad_norm": 2.7257778644561768, + "learning_rate": 0.0001371279767755008, + "loss": 3.185, + "step": 2570 + }, + { + "epoch": 3.77, + "grad_norm": 1.7737255096435547, + "learning_rate": 0.00013708442185858454, + "loss": 3.222, + "step": 2571 + }, + { + "epoch": 3.77, + "grad_norm": 3.827313184738159, + "learning_rate": 0.000137040858783344, + "loss": 3.1943, + "step": 2572 + }, + { + "epoch": 3.77, + "grad_norm": 2.6353366374969482, + "learning_rate": 0.00013699728755936277, + "loss": 3.2699, + "step": 2573 + }, + { + "epoch": 3.78, + "grad_norm": 4.652822017669678, + "learning_rate": 0.0001369537081962262, + "loss": 3.2313, + "step": 2574 + }, + { + "epoch": 3.78, + "grad_norm": 8.656730651855469, + "learning_rate": 0.0001369101207035215, + "loss": 3.1737, + "step": 2575 + }, + { + "epoch": 3.78, + "grad_norm": 1.588462471961975, + "learning_rate": 0.0001368665250908376, + "loss": 3.2057, + "step": 2576 + }, + { + "epoch": 3.78, + "grad_norm": 5.477829456329346, + "learning_rate": 0.00013682292136776524, + "loss": 3.2045, + "step": 2577 + }, + { + "epoch": 3.78, + "grad_norm": 6.158089637756348, + "learning_rate": 0.00013677930954389698, + "loss": 3.2183, + "step": 2578 + }, + { + "epoch": 3.78, + "grad_norm": 2.4769415855407715, + "learning_rate": 0.00013673568962882705, + "loss": 3.1601, + "step": 2579 + }, + { + "epoch": 3.78, + "grad_norm": 7.979514122009277, + "learning_rate": 0.00013669206163215164, + "loss": 3.2011, + "step": 2580 + }, + { + "epoch": 3.79, + "grad_norm": 11.770708084106445, + "learning_rate": 0.00013664842556346856, + "loss": 3.2215, + "step": 2581 + }, + { + "epoch": 3.79, + "grad_norm": 7.018968105316162, + "learning_rate": 0.00013660478143237746, + "loss": 3.2163, + "step": 2582 + }, + { + "epoch": 3.79, + "grad_norm": 4.448172569274902, + "learning_rate": 0.00013656112924847978, + "loss": 3.2027, + "step": 2583 + }, + { + "epoch": 3.79, + "grad_norm": 6.86191987991333, + "learning_rate": 0.00013651746902137868, + "loss": 3.1812, + "step": 2584 + }, + { + "epoch": 3.79, + "grad_norm": 4.056680202484131, + "learning_rate": 0.0001364738007606791, + "loss": 3.2857, + "step": 2585 + }, + { + "epoch": 3.79, + "grad_norm": 5.344080924987793, + "learning_rate": 0.0001364301244759878, + "loss": 3.255, + "step": 2586 + }, + { + "epoch": 3.79, + "grad_norm": 7.4425554275512695, + "learning_rate": 0.00013638644017691326, + "loss": 3.2032, + "step": 2587 + }, + { + "epoch": 3.8, + "grad_norm": 4.709530830383301, + "learning_rate": 0.00013634274787306577, + "loss": 3.25, + "step": 2588 + }, + { + "epoch": 3.8, + "grad_norm": 7.912161827087402, + "learning_rate": 0.00013629904757405723, + "loss": 3.2362, + "step": 2589 + }, + { + "epoch": 3.8, + "grad_norm": 24.812524795532227, + "learning_rate": 0.00013625533928950153, + "loss": 3.231, + "step": 2590 + }, + { + "epoch": 3.8, + "grad_norm": 9.660811424255371, + "learning_rate": 0.0001362116230290142, + "loss": 3.2175, + "step": 2591 + }, + { + "epoch": 3.8, + "grad_norm": 6.260350704193115, + "learning_rate": 0.00013616789880221242, + "loss": 3.2169, + "step": 2592 + }, + { + "epoch": 3.8, + "grad_norm": 2.6841564178466797, + "learning_rate": 0.00013612416661871533, + "loss": 3.2084, + "step": 2593 + }, + { + "epoch": 3.8, + "grad_norm": 6.153440475463867, + "learning_rate": 0.00013608042648814364, + "loss": 3.2029, + "step": 2594 + }, + { + "epoch": 3.81, + "grad_norm": 2.9339137077331543, + "learning_rate": 0.00013603667842011994, + "loss": 3.1789, + "step": 2595 + }, + { + "epoch": 3.81, + "grad_norm": 5.685370922088623, + "learning_rate": 0.00013599292242426851, + "loss": 3.2023, + "step": 2596 + }, + { + "epoch": 3.81, + "grad_norm": 12.19677734375, + "learning_rate": 0.00013594915851021532, + "loss": 3.2433, + "step": 2597 + }, + { + "epoch": 3.81, + "grad_norm": 5.506028652191162, + "learning_rate": 0.00013590538668758822, + "loss": 3.2088, + "step": 2598 + }, + { + "epoch": 3.81, + "grad_norm": 1.9152709245681763, + "learning_rate": 0.00013586160696601665, + "loss": 3.2067, + "step": 2599 + }, + { + "epoch": 3.81, + "grad_norm": 3.531026601791382, + "learning_rate": 0.00013581781935513188, + "loss": 3.2368, + "step": 2600 + }, + { + "epoch": 3.82, + "grad_norm": 1.68514883518219, + "learning_rate": 0.00013577402386456692, + "loss": 3.2567, + "step": 2601 + }, + { + "epoch": 3.82, + "grad_norm": 2.6391441822052, + "learning_rate": 0.00013573022050395643, + "loss": 3.1804, + "step": 2602 + }, + { + "epoch": 3.82, + "grad_norm": 2.4722888469696045, + "learning_rate": 0.0001356864092829369, + "loss": 3.2318, + "step": 2603 + }, + { + "epoch": 3.82, + "grad_norm": 1.5304458141326904, + "learning_rate": 0.0001356425902111465, + "loss": 3.2145, + "step": 2604 + }, + { + "epoch": 3.82, + "grad_norm": 4.6478705406188965, + "learning_rate": 0.0001355987632982251, + "loss": 3.2201, + "step": 2605 + }, + { + "epoch": 3.82, + "grad_norm": 3.80239200592041, + "learning_rate": 0.00013555492855381432, + "loss": 3.1819, + "step": 2606 + }, + { + "epoch": 3.82, + "grad_norm": 2.0871238708496094, + "learning_rate": 0.0001355110859875576, + "loss": 3.2169, + "step": 2607 + }, + { + "epoch": 3.83, + "grad_norm": 1.9589481353759766, + "learning_rate": 0.00013546723560909992, + "loss": 3.2077, + "step": 2608 + }, + { + "epoch": 3.83, + "grad_norm": 2.8086769580841064, + "learning_rate": 0.00013542337742808807, + "loss": 3.2386, + "step": 2609 + }, + { + "epoch": 3.83, + "grad_norm": 1.3195122480392456, + "learning_rate": 0.0001353795114541706, + "loss": 3.1651, + "step": 2610 + }, + { + "epoch": 3.83, + "grad_norm": 4.268999099731445, + "learning_rate": 0.00013533563769699774, + "loss": 3.2113, + "step": 2611 + }, + { + "epoch": 3.83, + "grad_norm": 2.7233781814575195, + "learning_rate": 0.00013529175616622133, + "loss": 3.2103, + "step": 2612 + }, + { + "epoch": 3.83, + "grad_norm": 3.8539392948150635, + "learning_rate": 0.00013524786687149512, + "loss": 3.2368, + "step": 2613 + }, + { + "epoch": 3.83, + "grad_norm": 3.89041805267334, + "learning_rate": 0.00013520396982247442, + "loss": 3.1865, + "step": 2614 + }, + { + "epoch": 3.84, + "grad_norm": 2.546916961669922, + "learning_rate": 0.0001351600650288163, + "loss": 3.192, + "step": 2615 + }, + { + "epoch": 3.84, + "grad_norm": 1.615480661392212, + "learning_rate": 0.0001351161525001795, + "loss": 3.2953, + "step": 2616 + }, + { + "epoch": 3.84, + "grad_norm": 3.4749913215637207, + "learning_rate": 0.00013507223224622445, + "loss": 3.1994, + "step": 2617 + }, + { + "epoch": 3.84, + "grad_norm": 1.6761186122894287, + "learning_rate": 0.00013502830427661338, + "loss": 3.1836, + "step": 2618 + }, + { + "epoch": 3.84, + "grad_norm": 6.091635704040527, + "learning_rate": 0.00013498436860101011, + "loss": 3.2183, + "step": 2619 + }, + { + "epoch": 3.84, + "grad_norm": 5.926030158996582, + "learning_rate": 0.00013494042522908022, + "loss": 3.2222, + "step": 2620 + }, + { + "epoch": 3.84, + "grad_norm": 12.00096321105957, + "learning_rate": 0.00013489647417049095, + "loss": 3.2073, + "step": 2621 + }, + { + "epoch": 3.85, + "grad_norm": 4.7124762535095215, + "learning_rate": 0.0001348525154349112, + "loss": 3.1859, + "step": 2622 + }, + { + "epoch": 3.85, + "grad_norm": 3.3024585247039795, + "learning_rate": 0.0001348085490320116, + "loss": 3.177, + "step": 2623 + }, + { + "epoch": 3.85, + "grad_norm": 4.006857395172119, + "learning_rate": 0.00013476457497146456, + "loss": 3.2317, + "step": 2624 + }, + { + "epoch": 3.85, + "grad_norm": 4.812809467315674, + "learning_rate": 0.00013472059326294396, + "loss": 3.2637, + "step": 2625 + }, + { + "epoch": 3.85, + "grad_norm": 1.2310017347335815, + "learning_rate": 0.00013467660391612555, + "loss": 3.2366, + "step": 2626 + }, + { + "epoch": 3.85, + "grad_norm": 2.896761894226074, + "learning_rate": 0.00013463260694068666, + "loss": 3.2069, + "step": 2627 + }, + { + "epoch": 3.86, + "grad_norm": 2.2707176208496094, + "learning_rate": 0.00013458860234630632, + "loss": 3.224, + "step": 2628 + }, + { + "epoch": 3.86, + "grad_norm": 2.2991745471954346, + "learning_rate": 0.00013454459014266526, + "loss": 3.2074, + "step": 2629 + }, + { + "epoch": 3.86, + "grad_norm": 2.838233470916748, + "learning_rate": 0.0001345005703394459, + "loss": 3.2019, + "step": 2630 + }, + { + "epoch": 3.86, + "grad_norm": 2.686645030975342, + "learning_rate": 0.0001344565429463323, + "loss": 3.1873, + "step": 2631 + }, + { + "epoch": 3.86, + "grad_norm": 3.610236883163452, + "learning_rate": 0.00013441250797301016, + "loss": 3.1667, + "step": 2632 + }, + { + "epoch": 3.86, + "grad_norm": 3.012873888015747, + "learning_rate": 0.00013436846542916686, + "loss": 3.2006, + "step": 2633 + }, + { + "epoch": 3.86, + "grad_norm": 2.223695755004883, + "learning_rate": 0.00013432441532449152, + "loss": 3.2004, + "step": 2634 + }, + { + "epoch": 3.87, + "grad_norm": 5.295112609863281, + "learning_rate": 0.00013428035766867484, + "loss": 3.1959, + "step": 2635 + }, + { + "epoch": 3.87, + "grad_norm": 1.9531642198562622, + "learning_rate": 0.00013423629247140919, + "loss": 3.1796, + "step": 2636 + }, + { + "epoch": 3.87, + "grad_norm": 11.508848190307617, + "learning_rate": 0.00013419221974238864, + "loss": 3.2325, + "step": 2637 + }, + { + "epoch": 3.87, + "grad_norm": 3.0934455394744873, + "learning_rate": 0.00013414813949130894, + "loss": 3.1904, + "step": 2638 + }, + { + "epoch": 3.87, + "grad_norm": 1.843153476715088, + "learning_rate": 0.00013410405172786734, + "loss": 3.177, + "step": 2639 + }, + { + "epoch": 3.87, + "grad_norm": 3.7216010093688965, + "learning_rate": 0.00013405995646176294, + "loss": 3.2278, + "step": 2640 + }, + { + "epoch": 3.87, + "grad_norm": 4.631261825561523, + "learning_rate": 0.00013401585370269637, + "loss": 3.2772, + "step": 2641 + }, + { + "epoch": 3.88, + "grad_norm": 2.928041696548462, + "learning_rate": 0.00013397174346036996, + "loss": 3.2489, + "step": 2642 + }, + { + "epoch": 3.88, + "grad_norm": 2.3705878257751465, + "learning_rate": 0.00013392762574448766, + "loss": 3.2577, + "step": 2643 + }, + { + "epoch": 3.88, + "grad_norm": 4.091426372528076, + "learning_rate": 0.00013388350056475504, + "loss": 3.1664, + "step": 2644 + }, + { + "epoch": 3.88, + "grad_norm": 1.9827549457550049, + "learning_rate": 0.0001338393679308794, + "loss": 3.1786, + "step": 2645 + }, + { + "epoch": 3.88, + "grad_norm": 4.402117729187012, + "learning_rate": 0.0001337952278525696, + "loss": 3.1966, + "step": 2646 + }, + { + "epoch": 3.88, + "grad_norm": 5.21814489364624, + "learning_rate": 0.00013375108033953614, + "loss": 3.2393, + "step": 2647 + }, + { + "epoch": 3.88, + "grad_norm": 2.4417850971221924, + "learning_rate": 0.0001337069254014912, + "loss": 3.1729, + "step": 2648 + }, + { + "epoch": 3.89, + "grad_norm": 3.521346092224121, + "learning_rate": 0.0001336627630481486, + "loss": 3.1997, + "step": 2649 + }, + { + "epoch": 3.89, + "grad_norm": 4.180556297302246, + "learning_rate": 0.0001336185932892237, + "loss": 3.1878, + "step": 2650 + }, + { + "epoch": 3.89, + "grad_norm": 2.869370460510254, + "learning_rate": 0.00013357441613443354, + "loss": 3.2052, + "step": 2651 + }, + { + "epoch": 3.89, + "grad_norm": 2.097275495529175, + "learning_rate": 0.0001335302315934969, + "loss": 3.1459, + "step": 2652 + }, + { + "epoch": 3.89, + "grad_norm": 3.70109486579895, + "learning_rate": 0.00013348603967613398, + "loss": 3.2187, + "step": 2653 + }, + { + "epoch": 3.89, + "grad_norm": 3.026033878326416, + "learning_rate": 0.00013344184039206677, + "loss": 3.2251, + "step": 2654 + }, + { + "epoch": 3.9, + "grad_norm": 2.825608015060425, + "learning_rate": 0.00013339763375101876, + "loss": 3.1845, + "step": 2655 + }, + { + "epoch": 3.9, + "grad_norm": 2.3940205574035645, + "learning_rate": 0.00013335341976271518, + "loss": 3.1909, + "step": 2656 + }, + { + "epoch": 3.9, + "grad_norm": 3.781881332397461, + "learning_rate": 0.00013330919843688275, + "loss": 3.1861, + "step": 2657 + }, + { + "epoch": 3.9, + "grad_norm": 3.5430831909179688, + "learning_rate": 0.0001332649697832499, + "loss": 3.2879, + "step": 2658 + }, + { + "epoch": 3.9, + "grad_norm": 3.327752113342285, + "learning_rate": 0.00013322073381154663, + "loss": 3.1407, + "step": 2659 + }, + { + "epoch": 3.9, + "grad_norm": 3.319768190383911, + "learning_rate": 0.00013317649053150456, + "loss": 3.2215, + "step": 2660 + }, + { + "epoch": 3.9, + "grad_norm": 2.317915201187134, + "learning_rate": 0.0001331322399528569, + "loss": 3.251, + "step": 2661 + }, + { + "epoch": 3.91, + "grad_norm": 2.999483108520508, + "learning_rate": 0.00013308798208533853, + "loss": 3.2182, + "step": 2662 + }, + { + "epoch": 3.91, + "grad_norm": 4.7933549880981445, + "learning_rate": 0.0001330437169386858, + "loss": 3.2272, + "step": 2663 + }, + { + "epoch": 3.91, + "grad_norm": 4.440158367156982, + "learning_rate": 0.00013299944452263682, + "loss": 3.202, + "step": 2664 + }, + { + "epoch": 3.91, + "grad_norm": 1.8913938999176025, + "learning_rate": 0.0001329551648469312, + "loss": 3.1705, + "step": 2665 + }, + { + "epoch": 3.91, + "grad_norm": 4.039271831512451, + "learning_rate": 0.00013291087792131017, + "loss": 3.213, + "step": 2666 + }, + { + "epoch": 3.91, + "grad_norm": 4.9916276931762695, + "learning_rate": 0.00013286658375551654, + "loss": 3.2316, + "step": 2667 + }, + { + "epoch": 3.91, + "grad_norm": 2.624239206314087, + "learning_rate": 0.00013282228235929474, + "loss": 3.1902, + "step": 2668 + }, + { + "epoch": 3.92, + "grad_norm": 2.3713390827178955, + "learning_rate": 0.0001327779737423908, + "loss": 3.1698, + "step": 2669 + }, + { + "epoch": 3.92, + "grad_norm": 3.9648921489715576, + "learning_rate": 0.0001327336579145523, + "loss": 3.1811, + "step": 2670 + }, + { + "epoch": 3.92, + "grad_norm": 3.6182374954223633, + "learning_rate": 0.00013268933488552845, + "loss": 3.2013, + "step": 2671 + }, + { + "epoch": 3.92, + "grad_norm": 1.5024638175964355, + "learning_rate": 0.00013264500466506998, + "loss": 3.1996, + "step": 2672 + }, + { + "epoch": 3.92, + "grad_norm": 1.957371473312378, + "learning_rate": 0.00013260066726292926, + "loss": 3.188, + "step": 2673 + }, + { + "epoch": 3.92, + "grad_norm": 1.9777625799179077, + "learning_rate": 0.00013255632268886023, + "loss": 3.2656, + "step": 2674 + }, + { + "epoch": 3.93, + "grad_norm": 1.5048279762268066, + "learning_rate": 0.0001325119709526184, + "loss": 3.2279, + "step": 2675 + }, + { + "epoch": 3.93, + "grad_norm": 35.774295806884766, + "learning_rate": 0.0001324676120639608, + "loss": 3.2366, + "step": 2676 + }, + { + "epoch": 3.93, + "grad_norm": 3.9443695545196533, + "learning_rate": 0.00013242324603264616, + "loss": 3.1658, + "step": 2677 + }, + { + "epoch": 3.93, + "grad_norm": 2.9231414794921875, + "learning_rate": 0.0001323788728684347, + "loss": 3.2058, + "step": 2678 + }, + { + "epoch": 3.93, + "grad_norm": 4.9291157722473145, + "learning_rate": 0.0001323344925810882, + "loss": 3.1528, + "step": 2679 + }, + { + "epoch": 3.93, + "grad_norm": 8.999177932739258, + "learning_rate": 0.00013229010518037002, + "loss": 3.2864, + "step": 2680 + }, + { + "epoch": 3.93, + "grad_norm": 4.351472854614258, + "learning_rate": 0.0001322457106760451, + "loss": 3.2048, + "step": 2681 + }, + { + "epoch": 3.94, + "grad_norm": 9.449458122253418, + "learning_rate": 0.00013220130907787994, + "loss": 3.2409, + "step": 2682 + }, + { + "epoch": 3.94, + "grad_norm": 2.7245359420776367, + "learning_rate": 0.00013215690039564257, + "loss": 3.1844, + "step": 2683 + }, + { + "epoch": 3.94, + "grad_norm": 1.7337415218353271, + "learning_rate": 0.00013211248463910262, + "loss": 3.1732, + "step": 2684 + }, + { + "epoch": 3.94, + "grad_norm": 2.845367431640625, + "learning_rate": 0.0001320680618180313, + "loss": 3.1608, + "step": 2685 + }, + { + "epoch": 3.94, + "grad_norm": 2.7872724533081055, + "learning_rate": 0.00013202363194220125, + "loss": 3.2057, + "step": 2686 + }, + { + "epoch": 3.94, + "grad_norm": 1.4545655250549316, + "learning_rate": 0.00013197919502138677, + "loss": 3.205, + "step": 2687 + }, + { + "epoch": 3.94, + "grad_norm": 3.010228157043457, + "learning_rate": 0.00013193475106536374, + "loss": 3.1844, + "step": 2688 + }, + { + "epoch": 3.95, + "grad_norm": 2.508854866027832, + "learning_rate": 0.00013189030008390945, + "loss": 3.2108, + "step": 2689 + }, + { + "epoch": 3.95, + "grad_norm": 3.1477272510528564, + "learning_rate": 0.0001318458420868029, + "loss": 3.2018, + "step": 2690 + }, + { + "epoch": 3.95, + "grad_norm": 2.543921947479248, + "learning_rate": 0.0001318013770838245, + "loss": 3.2235, + "step": 2691 + }, + { + "epoch": 3.95, + "grad_norm": 3.577669382095337, + "learning_rate": 0.00013175690508475626, + "loss": 3.2087, + "step": 2692 + }, + { + "epoch": 3.95, + "grad_norm": 2.2817633152008057, + "learning_rate": 0.00013171242609938175, + "loss": 3.1847, + "step": 2693 + }, + { + "epoch": 3.95, + "grad_norm": 7.939913272857666, + "learning_rate": 0.00013166794013748598, + "loss": 3.2269, + "step": 2694 + }, + { + "epoch": 3.95, + "grad_norm": 6.408430576324463, + "learning_rate": 0.00013162344720885565, + "loss": 3.1853, + "step": 2695 + }, + { + "epoch": 3.96, + "grad_norm": 95.63408660888672, + "learning_rate": 0.00013157894732327886, + "loss": 3.2441, + "step": 2696 + }, + { + "epoch": 3.96, + "grad_norm": 76.839111328125, + "learning_rate": 0.00013153444049054523, + "loss": 3.2624, + "step": 2697 + }, + { + "epoch": 3.96, + "grad_norm": 9.534677505493164, + "learning_rate": 0.00013148992672044608, + "loss": 3.2354, + "step": 2698 + }, + { + "epoch": 3.96, + "grad_norm": 29.769962310791016, + "learning_rate": 0.0001314454060227741, + "loss": 3.1857, + "step": 2699 + }, + { + "epoch": 3.96, + "grad_norm": 26.48052215576172, + "learning_rate": 0.00013140087840732355, + "loss": 3.2512, + "step": 2700 + }, + { + "epoch": 3.96, + "grad_norm": 12.384693145751953, + "learning_rate": 0.00013135634388389015, + "loss": 3.2054, + "step": 2701 + }, + { + "epoch": 3.97, + "grad_norm": 4.66050910949707, + "learning_rate": 0.00013131180246227125, + "loss": 3.2012, + "step": 2702 + }, + { + "epoch": 3.97, + "grad_norm": 28.54119300842285, + "learning_rate": 0.00013126725415226563, + "loss": 3.202, + "step": 2703 + }, + { + "epoch": 3.97, + "grad_norm": 69.08728790283203, + "learning_rate": 0.00013122269896367365, + "loss": 3.2423, + "step": 2704 + }, + { + "epoch": 3.97, + "eval_loss": 3.091963529586792, + "eval_runtime": 1165.9245, + "eval_samples_per_second": 14.702, + "eval_steps_per_second": 0.919, + "step": 2704 + }, + { + "epoch": 3.97, + "grad_norm": 22.657764434814453, + "learning_rate": 0.00013117813690629714, + "loss": 3.2586, + "step": 2705 + }, + { + "epoch": 3.97, + "grad_norm": 74.64742279052734, + "learning_rate": 0.00013113356798993947, + "loss": 3.2606, + "step": 2706 + }, + { + "epoch": 3.97, + "grad_norm": 14.374595642089844, + "learning_rate": 0.0001310889922244055, + "loss": 3.2525, + "step": 2707 + }, + { + "epoch": 3.97, + "grad_norm": 5.337686538696289, + "learning_rate": 0.00013104440961950154, + "loss": 3.2355, + "step": 2708 + }, + { + "epoch": 3.98, + "grad_norm": 3.205595016479492, + "learning_rate": 0.00013099982018503558, + "loss": 3.2323, + "step": 2709 + }, + { + "epoch": 3.98, + "grad_norm": 4.343560218811035, + "learning_rate": 0.0001309552239308169, + "loss": 3.2265, + "step": 2710 + }, + { + "epoch": 3.98, + "grad_norm": 68.85613250732422, + "learning_rate": 0.0001309106208666564, + "loss": 3.2412, + "step": 2711 + }, + { + "epoch": 3.98, + "grad_norm": 4.4929633140563965, + "learning_rate": 0.00013086601100236646, + "loss": 3.2243, + "step": 2712 + }, + { + "epoch": 3.98, + "grad_norm": 5.574833393096924, + "learning_rate": 0.000130821394347761, + "loss": 3.2, + "step": 2713 + }, + { + "epoch": 3.98, + "grad_norm": 2.189110040664673, + "learning_rate": 0.00013077677091265528, + "loss": 3.2177, + "step": 2714 + }, + { + "epoch": 3.98, + "grad_norm": 1.2807992696762085, + "learning_rate": 0.00013073214070686626, + "loss": 3.1971, + "step": 2715 + }, + { + "epoch": 3.99, + "grad_norm": 3.459458351135254, + "learning_rate": 0.0001306875037402122, + "loss": 3.186, + "step": 2716 + }, + { + "epoch": 3.99, + "grad_norm": 9.210981369018555, + "learning_rate": 0.000130642860022513, + "loss": 3.1966, + "step": 2717 + }, + { + "epoch": 3.99, + "grad_norm": 2.803443670272827, + "learning_rate": 0.00013059820956358998, + "loss": 3.1927, + "step": 2718 + }, + { + "epoch": 3.99, + "grad_norm": 2.078461170196533, + "learning_rate": 0.00013055355237326588, + "loss": 3.1776, + "step": 2719 + }, + { + "epoch": 3.99, + "grad_norm": 2.4591166973114014, + "learning_rate": 0.00013050888846136502, + "loss": 3.2126, + "step": 2720 + }, + { + "epoch": 3.99, + "grad_norm": 2.6859610080718994, + "learning_rate": 0.00013046421783771318, + "loss": 3.1956, + "step": 2721 + }, + { + "epoch": 3.99, + "grad_norm": 3.064992904663086, + "learning_rate": 0.00013041954051213757, + "loss": 3.25, + "step": 2722 + }, + { + "epoch": 4.0, + "grad_norm": 1.7159329652786255, + "learning_rate": 0.0001303748564944669, + "loss": 3.1734, + "step": 2723 + }, + { + "epoch": 4.0, + "grad_norm": 2.7244114875793457, + "learning_rate": 0.00013033016579453138, + "loss": 3.1743, + "step": 2724 + }, + { + "epoch": 4.0, + "grad_norm": 2.2309274673461914, + "learning_rate": 0.0001302854684221626, + "loss": 3.1751, + "step": 2725 + }, + { + "epoch": 4.0, + "grad_norm": 1.5704296827316284, + "learning_rate": 0.0001302407643871938, + "loss": 3.2018, + "step": 2726 + }, + { + "epoch": 4.0, + "grad_norm": 2.5336227416992188, + "learning_rate": 0.0001301960536994595, + "loss": 3.2139, + "step": 2727 + }, + { + "epoch": 4.0, + "grad_norm": 1.6835451126098633, + "learning_rate": 0.00013015133636879567, + "loss": 3.2198, + "step": 2728 + }, + { + "epoch": 4.01, + "grad_norm": 1.5716129541397095, + "learning_rate": 0.00013010661240503999, + "loss": 3.2108, + "step": 2729 + }, + { + "epoch": 4.01, + "grad_norm": 1.5545756816864014, + "learning_rate": 0.0001300618818180313, + "loss": 3.1592, + "step": 2730 + }, + { + "epoch": 4.01, + "grad_norm": 1.9438145160675049, + "learning_rate": 0.00013001714461761008, + "loss": 3.2145, + "step": 2731 + }, + { + "epoch": 4.01, + "grad_norm": 1.7333067655563354, + "learning_rate": 0.00012997240081361823, + "loss": 3.1362, + "step": 2732 + }, + { + "epoch": 4.01, + "grad_norm": 1.632812261581421, + "learning_rate": 0.00012992765041589904, + "loss": 3.2064, + "step": 2733 + }, + { + "epoch": 4.01, + "grad_norm": 1.1884607076644897, + "learning_rate": 0.00012988289343429733, + "loss": 3.1789, + "step": 2734 + }, + { + "epoch": 4.01, + "grad_norm": 2.9912478923797607, + "learning_rate": 0.00012983812987865935, + "loss": 3.1561, + "step": 2735 + }, + { + "epoch": 4.02, + "grad_norm": 1.3816310167312622, + "learning_rate": 0.00012979335975883276, + "loss": 3.1848, + "step": 2736 + }, + { + "epoch": 4.0, + "grad_norm": 19.01185417175293, + "learning_rate": 0.00012974858308466666, + "loss": 3.1801, + "step": 2737 + }, + { + "epoch": 4.0, + "grad_norm": 5.716503143310547, + "learning_rate": 0.0001297037998660117, + "loss": 3.2216, + "step": 2738 + }, + { + "epoch": 4.0, + "grad_norm": 2.985485792160034, + "learning_rate": 0.00012965901011271978, + "loss": 3.2136, + "step": 2739 + }, + { + "epoch": 4.01, + "grad_norm": 2.645477056503296, + "learning_rate": 0.00012961421383464445, + "loss": 3.176, + "step": 2740 + }, + { + "epoch": 4.01, + "grad_norm": 2.804504871368408, + "learning_rate": 0.0001295694110416405, + "loss": 3.1502, + "step": 2741 + }, + { + "epoch": 4.01, + "grad_norm": 2.257525682449341, + "learning_rate": 0.00012952460174356432, + "loss": 3.1909, + "step": 2742 + }, + { + "epoch": 4.01, + "grad_norm": 2.525448799133301, + "learning_rate": 0.0001294797859502736, + "loss": 3.2058, + "step": 2743 + }, + { + "epoch": 4.01, + "grad_norm": 2.589095115661621, + "learning_rate": 0.00012943496367162754, + "loss": 3.2137, + "step": 2744 + }, + { + "epoch": 4.01, + "grad_norm": 1.6353000402450562, + "learning_rate": 0.00012939013491748672, + "loss": 3.2001, + "step": 2745 + }, + { + "epoch": 4.01, + "grad_norm": 2.5494680404663086, + "learning_rate": 0.00012934529969771324, + "loss": 3.1529, + "step": 2746 + }, + { + "epoch": 4.02, + "grad_norm": 2.1689844131469727, + "learning_rate": 0.00012930045802217043, + "loss": 3.1931, + "step": 2747 + }, + { + "epoch": 4.02, + "grad_norm": 3.0633785724639893, + "learning_rate": 0.00012925560990072324, + "loss": 3.1941, + "step": 2748 + }, + { + "epoch": 4.02, + "grad_norm": 2.6445438861846924, + "learning_rate": 0.00012921075534323795, + "loss": 3.1673, + "step": 2749 + }, + { + "epoch": 4.02, + "grad_norm": 1.6328538656234741, + "learning_rate": 0.00012916589435958223, + "loss": 3.1533, + "step": 2750 + }, + { + "epoch": 4.02, + "grad_norm": 7.05519962310791, + "learning_rate": 0.0001291210269596252, + "loss": 3.1762, + "step": 2751 + }, + { + "epoch": 4.02, + "grad_norm": 10.450418472290039, + "learning_rate": 0.0001290761531532374, + "loss": 3.2032, + "step": 2752 + }, + { + "epoch": 4.02, + "grad_norm": 3.726336717605591, + "learning_rate": 0.00012903127295029078, + "loss": 3.1835, + "step": 2753 + }, + { + "epoch": 4.03, + "grad_norm": 2.4860239028930664, + "learning_rate": 0.00012898638636065868, + "loss": 3.1818, + "step": 2754 + }, + { + "epoch": 4.03, + "grad_norm": 2.166196584701538, + "learning_rate": 0.00012894149339421582, + "loss": 3.1609, + "step": 2755 + }, + { + "epoch": 4.03, + "grad_norm": 2.179659128189087, + "learning_rate": 0.00012889659406083837, + "loss": 3.2125, + "step": 2756 + }, + { + "epoch": 4.03, + "grad_norm": 2.2733445167541504, + "learning_rate": 0.0001288516883704039, + "loss": 3.2931, + "step": 2757 + }, + { + "epoch": 4.03, + "grad_norm": 1.3774925470352173, + "learning_rate": 0.00012880677633279132, + "loss": 3.1741, + "step": 2758 + }, + { + "epoch": 4.03, + "grad_norm": 4.090391159057617, + "learning_rate": 0.000128761857957881, + "loss": 3.2181, + "step": 2759 + }, + { + "epoch": 4.04, + "grad_norm": 3.762101411819458, + "learning_rate": 0.00012871693325555472, + "loss": 3.1991, + "step": 2760 + }, + { + "epoch": 4.04, + "grad_norm": 2.435084819793701, + "learning_rate": 0.00012867200223569555, + "loss": 3.1734, + "step": 2761 + }, + { + "epoch": 4.04, + "grad_norm": 11.30912971496582, + "learning_rate": 0.00012862706490818804, + "loss": 3.1755, + "step": 2762 + }, + { + "epoch": 4.04, + "grad_norm": 3.355214834213257, + "learning_rate": 0.0001285821212829181, + "loss": 3.1718, + "step": 2763 + }, + { + "epoch": 4.04, + "grad_norm": 2.4179513454437256, + "learning_rate": 0.00012853717136977305, + "loss": 3.1722, + "step": 2764 + }, + { + "epoch": 4.04, + "grad_norm": 2.18635630607605, + "learning_rate": 0.0001284922151786415, + "loss": 3.1982, + "step": 2765 + }, + { + "epoch": 4.04, + "grad_norm": 2.3029189109802246, + "learning_rate": 0.0001284472527194136, + "loss": 3.2427, + "step": 2766 + }, + { + "epoch": 4.05, + "grad_norm": 2.3949828147888184, + "learning_rate": 0.00012840228400198071, + "loss": 3.1944, + "step": 2767 + }, + { + "epoch": 4.05, + "grad_norm": 2.5206806659698486, + "learning_rate": 0.0001283573090362357, + "loss": 3.1899, + "step": 2768 + }, + { + "epoch": 4.05, + "grad_norm": 1.8850032091140747, + "learning_rate": 0.00012831232783207277, + "loss": 3.2212, + "step": 2769 + }, + { + "epoch": 4.05, + "grad_norm": 2.58017897605896, + "learning_rate": 0.0001282673403993874, + "loss": 3.2026, + "step": 2770 + }, + { + "epoch": 4.05, + "grad_norm": 2.990920305252075, + "learning_rate": 0.00012822234674807662, + "loss": 3.1852, + "step": 2771 + }, + { + "epoch": 4.05, + "grad_norm": 0.5992396473884583, + "learning_rate": 0.00012817734688803868, + "loss": 3.1736, + "step": 2772 + }, + { + "epoch": 4.05, + "grad_norm": 2.821469306945801, + "learning_rate": 0.0001281323408291733, + "loss": 3.1999, + "step": 2773 + }, + { + "epoch": 4.06, + "grad_norm": 15.83332633972168, + "learning_rate": 0.00012808732858138146, + "loss": 3.1844, + "step": 2774 + }, + { + "epoch": 4.06, + "grad_norm": 2.910820960998535, + "learning_rate": 0.0001280423101545655, + "loss": 3.1587, + "step": 2775 + }, + { + "epoch": 4.06, + "grad_norm": 17.933792114257812, + "learning_rate": 0.00012799728555862934, + "loss": 3.2201, + "step": 2776 + }, + { + "epoch": 4.06, + "grad_norm": 2.721127986907959, + "learning_rate": 0.0001279522548034779, + "loss": 3.1507, + "step": 2777 + }, + { + "epoch": 4.06, + "grad_norm": 1.5368844270706177, + "learning_rate": 0.00012790721789901775, + "loss": 3.1902, + "step": 2778 + }, + { + "epoch": 4.06, + "grad_norm": 3.0113532543182373, + "learning_rate": 0.00012786217485515674, + "loss": 3.149, + "step": 2779 + }, + { + "epoch": 4.06, + "grad_norm": 19.78055191040039, + "learning_rate": 0.00012781712568180393, + "loss": 3.1755, + "step": 2780 + }, + { + "epoch": 4.07, + "grad_norm": 2.508270025253296, + "learning_rate": 0.0001277720703888699, + "loss": 3.1915, + "step": 2781 + }, + { + "epoch": 4.07, + "grad_norm": 2.167954206466675, + "learning_rate": 0.00012772700898626652, + "loss": 3.2054, + "step": 2782 + }, + { + "epoch": 4.07, + "grad_norm": 4.016536235809326, + "learning_rate": 0.00012768194148390696, + "loss": 3.195, + "step": 2783 + }, + { + "epoch": 4.07, + "grad_norm": 3.441601276397705, + "learning_rate": 0.0001276368678917058, + "loss": 3.15, + "step": 2784 + }, + { + "epoch": 4.07, + "grad_norm": 7.370872974395752, + "learning_rate": 0.0001275917882195789, + "loss": 3.1633, + "step": 2785 + }, + { + "epoch": 4.07, + "grad_norm": 2.296081781387329, + "learning_rate": 0.00012754670247744354, + "loss": 3.2345, + "step": 2786 + }, + { + "epoch": 4.08, + "grad_norm": 2.009742259979248, + "learning_rate": 0.00012750161067521826, + "loss": 3.1938, + "step": 2787 + }, + { + "epoch": 4.08, + "grad_norm": 4.299360752105713, + "learning_rate": 0.00012745651282282288, + "loss": 3.2213, + "step": 2788 + }, + { + "epoch": 4.08, + "grad_norm": 3.636892080307007, + "learning_rate": 0.00012741140893017872, + "loss": 3.1713, + "step": 2789 + }, + { + "epoch": 4.08, + "grad_norm": 1.2036387920379639, + "learning_rate": 0.0001273662990072083, + "loss": 3.2107, + "step": 2790 + }, + { + "epoch": 4.08, + "grad_norm": 3.7308552265167236, + "learning_rate": 0.0001273211830638355, + "loss": 3.1955, + "step": 2791 + }, + { + "epoch": 4.08, + "grad_norm": 9.248310089111328, + "learning_rate": 0.00012727606110998549, + "loss": 3.1917, + "step": 2792 + }, + { + "epoch": 4.08, + "grad_norm": 15.789203643798828, + "learning_rate": 0.00012723093315558488, + "loss": 3.1611, + "step": 2793 + }, + { + "epoch": 4.09, + "grad_norm": 3.4315154552459717, + "learning_rate": 0.00012718579921056143, + "loss": 3.2364, + "step": 2794 + }, + { + "epoch": 4.09, + "grad_norm": 2.061835289001465, + "learning_rate": 0.00012714065928484436, + "loss": 3.2079, + "step": 2795 + }, + { + "epoch": 4.09, + "grad_norm": 2.4200713634490967, + "learning_rate": 0.00012709551338836413, + "loss": 3.1846, + "step": 2796 + }, + { + "epoch": 4.09, + "grad_norm": 3.561788558959961, + "learning_rate": 0.00012705036153105257, + "loss": 3.2151, + "step": 2797 + }, + { + "epoch": 4.09, + "grad_norm": 2.967207908630371, + "learning_rate": 0.00012700520372284273, + "loss": 3.2036, + "step": 2798 + }, + { + "epoch": 4.09, + "grad_norm": 3.490912437438965, + "learning_rate": 0.00012696003997366906, + "loss": 3.1893, + "step": 2799 + }, + { + "epoch": 4.09, + "grad_norm": 2.477809429168701, + "learning_rate": 0.00012691487029346728, + "loss": 3.1701, + "step": 2800 + }, + { + "epoch": 4.1, + "grad_norm": 3.558258295059204, + "learning_rate": 0.0001268696946921744, + "loss": 3.2343, + "step": 2801 + }, + { + "epoch": 4.1, + "grad_norm": 3.1721646785736084, + "learning_rate": 0.00012682451317972874, + "loss": 3.201, + "step": 2802 + }, + { + "epoch": 4.1, + "grad_norm": 2.562236785888672, + "learning_rate": 0.00012677932576606998, + "loss": 3.2291, + "step": 2803 + }, + { + "epoch": 4.1, + "grad_norm": 9.012580871582031, + "learning_rate": 0.00012673413246113905, + "loss": 3.2074, + "step": 2804 + }, + { + "epoch": 4.1, + "grad_norm": 2.1724753379821777, + "learning_rate": 0.00012668893327487805, + "loss": 3.2104, + "step": 2805 + }, + { + "epoch": 4.1, + "grad_norm": 1.8097708225250244, + "learning_rate": 0.00012664372821723064, + "loss": 3.1534, + "step": 2806 + }, + { + "epoch": 4.11, + "grad_norm": 2.409672498703003, + "learning_rate": 0.00012659851729814162, + "loss": 3.1748, + "step": 2807 + }, + { + "epoch": 4.11, + "grad_norm": 2.8095180988311768, + "learning_rate": 0.000126553300527557, + "loss": 3.1996, + "step": 2808 + }, + { + "epoch": 4.11, + "grad_norm": 2.353957176208496, + "learning_rate": 0.00012650807791542422, + "loss": 3.1945, + "step": 2809 + }, + { + "epoch": 4.11, + "grad_norm": 3.013678789138794, + "learning_rate": 0.00012646284947169195, + "loss": 3.1839, + "step": 2810 + }, + { + "epoch": 4.11, + "grad_norm": 2.154859781265259, + "learning_rate": 0.00012641761520631016, + "loss": 3.1541, + "step": 2811 + }, + { + "epoch": 4.11, + "grad_norm": 2.0929133892059326, + "learning_rate": 0.00012637237512923006, + "loss": 3.1655, + "step": 2812 + }, + { + "epoch": 4.11, + "grad_norm": 1.2277342081069946, + "learning_rate": 0.0001263271292504042, + "loss": 3.1933, + "step": 2813 + }, + { + "epoch": 4.12, + "grad_norm": 2.6938316822052, + "learning_rate": 0.0001262818775797863, + "loss": 3.1441, + "step": 2814 + }, + { + "epoch": 4.12, + "grad_norm": 1.9652117490768433, + "learning_rate": 0.00012623662012733146, + "loss": 3.1616, + "step": 2815 + }, + { + "epoch": 4.12, + "grad_norm": 1.6476244926452637, + "learning_rate": 0.00012619135690299603, + "loss": 3.1839, + "step": 2816 + }, + { + "epoch": 4.12, + "grad_norm": 1.534227967262268, + "learning_rate": 0.00012614608791673763, + "loss": 3.1646, + "step": 2817 + }, + { + "epoch": 4.12, + "grad_norm": 2.850703477859497, + "learning_rate": 0.00012610081317851511, + "loss": 3.207, + "step": 2818 + }, + { + "epoch": 4.12, + "grad_norm": 2.1084752082824707, + "learning_rate": 0.00012605553269828858, + "loss": 3.1688, + "step": 2819 + }, + { + "epoch": 4.12, + "grad_norm": 2.303504705429077, + "learning_rate": 0.0001260102464860195, + "loss": 3.229, + "step": 2820 + }, + { + "epoch": 4.13, + "grad_norm": 1.5470285415649414, + "learning_rate": 0.0001259649545516705, + "loss": 3.1814, + "step": 2821 + }, + { + "epoch": 4.13, + "grad_norm": 1.4991244077682495, + "learning_rate": 0.0001259196569052055, + "loss": 3.1927, + "step": 2822 + }, + { + "epoch": 4.13, + "grad_norm": 1.342314600944519, + "learning_rate": 0.0001258743535565897, + "loss": 3.1743, + "step": 2823 + }, + { + "epoch": 4.13, + "grad_norm": 2.662546396255493, + "learning_rate": 0.0001258290445157895, + "loss": 3.1884, + "step": 2824 + }, + { + "epoch": 4.13, + "grad_norm": 1.1698952913284302, + "learning_rate": 0.00012578372979277265, + "loss": 3.1841, + "step": 2825 + }, + { + "epoch": 4.13, + "grad_norm": 2.8144288063049316, + "learning_rate": 0.000125738409397508, + "loss": 3.1634, + "step": 2826 + }, + { + "epoch": 4.13, + "grad_norm": 1.0438013076782227, + "learning_rate": 0.00012569308333996578, + "loss": 3.1828, + "step": 2827 + }, + { + "epoch": 4.14, + "grad_norm": 5.697520732879639, + "learning_rate": 0.0001256477516301174, + "loss": 3.1897, + "step": 2828 + }, + { + "epoch": 4.14, + "grad_norm": 6.039661884307861, + "learning_rate": 0.00012560241427793555, + "loss": 3.2308, + "step": 2829 + }, + { + "epoch": 4.14, + "grad_norm": 51.954402923583984, + "learning_rate": 0.00012555707129339417, + "loss": 3.1693, + "step": 2830 + }, + { + "epoch": 4.14, + "grad_norm": 5.388270378112793, + "learning_rate": 0.00012551172268646834, + "loss": 3.2463, + "step": 2831 + }, + { + "epoch": 4.14, + "grad_norm": 4.189896583557129, + "learning_rate": 0.00012546636846713454, + "loss": 3.209, + "step": 2832 + }, + { + "epoch": 4.14, + "grad_norm": 2.421431064605713, + "learning_rate": 0.0001254210086453703, + "loss": 3.2312, + "step": 2833 + }, + { + "epoch": 4.15, + "grad_norm": 4.315710544586182, + "learning_rate": 0.00012537564323115454, + "loss": 3.2011, + "step": 2834 + }, + { + "epoch": 4.15, + "grad_norm": 1.7199636697769165, + "learning_rate": 0.00012533027223446733, + "loss": 3.1634, + "step": 2835 + }, + { + "epoch": 4.15, + "grad_norm": 4.567407608032227, + "learning_rate": 0.00012528489566528996, + "loss": 3.1999, + "step": 2836 + }, + { + "epoch": 4.15, + "grad_norm": 5.22501277923584, + "learning_rate": 0.00012523951353360504, + "loss": 3.1743, + "step": 2837 + }, + { + "epoch": 4.15, + "grad_norm": 2.197282314300537, + "learning_rate": 0.00012519412584939627, + "loss": 3.2332, + "step": 2838 + }, + { + "epoch": 4.15, + "grad_norm": 5.91029167175293, + "learning_rate": 0.00012514873262264864, + "loss": 3.1713, + "step": 2839 + }, + { + "epoch": 4.15, + "grad_norm": 6.8516716957092285, + "learning_rate": 0.0001251033338633484, + "loss": 3.1908, + "step": 2840 + }, + { + "epoch": 4.16, + "grad_norm": 3.321195125579834, + "learning_rate": 0.00012505792958148292, + "loss": 3.1693, + "step": 2841 + }, + { + "epoch": 4.16, + "grad_norm": 11.536995887756348, + "learning_rate": 0.00012501251978704088, + "loss": 3.1626, + "step": 2842 + }, + { + "epoch": 4.16, + "grad_norm": 9.267814636230469, + "learning_rate": 0.0001249671044900121, + "loss": 3.1807, + "step": 2843 + }, + { + "epoch": 4.16, + "grad_norm": 8.332723617553711, + "learning_rate": 0.00012492168370038765, + "loss": 3.1934, + "step": 2844 + }, + { + "epoch": 4.16, + "grad_norm": 3.4367361068725586, + "learning_rate": 0.00012487625742815985, + "loss": 3.1797, + "step": 2845 + }, + { + "epoch": 4.16, + "grad_norm": 6.636847019195557, + "learning_rate": 0.00012483082568332207, + "loss": 3.1851, + "step": 2846 + }, + { + "epoch": 4.16, + "grad_norm": 8.669164657592773, + "learning_rate": 0.00012478538847586907, + "loss": 3.2061, + "step": 2847 + }, + { + "epoch": 4.17, + "grad_norm": 6.34977912902832, + "learning_rate": 0.00012473994581579673, + "loss": 3.196, + "step": 2848 + }, + { + "epoch": 4.17, + "grad_norm": 4.2629780769348145, + "learning_rate": 0.00012469449771310208, + "loss": 3.2128, + "step": 2849 + }, + { + "epoch": 4.17, + "grad_norm": 5.895789623260498, + "learning_rate": 0.00012464904417778345, + "loss": 3.2769, + "step": 2850 + }, + { + "epoch": 4.17, + "grad_norm": 7.333654403686523, + "learning_rate": 0.0001246035852198403, + "loss": 3.2108, + "step": 2851 + }, + { + "epoch": 4.17, + "grad_norm": 2.965632915496826, + "learning_rate": 0.00012455812084927325, + "loss": 3.2012, + "step": 2852 + }, + { + "epoch": 4.17, + "grad_norm": 8.51198673248291, + "learning_rate": 0.0001245126510760842, + "loss": 3.2149, + "step": 2853 + }, + { + "epoch": 4.17, + "grad_norm": 7.021212100982666, + "learning_rate": 0.00012446717591027624, + "loss": 3.2624, + "step": 2854 + }, + { + "epoch": 4.18, + "grad_norm": 1.8178749084472656, + "learning_rate": 0.0001244216953618535, + "loss": 3.2099, + "step": 2855 + }, + { + "epoch": 4.18, + "grad_norm": 11.300747871398926, + "learning_rate": 0.00012437620944082145, + "loss": 3.1795, + "step": 2856 + }, + { + "epoch": 4.18, + "grad_norm": 9.994318962097168, + "learning_rate": 0.0001243307181571867, + "loss": 3.2285, + "step": 2857 + }, + { + "epoch": 4.18, + "grad_norm": 11.500404357910156, + "learning_rate": 0.00012428522152095703, + "loss": 3.1869, + "step": 2858 + }, + { + "epoch": 4.18, + "grad_norm": 5.326147556304932, + "learning_rate": 0.00012423971954214137, + "loss": 3.2226, + "step": 2859 + }, + { + "epoch": 4.18, + "grad_norm": 5.258158206939697, + "learning_rate": 0.00012419421223074982, + "loss": 3.1848, + "step": 2860 + }, + { + "epoch": 4.19, + "grad_norm": 7.557288646697998, + "learning_rate": 0.00012414869959679375, + "loss": 3.2594, + "step": 2861 + }, + { + "epoch": 4.19, + "grad_norm": 19.032909393310547, + "learning_rate": 0.0001241031816502856, + "loss": 3.2464, + "step": 2862 + }, + { + "epoch": 4.19, + "grad_norm": 9.38792896270752, + "learning_rate": 0.00012405765840123903, + "loss": 3.4169, + "step": 2863 + }, + { + "epoch": 4.19, + "grad_norm": 58.51255798339844, + "learning_rate": 0.0001240121298596688, + "loss": 3.3102, + "step": 2864 + }, + { + "epoch": 4.19, + "grad_norm": 9.871623992919922, + "learning_rate": 0.00012396659603559099, + "loss": 3.3069, + "step": 2865 + }, + { + "epoch": 4.19, + "grad_norm": 18.70452117919922, + "learning_rate": 0.00012392105693902262, + "loss": 3.2645, + "step": 2866 + }, + { + "epoch": 4.19, + "grad_norm": 6.495476245880127, + "learning_rate": 0.00012387551257998206, + "loss": 3.2188, + "step": 2867 + }, + { + "epoch": 4.2, + "grad_norm": 4.516813278198242, + "learning_rate": 0.00012382996296848875, + "loss": 3.2188, + "step": 2868 + }, + { + "epoch": 4.2, + "grad_norm": 9.024648666381836, + "learning_rate": 0.00012378440811456324, + "loss": 3.251, + "step": 2869 + }, + { + "epoch": 4.2, + "grad_norm": 10.055442810058594, + "learning_rate": 0.00012373884802822737, + "loss": 3.2687, + "step": 2870 + }, + { + "epoch": 4.2, + "grad_norm": 5.984691619873047, + "learning_rate": 0.00012369328271950405, + "loss": 3.2402, + "step": 2871 + }, + { + "epoch": 4.2, + "grad_norm": 4.42611026763916, + "learning_rate": 0.00012364771219841727, + "loss": 3.2241, + "step": 2872 + }, + { + "epoch": 4.2, + "grad_norm": 5.240525722503662, + "learning_rate": 0.0001236021364749923, + "loss": 3.2565, + "step": 2873 + }, + { + "epoch": 4.2, + "eval_loss": 3.031118154525757, + "eval_runtime": 1166.2283, + "eval_samples_per_second": 14.699, + "eval_steps_per_second": 0.919, + "step": 2873 + }, + { + "epoch": 4.2, + "grad_norm": 3.474372148513794, + "learning_rate": 0.00012355655555925548, + "loss": 3.1972, + "step": 2874 + }, + { + "epoch": 4.21, + "grad_norm": 3.6792542934417725, + "learning_rate": 0.00012351096946123431, + "loss": 3.1889, + "step": 2875 + }, + { + "epoch": 4.21, + "grad_norm": 3.1051828861236572, + "learning_rate": 0.00012346537819095743, + "loss": 3.206, + "step": 2876 + }, + { + "epoch": 4.21, + "grad_norm": 2.708695411682129, + "learning_rate": 0.00012341978175845462, + "loss": 3.2389, + "step": 2877 + }, + { + "epoch": 4.21, + "grad_norm": 2.4177584648132324, + "learning_rate": 0.00012337418017375676, + "loss": 3.1948, + "step": 2878 + }, + { + "epoch": 4.21, + "grad_norm": 3.8860323429107666, + "learning_rate": 0.00012332857344689587, + "loss": 3.2037, + "step": 2879 + }, + { + "epoch": 4.21, + "grad_norm": 4.779682159423828, + "learning_rate": 0.0001232829615879052, + "loss": 3.1901, + "step": 2880 + }, + { + "epoch": 4.21, + "grad_norm": 2.2922635078430176, + "learning_rate": 0.00012323734460681902, + "loss": 3.1367, + "step": 2881 + }, + { + "epoch": 4.22, + "grad_norm": 2.1851584911346436, + "learning_rate": 0.00012319172251367276, + "loss": 3.1782, + "step": 2882 + }, + { + "epoch": 4.22, + "grad_norm": 3.3458876609802246, + "learning_rate": 0.00012314609531850293, + "loss": 3.2361, + "step": 2883 + }, + { + "epoch": 4.22, + "grad_norm": 3.9705002307891846, + "learning_rate": 0.00012310046303134732, + "loss": 3.2227, + "step": 2884 + }, + { + "epoch": 4.22, + "grad_norm": 2.2800686359405518, + "learning_rate": 0.0001230548256622446, + "loss": 3.2052, + "step": 2885 + }, + { + "epoch": 4.22, + "grad_norm": 4.4462199211120605, + "learning_rate": 0.00012300918322123475, + "loss": 3.2142, + "step": 2886 + }, + { + "epoch": 4.22, + "grad_norm": 3.9958441257476807, + "learning_rate": 0.0001229635357183588, + "loss": 3.1989, + "step": 2887 + }, + { + "epoch": 4.23, + "grad_norm": 3.798068046569824, + "learning_rate": 0.00012291788316365888, + "loss": 3.1956, + "step": 2888 + }, + { + "epoch": 4.23, + "grad_norm": 5.135157108306885, + "learning_rate": 0.00012287222556717827, + "loss": 3.1854, + "step": 2889 + }, + { + "epoch": 4.23, + "grad_norm": 2.207172393798828, + "learning_rate": 0.0001228265629389613, + "loss": 3.1583, + "step": 2890 + }, + { + "epoch": 4.23, + "grad_norm": 8.646849632263184, + "learning_rate": 0.00012278089528905346, + "loss": 3.245, + "step": 2891 + }, + { + "epoch": 4.23, + "grad_norm": 7.601872444152832, + "learning_rate": 0.00012273522262750132, + "loss": 3.1968, + "step": 2892 + }, + { + "epoch": 4.23, + "grad_norm": 6.095008373260498, + "learning_rate": 0.00012268954496435258, + "loss": 3.1753, + "step": 2893 + }, + { + "epoch": 4.23, + "grad_norm": 3.311551094055176, + "learning_rate": 0.000122643862309656, + "loss": 3.2052, + "step": 2894 + }, + { + "epoch": 4.24, + "grad_norm": 6.019943714141846, + "learning_rate": 0.0001225981746734615, + "loss": 3.2064, + "step": 2895 + }, + { + "epoch": 4.24, + "grad_norm": 5.016195774078369, + "learning_rate": 0.00012255248206581996, + "loss": 3.2018, + "step": 2896 + }, + { + "epoch": 4.24, + "grad_norm": 2.2277472019195557, + "learning_rate": 0.00012250678449678354, + "loss": 3.1877, + "step": 2897 + }, + { + "epoch": 4.24, + "grad_norm": 5.388427734375, + "learning_rate": 0.00012246108197640538, + "loss": 3.2087, + "step": 2898 + }, + { + "epoch": 4.24, + "grad_norm": 3.7437267303466797, + "learning_rate": 0.0001224153745147397, + "loss": 3.1554, + "step": 2899 + }, + { + "epoch": 4.24, + "grad_norm": 2.8456854820251465, + "learning_rate": 0.0001223696621218418, + "loss": 3.1912, + "step": 2900 + }, + { + "epoch": 4.24, + "grad_norm": 2.9107625484466553, + "learning_rate": 0.0001223239448077682, + "loss": 3.1832, + "step": 2901 + }, + { + "epoch": 4.25, + "grad_norm": 2.9731972217559814, + "learning_rate": 0.00012227822258257636, + "loss": 3.1781, + "step": 2902 + }, + { + "epoch": 4.25, + "grad_norm": 3.2368147373199463, + "learning_rate": 0.00012223249545632482, + "loss": 3.1863, + "step": 2903 + }, + { + "epoch": 4.25, + "grad_norm": 1.3652539253234863, + "learning_rate": 0.00012218676343907328, + "loss": 3.1765, + "step": 2904 + }, + { + "epoch": 4.25, + "grad_norm": 6.007261753082275, + "learning_rate": 0.00012214102654088248, + "loss": 3.2073, + "step": 2905 + }, + { + "epoch": 4.25, + "grad_norm": 6.385366439819336, + "learning_rate": 0.00012209528477181422, + "loss": 3.1974, + "step": 2906 + }, + { + "epoch": 4.25, + "grad_norm": 1.5009405612945557, + "learning_rate": 0.00012204953814193139, + "loss": 3.1815, + "step": 2907 + }, + { + "epoch": 4.26, + "grad_norm": 4.717186450958252, + "learning_rate": 0.00012200378666129793, + "loss": 3.166, + "step": 2908 + }, + { + "epoch": 4.26, + "grad_norm": 4.571866035461426, + "learning_rate": 0.0001219580303399789, + "loss": 3.2043, + "step": 2909 + }, + { + "epoch": 4.26, + "grad_norm": 6.416868686676025, + "learning_rate": 0.0001219122691880403, + "loss": 3.1969, + "step": 2910 + }, + { + "epoch": 4.26, + "grad_norm": 4.813213348388672, + "learning_rate": 0.00012186650321554935, + "loss": 3.1706, + "step": 2911 + }, + { + "epoch": 4.26, + "grad_norm": 3.790310859680176, + "learning_rate": 0.00012182073243257426, + "loss": 3.178, + "step": 2912 + }, + { + "epoch": 4.26, + "grad_norm": 4.808967590332031, + "learning_rate": 0.00012177495684918423, + "loss": 3.2266, + "step": 2913 + }, + { + "epoch": 4.26, + "grad_norm": 8.275779724121094, + "learning_rate": 0.00012172917647544964, + "loss": 3.1736, + "step": 2914 + }, + { + "epoch": 4.27, + "grad_norm": 4.079070568084717, + "learning_rate": 0.00012168339132144186, + "loss": 3.1918, + "step": 2915 + }, + { + "epoch": 4.27, + "grad_norm": 2.108287811279297, + "learning_rate": 0.0001216376013972333, + "loss": 3.1543, + "step": 2916 + }, + { + "epoch": 4.27, + "grad_norm": 3.337580442428589, + "learning_rate": 0.00012159180671289743, + "loss": 3.1937, + "step": 2917 + }, + { + "epoch": 4.27, + "grad_norm": 2.6380503177642822, + "learning_rate": 0.0001215460072785088, + "loss": 3.1794, + "step": 2918 + }, + { + "epoch": 4.27, + "grad_norm": 2.371419906616211, + "learning_rate": 0.00012150020310414296, + "loss": 3.1682, + "step": 2919 + }, + { + "epoch": 4.27, + "grad_norm": 1.647721767425537, + "learning_rate": 0.00012145439419987652, + "loss": 3.1618, + "step": 2920 + }, + { + "epoch": 4.27, + "grad_norm": 0.846366822719574, + "learning_rate": 0.00012140858057578713, + "loss": 3.1883, + "step": 2921 + }, + { + "epoch": 4.28, + "grad_norm": 1.4289370775222778, + "learning_rate": 0.00012136276224195348, + "loss": 3.2119, + "step": 2922 + }, + { + "epoch": 4.28, + "grad_norm": 2.510512351989746, + "learning_rate": 0.00012131693920845533, + "loss": 3.1849, + "step": 2923 + }, + { + "epoch": 4.28, + "grad_norm": 1.3627164363861084, + "learning_rate": 0.0001212711114853734, + "loss": 3.1851, + "step": 2924 + }, + { + "epoch": 4.28, + "grad_norm": 2.383141279220581, + "learning_rate": 0.00012122527908278949, + "loss": 3.1599, + "step": 2925 + }, + { + "epoch": 4.28, + "grad_norm": 2.2744383811950684, + "learning_rate": 0.00012117944201078647, + "loss": 3.1814, + "step": 2926 + }, + { + "epoch": 4.28, + "grad_norm": 2.6775460243225098, + "learning_rate": 0.00012113360027944806, + "loss": 3.1495, + "step": 2927 + }, + { + "epoch": 4.28, + "grad_norm": 2.249178647994995, + "learning_rate": 0.00012108775389885929, + "loss": 3.1721, + "step": 2928 + }, + { + "epoch": 4.29, + "grad_norm": 1.3241461515426636, + "learning_rate": 0.00012104190287910597, + "loss": 3.1563, + "step": 2929 + }, + { + "epoch": 4.29, + "grad_norm": 3.977029323577881, + "learning_rate": 0.00012099604723027502, + "loss": 3.254, + "step": 2930 + }, + { + "epoch": 4.29, + "grad_norm": 3.9783012866973877, + "learning_rate": 0.00012095018696245441, + "loss": 3.1861, + "step": 2931 + }, + { + "epoch": 4.29, + "grad_norm": 1.687117338180542, + "learning_rate": 0.00012090432208573306, + "loss": 3.1259, + "step": 2932 + }, + { + "epoch": 4.29, + "grad_norm": 10.139452934265137, + "learning_rate": 0.00012085845261020095, + "loss": 3.1579, + "step": 2933 + }, + { + "epoch": 4.29, + "grad_norm": 4.6292500495910645, + "learning_rate": 0.00012081257854594906, + "loss": 3.2254, + "step": 2934 + }, + { + "epoch": 4.3, + "grad_norm": 3.5405800342559814, + "learning_rate": 0.00012076669990306934, + "loss": 3.1327, + "step": 2935 + }, + { + "epoch": 4.3, + "grad_norm": 1.5894381999969482, + "learning_rate": 0.00012072081669165482, + "loss": 3.1545, + "step": 2936 + }, + { + "epoch": 4.3, + "grad_norm": 3.921757221221924, + "learning_rate": 0.0001206749289217995, + "loss": 3.1646, + "step": 2937 + }, + { + "epoch": 4.3, + "grad_norm": 3.1364986896514893, + "learning_rate": 0.0001206290366035984, + "loss": 3.1835, + "step": 2938 + }, + { + "epoch": 4.3, + "grad_norm": 1.0131809711456299, + "learning_rate": 0.00012058313974714746, + "loss": 3.1446, + "step": 2939 + }, + { + "epoch": 4.3, + "grad_norm": 2.2197444438934326, + "learning_rate": 0.00012053723836254373, + "loss": 3.1805, + "step": 2940 + }, + { + "epoch": 4.3, + "grad_norm": 1.5680094957351685, + "learning_rate": 0.00012049133245988519, + "loss": 3.2005, + "step": 2941 + }, + { + "epoch": 4.31, + "grad_norm": 2.2786076068878174, + "learning_rate": 0.00012044542204927087, + "loss": 3.1663, + "step": 2942 + }, + { + "epoch": 4.31, + "grad_norm": 2.1755716800689697, + "learning_rate": 0.00012039950714080068, + "loss": 3.1709, + "step": 2943 + }, + { + "epoch": 4.31, + "grad_norm": 1.485142469406128, + "learning_rate": 0.00012035358774457563, + "loss": 3.2298, + "step": 2944 + }, + { + "epoch": 4.31, + "grad_norm": 1.9527829885482788, + "learning_rate": 0.00012030766387069771, + "loss": 3.1868, + "step": 2945 + }, + { + "epoch": 4.31, + "grad_norm": 1.5011379718780518, + "learning_rate": 0.00012026173552926981, + "loss": 3.1533, + "step": 2946 + }, + { + "epoch": 4.31, + "grad_norm": 1.8543486595153809, + "learning_rate": 0.00012021580273039585, + "loss": 3.1536, + "step": 2947 + }, + { + "epoch": 4.31, + "grad_norm": 2.353090763092041, + "learning_rate": 0.00012016986548418083, + "loss": 3.1675, + "step": 2948 + }, + { + "epoch": 4.32, + "grad_norm": 3.509500741958618, + "learning_rate": 0.00012012392380073056, + "loss": 3.1963, + "step": 2949 + }, + { + "epoch": 4.32, + "grad_norm": 4.018800258636475, + "learning_rate": 0.00012007797769015192, + "loss": 3.165, + "step": 2950 + }, + { + "epoch": 4.32, + "grad_norm": 2.620981454849243, + "learning_rate": 0.00012003202716255273, + "loss": 3.1379, + "step": 2951 + }, + { + "epoch": 4.32, + "grad_norm": 2501.735107421875, + "learning_rate": 0.00011998607222804181, + "loss": 3.2292, + "step": 2952 + }, + { + "epoch": 4.32, + "grad_norm": 7.561061859130859, + "learning_rate": 0.00011994011289672895, + "loss": 3.1922, + "step": 2953 + }, + { + "epoch": 4.32, + "grad_norm": 7.529016971588135, + "learning_rate": 0.00011989414917872488, + "loss": 3.1894, + "step": 2954 + }, + { + "epoch": 4.32, + "grad_norm": 2.2252986431121826, + "learning_rate": 0.00011984818108414133, + "loss": 3.1745, + "step": 2955 + }, + { + "epoch": 4.33, + "grad_norm": 6.323204040527344, + "learning_rate": 0.00011980220862309097, + "loss": 3.1978, + "step": 2956 + }, + { + "epoch": 4.33, + "grad_norm": 137.4775390625, + "learning_rate": 0.00011975623180568738, + "loss": 3.222, + "step": 2957 + }, + { + "epoch": 4.33, + "grad_norm": 9.54941177368164, + "learning_rate": 0.0001197102506420452, + "loss": 3.1733, + "step": 2958 + }, + { + "epoch": 4.33, + "grad_norm": 7.13044548034668, + "learning_rate": 0.00011966426514228, + "loss": 3.2191, + "step": 2959 + }, + { + "epoch": 4.33, + "grad_norm": 1.8224902153015137, + "learning_rate": 0.00011961827531650824, + "loss": 3.2066, + "step": 2960 + }, + { + "epoch": 4.33, + "grad_norm": 7.74959659576416, + "learning_rate": 0.00011957228117484736, + "loss": 3.2138, + "step": 2961 + }, + { + "epoch": 4.34, + "grad_norm": 8.487194061279297, + "learning_rate": 0.00011952628272741584, + "loss": 3.1951, + "step": 2962 + }, + { + "epoch": 4.34, + "grad_norm": 8.898509979248047, + "learning_rate": 0.00011948027998433296, + "loss": 3.1982, + "step": 2963 + }, + { + "epoch": 4.34, + "grad_norm": 11.637873649597168, + "learning_rate": 0.00011943427295571899, + "loss": 3.2156, + "step": 2964 + }, + { + "epoch": 4.34, + "grad_norm": 7.426854133605957, + "learning_rate": 0.00011938826165169525, + "loss": 3.2193, + "step": 2965 + }, + { + "epoch": 4.34, + "grad_norm": 12.111673355102539, + "learning_rate": 0.00011934224608238389, + "loss": 3.1818, + "step": 2966 + }, + { + "epoch": 4.34, + "grad_norm": 3.1284587383270264, + "learning_rate": 0.00011929622625790802, + "loss": 3.1583, + "step": 2967 + }, + { + "epoch": 4.34, + "grad_norm": 4.960999488830566, + "learning_rate": 0.00011925020218839167, + "loss": 3.1765, + "step": 2968 + }, + { + "epoch": 4.35, + "grad_norm": 7.320558071136475, + "learning_rate": 0.00011920417388395986, + "loss": 3.19, + "step": 2969 + }, + { + "epoch": 4.35, + "grad_norm": 4.303677082061768, + "learning_rate": 0.0001191581413547385, + "loss": 3.1875, + "step": 2970 + }, + { + "epoch": 4.35, + "grad_norm": 4.7796549797058105, + "learning_rate": 0.0001191121046108544, + "loss": 3.1843, + "step": 2971 + }, + { + "epoch": 4.35, + "grad_norm": 5.886425971984863, + "learning_rate": 0.0001190660636624354, + "loss": 3.1532, + "step": 2972 + }, + { + "epoch": 4.35, + "grad_norm": 2.2462821006774902, + "learning_rate": 0.00011902001851961017, + "loss": 3.1873, + "step": 2973 + }, + { + "epoch": 4.35, + "grad_norm": 6.365395545959473, + "learning_rate": 0.00011897396919250831, + "loss": 3.1509, + "step": 2974 + }, + { + "epoch": 4.35, + "grad_norm": 8.911288261413574, + "learning_rate": 0.0001189279156912604, + "loss": 3.1714, + "step": 2975 + }, + { + "epoch": 4.36, + "grad_norm": 5.056528091430664, + "learning_rate": 0.00011888185802599791, + "loss": 3.1708, + "step": 2976 + }, + { + "epoch": 4.36, + "grad_norm": 3.0920491218566895, + "learning_rate": 0.00011883579620685318, + "loss": 3.1315, + "step": 2977 + }, + { + "epoch": 4.36, + "grad_norm": 4.874855041503906, + "learning_rate": 0.00011878973024395952, + "loss": 3.1369, + "step": 2978 + }, + { + "epoch": 4.36, + "grad_norm": 2.600829601287842, + "learning_rate": 0.0001187436601474511, + "loss": 3.1673, + "step": 2979 + }, + { + "epoch": 4.36, + "grad_norm": 5.483844757080078, + "learning_rate": 0.00011869758592746309, + "loss": 3.1439, + "step": 2980 + }, + { + "epoch": 4.36, + "grad_norm": 7.73532772064209, + "learning_rate": 0.00011865150759413147, + "loss": 3.2182, + "step": 2981 + }, + { + "epoch": 4.36, + "grad_norm": 3.6422669887542725, + "learning_rate": 0.00011860542515759318, + "loss": 3.1551, + "step": 2982 + }, + { + "epoch": 4.37, + "grad_norm": 4.651947975158691, + "learning_rate": 0.00011855933862798605, + "loss": 3.1674, + "step": 2983 + }, + { + "epoch": 4.37, + "grad_norm": 6.519864559173584, + "learning_rate": 0.00011851324801544879, + "loss": 3.1561, + "step": 2984 + }, + { + "epoch": 4.37, + "grad_norm": 3.8000521659851074, + "learning_rate": 0.00011846715333012104, + "loss": 3.174, + "step": 2985 + }, + { + "epoch": 4.37, + "grad_norm": 4.087390899658203, + "learning_rate": 0.00011842105458214333, + "loss": 3.1715, + "step": 2986 + }, + { + "epoch": 4.37, + "grad_norm": 5.734311580657959, + "learning_rate": 0.00011837495178165706, + "loss": 3.1486, + "step": 2987 + }, + { + "epoch": 4.37, + "grad_norm": 2.4519381523132324, + "learning_rate": 0.00011832884493880452, + "loss": 3.1528, + "step": 2988 + }, + { + "epoch": 4.38, + "grad_norm": 4.925870895385742, + "learning_rate": 0.00011828273406372894, + "loss": 3.1785, + "step": 2989 + }, + { + "epoch": 4.38, + "grad_norm": 6.140622615814209, + "learning_rate": 0.0001182366191665744, + "loss": 3.207, + "step": 2990 + }, + { + "epoch": 4.38, + "grad_norm": 2.630859851837158, + "learning_rate": 0.00011819050025748583, + "loss": 3.1437, + "step": 2991 + }, + { + "epoch": 4.38, + "grad_norm": 6.897947311401367, + "learning_rate": 0.00011814437734660917, + "loss": 3.2066, + "step": 2992 + }, + { + "epoch": 4.38, + "grad_norm": 10.167567253112793, + "learning_rate": 0.00011809825044409106, + "loss": 3.1581, + "step": 2993 + }, + { + "epoch": 4.38, + "grad_norm": 13.395803451538086, + "learning_rate": 0.00011805211956007914, + "loss": 3.1348, + "step": 2994 + }, + { + "epoch": 4.38, + "grad_norm": 1.3696818351745605, + "learning_rate": 0.00011800598470472189, + "loss": 3.1391, + "step": 2995 + }, + { + "epoch": 4.39, + "grad_norm": 6.5317063331604, + "learning_rate": 0.00011795984588816869, + "loss": 3.1951, + "step": 2996 + }, + { + "epoch": 4.39, + "grad_norm": 63.888736724853516, + "learning_rate": 0.00011791370312056972, + "loss": 3.2487, + "step": 2997 + }, + { + "epoch": 4.39, + "grad_norm": 9.84652042388916, + "learning_rate": 0.00011786755641207614, + "loss": 3.1655, + "step": 2998 + }, + { + "epoch": 4.39, + "grad_norm": 7.442485809326172, + "learning_rate": 0.00011782140577283991, + "loss": 3.1921, + "step": 2999 + }, + { + "epoch": 4.39, + "grad_norm": 1.4993599653244019, + "learning_rate": 0.00011777525121301383, + "loss": 3.1366, + "step": 3000 + }, + { + "epoch": 4.39, + "grad_norm": 23.01314926147461, + "learning_rate": 0.00011772909274275159, + "loss": 3.1806, + "step": 3001 + }, + { + "epoch": 4.39, + "grad_norm": 13.442095756530762, + "learning_rate": 0.00011768293037220778, + "loss": 3.2192, + "step": 3002 + }, + { + "epoch": 4.4, + "grad_norm": 11.32374382019043, + "learning_rate": 0.0001176367641115378, + "loss": 3.1979, + "step": 3003 + }, + { + "epoch": 4.4, + "grad_norm": 4.027917385101318, + "learning_rate": 0.00011759059397089793, + "loss": 3.1963, + "step": 3004 + }, + { + "epoch": 4.4, + "grad_norm": 6.907668113708496, + "learning_rate": 0.00011754441996044523, + "loss": 3.168, + "step": 3005 + }, + { + "epoch": 4.4, + "grad_norm": 8.269980430603027, + "learning_rate": 0.00011749824209033775, + "loss": 3.1789, + "step": 3006 + }, + { + "epoch": 4.4, + "grad_norm": 6.633981227874756, + "learning_rate": 0.00011745206037073426, + "loss": 3.225, + "step": 3007 + }, + { + "epoch": 4.4, + "grad_norm": 5.4061994552612305, + "learning_rate": 0.00011740587481179442, + "loss": 3.2288, + "step": 3008 + }, + { + "epoch": 4.41, + "grad_norm": 5.786009311676025, + "learning_rate": 0.0001173596854236788, + "loss": 3.1826, + "step": 3009 + }, + { + "epoch": 4.41, + "grad_norm": 6.876216888427734, + "learning_rate": 0.00011731349221654875, + "loss": 3.1563, + "step": 3010 + }, + { + "epoch": 4.41, + "grad_norm": 6.093964099884033, + "learning_rate": 0.00011726729520056641, + "loss": 3.1879, + "step": 3011 + }, + { + "epoch": 4.41, + "grad_norm": 5.7057576179504395, + "learning_rate": 0.00011722109438589486, + "loss": 3.2176, + "step": 3012 + }, + { + "epoch": 4.41, + "grad_norm": 6.603635787963867, + "learning_rate": 0.00011717488978269795, + "loss": 3.1905, + "step": 3013 + }, + { + "epoch": 4.41, + "grad_norm": 4.3624067306518555, + "learning_rate": 0.00011712868140114037, + "loss": 3.1549, + "step": 3014 + }, + { + "epoch": 4.41, + "grad_norm": 3.682727098464966, + "learning_rate": 0.00011708246925138771, + "loss": 3.1376, + "step": 3015 + }, + { + "epoch": 4.42, + "grad_norm": 6.1669921875, + "learning_rate": 0.00011703625334360626, + "loss": 3.1513, + "step": 3016 + }, + { + "epoch": 4.42, + "grad_norm": 4.20526123046875, + "learning_rate": 0.00011699003368796328, + "loss": 3.1493, + "step": 3017 + }, + { + "epoch": 4.42, + "grad_norm": 2.534583568572998, + "learning_rate": 0.0001169438102946267, + "loss": 3.1625, + "step": 3018 + }, + { + "epoch": 4.42, + "grad_norm": 2.841315269470215, + "learning_rate": 0.00011689758317376547, + "loss": 3.2036, + "step": 3019 + }, + { + "epoch": 4.42, + "grad_norm": 14.288698196411133, + "learning_rate": 0.00011685135233554917, + "loss": 3.1499, + "step": 3020 + }, + { + "epoch": 4.42, + "grad_norm": 5.600762844085693, + "learning_rate": 0.00011680511779014827, + "loss": 3.1529, + "step": 3021 + }, + { + "epoch": 4.42, + "grad_norm": 2.945816993713379, + "learning_rate": 0.00011675887954773405, + "loss": 3.1832, + "step": 3022 + }, + { + "epoch": 4.43, + "grad_norm": 3.300783395767212, + "learning_rate": 0.0001167126376184787, + "loss": 3.16, + "step": 3023 + }, + { + "epoch": 4.43, + "grad_norm": 10.50096607208252, + "learning_rate": 0.00011666639201255506, + "loss": 3.128, + "step": 3024 + }, + { + "epoch": 4.43, + "grad_norm": 2.324991226196289, + "learning_rate": 0.00011662014274013684, + "loss": 3.1621, + "step": 3025 + }, + { + "epoch": 4.43, + "grad_norm": 3.7734105587005615, + "learning_rate": 0.00011657388981139865, + "loss": 3.1594, + "step": 3026 + }, + { + "epoch": 4.43, + "grad_norm": 3.324181079864502, + "learning_rate": 0.00011652763323651575, + "loss": 3.1495, + "step": 3027 + }, + { + "epoch": 4.43, + "grad_norm": 2.9058284759521484, + "learning_rate": 0.00011648137302566431, + "loss": 3.1342, + "step": 3028 + }, + { + "epoch": 4.43, + "grad_norm": 3.1446614265441895, + "learning_rate": 0.00011643510918902123, + "loss": 3.1408, + "step": 3029 + }, + { + "epoch": 4.44, + "grad_norm": 1.5861114263534546, + "learning_rate": 0.0001163888417367643, + "loss": 3.1833, + "step": 3030 + }, + { + "epoch": 4.44, + "grad_norm": 0.9234637022018433, + "learning_rate": 0.00011634257067907201, + "loss": 3.1692, + "step": 3031 + }, + { + "epoch": 4.44, + "grad_norm": 2.46486759185791, + "learning_rate": 0.00011629629602612368, + "loss": 3.1654, + "step": 3032 + }, + { + "epoch": 4.44, + "grad_norm": 1.57401442527771, + "learning_rate": 0.00011625001778809945, + "loss": 3.1469, + "step": 3033 + }, + { + "epoch": 4.44, + "grad_norm": 3.949265241622925, + "learning_rate": 0.00011620373597518024, + "loss": 3.2028, + "step": 3034 + }, + { + "epoch": 4.44, + "grad_norm": 4.033581256866455, + "learning_rate": 0.00011615745059754766, + "loss": 3.1827, + "step": 3035 + }, + { + "epoch": 4.45, + "grad_norm": 3.1215124130249023, + "learning_rate": 0.00011611116166538425, + "loss": 3.1908, + "step": 3036 + }, + { + "epoch": 4.45, + "grad_norm": 2.8856616020202637, + "learning_rate": 0.00011606486918887327, + "loss": 3.1631, + "step": 3037 + }, + { + "epoch": 4.45, + "grad_norm": 3.271087408065796, + "learning_rate": 0.00011601857317819871, + "loss": 3.1653, + "step": 3038 + }, + { + "epoch": 4.45, + "grad_norm": 3.887037515640259, + "learning_rate": 0.0001159722736435454, + "loss": 3.1656, + "step": 3039 + }, + { + "epoch": 4.45, + "grad_norm": 1.2171587944030762, + "learning_rate": 0.00011592597059509893, + "loss": 3.1612, + "step": 3040 + }, + { + "epoch": 4.45, + "grad_norm": 1.7865351438522339, + "learning_rate": 0.00011587966404304564, + "loss": 3.1736, + "step": 3041 + }, + { + "epoch": 4.45, + "grad_norm": 0.9533948302268982, + "learning_rate": 0.00011583335399757271, + "loss": 3.1167, + "step": 3042 + }, + { + "epoch": 4.45, + "eval_loss": 3.0038862228393555, + "eval_runtime": 1165.2502, + "eval_samples_per_second": 14.711, + "eval_steps_per_second": 0.92, + "step": 3042 + }, + { + "epoch": 4.46, + "grad_norm": 1.679763674736023, + "learning_rate": 0.000115787040468868, + "loss": 3.1827, + "step": 3043 + }, + { + "epoch": 4.46, + "grad_norm": 0.7595570683479309, + "learning_rate": 0.0001157407234671202, + "loss": 3.1615, + "step": 3044 + }, + { + "epoch": 4.46, + "grad_norm": 1.3758699893951416, + "learning_rate": 0.0001156944030025187, + "loss": 3.2312, + "step": 3045 + }, + { + "epoch": 4.46, + "grad_norm": 1.6547698974609375, + "learning_rate": 0.00011564807908525374, + "loss": 3.1552, + "step": 3046 + }, + { + "epoch": 4.46, + "grad_norm": 1.4767943620681763, + "learning_rate": 0.00011560175172551623, + "loss": 3.2091, + "step": 3047 + }, + { + "epoch": 4.46, + "grad_norm": 0.9520362615585327, + "learning_rate": 0.00011555542093349791, + "loss": 3.1592, + "step": 3048 + }, + { + "epoch": 4.46, + "grad_norm": 0.4594898223876953, + "learning_rate": 0.00011550908671939125, + "loss": 3.1426, + "step": 3049 + }, + { + "epoch": 4.47, + "grad_norm": 0.8685529232025146, + "learning_rate": 0.00011546274909338945, + "loss": 3.115, + "step": 3050 + }, + { + "epoch": 4.47, + "grad_norm": 0.986310601234436, + "learning_rate": 0.00011541640806568646, + "loss": 3.1239, + "step": 3051 + }, + { + "epoch": 4.47, + "grad_norm": 2.280745029449463, + "learning_rate": 0.000115370063646477, + "loss": 3.1833, + "step": 3052 + }, + { + "epoch": 4.47, + "grad_norm": 0.696453869342804, + "learning_rate": 0.00011532371584595656, + "loss": 3.1107, + "step": 3053 + }, + { + "epoch": 4.47, + "grad_norm": 3.4933245182037354, + "learning_rate": 0.00011527736467432133, + "loss": 3.1748, + "step": 3054 + }, + { + "epoch": 4.47, + "grad_norm": 2.052901268005371, + "learning_rate": 0.00011523101014176825, + "loss": 3.16, + "step": 3055 + }, + { + "epoch": 4.47, + "grad_norm": 3.9722251892089844, + "learning_rate": 0.00011518465225849502, + "loss": 3.1674, + "step": 3056 + }, + { + "epoch": 4.48, + "grad_norm": 3.952956438064575, + "learning_rate": 0.00011513829103470004, + "loss": 3.1506, + "step": 3057 + }, + { + "epoch": 4.48, + "grad_norm": 2.466555118560791, + "learning_rate": 0.00011509192648058249, + "loss": 3.1843, + "step": 3058 + }, + { + "epoch": 4.48, + "grad_norm": 2.9880712032318115, + "learning_rate": 0.00011504555860634225, + "loss": 3.2101, + "step": 3059 + }, + { + "epoch": 4.48, + "grad_norm": 3.0772719383239746, + "learning_rate": 0.00011499918742217996, + "loss": 3.1815, + "step": 3060 + }, + { + "epoch": 4.48, + "grad_norm": 3.1504437923431396, + "learning_rate": 0.00011495281293829697, + "loss": 3.1493, + "step": 3061 + }, + { + "epoch": 4.48, + "grad_norm": 1.549119472503662, + "learning_rate": 0.00011490643516489532, + "loss": 3.154, + "step": 3062 + }, + { + "epoch": 4.49, + "grad_norm": 5.081751346588135, + "learning_rate": 0.00011486005411217784, + "loss": 3.2329, + "step": 3063 + }, + { + "epoch": 4.49, + "grad_norm": 5.428029537200928, + "learning_rate": 0.00011481366979034806, + "loss": 3.1709, + "step": 3064 + }, + { + "epoch": 4.49, + "grad_norm": 1.1052792072296143, + "learning_rate": 0.00011476728220961015, + "loss": 3.1713, + "step": 3065 + }, + { + "epoch": 4.49, + "grad_norm": 4.415512561798096, + "learning_rate": 0.00011472089138016917, + "loss": 3.1753, + "step": 3066 + }, + { + "epoch": 4.49, + "grad_norm": 4.5699143409729, + "learning_rate": 0.00011467449731223075, + "loss": 3.209, + "step": 3067 + }, + { + "epoch": 4.49, + "grad_norm": 3.3557825088500977, + "learning_rate": 0.00011462810001600123, + "loss": 3.1721, + "step": 3068 + }, + { + "epoch": 4.49, + "grad_norm": 3.846703290939331, + "learning_rate": 0.00011458169950168774, + "loss": 3.1592, + "step": 3069 + }, + { + "epoch": 4.5, + "grad_norm": 3.0023891925811768, + "learning_rate": 0.0001145352957794981, + "loss": 3.1636, + "step": 3070 + }, + { + "epoch": 4.5, + "grad_norm": 6.419353485107422, + "learning_rate": 0.0001144888888596408, + "loss": 3.1472, + "step": 3071 + }, + { + "epoch": 4.5, + "grad_norm": 6.247819900512695, + "learning_rate": 0.00011444247875232503, + "loss": 3.1901, + "step": 3072 + }, + { + "epoch": 4.5, + "grad_norm": 4.539994239807129, + "learning_rate": 0.00011439606546776076, + "loss": 3.1452, + "step": 3073 + }, + { + "epoch": 4.5, + "grad_norm": 3.100914716720581, + "learning_rate": 0.00011434964901615852, + "loss": 3.1582, + "step": 3074 + }, + { + "epoch": 4.5, + "grad_norm": 8.092155456542969, + "learning_rate": 0.0001143032294077297, + "loss": 3.2045, + "step": 3075 + }, + { + "epoch": 4.5, + "grad_norm": 2.9788618087768555, + "learning_rate": 0.00011425680665268626, + "loss": 3.1545, + "step": 3076 + }, + { + "epoch": 4.51, + "grad_norm": 2.915862560272217, + "learning_rate": 0.00011421038076124089, + "loss": 3.1752, + "step": 3077 + }, + { + "epoch": 4.51, + "grad_norm": 3.0964643955230713, + "learning_rate": 0.000114163951743607, + "loss": 3.1225, + "step": 3078 + }, + { + "epoch": 4.51, + "grad_norm": 2.3280699253082275, + "learning_rate": 0.00011411751960999862, + "loss": 3.179, + "step": 3079 + }, + { + "epoch": 4.51, + "grad_norm": 1.9851235151290894, + "learning_rate": 0.00011407108437063057, + "loss": 3.1557, + "step": 3080 + }, + { + "epoch": 4.51, + "grad_norm": 2.6399571895599365, + "learning_rate": 0.00011402464603571825, + "loss": 3.131, + "step": 3081 + }, + { + "epoch": 4.51, + "grad_norm": 10.382826805114746, + "learning_rate": 0.00011397820461547777, + "loss": 3.1541, + "step": 3082 + }, + { + "epoch": 4.51, + "grad_norm": 1.8169869184494019, + "learning_rate": 0.00011393176012012596, + "loss": 3.2169, + "step": 3083 + }, + { + "epoch": 4.52, + "grad_norm": 3.3234314918518066, + "learning_rate": 0.00011388531255988033, + "loss": 3.1631, + "step": 3084 + }, + { + "epoch": 4.52, + "grad_norm": 1.7650176286697388, + "learning_rate": 0.00011383886194495896, + "loss": 3.1561, + "step": 3085 + }, + { + "epoch": 4.52, + "grad_norm": 5.759214878082275, + "learning_rate": 0.00011379240828558072, + "loss": 3.1424, + "step": 3086 + }, + { + "epoch": 4.52, + "grad_norm": 5.468089580535889, + "learning_rate": 0.00011374595159196506, + "loss": 3.1697, + "step": 3087 + }, + { + "epoch": 4.52, + "grad_norm": 25.312021255493164, + "learning_rate": 0.00011369949187433218, + "loss": 3.1632, + "step": 3088 + }, + { + "epoch": 4.52, + "grad_norm": 3.2354679107666016, + "learning_rate": 0.0001136530291429029, + "loss": 3.166, + "step": 3089 + }, + { + "epoch": 4.53, + "grad_norm": 11.292868614196777, + "learning_rate": 0.00011360656340789873, + "loss": 3.1495, + "step": 3090 + }, + { + "epoch": 4.53, + "grad_norm": 9.209776878356934, + "learning_rate": 0.00011356009467954176, + "loss": 3.189, + "step": 3091 + }, + { + "epoch": 4.53, + "grad_norm": 40.07894515991211, + "learning_rate": 0.00011351362296805485, + "loss": 3.1709, + "step": 3092 + }, + { + "epoch": 4.53, + "grad_norm": 4.195234298706055, + "learning_rate": 0.00011346714828366148, + "loss": 3.2041, + "step": 3093 + }, + { + "epoch": 4.53, + "grad_norm": 154.33792114257812, + "learning_rate": 0.0001134206706365857, + "loss": 3.1673, + "step": 3094 + }, + { + "epoch": 4.53, + "grad_norm": 3.392430305480957, + "learning_rate": 0.00011337419003705237, + "loss": 3.1747, + "step": 3095 + }, + { + "epoch": 4.53, + "grad_norm": 156.2117462158203, + "learning_rate": 0.00011332770649528681, + "loss": 3.1426, + "step": 3096 + }, + { + "epoch": 4.54, + "grad_norm": 3.3672966957092285, + "learning_rate": 0.00011328122002151516, + "loss": 3.1428, + "step": 3097 + }, + { + "epoch": 4.54, + "grad_norm": 1.9393489360809326, + "learning_rate": 0.00011323473062596413, + "loss": 3.1491, + "step": 3098 + }, + { + "epoch": 4.54, + "grad_norm": 4.762568473815918, + "learning_rate": 0.00011318823831886103, + "loss": 3.1408, + "step": 3099 + }, + { + "epoch": 4.54, + "grad_norm": 4.439011573791504, + "learning_rate": 0.00011314174311043391, + "loss": 3.149, + "step": 3100 + }, + { + "epoch": 4.54, + "grad_norm": 1099.63525390625, + "learning_rate": 0.00011309524501091135, + "loss": 3.1883, + "step": 3101 + }, + { + "epoch": 4.54, + "grad_norm": 14.831306457519531, + "learning_rate": 0.00011304874403052266, + "loss": 3.2264, + "step": 3102 + }, + { + "epoch": 4.54, + "grad_norm": 1.7882112264633179, + "learning_rate": 0.00011300224017949772, + "loss": 3.1552, + "step": 3103 + }, + { + "epoch": 4.55, + "grad_norm": 5.061419486999512, + "learning_rate": 0.00011295573346806709, + "loss": 3.1629, + "step": 3104 + }, + { + "epoch": 4.55, + "grad_norm": 2.1199474334716797, + "learning_rate": 0.00011290922390646192, + "loss": 3.1672, + "step": 3105 + }, + { + "epoch": 4.55, + "grad_norm": 2.093383312225342, + "learning_rate": 0.000112862711504914, + "loss": 3.1872, + "step": 3106 + }, + { + "epoch": 4.55, + "grad_norm": 1.5643643140792847, + "learning_rate": 0.00011281619627365576, + "loss": 3.2203, + "step": 3107 + }, + { + "epoch": 4.55, + "grad_norm": 3.1214048862457275, + "learning_rate": 0.00011276967822292024, + "loss": 3.1508, + "step": 3108 + }, + { + "epoch": 4.55, + "grad_norm": 17.56413459777832, + "learning_rate": 0.00011272315736294108, + "loss": 3.1855, + "step": 3109 + }, + { + "epoch": 4.56, + "grad_norm": 3.3824844360351562, + "learning_rate": 0.00011267663370395256, + "loss": 3.1449, + "step": 3110 + }, + { + "epoch": 4.56, + "grad_norm": 2.0639045238494873, + "learning_rate": 0.00011263010725618959, + "loss": 3.1227, + "step": 3111 + }, + { + "epoch": 4.56, + "grad_norm": 2.0494930744171143, + "learning_rate": 0.00011258357802988766, + "loss": 3.1386, + "step": 3112 + }, + { + "epoch": 4.56, + "grad_norm": 2.3310718536376953, + "learning_rate": 0.00011253704603528288, + "loss": 3.1649, + "step": 3113 + }, + { + "epoch": 4.56, + "grad_norm": 1.5938396453857422, + "learning_rate": 0.00011249051128261203, + "loss": 3.1524, + "step": 3114 + }, + { + "epoch": 4.56, + "grad_norm": 1.9962538480758667, + "learning_rate": 0.00011244397378211238, + "loss": 3.1949, + "step": 3115 + }, + { + "epoch": 4.56, + "grad_norm": 2.338400363922119, + "learning_rate": 0.00011239743354402188, + "loss": 3.1258, + "step": 3116 + }, + { + "epoch": 4.57, + "grad_norm": 1.9446300268173218, + "learning_rate": 0.00011235089057857911, + "loss": 3.1829, + "step": 3117 + }, + { + "epoch": 4.57, + "grad_norm": 2.5406713485717773, + "learning_rate": 0.00011230434489602317, + "loss": 3.1782, + "step": 3118 + }, + { + "epoch": 4.57, + "grad_norm": 1.6750293970108032, + "learning_rate": 0.00011225779650659383, + "loss": 3.139, + "step": 3119 + }, + { + "epoch": 4.57, + "grad_norm": 10.995355606079102, + "learning_rate": 0.00011221124542053139, + "loss": 3.152, + "step": 3120 + }, + { + "epoch": 4.57, + "grad_norm": 3.1163713932037354, + "learning_rate": 0.00011216469164807679, + "loss": 3.1542, + "step": 3121 + }, + { + "epoch": 4.57, + "grad_norm": 1.6949200630187988, + "learning_rate": 0.00011211813519947159, + "loss": 3.1438, + "step": 3122 + }, + { + "epoch": 4.57, + "grad_norm": 3.045480251312256, + "learning_rate": 0.00011207157608495779, + "loss": 3.1611, + "step": 3123 + }, + { + "epoch": 4.58, + "grad_norm": 2.5979998111724854, + "learning_rate": 0.0001120250143147782, + "loss": 3.1628, + "step": 3124 + }, + { + "epoch": 4.58, + "grad_norm": 2.419813632965088, + "learning_rate": 0.00011197844989917605, + "loss": 3.1876, + "step": 3125 + }, + { + "epoch": 4.58, + "grad_norm": 2.4058735370635986, + "learning_rate": 0.00011193188284839517, + "loss": 3.1617, + "step": 3126 + }, + { + "epoch": 4.58, + "grad_norm": 1.7974401712417603, + "learning_rate": 0.00011188531317268005, + "loss": 3.1276, + "step": 3127 + }, + { + "epoch": 4.58, + "grad_norm": 2.3527932167053223, + "learning_rate": 0.0001118387408822757, + "loss": 3.1843, + "step": 3128 + }, + { + "epoch": 4.58, + "grad_norm": 2.5821757316589355, + "learning_rate": 0.00011179216598742768, + "loss": 3.1492, + "step": 3129 + }, + { + "epoch": 4.58, + "grad_norm": 0.7473565936088562, + "learning_rate": 0.00011174558849838213, + "loss": 3.1384, + "step": 3130 + }, + { + "epoch": 4.59, + "grad_norm": 2.998114585876465, + "learning_rate": 0.00011169900842538588, + "loss": 3.173, + "step": 3131 + }, + { + "epoch": 4.59, + "grad_norm": 8.924659729003906, + "learning_rate": 0.00011165242577868616, + "loss": 3.1929, + "step": 3132 + }, + { + "epoch": 4.59, + "grad_norm": 13.762849807739258, + "learning_rate": 0.00011160584056853082, + "loss": 3.165, + "step": 3133 + }, + { + "epoch": 4.59, + "grad_norm": 2.115877151489258, + "learning_rate": 0.00011155925280516841, + "loss": 3.119, + "step": 3134 + }, + { + "epoch": 4.59, + "grad_norm": 3.3011577129364014, + "learning_rate": 0.00011151266249884783, + "loss": 3.1905, + "step": 3135 + }, + { + "epoch": 4.59, + "grad_norm": 26.584270477294922, + "learning_rate": 0.00011146606965981862, + "loss": 3.2361, + "step": 3136 + }, + { + "epoch": 4.6, + "grad_norm": 1.8690835237503052, + "learning_rate": 0.00011141947429833096, + "loss": 3.1527, + "step": 3137 + }, + { + "epoch": 4.6, + "grad_norm": 3.516594886779785, + "learning_rate": 0.00011137287642463545, + "loss": 3.1821, + "step": 3138 + }, + { + "epoch": 4.6, + "grad_norm": 2.7472705841064453, + "learning_rate": 0.00011132627604898338, + "loss": 3.1594, + "step": 3139 + }, + { + "epoch": 4.6, + "grad_norm": 1.927894115447998, + "learning_rate": 0.00011127967318162644, + "loss": 3.1446, + "step": 3140 + }, + { + "epoch": 4.6, + "grad_norm": 1.823498249053955, + "learning_rate": 0.00011123306783281701, + "loss": 3.1621, + "step": 3141 + }, + { + "epoch": 4.6, + "grad_norm": 2.379281759262085, + "learning_rate": 0.00011118646001280799, + "loss": 3.1441, + "step": 3142 + }, + { + "epoch": 4.6, + "grad_norm": 1.6748161315917969, + "learning_rate": 0.00011113984973185268, + "loss": 3.1687, + "step": 3143 + }, + { + "epoch": 4.61, + "grad_norm": 4.564174652099609, + "learning_rate": 0.00011109323700020508, + "loss": 3.1976, + "step": 3144 + }, + { + "epoch": 4.61, + "grad_norm": 4.5107550621032715, + "learning_rate": 0.00011104662182811974, + "loss": 3.1711, + "step": 3145 + }, + { + "epoch": 4.61, + "grad_norm": 1.1205596923828125, + "learning_rate": 0.00011100000422585158, + "loss": 3.1573, + "step": 3146 + }, + { + "epoch": 4.61, + "grad_norm": 2.339171886444092, + "learning_rate": 0.00011095338420365621, + "loss": 3.1498, + "step": 3147 + }, + { + "epoch": 4.61, + "grad_norm": 2.139634609222412, + "learning_rate": 0.00011090676177178973, + "loss": 3.1493, + "step": 3148 + }, + { + "epoch": 4.61, + "grad_norm": 0.9772034287452698, + "learning_rate": 0.00011086013694050876, + "loss": 3.1551, + "step": 3149 + }, + { + "epoch": 4.61, + "grad_norm": 3.0146749019622803, + "learning_rate": 0.00011081350972007043, + "loss": 3.1186, + "step": 3150 + }, + { + "epoch": 4.62, + "grad_norm": 3.702209711074829, + "learning_rate": 0.00011076688012073244, + "loss": 3.2039, + "step": 3151 + }, + { + "epoch": 4.62, + "grad_norm": 6.443127155303955, + "learning_rate": 0.000110720248152753, + "loss": 3.1794, + "step": 3152 + }, + { + "epoch": 4.62, + "grad_norm": 3.7430996894836426, + "learning_rate": 0.00011067361382639079, + "loss": 3.172, + "step": 3153 + }, + { + "epoch": 4.62, + "grad_norm": 0.7498925924301147, + "learning_rate": 0.00011062697715190507, + "loss": 3.1905, + "step": 3154 + }, + { + "epoch": 4.62, + "grad_norm": 5.21573543548584, + "learning_rate": 0.00011058033813955557, + "loss": 3.1483, + "step": 3155 + }, + { + "epoch": 4.62, + "grad_norm": 5.235692977905273, + "learning_rate": 0.00011053369679960264, + "loss": 3.1197, + "step": 3156 + }, + { + "epoch": 4.62, + "grad_norm": 2.265049934387207, + "learning_rate": 0.00011048705314230696, + "loss": 3.1359, + "step": 3157 + }, + { + "epoch": 4.63, + "grad_norm": 6.700308799743652, + "learning_rate": 0.00011044040717792989, + "loss": 3.141, + "step": 3158 + }, + { + "epoch": 4.63, + "grad_norm": 7.944544792175293, + "learning_rate": 0.00011039375891673322, + "loss": 3.2029, + "step": 3159 + }, + { + "epoch": 4.63, + "grad_norm": 3.51708984375, + "learning_rate": 0.00011034710836897921, + "loss": 3.1747, + "step": 3160 + }, + { + "epoch": 4.63, + "grad_norm": 4.683743000030518, + "learning_rate": 0.00011030045554493075, + "loss": 3.181, + "step": 3161 + }, + { + "epoch": 4.63, + "grad_norm": 7.184266090393066, + "learning_rate": 0.00011025380045485107, + "loss": 3.1523, + "step": 3162 + }, + { + "epoch": 4.63, + "grad_norm": 5.593520164489746, + "learning_rate": 0.00011020714310900401, + "loss": 3.1564, + "step": 3163 + }, + { + "epoch": 4.64, + "grad_norm": 1.7525691986083984, + "learning_rate": 0.0001101604835176539, + "loss": 3.1461, + "step": 3164 + }, + { + "epoch": 4.64, + "grad_norm": 6.076540470123291, + "learning_rate": 0.00011011382169106546, + "loss": 3.1734, + "step": 3165 + }, + { + "epoch": 4.64, + "grad_norm": 7.846095561981201, + "learning_rate": 0.00011006715763950406, + "loss": 3.1622, + "step": 3166 + }, + { + "epoch": 4.64, + "grad_norm": 4.10769510269165, + "learning_rate": 0.00011002049137323543, + "loss": 3.1823, + "step": 3167 + }, + { + "epoch": 4.64, + "grad_norm": 4.750815391540527, + "learning_rate": 0.00010997382290252585, + "loss": 3.1425, + "step": 3168 + }, + { + "epoch": 4.64, + "grad_norm": 4.68991231918335, + "learning_rate": 0.0001099271522376421, + "loss": 3.1231, + "step": 3169 + }, + { + "epoch": 4.64, + "grad_norm": 3.5092368125915527, + "learning_rate": 0.00010988047938885135, + "loss": 3.2153, + "step": 3170 + }, + { + "epoch": 4.65, + "grad_norm": 4.325351715087891, + "learning_rate": 0.00010983380436642136, + "loss": 3.1567, + "step": 3171 + }, + { + "epoch": 4.65, + "grad_norm": 3.288820505142212, + "learning_rate": 0.00010978712718062036, + "loss": 3.1142, + "step": 3172 + }, + { + "epoch": 4.65, + "grad_norm": 3.504354238510132, + "learning_rate": 0.00010974044784171693, + "loss": 3.143, + "step": 3173 + }, + { + "epoch": 4.65, + "grad_norm": 4.500937461853027, + "learning_rate": 0.00010969376635998023, + "loss": 3.1615, + "step": 3174 + }, + { + "epoch": 4.65, + "grad_norm": 39.75580596923828, + "learning_rate": 0.00010964708274567994, + "loss": 3.2307, + "step": 3175 + }, + { + "epoch": 4.65, + "grad_norm": 3.0470621585845947, + "learning_rate": 0.00010960039700908608, + "loss": 3.1374, + "step": 3176 + }, + { + "epoch": 4.65, + "grad_norm": 2.1173510551452637, + "learning_rate": 0.00010955370916046918, + "loss": 3.1752, + "step": 3177 + }, + { + "epoch": 4.66, + "grad_norm": 2.3193938732147217, + "learning_rate": 0.00010950701921010036, + "loss": 3.1474, + "step": 3178 + }, + { + "epoch": 4.66, + "grad_norm": 2.5212008953094482, + "learning_rate": 0.00010946032716825099, + "loss": 3.2541, + "step": 3179 + }, + { + "epoch": 4.66, + "grad_norm": 1.1363469362258911, + "learning_rate": 0.00010941363304519305, + "loss": 3.1424, + "step": 3180 + }, + { + "epoch": 4.66, + "grad_norm": 2.563216209411621, + "learning_rate": 0.00010936693685119892, + "loss": 3.1572, + "step": 3181 + }, + { + "epoch": 4.66, + "grad_norm": 1.520682454109192, + "learning_rate": 0.00010932023859654145, + "loss": 3.1507, + "step": 3182 + }, + { + "epoch": 4.66, + "grad_norm": 3.5463900566101074, + "learning_rate": 0.00010927353829149398, + "loss": 3.1672, + "step": 3183 + }, + { + "epoch": 4.66, + "grad_norm": 4.035703659057617, + "learning_rate": 0.00010922683594633021, + "loss": 3.2053, + "step": 3184 + }, + { + "epoch": 4.67, + "grad_norm": 3.0087568759918213, + "learning_rate": 0.00010918013157132436, + "loss": 3.1566, + "step": 3185 + }, + { + "epoch": 4.67, + "grad_norm": 2.657742500305176, + "learning_rate": 0.00010913342517675113, + "loss": 3.1382, + "step": 3186 + }, + { + "epoch": 4.67, + "grad_norm": 4.097416877746582, + "learning_rate": 0.00010908671677288552, + "loss": 3.1695, + "step": 3187 + }, + { + "epoch": 4.67, + "grad_norm": 3.2829132080078125, + "learning_rate": 0.00010904000637000313, + "loss": 3.2016, + "step": 3188 + }, + { + "epoch": 4.67, + "grad_norm": 0.8120042681694031, + "learning_rate": 0.00010899329397837996, + "loss": 3.1287, + "step": 3189 + }, + { + "epoch": 4.67, + "grad_norm": 1.4076851606369019, + "learning_rate": 0.00010894657960829235, + "loss": 3.1302, + "step": 3190 + }, + { + "epoch": 4.68, + "grad_norm": 2.486541986465454, + "learning_rate": 0.00010889986327001716, + "loss": 3.1751, + "step": 3191 + }, + { + "epoch": 4.68, + "grad_norm": 1.665196418762207, + "learning_rate": 0.00010885314497383175, + "loss": 3.1353, + "step": 3192 + }, + { + "epoch": 4.68, + "grad_norm": 3.312460422515869, + "learning_rate": 0.00010880642473001375, + "loss": 3.1851, + "step": 3193 + }, + { + "epoch": 4.68, + "grad_norm": 6.796298027038574, + "learning_rate": 0.0001087597025488413, + "loss": 3.1513, + "step": 3194 + }, + { + "epoch": 4.68, + "grad_norm": 1.5955018997192383, + "learning_rate": 0.00010871297844059304, + "loss": 3.1609, + "step": 3195 + }, + { + "epoch": 4.68, + "grad_norm": 4.358046531677246, + "learning_rate": 0.00010866625241554789, + "loss": 3.1224, + "step": 3196 + }, + { + "epoch": 4.68, + "grad_norm": 4.621119976043701, + "learning_rate": 0.00010861952448398529, + "loss": 3.1221, + "step": 3197 + }, + { + "epoch": 4.69, + "grad_norm": 1.6193621158599854, + "learning_rate": 0.00010857279465618508, + "loss": 3.1495, + "step": 3198 + }, + { + "epoch": 4.69, + "grad_norm": 3.5418941974639893, + "learning_rate": 0.00010852606294242749, + "loss": 3.182, + "step": 3199 + }, + { + "epoch": 4.69, + "grad_norm": 3.8759021759033203, + "learning_rate": 0.00010847932935299319, + "loss": 3.1323, + "step": 3200 + }, + { + "epoch": 4.69, + "grad_norm": 1.662946343421936, + "learning_rate": 0.00010843259389816325, + "loss": 3.1605, + "step": 3201 + }, + { + "epoch": 4.69, + "grad_norm": 3.667335271835327, + "learning_rate": 0.00010838585658821918, + "loss": 3.1713, + "step": 3202 + }, + { + "epoch": 4.69, + "grad_norm": 4.988526344299316, + "learning_rate": 0.00010833911743344287, + "loss": 3.1812, + "step": 3203 + }, + { + "epoch": 4.69, + "grad_norm": 3.444049596786499, + "learning_rate": 0.00010829237644411659, + "loss": 3.1978, + "step": 3204 + }, + { + "epoch": 4.7, + "grad_norm": 2.074106216430664, + "learning_rate": 0.00010824563363052307, + "loss": 3.1567, + "step": 3205 + }, + { + "epoch": 4.7, + "grad_norm": 5.682570934295654, + "learning_rate": 0.00010819888900294542, + "loss": 3.1777, + "step": 3206 + }, + { + "epoch": 4.7, + "grad_norm": 7.264477729797363, + "learning_rate": 0.00010815214257166712, + "loss": 3.2065, + "step": 3207 + }, + { + "epoch": 4.7, + "grad_norm": 4.073165416717529, + "learning_rate": 0.0001081053943469721, + "loss": 3.1564, + "step": 3208 + }, + { + "epoch": 4.7, + "grad_norm": 3.7732391357421875, + "learning_rate": 0.00010805864433914465, + "loss": 3.1654, + "step": 3209 + }, + { + "epoch": 4.7, + "grad_norm": 4.149001598358154, + "learning_rate": 0.00010801189255846944, + "loss": 3.1563, + "step": 3210 + }, + { + "epoch": 4.71, + "grad_norm": 4.86202335357666, + "learning_rate": 0.00010796513901523155, + "loss": 3.1624, + "step": 3211 + }, + { + "epoch": 4.71, + "eval_loss": 3.0107593536376953, + "eval_runtime": 1169.2042, + "eval_samples_per_second": 14.661, + "eval_steps_per_second": 0.917, + "step": 3211 + }, + { + "epoch": 4.71, + "grad_norm": 4.124433994293213, + "learning_rate": 0.00010791838371971648, + "loss": 3.1329, + "step": 3212 + }, + { + "epoch": 4.71, + "grad_norm": 1.8730932474136353, + "learning_rate": 0.00010787162668221004, + "loss": 3.1344, + "step": 3213 + }, + { + "epoch": 4.71, + "grad_norm": 3.890324831008911, + "learning_rate": 0.00010782486791299849, + "loss": 3.1527, + "step": 3214 + }, + { + "epoch": 4.71, + "grad_norm": 3.587141513824463, + "learning_rate": 0.00010777810742236844, + "loss": 3.1435, + "step": 3215 + }, + { + "epoch": 4.71, + "grad_norm": 1.505163311958313, + "learning_rate": 0.00010773134522060688, + "loss": 3.1474, + "step": 3216 + }, + { + "epoch": 4.71, + "grad_norm": 3.0347487926483154, + "learning_rate": 0.00010768458131800119, + "loss": 3.1363, + "step": 3217 + }, + { + "epoch": 4.72, + "grad_norm": 3.127628803253174, + "learning_rate": 0.0001076378157248391, + "loss": 3.118, + "step": 3218 + }, + { + "epoch": 4.72, + "grad_norm": 2.165717363357544, + "learning_rate": 0.00010759104845140877, + "loss": 3.1966, + "step": 3219 + }, + { + "epoch": 4.72, + "grad_norm": 2.3362936973571777, + "learning_rate": 0.00010754427950799861, + "loss": 3.1494, + "step": 3220 + }, + { + "epoch": 4.72, + "grad_norm": 2.43580961227417, + "learning_rate": 0.00010749750890489754, + "loss": 3.1281, + "step": 3221 + }, + { + "epoch": 4.72, + "grad_norm": 23.91058349609375, + "learning_rate": 0.00010745073665239476, + "loss": 3.1482, + "step": 3222 + }, + { + "epoch": 4.72, + "grad_norm": 3.7118659019470215, + "learning_rate": 0.00010740396276077985, + "loss": 3.1884, + "step": 3223 + }, + { + "epoch": 4.72, + "grad_norm": 3.4350130558013916, + "learning_rate": 0.00010735718724034273, + "loss": 3.1709, + "step": 3224 + }, + { + "epoch": 4.73, + "grad_norm": 0.7842398285865784, + "learning_rate": 0.00010731041010137375, + "loss": 3.162, + "step": 3225 + }, + { + "epoch": 4.73, + "grad_norm": 4.245161533355713, + "learning_rate": 0.00010726363135416351, + "loss": 3.1882, + "step": 3226 + }, + { + "epoch": 4.73, + "grad_norm": 6.0483622550964355, + "learning_rate": 0.00010721685100900308, + "loss": 3.1815, + "step": 3227 + }, + { + "epoch": 4.73, + "grad_norm": 4.542155742645264, + "learning_rate": 0.00010717006907618377, + "loss": 3.1413, + "step": 3228 + }, + { + "epoch": 4.73, + "grad_norm": 1.648128867149353, + "learning_rate": 0.00010712328556599733, + "loss": 3.1296, + "step": 3229 + }, + { + "epoch": 4.73, + "grad_norm": 5.657898902893066, + "learning_rate": 0.00010707650048873579, + "loss": 3.178, + "step": 3230 + }, + { + "epoch": 4.73, + "grad_norm": 5.276024341583252, + "learning_rate": 0.00010702971385469158, + "loss": 3.192, + "step": 3231 + }, + { + "epoch": 4.74, + "grad_norm": 4.145871639251709, + "learning_rate": 0.00010698292567415742, + "loss": 3.134, + "step": 3232 + }, + { + "epoch": 4.74, + "grad_norm": 12.922083854675293, + "learning_rate": 0.00010693613595742644, + "loss": 3.1523, + "step": 3233 + }, + { + "epoch": 4.74, + "grad_norm": 3.349857807159424, + "learning_rate": 0.00010688934471479199, + "loss": 3.1872, + "step": 3234 + }, + { + "epoch": 4.74, + "grad_norm": 2.6229217052459717, + "learning_rate": 0.00010684255195654788, + "loss": 3.1266, + "step": 3235 + }, + { + "epoch": 4.74, + "grad_norm": 3.148838520050049, + "learning_rate": 0.00010679575769298824, + "loss": 3.2267, + "step": 3236 + }, + { + "epoch": 4.74, + "grad_norm": 3.843184471130371, + "learning_rate": 0.00010674896193440742, + "loss": 3.1531, + "step": 3237 + }, + { + "epoch": 4.75, + "grad_norm": 2.107234239578247, + "learning_rate": 0.00010670216469110019, + "loss": 3.1549, + "step": 3238 + }, + { + "epoch": 4.75, + "grad_norm": 1.9575111865997314, + "learning_rate": 0.00010665536597336167, + "loss": 3.1334, + "step": 3239 + }, + { + "epoch": 4.75, + "grad_norm": 2.4358248710632324, + "learning_rate": 0.00010660856579148724, + "loss": 3.1869, + "step": 3240 + }, + { + "epoch": 4.75, + "grad_norm": 1.9233509302139282, + "learning_rate": 0.00010656176415577262, + "loss": 3.1206, + "step": 3241 + }, + { + "epoch": 4.75, + "grad_norm": 2.0048117637634277, + "learning_rate": 0.00010651496107651384, + "loss": 3.1099, + "step": 3242 + }, + { + "epoch": 4.75, + "grad_norm": 4.973776817321777, + "learning_rate": 0.00010646815656400731, + "loss": 3.1621, + "step": 3243 + }, + { + "epoch": 4.75, + "grad_norm": 3.1254444122314453, + "learning_rate": 0.00010642135062854965, + "loss": 3.1564, + "step": 3244 + }, + { + "epoch": 4.76, + "grad_norm": 2.404599666595459, + "learning_rate": 0.00010637454328043793, + "loss": 3.1772, + "step": 3245 + }, + { + "epoch": 4.76, + "grad_norm": 2.372962474822998, + "learning_rate": 0.00010632773452996936, + "loss": 3.1053, + "step": 3246 + }, + { + "epoch": 4.76, + "grad_norm": 1.7531036138534546, + "learning_rate": 0.00010628092438744167, + "loss": 3.1678, + "step": 3247 + }, + { + "epoch": 4.76, + "grad_norm": 15.43344783782959, + "learning_rate": 0.00010623411286315261, + "loss": 3.1616, + "step": 3248 + }, + { + "epoch": 4.76, + "grad_norm": 3.320650577545166, + "learning_rate": 0.00010618729996740056, + "loss": 3.2037, + "step": 3249 + }, + { + "epoch": 4.76, + "grad_norm": 14.557812690734863, + "learning_rate": 0.00010614048571048401, + "loss": 3.221, + "step": 3250 + }, + { + "epoch": 4.76, + "grad_norm": 1.5140577554702759, + "learning_rate": 0.00010609367010270168, + "loss": 3.1666, + "step": 3251 + }, + { + "epoch": 4.77, + "grad_norm": 2.14962100982666, + "learning_rate": 0.00010604685315435281, + "loss": 3.1691, + "step": 3252 + }, + { + "epoch": 4.77, + "grad_norm": 0.9101262092590332, + "learning_rate": 0.0001060000348757368, + "loss": 3.1835, + "step": 3253 + }, + { + "epoch": 4.77, + "grad_norm": 1.9691673517227173, + "learning_rate": 0.00010595321527715327, + "loss": 3.1425, + "step": 3254 + }, + { + "epoch": 4.77, + "grad_norm": 6.10009241104126, + "learning_rate": 0.0001059063943689023, + "loss": 3.163, + "step": 3255 + }, + { + "epoch": 4.77, + "grad_norm": 1.81471848487854, + "learning_rate": 0.00010585957216128415, + "loss": 3.1742, + "step": 3256 + }, + { + "epoch": 4.77, + "grad_norm": 1.8879374265670776, + "learning_rate": 0.0001058127486645994, + "loss": 3.1732, + "step": 3257 + }, + { + "epoch": 4.77, + "grad_norm": 0.9328217506408691, + "learning_rate": 0.0001057659238891489, + "loss": 3.1223, + "step": 3258 + }, + { + "epoch": 4.78, + "grad_norm": 1.236288070678711, + "learning_rate": 0.00010571909784523377, + "loss": 3.114, + "step": 3259 + }, + { + "epoch": 4.78, + "grad_norm": 2.6463773250579834, + "learning_rate": 0.00010567227054315545, + "loss": 3.1444, + "step": 3260 + }, + { + "epoch": 4.78, + "grad_norm": 1.5447609424591064, + "learning_rate": 0.00010562544199321562, + "loss": 3.1348, + "step": 3261 + }, + { + "epoch": 4.78, + "grad_norm": 3.4742603302001953, + "learning_rate": 0.00010557861220571625, + "loss": 3.1773, + "step": 3262 + }, + { + "epoch": 4.78, + "grad_norm": 2.8108527660369873, + "learning_rate": 0.0001055317811909596, + "loss": 3.1871, + "step": 3263 + }, + { + "epoch": 4.78, + "grad_norm": 2.6389920711517334, + "learning_rate": 0.00010548494895924816, + "loss": 3.1609, + "step": 3264 + }, + { + "epoch": 4.79, + "grad_norm": 2.764688491821289, + "learning_rate": 0.00010543811552088468, + "loss": 3.1937, + "step": 3265 + }, + { + "epoch": 4.79, + "grad_norm": 3.0502400398254395, + "learning_rate": 0.00010539128088617225, + "loss": 3.1669, + "step": 3266 + }, + { + "epoch": 4.79, + "grad_norm": 6.166628837585449, + "learning_rate": 0.00010534444506541417, + "loss": 3.142, + "step": 3267 + }, + { + "epoch": 4.79, + "grad_norm": 47.169883728027344, + "learning_rate": 0.00010529760806891395, + "loss": 3.1328, + "step": 3268 + }, + { + "epoch": 4.79, + "grad_norm": 3.0846803188323975, + "learning_rate": 0.0001052507699069755, + "loss": 3.1455, + "step": 3269 + }, + { + "epoch": 4.79, + "grad_norm": 2.2564151287078857, + "learning_rate": 0.00010520393058990285, + "loss": 3.1306, + "step": 3270 + }, + { + "epoch": 4.79, + "grad_norm": 1.4548720121383667, + "learning_rate": 0.00010515709012800033, + "loss": 3.1312, + "step": 3271 + }, + { + "epoch": 4.8, + "grad_norm": 1.8430453538894653, + "learning_rate": 0.00010511024853157255, + "loss": 3.1276, + "step": 3272 + }, + { + "epoch": 4.8, + "grad_norm": 1.2610652446746826, + "learning_rate": 0.00010506340581092433, + "loss": 3.1682, + "step": 3273 + }, + { + "epoch": 4.8, + "grad_norm": 0.8417081832885742, + "learning_rate": 0.00010501656197636078, + "loss": 3.1505, + "step": 3274 + }, + { + "epoch": 4.8, + "grad_norm": 1.3054739236831665, + "learning_rate": 0.00010496971703818721, + "loss": 3.1465, + "step": 3275 + }, + { + "epoch": 4.8, + "grad_norm": 2.4543936252593994, + "learning_rate": 0.00010492287100670919, + "loss": 3.2098, + "step": 3276 + }, + { + "epoch": 4.8, + "grad_norm": 1.5280994176864624, + "learning_rate": 0.00010487602389223257, + "loss": 3.1443, + "step": 3277 + }, + { + "epoch": 4.8, + "grad_norm": 1.9430744647979736, + "learning_rate": 0.00010482917570506334, + "loss": 3.1531, + "step": 3278 + }, + { + "epoch": 4.81, + "grad_norm": 1.8071426153182983, + "learning_rate": 0.00010478232645550782, + "loss": 3.171, + "step": 3279 + }, + { + "epoch": 4.81, + "grad_norm": 3.768815517425537, + "learning_rate": 0.00010473547615387256, + "loss": 3.1586, + "step": 3280 + }, + { + "epoch": 4.81, + "grad_norm": 2.8686270713806152, + "learning_rate": 0.00010468862481046427, + "loss": 3.112, + "step": 3281 + }, + { + "epoch": 4.81, + "grad_norm": 2.201878070831299, + "learning_rate": 0.0001046417724355899, + "loss": 3.1815, + "step": 3282 + }, + { + "epoch": 4.81, + "grad_norm": 1.5182592868804932, + "learning_rate": 0.00010459491903955676, + "loss": 3.1391, + "step": 3283 + }, + { + "epoch": 4.81, + "grad_norm": 23.839717864990234, + "learning_rate": 0.0001045480646326722, + "loss": 3.1331, + "step": 3284 + }, + { + "epoch": 4.81, + "grad_norm": 1.8520891666412354, + "learning_rate": 0.00010450120922524387, + "loss": 3.1622, + "step": 3285 + }, + { + "epoch": 4.82, + "grad_norm": 0.8244103789329529, + "learning_rate": 0.00010445435282757972, + "loss": 3.1308, + "step": 3286 + }, + { + "epoch": 4.82, + "grad_norm": 2.278554677963257, + "learning_rate": 0.00010440749544998778, + "loss": 3.1283, + "step": 3287 + }, + { + "epoch": 4.82, + "grad_norm": 3.664557933807373, + "learning_rate": 0.00010436063710277639, + "loss": 3.1424, + "step": 3288 + }, + { + "epoch": 4.82, + "grad_norm": 1.274968147277832, + "learning_rate": 0.00010431377779625407, + "loss": 3.1358, + "step": 3289 + }, + { + "epoch": 4.82, + "grad_norm": 0.9037652611732483, + "learning_rate": 0.00010426691754072952, + "loss": 3.1329, + "step": 3290 + }, + { + "epoch": 4.82, + "grad_norm": 2.134868860244751, + "learning_rate": 0.00010422005634651173, + "loss": 3.0978, + "step": 3291 + }, + { + "epoch": 4.83, + "grad_norm": 4.81277322769165, + "learning_rate": 0.00010417319422390979, + "loss": 3.118, + "step": 3292 + }, + { + "epoch": 4.83, + "grad_norm": 1.5502432584762573, + "learning_rate": 0.00010412633118323311, + "loss": 3.1258, + "step": 3293 + }, + { + "epoch": 4.83, + "grad_norm": 1.0471981763839722, + "learning_rate": 0.00010407946723479125, + "loss": 3.1163, + "step": 3294 + }, + { + "epoch": 4.83, + "grad_norm": 2.2549943923950195, + "learning_rate": 0.00010403260238889389, + "loss": 3.1441, + "step": 3295 + }, + { + "epoch": 4.83, + "grad_norm": 1.7782256603240967, + "learning_rate": 0.00010398573665585105, + "loss": 3.1251, + "step": 3296 + }, + { + "epoch": 4.83, + "grad_norm": 3.3199939727783203, + "learning_rate": 0.00010393887004597287, + "loss": 3.1731, + "step": 3297 + }, + { + "epoch": 4.83, + "grad_norm": 3.324904441833496, + "learning_rate": 0.00010389200256956968, + "loss": 3.1439, + "step": 3298 + }, + { + "epoch": 4.84, + "grad_norm": 2.5700957775115967, + "learning_rate": 0.00010384513423695196, + "loss": 3.138, + "step": 3299 + }, + { + "epoch": 4.84, + "grad_norm": 3.218172550201416, + "learning_rate": 0.00010379826505843054, + "loss": 3.1335, + "step": 3300 + }, + { + "epoch": 4.84, + "grad_norm": 1.4542341232299805, + "learning_rate": 0.00010375139504431622, + "loss": 3.1655, + "step": 3301 + }, + { + "epoch": 4.84, + "grad_norm": 21.637861251831055, + "learning_rate": 0.00010370452420492012, + "loss": 3.1785, + "step": 3302 + }, + { + "epoch": 4.84, + "grad_norm": 3.2379684448242188, + "learning_rate": 0.00010365765255055358, + "loss": 3.133, + "step": 3303 + }, + { + "epoch": 4.84, + "grad_norm": 1.356064796447754, + "learning_rate": 0.00010361078009152793, + "loss": 3.1157, + "step": 3304 + }, + { + "epoch": 4.84, + "grad_norm": 1.2905539274215698, + "learning_rate": 0.00010356390683815489, + "loss": 3.1529, + "step": 3305 + }, + { + "epoch": 4.85, + "grad_norm": 2.5484352111816406, + "learning_rate": 0.00010351703280074623, + "loss": 3.1459, + "step": 3306 + }, + { + "epoch": 4.85, + "grad_norm": 1.1913775205612183, + "learning_rate": 0.0001034701579896139, + "loss": 3.2115, + "step": 3307 + }, + { + "epoch": 4.85, + "grad_norm": 1.7160303592681885, + "learning_rate": 0.0001034232824150701, + "loss": 3.1572, + "step": 3308 + }, + { + "epoch": 4.85, + "grad_norm": 2.059537887573242, + "learning_rate": 0.0001033764060874271, + "loss": 3.1361, + "step": 3309 + }, + { + "epoch": 4.85, + "grad_norm": 1.258538007736206, + "learning_rate": 0.00010332952901699738, + "loss": 3.1483, + "step": 3310 + }, + { + "epoch": 4.85, + "grad_norm": 2.301053285598755, + "learning_rate": 0.00010328265121409361, + "loss": 3.214, + "step": 3311 + }, + { + "epoch": 4.86, + "grad_norm": 1.366959810256958, + "learning_rate": 0.00010323577268902853, + "loss": 3.1272, + "step": 3312 + }, + { + "epoch": 4.86, + "grad_norm": 1.7083238363265991, + "learning_rate": 0.00010318889345211518, + "loss": 3.1432, + "step": 3313 + }, + { + "epoch": 4.86, + "grad_norm": 1.0292158126831055, + "learning_rate": 0.00010314201351366666, + "loss": 3.16, + "step": 3314 + }, + { + "epoch": 4.86, + "grad_norm": 1.8825881481170654, + "learning_rate": 0.0001030951328839962, + "loss": 3.1021, + "step": 3315 + }, + { + "epoch": 4.86, + "grad_norm": 1.2715975046157837, + "learning_rate": 0.00010304825157341722, + "loss": 3.1305, + "step": 3316 + }, + { + "epoch": 4.86, + "grad_norm": 1.972051978111267, + "learning_rate": 0.00010300136959224335, + "loss": 3.1384, + "step": 3317 + }, + { + "epoch": 4.86, + "grad_norm": 2.031904697418213, + "learning_rate": 0.00010295448695078829, + "loss": 3.1444, + "step": 3318 + }, + { + "epoch": 4.87, + "grad_norm": 1.3081032037734985, + "learning_rate": 0.00010290760365936587, + "loss": 3.1418, + "step": 3319 + }, + { + "epoch": 4.87, + "grad_norm": 1.2461774349212646, + "learning_rate": 0.00010286071972829016, + "loss": 3.1776, + "step": 3320 + }, + { + "epoch": 4.87, + "grad_norm": 614.052978515625, + "learning_rate": 0.00010281383516787528, + "loss": 3.162, + "step": 3321 + }, + { + "epoch": 4.87, + "grad_norm": 16.35662269592285, + "learning_rate": 0.0001027669499884355, + "loss": 3.1498, + "step": 3322 + }, + { + "epoch": 4.87, + "grad_norm": 5.687806606292725, + "learning_rate": 0.00010272006420028528, + "loss": 3.1927, + "step": 3323 + }, + { + "epoch": 4.87, + "grad_norm": 4.659135341644287, + "learning_rate": 0.00010267317781373918, + "loss": 3.1678, + "step": 3324 + }, + { + "epoch": 4.87, + "grad_norm": 6.275466442108154, + "learning_rate": 0.00010262629083911186, + "loss": 3.1716, + "step": 3325 + }, + { + "epoch": 4.88, + "grad_norm": 3.4971659183502197, + "learning_rate": 0.00010257940328671813, + "loss": 3.1567, + "step": 3326 + }, + { + "epoch": 4.88, + "grad_norm": 4.331486225128174, + "learning_rate": 0.00010253251516687303, + "loss": 3.164, + "step": 3327 + }, + { + "epoch": 4.88, + "grad_norm": 2.695127010345459, + "learning_rate": 0.00010248562648989153, + "loss": 3.1222, + "step": 3328 + }, + { + "epoch": 4.88, + "grad_norm": 2.059220314025879, + "learning_rate": 0.00010243873726608881, + "loss": 3.1783, + "step": 3329 + }, + { + "epoch": 4.88, + "grad_norm": 2.7665822505950928, + "learning_rate": 0.0001023918475057803, + "loss": 3.1525, + "step": 3330 + }, + { + "epoch": 4.88, + "grad_norm": 1.4520095586776733, + "learning_rate": 0.00010234495721928137, + "loss": 3.1559, + "step": 3331 + }, + { + "epoch": 4.88, + "grad_norm": 3.7544808387756348, + "learning_rate": 0.00010229806641690754, + "loss": 3.196, + "step": 3332 + }, + { + "epoch": 4.89, + "grad_norm": 4.363980770111084, + "learning_rate": 0.0001022511751089745, + "loss": 3.1835, + "step": 3333 + }, + { + "epoch": 4.89, + "grad_norm": 2.252725124359131, + "learning_rate": 0.00010220428330579804, + "loss": 3.1515, + "step": 3334 + }, + { + "epoch": 4.89, + "grad_norm": 2.367949962615967, + "learning_rate": 0.00010215739101769399, + "loss": 3.1599, + "step": 3335 + }, + { + "epoch": 4.89, + "grad_norm": 2.9788756370544434, + "learning_rate": 0.00010211049825497841, + "loss": 3.147, + "step": 3336 + }, + { + "epoch": 4.89, + "grad_norm": 1.9200280904769897, + "learning_rate": 0.00010206360502796734, + "loss": 3.1723, + "step": 3337 + }, + { + "epoch": 4.89, + "grad_norm": 2.547455072402954, + "learning_rate": 0.00010201671134697699, + "loss": 3.1607, + "step": 3338 + }, + { + "epoch": 4.9, + "grad_norm": 3.103799343109131, + "learning_rate": 0.00010196981722232368, + "loss": 3.1501, + "step": 3339 + }, + { + "epoch": 4.9, + "grad_norm": 0.7216553092002869, + "learning_rate": 0.00010192292266432377, + "loss": 3.1397, + "step": 3340 + }, + { + "epoch": 4.9, + "grad_norm": 3.535027265548706, + "learning_rate": 0.00010187602768329378, + "loss": 3.1029, + "step": 3341 + }, + { + "epoch": 4.9, + "grad_norm": 3.598963975906372, + "learning_rate": 0.00010182913228955025, + "loss": 3.1537, + "step": 3342 + }, + { + "epoch": 4.9, + "grad_norm": 1.5128989219665527, + "learning_rate": 0.00010178223649340985, + "loss": 3.1342, + "step": 3343 + }, + { + "epoch": 4.9, + "grad_norm": 1.6262668371200562, + "learning_rate": 0.00010173534030518941, + "loss": 3.1654, + "step": 3344 + }, + { + "epoch": 4.9, + "grad_norm": 1.0745606422424316, + "learning_rate": 0.00010168844373520572, + "loss": 3.1105, + "step": 3345 + }, + { + "epoch": 4.91, + "grad_norm": 1.7129157781600952, + "learning_rate": 0.00010164154679377569, + "loss": 3.1317, + "step": 3346 + }, + { + "epoch": 4.91, + "grad_norm": 1.3464902639389038, + "learning_rate": 0.00010159464949121641, + "loss": 3.1503, + "step": 3347 + }, + { + "epoch": 4.91, + "grad_norm": 8.157245635986328, + "learning_rate": 0.00010154775183784493, + "loss": 3.1043, + "step": 3348 + }, + { + "epoch": 4.91, + "grad_norm": 4.697030544281006, + "learning_rate": 0.0001015008538439784, + "loss": 3.194, + "step": 3349 + }, + { + "epoch": 4.91, + "grad_norm": 34.21104049682617, + "learning_rate": 0.00010145395551993408, + "loss": 3.171, + "step": 3350 + }, + { + "epoch": 4.91, + "grad_norm": 3.0200319290161133, + "learning_rate": 0.00010140705687602929, + "loss": 3.1502, + "step": 3351 + }, + { + "epoch": 4.91, + "grad_norm": 0.9504959583282471, + "learning_rate": 0.00010136015792258142, + "loss": 3.1142, + "step": 3352 + }, + { + "epoch": 4.92, + "grad_norm": 2.7994866371154785, + "learning_rate": 0.00010131325866990793, + "loss": 3.1242, + "step": 3353 + }, + { + "epoch": 4.92, + "grad_norm": 2.5278077125549316, + "learning_rate": 0.00010126635912832631, + "loss": 3.0951, + "step": 3354 + }, + { + "epoch": 4.92, + "grad_norm": 3.453552007675171, + "learning_rate": 0.00010121945930815422, + "loss": 3.1458, + "step": 3355 + }, + { + "epoch": 4.92, + "grad_norm": 3.2152633666992188, + "learning_rate": 0.00010117255921970921, + "loss": 3.1589, + "step": 3356 + }, + { + "epoch": 4.92, + "grad_norm": 1.388268232345581, + "learning_rate": 0.00010112565887330903, + "loss": 3.1496, + "step": 3357 + }, + { + "epoch": 4.92, + "grad_norm": 3.23463773727417, + "learning_rate": 0.00010107875827927147, + "loss": 3.1184, + "step": 3358 + }, + { + "epoch": 4.92, + "grad_norm": 2.970151662826538, + "learning_rate": 0.00010103185744791431, + "loss": 3.0931, + "step": 3359 + }, + { + "epoch": 4.93, + "grad_norm": 7.270925998687744, + "learning_rate": 0.00010098495638955538, + "loss": 3.1356, + "step": 3360 + }, + { + "epoch": 4.93, + "grad_norm": 18.72118377685547, + "learning_rate": 0.00010093805511451271, + "loss": 3.1538, + "step": 3361 + }, + { + "epoch": 4.93, + "grad_norm": 52.44887161254883, + "learning_rate": 0.00010089115363310417, + "loss": 3.1924, + "step": 3362 + }, + { + "epoch": 4.93, + "grad_norm": 7.027931213378906, + "learning_rate": 0.00010084425195564782, + "loss": 3.1726, + "step": 3363 + }, + { + "epoch": 4.93, + "grad_norm": 17.143966674804688, + "learning_rate": 0.00010079735009246167, + "loss": 3.1921, + "step": 3364 + }, + { + "epoch": 4.93, + "grad_norm": 3.896998167037964, + "learning_rate": 0.00010075044805386386, + "loss": 3.1341, + "step": 3365 + }, + { + "epoch": 4.94, + "grad_norm": 4.068105697631836, + "learning_rate": 0.0001007035458501725, + "loss": 3.1548, + "step": 3366 + }, + { + "epoch": 4.94, + "grad_norm": 3.3968491554260254, + "learning_rate": 0.00010065664349170578, + "loss": 3.146, + "step": 3367 + }, + { + "epoch": 4.94, + "grad_norm": 3.594057083129883, + "learning_rate": 0.00010060974098878189, + "loss": 3.1631, + "step": 3368 + }, + { + "epoch": 4.94, + "grad_norm": 0.987125813961029, + "learning_rate": 0.00010056283835171906, + "loss": 3.1319, + "step": 3369 + }, + { + "epoch": 4.94, + "grad_norm": 4.898683547973633, + "learning_rate": 0.00010051593559083557, + "loss": 3.1316, + "step": 3370 + }, + { + "epoch": 4.94, + "grad_norm": 6.864829063415527, + "learning_rate": 0.0001004690327164497, + "loss": 3.2037, + "step": 3371 + }, + { + "epoch": 4.94, + "grad_norm": 5.136176109313965, + "learning_rate": 0.00010042212973887982, + "loss": 3.1921, + "step": 3372 + }, + { + "epoch": 4.95, + "grad_norm": 2.621640205383301, + "learning_rate": 0.00010037522666844415, + "loss": 3.1716, + "step": 3373 + }, + { + "epoch": 4.95, + "grad_norm": 4.9672746658325195, + "learning_rate": 0.00010032832351546117, + "loss": 3.1374, + "step": 3374 + }, + { + "epoch": 4.95, + "grad_norm": 6.860297203063965, + "learning_rate": 0.00010028142029024923, + "loss": 3.1863, + "step": 3375 + }, + { + "epoch": 4.95, + "grad_norm": 131.51553344726562, + "learning_rate": 0.00010023451700312669, + "loss": 3.1629, + "step": 3376 + }, + { + "epoch": 4.95, + "grad_norm": 4.443822860717773, + "learning_rate": 0.00010018761366441198, + "loss": 3.1513, + "step": 3377 + }, + { + "epoch": 4.95, + "grad_norm": 1.4408992528915405, + "learning_rate": 0.00010014071028442354, + "loss": 3.1317, + "step": 3378 + }, + { + "epoch": 4.95, + "grad_norm": 8.419514656066895, + "learning_rate": 0.00010009380687347975, + "loss": 3.1649, + "step": 3379 + }, + { + "epoch": 4.96, + "grad_norm": 19.037832260131836, + "learning_rate": 0.00010004690344189912, + "loss": 3.1697, + "step": 3380 + }, + { + "epoch": 4.96, + "eval_loss": 3.100813865661621, + "eval_runtime": 1176.1523, + "eval_samples_per_second": 14.575, + "eval_steps_per_second": 0.911, + "step": 3380 + }, + { + "epoch": 4.96, + "grad_norm": 11.52230167388916, + "learning_rate": 0.0001, + "loss": 3.2739, + "step": 3381 + }, + { + "epoch": 4.96, + "grad_norm": 8.755014419555664, + "learning_rate": 9.995309655810093e-05, + "loss": 3.2172, + "step": 3382 + }, + { + "epoch": 4.96, + "grad_norm": 5.496139049530029, + "learning_rate": 9.990619312652026e-05, + "loss": 3.1656, + "step": 3383 + }, + { + "epoch": 4.96, + "grad_norm": 3.737701416015625, + "learning_rate": 9.985928971557649e-05, + "loss": 3.2014, + "step": 3384 + }, + { + "epoch": 4.96, + "grad_norm": 6.020711898803711, + "learning_rate": 9.981238633558805e-05, + "loss": 3.2036, + "step": 3385 + }, + { + "epoch": 4.96, + "grad_norm": 7.874726295471191, + "learning_rate": 9.976548299687332e-05, + "loss": 3.1735, + "step": 3386 + }, + { + "epoch": 4.97, + "grad_norm": 7.075169563293457, + "learning_rate": 9.97185797097508e-05, + "loss": 3.1674, + "step": 3387 + }, + { + "epoch": 4.97, + "grad_norm": 3.235295057296753, + "learning_rate": 9.967167648453885e-05, + "loss": 3.1999, + "step": 3388 + }, + { + "epoch": 4.97, + "grad_norm": 3.599179744720459, + "learning_rate": 9.962477333155585e-05, + "loss": 3.1829, + "step": 3389 + }, + { + "epoch": 4.97, + "grad_norm": 6.6362528800964355, + "learning_rate": 9.95778702611202e-05, + "loss": 3.2165, + "step": 3390 + }, + { + "epoch": 4.97, + "grad_norm": 7.1437153816223145, + "learning_rate": 9.953096728355031e-05, + "loss": 3.1884, + "step": 3391 + }, + { + "epoch": 4.97, + "grad_norm": 4.677809715270996, + "learning_rate": 9.948406440916443e-05, + "loss": 3.1899, + "step": 3392 + }, + { + "epoch": 4.98, + "grad_norm": 1.904357671737671, + "learning_rate": 9.943716164828096e-05, + "loss": 3.1357, + "step": 3393 + }, + { + "epoch": 4.98, + "grad_norm": 5.046501636505127, + "learning_rate": 9.939025901121814e-05, + "loss": 3.1666, + "step": 3394 + }, + { + "epoch": 4.98, + "grad_norm": 6.0771074295043945, + "learning_rate": 9.934335650829425e-05, + "loss": 3.1961, + "step": 3395 + }, + { + "epoch": 4.98, + "grad_norm": 5.269542694091797, + "learning_rate": 9.92964541498275e-05, + "loss": 3.1937, + "step": 3396 + }, + { + "epoch": 4.98, + "grad_norm": 2.9679410457611084, + "learning_rate": 9.924955194613616e-05, + "loss": 3.1376, + "step": 3397 + }, + { + "epoch": 4.98, + "grad_norm": 3.366711139678955, + "learning_rate": 9.920264990753837e-05, + "loss": 3.1116, + "step": 3398 + }, + { + "epoch": 4.98, + "grad_norm": 6.38279390335083, + "learning_rate": 9.91557480443522e-05, + "loss": 3.1385, + "step": 3399 + }, + { + "epoch": 4.99, + "grad_norm": 5.471691608428955, + "learning_rate": 9.910884636689585e-05, + "loss": 3.1874, + "step": 3400 + }, + { + "epoch": 4.99, + "grad_norm": 2.8460099697113037, + "learning_rate": 9.906194488548734e-05, + "loss": 3.1898, + "step": 3401 + }, + { + "epoch": 4.99, + "grad_norm": 3.193862199783325, + "learning_rate": 9.901504361044461e-05, + "loss": 3.1727, + "step": 3402 + }, + { + "epoch": 4.99, + "grad_norm": 4.644516944885254, + "learning_rate": 9.896814255208571e-05, + "loss": 3.1828, + "step": 3403 + }, + { + "epoch": 4.99, + "grad_norm": 4.342918872833252, + "learning_rate": 9.892124172072857e-05, + "loss": 3.1714, + "step": 3404 + }, + { + "epoch": 4.99, + "grad_norm": 2.4274344444274902, + "learning_rate": 9.887434112669097e-05, + "loss": 3.1509, + "step": 3405 + }, + { + "epoch": 4.99, + "grad_norm": 3.326143264770508, + "learning_rate": 9.882744078029081e-05, + "loss": 3.1433, + "step": 3406 + }, + { + "epoch": 5.0, + "grad_norm": 5.951056003570557, + "learning_rate": 9.878054069184583e-05, + "loss": 3.1543, + "step": 3407 + }, + { + "epoch": 5.0, + "grad_norm": 5.0941972732543945, + "learning_rate": 9.873364087167367e-05, + "loss": 3.1558, + "step": 3408 + }, + { + "epoch": 5.0, + "grad_norm": 2.075986623764038, + "learning_rate": 9.868674133009209e-05, + "loss": 3.144, + "step": 3409 + }, + { + "epoch": 5.0, + "grad_norm": 3.3112316131591797, + "learning_rate": 9.863984207741859e-05, + "loss": 3.1444, + "step": 3410 + }, + { + "epoch": 5.0, + "grad_norm": 4.910267353057861, + "learning_rate": 9.85929431239707e-05, + "loss": 3.1823, + "step": 3411 + }, + { + "epoch": 5.0, + "grad_norm": 4.1945624351501465, + "learning_rate": 9.854604448006594e-05, + "loss": 3.1945, + "step": 3412 + }, + { + "epoch": 5.01, + "grad_norm": 1.877642273902893, + "learning_rate": 9.849914615602163e-05, + "loss": 3.1357, + "step": 3413 + }, + { + "epoch": 5.01, + "grad_norm": 3.460303783416748, + "learning_rate": 9.845224816215508e-05, + "loss": 3.1039, + "step": 3414 + }, + { + "epoch": 5.01, + "grad_norm": 5.127685070037842, + "learning_rate": 9.84053505087836e-05, + "loss": 3.1041, + "step": 3415 + }, + { + "epoch": 5.01, + "grad_norm": 3.685288906097412, + "learning_rate": 9.835845320622434e-05, + "loss": 3.1874, + "step": 3416 + }, + { + "epoch": 5.01, + "grad_norm": 1.9846138954162598, + "learning_rate": 9.83115562647943e-05, + "loss": 3.1365, + "step": 3417 + }, + { + "epoch": 5.01, + "grad_norm": 5.992250442504883, + "learning_rate": 9.826465969481061e-05, + "loss": 3.0993, + "step": 3418 + }, + { + "epoch": 5.01, + "grad_norm": 4.008816242218018, + "learning_rate": 9.821776350659017e-05, + "loss": 3.1471, + "step": 3419 + }, + { + "epoch": 5.0, + "grad_norm": 2.797140598297119, + "learning_rate": 9.81708677104498e-05, + "loss": 3.1822, + "step": 3420 + }, + { + "epoch": 5.0, + "grad_norm": 3.2682175636291504, + "learning_rate": 9.812397231670624e-05, + "loss": 3.1501, + "step": 3421 + }, + { + "epoch": 5.0, + "grad_norm": 4.154157638549805, + "learning_rate": 9.807707733567626e-05, + "loss": 3.1572, + "step": 3422 + }, + { + "epoch": 5.01, + "grad_norm": 2.315260648727417, + "learning_rate": 9.803018277767634e-05, + "loss": 3.1305, + "step": 3423 + }, + { + "epoch": 5.01, + "grad_norm": 4.300922870635986, + "learning_rate": 9.798328865302299e-05, + "loss": 3.2001, + "step": 3424 + }, + { + "epoch": 5.01, + "grad_norm": 4.946724891662598, + "learning_rate": 9.793639497203268e-05, + "loss": 3.1431, + "step": 3425 + }, + { + "epoch": 5.01, + "grad_norm": 1.1593897342681885, + "learning_rate": 9.788950174502164e-05, + "loss": 3.1631, + "step": 3426 + }, + { + "epoch": 5.01, + "grad_norm": 88.28909301757812, + "learning_rate": 9.784260898230603e-05, + "loss": 3.1888, + "step": 3427 + }, + { + "epoch": 5.01, + "grad_norm": 7.40468168258667, + "learning_rate": 9.7795716694202e-05, + "loss": 3.1605, + "step": 3428 + }, + { + "epoch": 5.01, + "grad_norm": 8.019392967224121, + "learning_rate": 9.774882489102555e-05, + "loss": 3.2278, + "step": 3429 + }, + { + "epoch": 5.02, + "grad_norm": 5.9179205894470215, + "learning_rate": 9.770193358309248e-05, + "loss": 3.1715, + "step": 3430 + }, + { + "epoch": 5.02, + "grad_norm": 2.9003238677978516, + "learning_rate": 9.765504278071866e-05, + "loss": 3.1651, + "step": 3431 + }, + { + "epoch": 5.02, + "grad_norm": 4.174776554107666, + "learning_rate": 9.760815249421973e-05, + "loss": 3.1573, + "step": 3432 + }, + { + "epoch": 5.02, + "grad_norm": 5.618264198303223, + "learning_rate": 9.756126273391118e-05, + "loss": 3.1799, + "step": 3433 + }, + { + "epoch": 5.02, + "grad_norm": 5.217837333679199, + "learning_rate": 9.75143735101085e-05, + "loss": 3.1674, + "step": 3434 + }, + { + "epoch": 5.02, + "grad_norm": 6.881947994232178, + "learning_rate": 9.746748483312702e-05, + "loss": 3.2233, + "step": 3435 + }, + { + "epoch": 5.02, + "grad_norm": 4.155002117156982, + "learning_rate": 9.742059671328185e-05, + "loss": 3.1715, + "step": 3436 + }, + { + "epoch": 5.03, + "grad_norm": 3.3155972957611084, + "learning_rate": 9.737370916088815e-05, + "loss": 3.1688, + "step": 3437 + }, + { + "epoch": 5.03, + "grad_norm": 4.904897689819336, + "learning_rate": 9.732682218626086e-05, + "loss": 3.1702, + "step": 3438 + }, + { + "epoch": 5.03, + "grad_norm": 27.04771614074707, + "learning_rate": 9.727993579971471e-05, + "loss": 3.1816, + "step": 3439 + }, + { + "epoch": 5.03, + "grad_norm": 3.3083443641662598, + "learning_rate": 9.723305001156452e-05, + "loss": 3.1635, + "step": 3440 + }, + { + "epoch": 5.03, + "grad_norm": 3.0965356826782227, + "learning_rate": 9.718616483212474e-05, + "loss": 3.1631, + "step": 3441 + }, + { + "epoch": 5.03, + "grad_norm": 2.470313310623169, + "learning_rate": 9.713928027170989e-05, + "loss": 3.1788, + "step": 3442 + }, + { + "epoch": 5.04, + "grad_norm": 1.455441951751709, + "learning_rate": 9.709239634063414e-05, + "loss": 3.1543, + "step": 3443 + }, + { + "epoch": 5.04, + "grad_norm": 3.4748833179473877, + "learning_rate": 9.704551304921175e-05, + "loss": 3.1228, + "step": 3444 + }, + { + "epoch": 5.04, + "grad_norm": 3.481116771697998, + "learning_rate": 9.69986304077567e-05, + "loss": 3.1595, + "step": 3445 + }, + { + "epoch": 5.04, + "grad_norm": 4.942490577697754, + "learning_rate": 9.69517484265828e-05, + "loss": 3.1443, + "step": 3446 + }, + { + "epoch": 5.04, + "grad_norm": 3.2956652641296387, + "learning_rate": 9.690486711600384e-05, + "loss": 3.1699, + "step": 3447 + }, + { + "epoch": 5.04, + "grad_norm": 2.4343531131744385, + "learning_rate": 9.68579864863334e-05, + "loss": 3.1488, + "step": 3448 + }, + { + "epoch": 5.04, + "grad_norm": 2.82985782623291, + "learning_rate": 9.681110654788482e-05, + "loss": 3.1032, + "step": 3449 + }, + { + "epoch": 5.05, + "grad_norm": 1.59348464012146, + "learning_rate": 9.676422731097149e-05, + "loss": 3.1256, + "step": 3450 + }, + { + "epoch": 5.05, + "grad_norm": 4.060868740081787, + "learning_rate": 9.671734878590644e-05, + "loss": 3.1592, + "step": 3451 + }, + { + "epoch": 5.05, + "grad_norm": 4.718034744262695, + "learning_rate": 9.667047098300262e-05, + "loss": 3.1666, + "step": 3452 + }, + { + "epoch": 5.05, + "grad_norm": 1.5934293270111084, + "learning_rate": 9.662359391257293e-05, + "loss": 3.1592, + "step": 3453 + }, + { + "epoch": 5.05, + "grad_norm": 4.770664215087891, + "learning_rate": 9.657671758492992e-05, + "loss": 3.1303, + "step": 3454 + }, + { + "epoch": 5.05, + "grad_norm": 5.743010520935059, + "learning_rate": 9.652984201038608e-05, + "loss": 3.1736, + "step": 3455 + }, + { + "epoch": 5.05, + "grad_norm": 22.09770393371582, + "learning_rate": 9.64829671992538e-05, + "loss": 3.1555, + "step": 3456 + }, + { + "epoch": 5.06, + "grad_norm": 4.085333347320557, + "learning_rate": 9.643609316184514e-05, + "loss": 3.1203, + "step": 3457 + }, + { + "epoch": 5.06, + "grad_norm": 3.675048589706421, + "learning_rate": 9.638921990847206e-05, + "loss": 3.128, + "step": 3458 + }, + { + "epoch": 5.06, + "grad_norm": 1.0433051586151123, + "learning_rate": 9.634234744944646e-05, + "loss": 3.1206, + "step": 3459 + }, + { + "epoch": 5.06, + "grad_norm": 4.096065521240234, + "learning_rate": 9.62954757950799e-05, + "loss": 3.1567, + "step": 3460 + }, + { + "epoch": 5.06, + "grad_norm": 5.4872870445251465, + "learning_rate": 9.624860495568379e-05, + "loss": 3.1482, + "step": 3461 + }, + { + "epoch": 5.06, + "grad_norm": 3.236421585083008, + "learning_rate": 9.620173494156949e-05, + "loss": 3.1172, + "step": 3462 + }, + { + "epoch": 5.06, + "grad_norm": 3.7397096157073975, + "learning_rate": 9.615486576304806e-05, + "loss": 3.1485, + "step": 3463 + }, + { + "epoch": 5.07, + "grad_norm": 2.640143632888794, + "learning_rate": 9.610799743043036e-05, + "loss": 3.1444, + "step": 3464 + }, + { + "epoch": 5.07, + "grad_norm": 2.277892589569092, + "learning_rate": 9.606112995402714e-05, + "loss": 3.117, + "step": 3465 + }, + { + "epoch": 5.07, + "grad_norm": 2.7552220821380615, + "learning_rate": 9.601426334414898e-05, + "loss": 3.1557, + "step": 3466 + }, + { + "epoch": 5.07, + "grad_norm": 3.1442368030548096, + "learning_rate": 9.596739761110613e-05, + "loss": 3.1161, + "step": 3467 + }, + { + "epoch": 5.07, + "grad_norm": 1.997355341911316, + "learning_rate": 9.592053276520877e-05, + "loss": 3.1332, + "step": 3468 + }, + { + "epoch": 5.07, + "grad_norm": 1.959526777267456, + "learning_rate": 9.58736688167669e-05, + "loss": 3.141, + "step": 3469 + }, + { + "epoch": 5.08, + "grad_norm": 2.0642502307891846, + "learning_rate": 9.582680577609022e-05, + "loss": 3.1532, + "step": 3470 + }, + { + "epoch": 5.08, + "grad_norm": 1.1112334728240967, + "learning_rate": 9.57799436534883e-05, + "loss": 3.1402, + "step": 3471 + }, + { + "epoch": 5.08, + "grad_norm": 2.348271131515503, + "learning_rate": 9.57330824592705e-05, + "loss": 3.162, + "step": 3472 + }, + { + "epoch": 5.08, + "grad_norm": 1.7725739479064941, + "learning_rate": 9.5686222203746e-05, + "loss": 3.159, + "step": 3473 + }, + { + "epoch": 5.08, + "grad_norm": 1.3065656423568726, + "learning_rate": 9.563936289722363e-05, + "loss": 3.155, + "step": 3474 + }, + { + "epoch": 5.08, + "grad_norm": 2.1432132720947266, + "learning_rate": 9.559250455001223e-05, + "loss": 3.0939, + "step": 3475 + }, + { + "epoch": 5.08, + "grad_norm": 1.2344911098480225, + "learning_rate": 9.554564717242032e-05, + "loss": 3.1094, + "step": 3476 + }, + { + "epoch": 5.09, + "grad_norm": 1.591896653175354, + "learning_rate": 9.549879077475613e-05, + "loss": 3.1632, + "step": 3477 + }, + { + "epoch": 5.09, + "grad_norm": 1.2728568315505981, + "learning_rate": 9.545193536732783e-05, + "loss": 3.1537, + "step": 3478 + }, + { + "epoch": 5.09, + "grad_norm": 0.9794215559959412, + "learning_rate": 9.540508096044329e-05, + "loss": 3.1012, + "step": 3479 + }, + { + "epoch": 5.09, + "grad_norm": 2.092595100402832, + "learning_rate": 9.535822756441011e-05, + "loss": 3.1352, + "step": 3480 + }, + { + "epoch": 5.09, + "grad_norm": 1.6915316581726074, + "learning_rate": 9.531137518953577e-05, + "loss": 3.1332, + "step": 3481 + }, + { + "epoch": 5.09, + "grad_norm": 6.075295925140381, + "learning_rate": 9.526452384612749e-05, + "loss": 3.1654, + "step": 3482 + }, + { + "epoch": 5.09, + "grad_norm": 2.9814789295196533, + "learning_rate": 9.521767354449219e-05, + "loss": 3.1196, + "step": 3483 + }, + { + "epoch": 5.1, + "grad_norm": 1.2538803815841675, + "learning_rate": 9.51708242949367e-05, + "loss": 3.1304, + "step": 3484 + }, + { + "epoch": 5.1, + "grad_norm": 1.274577260017395, + "learning_rate": 9.512397610776747e-05, + "loss": 3.1293, + "step": 3485 + }, + { + "epoch": 5.1, + "grad_norm": 10.467679023742676, + "learning_rate": 9.507712899329079e-05, + "loss": 3.1396, + "step": 3486 + }, + { + "epoch": 5.1, + "grad_norm": 3.3941073417663574, + "learning_rate": 9.503028296181281e-05, + "loss": 3.1455, + "step": 3487 + }, + { + "epoch": 5.1, + "grad_norm": 1.5310273170471191, + "learning_rate": 9.498343802363923e-05, + "loss": 3.141, + "step": 3488 + }, + { + "epoch": 5.1, + "grad_norm": 4.921885967254639, + "learning_rate": 9.493659418907566e-05, + "loss": 3.1257, + "step": 3489 + }, + { + "epoch": 5.1, + "grad_norm": 6.093456268310547, + "learning_rate": 9.488975146842746e-05, + "loss": 3.1542, + "step": 3490 + }, + { + "epoch": 5.11, + "grad_norm": 2.32529616355896, + "learning_rate": 9.484290987199968e-05, + "loss": 3.1527, + "step": 3491 + }, + { + "epoch": 5.11, + "grad_norm": 4.923595905303955, + "learning_rate": 9.47960694100972e-05, + "loss": 3.1667, + "step": 3492 + }, + { + "epoch": 5.11, + "grad_norm": 5.732090950012207, + "learning_rate": 9.474923009302452e-05, + "loss": 3.1218, + "step": 3493 + }, + { + "epoch": 5.11, + "grad_norm": 3.330801486968994, + "learning_rate": 9.470239193108607e-05, + "loss": 3.1252, + "step": 3494 + }, + { + "epoch": 5.11, + "grad_norm": 3.7033092975616455, + "learning_rate": 9.465555493458588e-05, + "loss": 3.1358, + "step": 3495 + }, + { + "epoch": 5.11, + "grad_norm": 6.471573352813721, + "learning_rate": 9.460871911382774e-05, + "loss": 3.1961, + "step": 3496 + }, + { + "epoch": 5.12, + "grad_norm": 6.723712921142578, + "learning_rate": 9.456188447911533e-05, + "loss": 3.0915, + "step": 3497 + }, + { + "epoch": 5.12, + "grad_norm": 1.7002429962158203, + "learning_rate": 9.451505104075188e-05, + "loss": 3.1452, + "step": 3498 + }, + { + "epoch": 5.12, + "grad_norm": 2.4126813411712646, + "learning_rate": 9.44682188090404e-05, + "loss": 3.1457, + "step": 3499 + }, + { + "epoch": 5.12, + "grad_norm": 1.8296704292297363, + "learning_rate": 9.442138779428376e-05, + "loss": 3.1215, + "step": 3500 + }, + { + "epoch": 5.12, + "grad_norm": 1.7919952869415283, + "learning_rate": 9.43745580067844e-05, + "loss": 3.179, + "step": 3501 + }, + { + "epoch": 5.12, + "grad_norm": 1.8230031728744507, + "learning_rate": 9.432772945684454e-05, + "loss": 3.1339, + "step": 3502 + }, + { + "epoch": 5.12, + "grad_norm": 1.5876286029815674, + "learning_rate": 9.428090215476625e-05, + "loss": 3.1248, + "step": 3503 + }, + { + "epoch": 5.13, + "grad_norm": 2.1340315341949463, + "learning_rate": 9.423407611085113e-05, + "loss": 3.1396, + "step": 3504 + }, + { + "epoch": 5.13, + "grad_norm": 1.6990450620651245, + "learning_rate": 9.41872513354006e-05, + "loss": 3.1119, + "step": 3505 + }, + { + "epoch": 5.13, + "grad_norm": 1.792682409286499, + "learning_rate": 9.414042783871586e-05, + "loss": 3.1476, + "step": 3506 + }, + { + "epoch": 5.13, + "grad_norm": 2.685699701309204, + "learning_rate": 9.409360563109773e-05, + "loss": 3.1468, + "step": 3507 + }, + { + "epoch": 5.13, + "grad_norm": 4.750230312347412, + "learning_rate": 9.404678472284673e-05, + "loss": 3.1361, + "step": 3508 + }, + { + "epoch": 5.13, + "grad_norm": 1.0667452812194824, + "learning_rate": 9.399996512426323e-05, + "loss": 3.1625, + "step": 3509 + }, + { + "epoch": 5.13, + "grad_norm": 2.092228889465332, + "learning_rate": 9.395314684564721e-05, + "loss": 3.1709, + "step": 3510 + }, + { + "epoch": 5.14, + "grad_norm": 1.991498589515686, + "learning_rate": 9.390632989729831e-05, + "loss": 3.1122, + "step": 3511 + }, + { + "epoch": 5.14, + "grad_norm": 10.466729164123535, + "learning_rate": 9.385951428951603e-05, + "loss": 3.1214, + "step": 3512 + }, + { + "epoch": 5.14, + "grad_norm": 3.6561026573181152, + "learning_rate": 9.381270003259945e-05, + "loss": 3.1363, + "step": 3513 + }, + { + "epoch": 5.14, + "grad_norm": 1.4475880861282349, + "learning_rate": 9.37658871368474e-05, + "loss": 3.1801, + "step": 3514 + }, + { + "epoch": 5.14, + "grad_norm": 5.58400821685791, + "learning_rate": 9.371907561255837e-05, + "loss": 3.1223, + "step": 3515 + }, + { + "epoch": 5.14, + "grad_norm": 9.835494041442871, + "learning_rate": 9.367226547003065e-05, + "loss": 3.2195, + "step": 3516 + }, + { + "epoch": 5.14, + "grad_norm": 6.080451011657715, + "learning_rate": 9.362545671956214e-05, + "loss": 3.1594, + "step": 3517 + }, + { + "epoch": 5.15, + "grad_norm": 1.3072457313537598, + "learning_rate": 9.357864937145036e-05, + "loss": 3.147, + "step": 3518 + }, + { + "epoch": 5.15, + "grad_norm": 5.248061180114746, + "learning_rate": 9.353184343599273e-05, + "loss": 3.1497, + "step": 3519 + }, + { + "epoch": 5.15, + "grad_norm": 5.6261420249938965, + "learning_rate": 9.34850389234862e-05, + "loss": 3.1824, + "step": 3520 + }, + { + "epoch": 5.15, + "grad_norm": 3.202202320098877, + "learning_rate": 9.343823584422742e-05, + "loss": 3.1609, + "step": 3521 + }, + { + "epoch": 5.15, + "grad_norm": 4.45296573638916, + "learning_rate": 9.339143420851279e-05, + "loss": 3.1792, + "step": 3522 + }, + { + "epoch": 5.15, + "grad_norm": 4.815643310546875, + "learning_rate": 9.334463402663837e-05, + "loss": 3.2226, + "step": 3523 + }, + { + "epoch": 5.16, + "grad_norm": 12.902685165405273, + "learning_rate": 9.32978353088998e-05, + "loss": 3.1462, + "step": 3524 + }, + { + "epoch": 5.16, + "grad_norm": 3.254309892654419, + "learning_rate": 9.32510380655926e-05, + "loss": 3.1306, + "step": 3525 + }, + { + "epoch": 5.16, + "grad_norm": 8.259904861450195, + "learning_rate": 9.320424230701179e-05, + "loss": 3.2094, + "step": 3526 + }, + { + "epoch": 5.16, + "grad_norm": 2.651756525039673, + "learning_rate": 9.31574480434521e-05, + "loss": 3.1366, + "step": 3527 + }, + { + "epoch": 5.16, + "grad_norm": 4.296948432922363, + "learning_rate": 9.311065528520802e-05, + "loss": 3.1398, + "step": 3528 + }, + { + "epoch": 5.16, + "grad_norm": 4.362511157989502, + "learning_rate": 9.30638640425736e-05, + "loss": 3.163, + "step": 3529 + }, + { + "epoch": 5.16, + "grad_norm": 1.9980674982070923, + "learning_rate": 9.301707432584257e-05, + "loss": 3.1884, + "step": 3530 + }, + { + "epoch": 5.17, + "grad_norm": 4.281768321990967, + "learning_rate": 9.297028614530844e-05, + "loss": 3.1488, + "step": 3531 + }, + { + "epoch": 5.17, + "grad_norm": 5.395291328430176, + "learning_rate": 9.292349951126424e-05, + "loss": 3.1816, + "step": 3532 + }, + { + "epoch": 5.17, + "grad_norm": 1.9070907831192017, + "learning_rate": 9.287671443400268e-05, + "loss": 3.1082, + "step": 3533 + }, + { + "epoch": 5.17, + "grad_norm": 5.979832649230957, + "learning_rate": 9.282993092381625e-05, + "loss": 3.1332, + "step": 3534 + }, + { + "epoch": 5.17, + "grad_norm": 7.1100382804870605, + "learning_rate": 9.278314899099694e-05, + "loss": 3.2056, + "step": 3535 + }, + { + "epoch": 5.17, + "grad_norm": 22.313732147216797, + "learning_rate": 9.273636864583647e-05, + "loss": 3.1634, + "step": 3536 + }, + { + "epoch": 5.17, + "grad_norm": 11.737483978271484, + "learning_rate": 9.268958989862628e-05, + "loss": 3.1481, + "step": 3537 + }, + { + "epoch": 5.18, + "grad_norm": 2.638352394104004, + "learning_rate": 9.264281275965728e-05, + "loss": 3.1341, + "step": 3538 + }, + { + "epoch": 5.18, + "grad_norm": 2.807657241821289, + "learning_rate": 9.25960372392202e-05, + "loss": 3.1505, + "step": 3539 + }, + { + "epoch": 5.18, + "grad_norm": 20.156600952148438, + "learning_rate": 9.254926334760526e-05, + "loss": 3.1956, + "step": 3540 + }, + { + "epoch": 5.18, + "grad_norm": 2.164813280105591, + "learning_rate": 9.25024910951025e-05, + "loss": 3.1844, + "step": 3541 + }, + { + "epoch": 5.18, + "grad_norm": 1.7954508066177368, + "learning_rate": 9.245572049200142e-05, + "loss": 3.222, + "step": 3542 + }, + { + "epoch": 5.18, + "grad_norm": 4.100772380828857, + "learning_rate": 9.240895154859125e-05, + "loss": 3.1239, + "step": 3543 + }, + { + "epoch": 5.19, + "grad_norm": 3.8500773906707764, + "learning_rate": 9.236218427516092e-05, + "loss": 3.1278, + "step": 3544 + }, + { + "epoch": 5.19, + "grad_norm": 2.076373815536499, + "learning_rate": 9.231541868199883e-05, + "loss": 3.164, + "step": 3545 + }, + { + "epoch": 5.19, + "grad_norm": 1.8136000633239746, + "learning_rate": 9.226865477939313e-05, + "loss": 3.1479, + "step": 3546 + }, + { + "epoch": 5.19, + "grad_norm": 128.6366424560547, + "learning_rate": 9.222189257763158e-05, + "loss": 3.1239, + "step": 3547 + }, + { + "epoch": 5.19, + "grad_norm": 3.7925548553466797, + "learning_rate": 9.217513208700154e-05, + "loss": 3.1402, + "step": 3548 + }, + { + "epoch": 5.19, + "grad_norm": 11.528712272644043, + "learning_rate": 9.212837331778997e-05, + "loss": 3.1434, + "step": 3549 + }, + { + "epoch": 5.19, + "eval_loss": 2.9914677143096924, + "eval_runtime": 1173.7001, + "eval_samples_per_second": 14.605, + "eval_steps_per_second": 0.913, + "step": 3549 + }, + { + "epoch": 5.19, + "grad_norm": 3.5686726570129395, + "learning_rate": 9.208161628028355e-05, + "loss": 3.1025, + "step": 3550 + }, + { + "epoch": 5.2, + "grad_norm": 19.285165786743164, + "learning_rate": 9.203486098476849e-05, + "loss": 3.1957, + "step": 3551 + }, + { + "epoch": 5.2, + "grad_norm": 7.824908256530762, + "learning_rate": 9.19881074415306e-05, + "loss": 3.1752, + "step": 3552 + }, + { + "epoch": 5.2, + "grad_norm": 3.9910645484924316, + "learning_rate": 9.194135566085537e-05, + "loss": 3.1751, + "step": 3553 + }, + { + "epoch": 5.2, + "grad_norm": 32.02030944824219, + "learning_rate": 9.189460565302794e-05, + "loss": 3.1539, + "step": 3554 + }, + { + "epoch": 5.2, + "grad_norm": 10.703150749206543, + "learning_rate": 9.18478574283329e-05, + "loss": 3.1623, + "step": 3555 + }, + { + "epoch": 5.2, + "grad_norm": 37.447715759277344, + "learning_rate": 9.180111099705459e-05, + "loss": 3.1512, + "step": 3556 + }, + { + "epoch": 5.2, + "grad_norm": 4.359159469604492, + "learning_rate": 9.175436636947697e-05, + "loss": 3.1754, + "step": 3557 + }, + { + "epoch": 5.21, + "grad_norm": 11.65587329864502, + "learning_rate": 9.170762355588342e-05, + "loss": 3.1365, + "step": 3558 + }, + { + "epoch": 5.21, + "grad_norm": 6.701794147491455, + "learning_rate": 9.166088256655715e-05, + "loss": 3.1705, + "step": 3559 + }, + { + "epoch": 5.21, + "grad_norm": 12.4017915725708, + "learning_rate": 9.161414341178084e-05, + "loss": 3.1724, + "step": 3560 + }, + { + "epoch": 5.21, + "grad_norm": 2.782519578933716, + "learning_rate": 9.156740610183674e-05, + "loss": 3.1809, + "step": 3561 + }, + { + "epoch": 5.21, + "grad_norm": 3.470179319381714, + "learning_rate": 9.152067064700683e-05, + "loss": 3.1531, + "step": 3562 + }, + { + "epoch": 5.21, + "grad_norm": 12.016058921813965, + "learning_rate": 9.147393705757254e-05, + "loss": 3.1714, + "step": 3563 + }, + { + "epoch": 5.21, + "grad_norm": 8.318673133850098, + "learning_rate": 9.142720534381497e-05, + "loss": 3.178, + "step": 3564 + }, + { + "epoch": 5.22, + "grad_norm": 26.07087516784668, + "learning_rate": 9.138047551601473e-05, + "loss": 3.1636, + "step": 3565 + }, + { + "epoch": 5.22, + "grad_norm": 55.82182312011719, + "learning_rate": 9.133374758445212e-05, + "loss": 3.1598, + "step": 3566 + }, + { + "epoch": 5.22, + "grad_norm": 3.5638935565948486, + "learning_rate": 9.1287021559407e-05, + "loss": 3.1711, + "step": 3567 + }, + { + "epoch": 5.22, + "grad_norm": 4.856345176696777, + "learning_rate": 9.12402974511587e-05, + "loss": 3.1686, + "step": 3568 + }, + { + "epoch": 5.22, + "grad_norm": 6.552631855010986, + "learning_rate": 9.119357526998628e-05, + "loss": 3.1437, + "step": 3569 + }, + { + "epoch": 5.22, + "grad_norm": 62.54035186767578, + "learning_rate": 9.11468550261683e-05, + "loss": 3.1711, + "step": 3570 + }, + { + "epoch": 5.23, + "grad_norm": 19.124649047851562, + "learning_rate": 9.110013672998284e-05, + "loss": 3.2002, + "step": 3571 + }, + { + "epoch": 5.23, + "grad_norm": 14.90050983428955, + "learning_rate": 9.105342039170768e-05, + "loss": 3.2217, + "step": 3572 + }, + { + "epoch": 5.23, + "grad_norm": 4.7844109535217285, + "learning_rate": 9.100670602162009e-05, + "loss": 3.1506, + "step": 3573 + }, + { + "epoch": 5.23, + "grad_norm": 1.9373784065246582, + "learning_rate": 9.095999362999687e-05, + "loss": 3.1288, + "step": 3574 + }, + { + "epoch": 5.23, + "grad_norm": 60.557308197021484, + "learning_rate": 9.091328322711449e-05, + "loss": 3.1368, + "step": 3575 + }, + { + "epoch": 5.23, + "grad_norm": 6.0116658210754395, + "learning_rate": 9.086657482324892e-05, + "loss": 3.1685, + "step": 3576 + }, + { + "epoch": 5.23, + "grad_norm": 7.286896228790283, + "learning_rate": 9.081986842867562e-05, + "loss": 3.1756, + "step": 3577 + }, + { + "epoch": 5.24, + "grad_norm": 2.469630479812622, + "learning_rate": 9.077316405366981e-05, + "loss": 3.1425, + "step": 3578 + }, + { + "epoch": 5.24, + "grad_norm": 3.226400136947632, + "learning_rate": 9.072646170850605e-05, + "loss": 3.1015, + "step": 3579 + }, + { + "epoch": 5.24, + "grad_norm": 19.016260147094727, + "learning_rate": 9.067976140345854e-05, + "loss": 3.1746, + "step": 3580 + }, + { + "epoch": 5.24, + "grad_norm": 5.787333011627197, + "learning_rate": 9.06330631488011e-05, + "loss": 3.1975, + "step": 3581 + }, + { + "epoch": 5.24, + "grad_norm": 5.138857841491699, + "learning_rate": 9.058636695480698e-05, + "loss": 3.181, + "step": 3582 + }, + { + "epoch": 5.24, + "grad_norm": 11.81260871887207, + "learning_rate": 9.053967283174901e-05, + "loss": 3.1447, + "step": 3583 + }, + { + "epoch": 5.24, + "grad_norm": 6.516455173492432, + "learning_rate": 9.049298078989967e-05, + "loss": 3.1348, + "step": 3584 + }, + { + "epoch": 5.25, + "grad_norm": 6.460428237915039, + "learning_rate": 9.044629083953082e-05, + "loss": 3.1878, + "step": 3585 + }, + { + "epoch": 5.25, + "grad_norm": 5.189371585845947, + "learning_rate": 9.039960299091397e-05, + "loss": 3.1585, + "step": 3586 + }, + { + "epoch": 5.25, + "grad_norm": 4.265483379364014, + "learning_rate": 9.035291725432008e-05, + "loss": 3.1688, + "step": 3587 + }, + { + "epoch": 5.25, + "grad_norm": 2.625492572784424, + "learning_rate": 9.030623364001979e-05, + "loss": 3.1438, + "step": 3588 + }, + { + "epoch": 5.25, + "grad_norm": 3.3770155906677246, + "learning_rate": 9.025955215828312e-05, + "loss": 3.1574, + "step": 3589 + }, + { + "epoch": 5.25, + "grad_norm": 21.14669418334961, + "learning_rate": 9.021287281937967e-05, + "loss": 3.1473, + "step": 3590 + }, + { + "epoch": 5.25, + "grad_norm": 30.993927001953125, + "learning_rate": 9.016619563357865e-05, + "loss": 3.2186, + "step": 3591 + }, + { + "epoch": 5.26, + "grad_norm": 5.043474197387695, + "learning_rate": 9.011952061114867e-05, + "loss": 3.1602, + "step": 3592 + }, + { + "epoch": 5.26, + "grad_norm": 17.649402618408203, + "learning_rate": 9.007284776235791e-05, + "loss": 3.139, + "step": 3593 + }, + { + "epoch": 5.26, + "grad_norm": 50.776214599609375, + "learning_rate": 9.002617709747416e-05, + "loss": 3.1279, + "step": 3594 + }, + { + "epoch": 5.26, + "grad_norm": 12.535754203796387, + "learning_rate": 8.997950862676462e-05, + "loss": 3.174, + "step": 3595 + }, + { + "epoch": 5.26, + "grad_norm": 4.332036972045898, + "learning_rate": 8.993284236049598e-05, + "loss": 3.106, + "step": 3596 + }, + { + "epoch": 5.26, + "grad_norm": 1.4636400938034058, + "learning_rate": 8.988617830893455e-05, + "loss": 3.1456, + "step": 3597 + }, + { + "epoch": 5.27, + "grad_norm": 4.1321024894714355, + "learning_rate": 8.983951648234617e-05, + "loss": 3.102, + "step": 3598 + }, + { + "epoch": 5.27, + "grad_norm": 4.079101085662842, + "learning_rate": 8.979285689099601e-05, + "loss": 3.1632, + "step": 3599 + }, + { + "epoch": 5.27, + "grad_norm": 7.1509690284729, + "learning_rate": 8.974619954514895e-05, + "loss": 3.1862, + "step": 3600 + }, + { + "epoch": 5.27, + "grad_norm": 3.1132075786590576, + "learning_rate": 8.96995444550693e-05, + "loss": 3.1587, + "step": 3601 + }, + { + "epoch": 5.27, + "grad_norm": 15.199933052062988, + "learning_rate": 8.965289163102078e-05, + "loss": 3.1385, + "step": 3602 + }, + { + "epoch": 5.27, + "grad_norm": 3.226987600326538, + "learning_rate": 8.960624108326679e-05, + "loss": 3.1793, + "step": 3603 + }, + { + "epoch": 5.27, + "grad_norm": 3.0855495929718018, + "learning_rate": 8.955959282207013e-05, + "loss": 3.121, + "step": 3604 + }, + { + "epoch": 5.28, + "grad_norm": 1.8328437805175781, + "learning_rate": 8.951294685769303e-05, + "loss": 3.1515, + "step": 3605 + }, + { + "epoch": 5.28, + "grad_norm": 2.6738593578338623, + "learning_rate": 8.946630320039737e-05, + "loss": 3.1412, + "step": 3606 + }, + { + "epoch": 5.28, + "grad_norm": 2.790065288543701, + "learning_rate": 8.941966186044444e-05, + "loss": 3.1307, + "step": 3607 + }, + { + "epoch": 5.28, + "grad_norm": 3.6638455390930176, + "learning_rate": 8.937302284809494e-05, + "loss": 3.1155, + "step": 3608 + }, + { + "epoch": 5.28, + "grad_norm": 3.6409106254577637, + "learning_rate": 8.932638617360925e-05, + "loss": 3.1242, + "step": 3609 + }, + { + "epoch": 5.28, + "grad_norm": 2.8414742946624756, + "learning_rate": 8.927975184724704e-05, + "loss": 3.1433, + "step": 3610 + }, + { + "epoch": 5.28, + "grad_norm": 4.518361568450928, + "learning_rate": 8.92331198792676e-05, + "loss": 3.1883, + "step": 3611 + }, + { + "epoch": 5.29, + "grad_norm": 7.363691806793213, + "learning_rate": 8.918649027992958e-05, + "loss": 3.1231, + "step": 3612 + }, + { + "epoch": 5.29, + "grad_norm": 2.302523136138916, + "learning_rate": 8.913986305949125e-05, + "loss": 3.0972, + "step": 3613 + }, + { + "epoch": 5.29, + "grad_norm": 1.6068569421768188, + "learning_rate": 8.909323822821031e-05, + "loss": 3.1498, + "step": 3614 + }, + { + "epoch": 5.29, + "grad_norm": 1.9950542449951172, + "learning_rate": 8.90466157963438e-05, + "loss": 3.1339, + "step": 3615 + }, + { + "epoch": 5.29, + "grad_norm": 1.7338231801986694, + "learning_rate": 8.899999577414845e-05, + "loss": 3.1382, + "step": 3616 + }, + { + "epoch": 5.29, + "grad_norm": 2.231691598892212, + "learning_rate": 8.895337817188031e-05, + "loss": 3.1417, + "step": 3617 + }, + { + "epoch": 5.29, + "grad_norm": 1.9115151166915894, + "learning_rate": 8.890676299979492e-05, + "loss": 3.1338, + "step": 3618 + }, + { + "epoch": 5.3, + "grad_norm": 2.1142683029174805, + "learning_rate": 8.886015026814736e-05, + "loss": 3.1768, + "step": 3619 + }, + { + "epoch": 5.3, + "grad_norm": 2.7301149368286133, + "learning_rate": 8.881353998719206e-05, + "loss": 3.1843, + "step": 3620 + }, + { + "epoch": 5.3, + "grad_norm": 3.0939254760742188, + "learning_rate": 8.876693216718297e-05, + "loss": 3.1582, + "step": 3621 + }, + { + "epoch": 5.3, + "grad_norm": 1.2294076681137085, + "learning_rate": 8.872032681837358e-05, + "loss": 3.1484, + "step": 3622 + }, + { + "epoch": 5.3, + "grad_norm": 2.869823694229126, + "learning_rate": 8.867372395101666e-05, + "loss": 3.1455, + "step": 3623 + }, + { + "epoch": 5.3, + "grad_norm": 3.04744291305542, + "learning_rate": 8.862712357536454e-05, + "loss": 3.151, + "step": 3624 + }, + { + "epoch": 5.31, + "grad_norm": 1.6261351108551025, + "learning_rate": 8.858052570166908e-05, + "loss": 3.1434, + "step": 3625 + }, + { + "epoch": 5.31, + "grad_norm": 2.8542559146881104, + "learning_rate": 8.85339303401814e-05, + "loss": 3.1384, + "step": 3626 + }, + { + "epoch": 5.31, + "grad_norm": 0.9066136479377747, + "learning_rate": 8.848733750115218e-05, + "loss": 3.1576, + "step": 3627 + }, + { + "epoch": 5.31, + "grad_norm": 2.466120719909668, + "learning_rate": 8.84407471948316e-05, + "loss": 3.1478, + "step": 3628 + }, + { + "epoch": 5.31, + "grad_norm": 1.9956951141357422, + "learning_rate": 8.839415943146919e-05, + "loss": 3.1492, + "step": 3629 + }, + { + "epoch": 5.31, + "grad_norm": 10.860211372375488, + "learning_rate": 8.834757422131385e-05, + "loss": 3.198, + "step": 3630 + }, + { + "epoch": 5.31, + "grad_norm": 4.2302327156066895, + "learning_rate": 8.830099157461413e-05, + "loss": 3.146, + "step": 3631 + }, + { + "epoch": 5.32, + "grad_norm": 13.344452857971191, + "learning_rate": 8.825441150161788e-05, + "loss": 3.1234, + "step": 3632 + }, + { + "epoch": 5.32, + "grad_norm": 3.983452558517456, + "learning_rate": 8.820783401257236e-05, + "loss": 3.1775, + "step": 3633 + }, + { + "epoch": 5.32, + "grad_norm": 2.76181960105896, + "learning_rate": 8.816125911772433e-05, + "loss": 3.1776, + "step": 3634 + }, + { + "epoch": 5.32, + "grad_norm": 5.169229030609131, + "learning_rate": 8.811468682731997e-05, + "loss": 3.1174, + "step": 3635 + }, + { + "epoch": 5.32, + "grad_norm": 2.876774549484253, + "learning_rate": 8.806811715160485e-05, + "loss": 3.1506, + "step": 3636 + }, + { + "epoch": 5.32, + "grad_norm": 1.740065336227417, + "learning_rate": 8.802155010082396e-05, + "loss": 3.1375, + "step": 3637 + }, + { + "epoch": 5.32, + "grad_norm": 2.2916555404663086, + "learning_rate": 8.797498568522182e-05, + "loss": 3.1736, + "step": 3638 + }, + { + "epoch": 5.33, + "grad_norm": 13.75604248046875, + "learning_rate": 8.792842391504222e-05, + "loss": 3.1382, + "step": 3639 + }, + { + "epoch": 5.33, + "grad_norm": 8.454766273498535, + "learning_rate": 8.788186480052844e-05, + "loss": 3.1663, + "step": 3640 + }, + { + "epoch": 5.33, + "grad_norm": 3.627807140350342, + "learning_rate": 8.783530835192322e-05, + "loss": 3.1633, + "step": 3641 + }, + { + "epoch": 5.33, + "grad_norm": 1.9736120700836182, + "learning_rate": 8.778875457946866e-05, + "loss": 3.2027, + "step": 3642 + }, + { + "epoch": 5.33, + "grad_norm": 4.044384479522705, + "learning_rate": 8.77422034934062e-05, + "loss": 3.1484, + "step": 3643 + }, + { + "epoch": 5.33, + "grad_norm": 4.4502763748168945, + "learning_rate": 8.769565510397684e-05, + "loss": 3.1589, + "step": 3644 + }, + { + "epoch": 5.34, + "grad_norm": 2.187070369720459, + "learning_rate": 8.764910942142093e-05, + "loss": 3.1593, + "step": 3645 + }, + { + "epoch": 5.34, + "grad_norm": 2.6429078578948975, + "learning_rate": 8.760256645597813e-05, + "loss": 3.1466, + "step": 3646 + }, + { + "epoch": 5.34, + "grad_norm": 1.6300479173660278, + "learning_rate": 8.755602621788764e-05, + "loss": 3.1133, + "step": 3647 + }, + { + "epoch": 5.34, + "grad_norm": 2.9914662837982178, + "learning_rate": 8.750948871738802e-05, + "loss": 3.114, + "step": 3648 + }, + { + "epoch": 5.34, + "grad_norm": 2.315000534057617, + "learning_rate": 8.746295396471711e-05, + "loss": 3.1427, + "step": 3649 + }, + { + "epoch": 5.34, + "grad_norm": 3.0107791423797607, + "learning_rate": 8.741642197011235e-05, + "loss": 3.1085, + "step": 3650 + }, + { + "epoch": 5.34, + "grad_norm": 11.682272911071777, + "learning_rate": 8.736989274381045e-05, + "loss": 3.1428, + "step": 3651 + }, + { + "epoch": 5.35, + "grad_norm": 2.8496906757354736, + "learning_rate": 8.732336629604745e-05, + "loss": 3.1269, + "step": 3652 + }, + { + "epoch": 5.35, + "grad_norm": 2.9771509170532227, + "learning_rate": 8.727684263705895e-05, + "loss": 3.147, + "step": 3653 + }, + { + "epoch": 5.35, + "grad_norm": 2.937957525253296, + "learning_rate": 8.723032177707978e-05, + "loss": 3.1731, + "step": 3654 + }, + { + "epoch": 5.35, + "grad_norm": 3.5484066009521484, + "learning_rate": 8.718380372634422e-05, + "loss": 3.1602, + "step": 3655 + }, + { + "epoch": 5.35, + "grad_norm": 9.137898445129395, + "learning_rate": 8.7137288495086e-05, + "loss": 3.1692, + "step": 3656 + }, + { + "epoch": 5.35, + "grad_norm": 10.832776069641113, + "learning_rate": 8.709077609353809e-05, + "loss": 3.1829, + "step": 3657 + }, + { + "epoch": 5.35, + "grad_norm": 3.5576894283294678, + "learning_rate": 8.704426653193289e-05, + "loss": 3.1439, + "step": 3658 + }, + { + "epoch": 5.36, + "grad_norm": 2.771862745285034, + "learning_rate": 8.699775982050229e-05, + "loss": 3.1435, + "step": 3659 + }, + { + "epoch": 5.36, + "grad_norm": 7.5768046379089355, + "learning_rate": 8.695125596947736e-05, + "loss": 3.1898, + "step": 3660 + }, + { + "epoch": 5.36, + "grad_norm": 3.332085132598877, + "learning_rate": 8.690475498908868e-05, + "loss": 3.1365, + "step": 3661 + }, + { + "epoch": 5.36, + "grad_norm": 8.380770683288574, + "learning_rate": 8.685825688956613e-05, + "loss": 3.1733, + "step": 3662 + }, + { + "epoch": 5.36, + "grad_norm": 9.0435791015625, + "learning_rate": 8.6811761681139e-05, + "loss": 3.1752, + "step": 3663 + }, + { + "epoch": 5.36, + "grad_norm": 12.012619018554688, + "learning_rate": 8.676526937403591e-05, + "loss": 3.1615, + "step": 3664 + }, + { + "epoch": 5.36, + "grad_norm": 6.051881313323975, + "learning_rate": 8.671877997848484e-05, + "loss": 3.1719, + "step": 3665 + }, + { + "epoch": 5.37, + "grad_norm": 5.059371471405029, + "learning_rate": 8.667229350471322e-05, + "loss": 3.1775, + "step": 3666 + }, + { + "epoch": 5.37, + "grad_norm": 9.862431526184082, + "learning_rate": 8.662580996294767e-05, + "loss": 3.1795, + "step": 3667 + }, + { + "epoch": 5.37, + "grad_norm": 15.149982452392578, + "learning_rate": 8.657932936341428e-05, + "loss": 3.2141, + "step": 3668 + }, + { + "epoch": 5.37, + "grad_norm": 9.565960884094238, + "learning_rate": 8.653285171633855e-05, + "loss": 3.1831, + "step": 3669 + }, + { + "epoch": 5.37, + "grad_norm": 19.23344612121582, + "learning_rate": 8.648637703194516e-05, + "loss": 3.1925, + "step": 3670 + }, + { + "epoch": 5.37, + "grad_norm": 6.268587589263916, + "learning_rate": 8.643990532045822e-05, + "loss": 3.1621, + "step": 3671 + }, + { + "epoch": 5.38, + "grad_norm": 56.983970642089844, + "learning_rate": 8.639343659210131e-05, + "loss": 3.1715, + "step": 3672 + }, + { + "epoch": 5.38, + "grad_norm": 50.29884719848633, + "learning_rate": 8.634697085709711e-05, + "loss": 3.1474, + "step": 3673 + }, + { + "epoch": 5.38, + "grad_norm": 13.747417449951172, + "learning_rate": 8.630050812566781e-05, + "loss": 3.1805, + "step": 3674 + }, + { + "epoch": 5.38, + "grad_norm": 20.72515106201172, + "learning_rate": 8.625404840803498e-05, + "loss": 3.1528, + "step": 3675 + }, + { + "epoch": 5.38, + "grad_norm": 12.049095153808594, + "learning_rate": 8.620759171441935e-05, + "loss": 3.181, + "step": 3676 + }, + { + "epoch": 5.38, + "grad_norm": 11.251856803894043, + "learning_rate": 8.616113805504106e-05, + "loss": 3.1962, + "step": 3677 + }, + { + "epoch": 5.38, + "grad_norm": 6.218480587005615, + "learning_rate": 8.61146874401197e-05, + "loss": 3.2132, + "step": 3678 + }, + { + "epoch": 5.39, + "grad_norm": 10.480876922607422, + "learning_rate": 8.606823987987405e-05, + "loss": 3.1913, + "step": 3679 + }, + { + "epoch": 5.39, + "grad_norm": 12.099661827087402, + "learning_rate": 8.602179538452222e-05, + "loss": 3.178, + "step": 3680 + }, + { + "epoch": 5.39, + "grad_norm": 23.720500946044922, + "learning_rate": 8.597535396428177e-05, + "loss": 3.1558, + "step": 3681 + }, + { + "epoch": 5.39, + "grad_norm": 16.246944427490234, + "learning_rate": 8.592891562936947e-05, + "loss": 3.1834, + "step": 3682 + }, + { + "epoch": 5.39, + "grad_norm": 37.614784240722656, + "learning_rate": 8.58824803900014e-05, + "loss": 3.174, + "step": 3683 + }, + { + "epoch": 5.39, + "grad_norm": 9.572596549987793, + "learning_rate": 8.583604825639302e-05, + "loss": 3.1871, + "step": 3684 + }, + { + "epoch": 5.39, + "grad_norm": 6.099634170532227, + "learning_rate": 8.578961923875915e-05, + "loss": 3.1663, + "step": 3685 + }, + { + "epoch": 5.4, + "grad_norm": 18.694616317749023, + "learning_rate": 8.57431933473138e-05, + "loss": 3.1668, + "step": 3686 + }, + { + "epoch": 5.4, + "grad_norm": 133.21104431152344, + "learning_rate": 8.569677059227032e-05, + "loss": 3.1881, + "step": 3687 + }, + { + "epoch": 5.4, + "grad_norm": 33.713504791259766, + "learning_rate": 8.565035098384149e-05, + "loss": 3.1648, + "step": 3688 + }, + { + "epoch": 5.4, + "grad_norm": 9.273075103759766, + "learning_rate": 8.56039345322393e-05, + "loss": 3.2306, + "step": 3689 + }, + { + "epoch": 5.4, + "grad_norm": 18.504985809326172, + "learning_rate": 8.555752124767498e-05, + "loss": 3.2082, + "step": 3690 + }, + { + "epoch": 5.4, + "grad_norm": 13.903569221496582, + "learning_rate": 8.551111114035922e-05, + "loss": 3.1519, + "step": 3691 + }, + { + "epoch": 5.4, + "grad_norm": 48.394412994384766, + "learning_rate": 8.546470422050193e-05, + "loss": 3.1922, + "step": 3692 + }, + { + "epoch": 5.41, + "grad_norm": 9.13818073272705, + "learning_rate": 8.541830049831227e-05, + "loss": 3.2432, + "step": 3693 + }, + { + "epoch": 5.41, + "grad_norm": 7.568233013153076, + "learning_rate": 8.537189998399879e-05, + "loss": 3.2007, + "step": 3694 + }, + { + "epoch": 5.41, + "grad_norm": 110.37165832519531, + "learning_rate": 8.53255026877693e-05, + "loss": 3.1837, + "step": 3695 + }, + { + "epoch": 5.41, + "grad_norm": 8.272992134094238, + "learning_rate": 8.527910861983084e-05, + "loss": 3.169, + "step": 3696 + }, + { + "epoch": 5.41, + "grad_norm": 29.426700592041016, + "learning_rate": 8.523271779038986e-05, + "loss": 3.1793, + "step": 3697 + }, + { + "epoch": 5.41, + "grad_norm": 9.58652400970459, + "learning_rate": 8.518633020965199e-05, + "loss": 3.1662, + "step": 3698 + }, + { + "epoch": 5.42, + "grad_norm": 5.326894283294678, + "learning_rate": 8.513994588782217e-05, + "loss": 3.1817, + "step": 3699 + }, + { + "epoch": 5.42, + "grad_norm": 8.72939395904541, + "learning_rate": 8.509356483510471e-05, + "loss": 3.1888, + "step": 3700 + }, + { + "epoch": 5.42, + "grad_norm": 9.870657920837402, + "learning_rate": 8.504718706170306e-05, + "loss": 3.1772, + "step": 3701 + }, + { + "epoch": 5.42, + "grad_norm": 39.32550811767578, + "learning_rate": 8.500081257782003e-05, + "loss": 3.203, + "step": 3702 + }, + { + "epoch": 5.42, + "grad_norm": 20.613840103149414, + "learning_rate": 8.495444139365776e-05, + "loss": 3.1721, + "step": 3703 + }, + { + "epoch": 5.42, + "grad_norm": 10.104787826538086, + "learning_rate": 8.490807351941753e-05, + "loss": 3.1791, + "step": 3704 + }, + { + "epoch": 5.42, + "grad_norm": 8.861245155334473, + "learning_rate": 8.486170896529996e-05, + "loss": 3.2222, + "step": 3705 + }, + { + "epoch": 5.43, + "grad_norm": 40.4407844543457, + "learning_rate": 8.481534774150502e-05, + "loss": 3.3014, + "step": 3706 + }, + { + "epoch": 5.43, + "grad_norm": 4.760523319244385, + "learning_rate": 8.476898985823177e-05, + "loss": 3.1825, + "step": 3707 + }, + { + "epoch": 5.43, + "grad_norm": 1924.5662841796875, + "learning_rate": 8.472263532567871e-05, + "loss": 3.1802, + "step": 3708 + }, + { + "epoch": 5.43, + "grad_norm": 5.515054702758789, + "learning_rate": 8.467628415404345e-05, + "loss": 3.1833, + "step": 3709 + }, + { + "epoch": 5.43, + "grad_norm": 2.0650458335876465, + "learning_rate": 8.462993635352303e-05, + "loss": 3.184, + "step": 3710 + }, + { + "epoch": 5.43, + "grad_norm": 4.477024078369141, + "learning_rate": 8.458359193431358e-05, + "loss": 3.1489, + "step": 3711 + }, + { + "epoch": 5.43, + "grad_norm": 8.2962646484375, + "learning_rate": 8.453725090661057e-05, + "loss": 3.1318, + "step": 3712 + }, + { + "epoch": 5.44, + "grad_norm": 4.55861234664917, + "learning_rate": 8.449091328060877e-05, + "loss": 3.185, + "step": 3713 + }, + { + "epoch": 5.44, + "grad_norm": 6.914697647094727, + "learning_rate": 8.44445790665021e-05, + "loss": 3.1659, + "step": 3714 + }, + { + "epoch": 5.44, + "grad_norm": 5.971372127532959, + "learning_rate": 8.439824827448376e-05, + "loss": 3.1691, + "step": 3715 + }, + { + "epoch": 5.44, + "grad_norm": 3.358804702758789, + "learning_rate": 8.435192091474628e-05, + "loss": 3.187, + "step": 3716 + }, + { + "epoch": 5.44, + "grad_norm": 50.52218246459961, + "learning_rate": 8.430559699748132e-05, + "loss": 3.1595, + "step": 3717 + }, + { + "epoch": 5.44, + "grad_norm": 4.3211774826049805, + "learning_rate": 8.425927653287981e-05, + "loss": 3.2301, + "step": 3718 + }, + { + "epoch": 5.44, + "eval_loss": 3.003319025039673, + "eval_runtime": 1168.0997, + "eval_samples_per_second": 14.675, + "eval_steps_per_second": 0.918, + "step": 3718 + }, + { + "epoch": 5.44, + "grad_norm": 3.5544116497039795, + "learning_rate": 8.421295953113201e-05, + "loss": 3.1628, + "step": 3719 + }, + { + "epoch": 5.45, + "grad_norm": 4.997739791870117, + "learning_rate": 8.416664600242733e-05, + "loss": 3.1741, + "step": 3720 + }, + { + "epoch": 5.45, + "grad_norm": 2.1352174282073975, + "learning_rate": 8.412033595695437e-05, + "loss": 3.1518, + "step": 3721 + }, + { + "epoch": 5.45, + "grad_norm": 7.084630489349365, + "learning_rate": 8.40740294049011e-05, + "loss": 3.1283, + "step": 3722 + }, + { + "epoch": 5.45, + "grad_norm": 2.8684113025665283, + "learning_rate": 8.402772635645465e-05, + "loss": 3.1823, + "step": 3723 + }, + { + "epoch": 5.45, + "grad_norm": 14.201617240905762, + "learning_rate": 8.398142682180131e-05, + "loss": 3.14, + "step": 3724 + }, + { + "epoch": 5.45, + "grad_norm": 4.209918975830078, + "learning_rate": 8.393513081112674e-05, + "loss": 3.1867, + "step": 3725 + }, + { + "epoch": 5.46, + "grad_norm": 2.907388687133789, + "learning_rate": 8.388883833461578e-05, + "loss": 3.2095, + "step": 3726 + }, + { + "epoch": 5.46, + "grad_norm": 11.771166801452637, + "learning_rate": 8.384254940245233e-05, + "loss": 3.1938, + "step": 3727 + }, + { + "epoch": 5.46, + "grad_norm": 15.321097373962402, + "learning_rate": 8.379626402481977e-05, + "loss": 3.1755, + "step": 3728 + }, + { + "epoch": 5.46, + "grad_norm": 3.204758882522583, + "learning_rate": 8.374998221190056e-05, + "loss": 3.1617, + "step": 3729 + }, + { + "epoch": 5.46, + "grad_norm": 7.3266754150390625, + "learning_rate": 8.37037039738763e-05, + "loss": 3.1647, + "step": 3730 + }, + { + "epoch": 5.46, + "grad_norm": 108.37910461425781, + "learning_rate": 8.365742932092801e-05, + "loss": 3.1807, + "step": 3731 + }, + { + "epoch": 5.46, + "grad_norm": 4.3307623863220215, + "learning_rate": 8.361115826323572e-05, + "loss": 3.2109, + "step": 3732 + }, + { + "epoch": 5.47, + "grad_norm": 1.9860690832138062, + "learning_rate": 8.35648908109788e-05, + "loss": 3.1484, + "step": 3733 + }, + { + "epoch": 5.47, + "grad_norm": 3.039022207260132, + "learning_rate": 8.351862697433573e-05, + "loss": 3.175, + "step": 3734 + }, + { + "epoch": 5.47, + "grad_norm": 7.283559799194336, + "learning_rate": 8.347236676348426e-05, + "loss": 3.219, + "step": 3735 + }, + { + "epoch": 5.47, + "grad_norm": 3.774641513824463, + "learning_rate": 8.342611018860139e-05, + "loss": 3.1932, + "step": 3736 + }, + { + "epoch": 5.47, + "grad_norm": 4.264026641845703, + "learning_rate": 8.337985725986314e-05, + "loss": 3.193, + "step": 3737 + }, + { + "epoch": 5.47, + "grad_norm": 1.858871340751648, + "learning_rate": 8.333360798744496e-05, + "loss": 3.122, + "step": 3738 + }, + { + "epoch": 5.47, + "grad_norm": 4.920361518859863, + "learning_rate": 8.328736238152133e-05, + "loss": 3.1425, + "step": 3739 + }, + { + "epoch": 5.48, + "grad_norm": 4.10701847076416, + "learning_rate": 8.324112045226593e-05, + "loss": 3.1595, + "step": 3740 + }, + { + "epoch": 5.48, + "grad_norm": 5.043720245361328, + "learning_rate": 8.319488220985174e-05, + "loss": 3.1407, + "step": 3741 + }, + { + "epoch": 5.48, + "grad_norm": 3.795624256134033, + "learning_rate": 8.314864766445088e-05, + "loss": 3.2385, + "step": 3742 + }, + { + "epoch": 5.48, + "grad_norm": 3.801236867904663, + "learning_rate": 8.310241682623454e-05, + "loss": 3.152, + "step": 3743 + }, + { + "epoch": 5.48, + "grad_norm": 2.491832971572876, + "learning_rate": 8.30561897053733e-05, + "loss": 3.1737, + "step": 3744 + }, + { + "epoch": 5.48, + "grad_norm": 2.1997487545013428, + "learning_rate": 8.300996631203675e-05, + "loss": 3.1273, + "step": 3745 + }, + { + "epoch": 5.49, + "grad_norm": 2.1604340076446533, + "learning_rate": 8.296374665639373e-05, + "loss": 3.2015, + "step": 3746 + }, + { + "epoch": 5.49, + "grad_norm": 14.758793830871582, + "learning_rate": 8.291753074861233e-05, + "loss": 3.1594, + "step": 3747 + }, + { + "epoch": 5.49, + "grad_norm": 1.7705718278884888, + "learning_rate": 8.287131859885964e-05, + "loss": 3.1615, + "step": 3748 + }, + { + "epoch": 5.49, + "grad_norm": 3.673743724822998, + "learning_rate": 8.282511021730206e-05, + "loss": 3.1665, + "step": 3749 + }, + { + "epoch": 5.49, + "grad_norm": 2.580552101135254, + "learning_rate": 8.277890561410518e-05, + "loss": 3.1972, + "step": 3750 + }, + { + "epoch": 5.49, + "grad_norm": 1.8429394960403442, + "learning_rate": 8.273270479943361e-05, + "loss": 3.1316, + "step": 3751 + }, + { + "epoch": 5.49, + "grad_norm": 1.4963501691818237, + "learning_rate": 8.268650778345126e-05, + "loss": 3.1498, + "step": 3752 + }, + { + "epoch": 5.5, + "grad_norm": 3.3234293460845947, + "learning_rate": 8.26403145763212e-05, + "loss": 3.1475, + "step": 3753 + }, + { + "epoch": 5.5, + "grad_norm": 3.2355213165283203, + "learning_rate": 8.25941251882056e-05, + "loss": 3.1404, + "step": 3754 + }, + { + "epoch": 5.5, + "grad_norm": 1.3712440729141235, + "learning_rate": 8.254793962926579e-05, + "loss": 3.1316, + "step": 3755 + }, + { + "epoch": 5.5, + "grad_norm": 1.9082093238830566, + "learning_rate": 8.250175790966228e-05, + "loss": 3.1388, + "step": 3756 + }, + { + "epoch": 5.5, + "grad_norm": 2.096512794494629, + "learning_rate": 8.245558003955481e-05, + "loss": 3.1753, + "step": 3757 + }, + { + "epoch": 5.5, + "grad_norm": 1.1812965869903564, + "learning_rate": 8.240940602910212e-05, + "loss": 3.1647, + "step": 3758 + }, + { + "epoch": 5.5, + "grad_norm": 2.799182415008545, + "learning_rate": 8.23632358884622e-05, + "loss": 3.1474, + "step": 3759 + }, + { + "epoch": 5.51, + "grad_norm": 2.4367105960845947, + "learning_rate": 8.231706962779222e-05, + "loss": 3.1312, + "step": 3760 + }, + { + "epoch": 5.51, + "grad_norm": 2.1872997283935547, + "learning_rate": 8.227090725724842e-05, + "loss": 3.0935, + "step": 3761 + }, + { + "epoch": 5.51, + "grad_norm": 3.3611397743225098, + "learning_rate": 8.222474878698617e-05, + "loss": 3.0987, + "step": 3762 + }, + { + "epoch": 5.51, + "grad_norm": 1.0651507377624512, + "learning_rate": 8.217859422716012e-05, + "loss": 3.183, + "step": 3763 + }, + { + "epoch": 5.51, + "grad_norm": 0.8456958532333374, + "learning_rate": 8.213244358792389e-05, + "loss": 3.123, + "step": 3764 + }, + { + "epoch": 5.51, + "grad_norm": 0.9663138389587402, + "learning_rate": 8.208629687943029e-05, + "loss": 3.16, + "step": 3765 + }, + { + "epoch": 5.51, + "grad_norm": 135.12335205078125, + "learning_rate": 8.204015411183135e-05, + "loss": 3.1066, + "step": 3766 + }, + { + "epoch": 5.52, + "grad_norm": 1.7855501174926758, + "learning_rate": 8.199401529527816e-05, + "loss": 3.0972, + "step": 3767 + }, + { + "epoch": 5.52, + "grad_norm": 1.187641978263855, + "learning_rate": 8.194788043992089e-05, + "loss": 3.1476, + "step": 3768 + }, + { + "epoch": 5.52, + "grad_norm": 6.5035834312438965, + "learning_rate": 8.190174955590897e-05, + "loss": 3.134, + "step": 3769 + }, + { + "epoch": 5.52, + "grad_norm": 3.347416400909424, + "learning_rate": 8.185562265339088e-05, + "loss": 3.1592, + "step": 3770 + }, + { + "epoch": 5.52, + "grad_norm": 2.2179319858551025, + "learning_rate": 8.180949974251416e-05, + "loss": 3.1656, + "step": 3771 + }, + { + "epoch": 5.52, + "grad_norm": 3.2867071628570557, + "learning_rate": 8.176338083342561e-05, + "loss": 3.1231, + "step": 3772 + }, + { + "epoch": 5.53, + "grad_norm": 4.15543794631958, + "learning_rate": 8.17172659362711e-05, + "loss": 3.1501, + "step": 3773 + }, + { + "epoch": 5.53, + "grad_norm": 3.2263450622558594, + "learning_rate": 8.167115506119549e-05, + "loss": 3.1269, + "step": 3774 + }, + { + "epoch": 5.53, + "grad_norm": 1.2961052656173706, + "learning_rate": 8.162504821834295e-05, + "loss": 3.1204, + "step": 3775 + }, + { + "epoch": 5.53, + "grad_norm": 2.9037585258483887, + "learning_rate": 8.157894541785671e-05, + "loss": 3.1801, + "step": 3776 + }, + { + "epoch": 5.53, + "grad_norm": 2.3933663368225098, + "learning_rate": 8.153284666987897e-05, + "loss": 3.1537, + "step": 3777 + }, + { + "epoch": 5.53, + "grad_norm": 1.6720260381698608, + "learning_rate": 8.148675198455123e-05, + "loss": 3.1501, + "step": 3778 + }, + { + "epoch": 5.53, + "grad_norm": 11.296658515930176, + "learning_rate": 8.144066137201398e-05, + "loss": 3.1364, + "step": 3779 + }, + { + "epoch": 5.54, + "grad_norm": 14.168936729431152, + "learning_rate": 8.139457484240687e-05, + "loss": 3.1629, + "step": 3780 + }, + { + "epoch": 5.54, + "grad_norm": 5.073040008544922, + "learning_rate": 8.134849240586854e-05, + "loss": 3.1765, + "step": 3781 + }, + { + "epoch": 5.54, + "grad_norm": 2.859703779220581, + "learning_rate": 8.130241407253693e-05, + "loss": 3.1252, + "step": 3782 + }, + { + "epoch": 5.54, + "grad_norm": 6.422140121459961, + "learning_rate": 8.125633985254893e-05, + "loss": 3.126, + "step": 3783 + }, + { + "epoch": 5.54, + "grad_norm": 16.3258056640625, + "learning_rate": 8.121026975604052e-05, + "loss": 3.1953, + "step": 3784 + }, + { + "epoch": 5.54, + "grad_norm": 7.476664066314697, + "learning_rate": 8.116420379314684e-05, + "loss": 3.1981, + "step": 3785 + }, + { + "epoch": 5.54, + "grad_norm": 57.822113037109375, + "learning_rate": 8.111814197400213e-05, + "loss": 3.1517, + "step": 3786 + }, + { + "epoch": 5.55, + "grad_norm": 6.158851146697998, + "learning_rate": 8.10720843087396e-05, + "loss": 3.1655, + "step": 3787 + }, + { + "epoch": 5.55, + "grad_norm": 7.291560173034668, + "learning_rate": 8.102603080749171e-05, + "loss": 3.113, + "step": 3788 + }, + { + "epoch": 5.55, + "grad_norm": 6.275900840759277, + "learning_rate": 8.097998148038985e-05, + "loss": 3.1209, + "step": 3789 + }, + { + "epoch": 5.55, + "grad_norm": 4.1805338859558105, + "learning_rate": 8.093393633756459e-05, + "loss": 3.1049, + "step": 3790 + }, + { + "epoch": 5.55, + "grad_norm": 6.27461051940918, + "learning_rate": 8.088789538914561e-05, + "loss": 3.169, + "step": 3791 + }, + { + "epoch": 5.55, + "grad_norm": 63.376373291015625, + "learning_rate": 8.084185864526153e-05, + "loss": 3.2346, + "step": 3792 + }, + { + "epoch": 5.55, + "grad_norm": 17.007471084594727, + "learning_rate": 8.079582611604015e-05, + "loss": 3.1799, + "step": 3793 + }, + { + "epoch": 5.56, + "grad_norm": 3.8963406085968018, + "learning_rate": 8.074979781160836e-05, + "loss": 3.1546, + "step": 3794 + }, + { + "epoch": 5.56, + "grad_norm": 1.5541036128997803, + "learning_rate": 8.070377374209202e-05, + "loss": 3.17, + "step": 3795 + }, + { + "epoch": 5.56, + "grad_norm": 4.511335372924805, + "learning_rate": 8.06577539176161e-05, + "loss": 3.1307, + "step": 3796 + }, + { + "epoch": 5.56, + "grad_norm": 3.2192275524139404, + "learning_rate": 8.061173834830476e-05, + "loss": 3.1262, + "step": 3797 + }, + { + "epoch": 5.56, + "grad_norm": 10.192057609558105, + "learning_rate": 8.056572704428103e-05, + "loss": 3.1341, + "step": 3798 + }, + { + "epoch": 5.56, + "grad_norm": 10.097148895263672, + "learning_rate": 8.051972001566708e-05, + "loss": 3.1627, + "step": 3799 + }, + { + "epoch": 5.57, + "grad_norm": 5.317000865936279, + "learning_rate": 8.04737172725842e-05, + "loss": 3.1695, + "step": 3800 + }, + { + "epoch": 5.57, + "grad_norm": 3.3502695560455322, + "learning_rate": 8.042771882515265e-05, + "loss": 3.1111, + "step": 3801 + }, + { + "epoch": 5.57, + "grad_norm": 3.3459572792053223, + "learning_rate": 8.038172468349178e-05, + "loss": 3.1737, + "step": 3802 + }, + { + "epoch": 5.57, + "grad_norm": 4.4290900230407715, + "learning_rate": 8.033573485772001e-05, + "loss": 3.1335, + "step": 3803 + }, + { + "epoch": 5.57, + "grad_norm": 3.444740056991577, + "learning_rate": 8.028974935795481e-05, + "loss": 3.1575, + "step": 3804 + }, + { + "epoch": 5.57, + "grad_norm": 6.389871120452881, + "learning_rate": 8.024376819431266e-05, + "loss": 3.1654, + "step": 3805 + }, + { + "epoch": 5.57, + "grad_norm": 4.629978179931641, + "learning_rate": 8.019779137690906e-05, + "loss": 3.1554, + "step": 3806 + }, + { + "epoch": 5.58, + "grad_norm": 2.2746245861053467, + "learning_rate": 8.015181891585871e-05, + "loss": 3.1065, + "step": 3807 + }, + { + "epoch": 5.58, + "grad_norm": 4.390033721923828, + "learning_rate": 8.010585082127513e-05, + "loss": 3.142, + "step": 3808 + }, + { + "epoch": 5.58, + "grad_norm": 8.950037956237793, + "learning_rate": 8.005988710327105e-05, + "loss": 3.1759, + "step": 3809 + }, + { + "epoch": 5.58, + "grad_norm": 7.786754608154297, + "learning_rate": 8.001392777195821e-05, + "loss": 3.1518, + "step": 3810 + }, + { + "epoch": 5.58, + "grad_norm": 1.803338646888733, + "learning_rate": 7.996797283744732e-05, + "loss": 3.1452, + "step": 3811 + }, + { + "epoch": 5.58, + "grad_norm": 5.436160087585449, + "learning_rate": 7.992202230984811e-05, + "loss": 3.1284, + "step": 3812 + }, + { + "epoch": 5.58, + "grad_norm": 5.9366912841796875, + "learning_rate": 7.987607619926946e-05, + "loss": 3.1547, + "step": 3813 + }, + { + "epoch": 5.59, + "grad_norm": 5.989798545837402, + "learning_rate": 7.98301345158192e-05, + "loss": 3.1264, + "step": 3814 + }, + { + "epoch": 5.59, + "grad_norm": 2.005276679992676, + "learning_rate": 7.978419726960413e-05, + "loss": 3.1387, + "step": 3815 + }, + { + "epoch": 5.59, + "grad_norm": 4.961897373199463, + "learning_rate": 7.97382644707302e-05, + "loss": 3.1506, + "step": 3816 + }, + { + "epoch": 5.59, + "grad_norm": 5.611112594604492, + "learning_rate": 7.969233612930234e-05, + "loss": 3.1237, + "step": 3817 + }, + { + "epoch": 5.59, + "grad_norm": 2.2314023971557617, + "learning_rate": 7.964641225542438e-05, + "loss": 3.1433, + "step": 3818 + }, + { + "epoch": 5.59, + "grad_norm": 27.2403564453125, + "learning_rate": 7.960049285919933e-05, + "loss": 3.1587, + "step": 3819 + }, + { + "epoch": 5.59, + "grad_norm": 7.609023571014404, + "learning_rate": 7.955457795072918e-05, + "loss": 3.2031, + "step": 3820 + }, + { + "epoch": 5.6, + "grad_norm": 6.942310810089111, + "learning_rate": 7.95086675401148e-05, + "loss": 3.1571, + "step": 3821 + }, + { + "epoch": 5.6, + "grad_norm": 5.55133056640625, + "learning_rate": 7.94627616374563e-05, + "loss": 3.1465, + "step": 3822 + }, + { + "epoch": 5.6, + "grad_norm": 2.6575374603271484, + "learning_rate": 7.941686025285257e-05, + "loss": 3.1384, + "step": 3823 + }, + { + "epoch": 5.6, + "grad_norm": 4.183603763580322, + "learning_rate": 7.937096339640162e-05, + "loss": 3.1551, + "step": 3824 + }, + { + "epoch": 5.6, + "grad_norm": 203.84231567382812, + "learning_rate": 7.932507107820051e-05, + "loss": 3.1756, + "step": 3825 + }, + { + "epoch": 5.6, + "grad_norm": 1.8645808696746826, + "learning_rate": 7.92791833083452e-05, + "loss": 3.1254, + "step": 3826 + }, + { + "epoch": 5.61, + "grad_norm": 3.2966809272766113, + "learning_rate": 7.923330009693071e-05, + "loss": 3.1409, + "step": 3827 + }, + { + "epoch": 5.61, + "grad_norm": 16.701486587524414, + "learning_rate": 7.918742145405098e-05, + "loss": 3.1687, + "step": 3828 + }, + { + "epoch": 5.61, + "grad_norm": 5.320707321166992, + "learning_rate": 7.914154738979908e-05, + "loss": 3.166, + "step": 3829 + }, + { + "epoch": 5.61, + "grad_norm": 3.770569086074829, + "learning_rate": 7.909567791426697e-05, + "loss": 3.129, + "step": 3830 + }, + { + "epoch": 5.61, + "grad_norm": 1.9080885648727417, + "learning_rate": 7.904981303754561e-05, + "loss": 3.1792, + "step": 3831 + }, + { + "epoch": 5.61, + "grad_norm": 2.2089335918426514, + "learning_rate": 7.900395276972502e-05, + "loss": 3.1417, + "step": 3832 + }, + { + "epoch": 5.61, + "grad_norm": 2.3139476776123047, + "learning_rate": 7.895809712089406e-05, + "loss": 3.1278, + "step": 3833 + }, + { + "epoch": 5.62, + "grad_norm": 3.4870378971099854, + "learning_rate": 7.891224610114072e-05, + "loss": 3.1562, + "step": 3834 + }, + { + "epoch": 5.62, + "grad_norm": 4.3670244216918945, + "learning_rate": 7.886639972055195e-05, + "loss": 3.1389, + "step": 3835 + }, + { + "epoch": 5.62, + "grad_norm": 5.9168596267700195, + "learning_rate": 7.882055798921358e-05, + "loss": 3.1104, + "step": 3836 + }, + { + "epoch": 5.62, + "grad_norm": 7.184929370880127, + "learning_rate": 7.87747209172105e-05, + "loss": 3.1555, + "step": 3837 + }, + { + "epoch": 5.62, + "grad_norm": 4.011050701141357, + "learning_rate": 7.872888851462663e-05, + "loss": 3.1192, + "step": 3838 + }, + { + "epoch": 5.62, + "grad_norm": 1.6208895444869995, + "learning_rate": 7.86830607915447e-05, + "loss": 3.1861, + "step": 3839 + }, + { + "epoch": 5.62, + "grad_norm": 1.3240761756896973, + "learning_rate": 7.863723775804651e-05, + "loss": 3.1656, + "step": 3840 + }, + { + "epoch": 5.63, + "grad_norm": 2.556436538696289, + "learning_rate": 7.85914194242129e-05, + "loss": 3.1392, + "step": 3841 + }, + { + "epoch": 5.63, + "grad_norm": 1.8541210889816284, + "learning_rate": 7.854560580012351e-05, + "loss": 3.1391, + "step": 3842 + }, + { + "epoch": 5.63, + "grad_norm": 2.840593099594116, + "learning_rate": 7.849979689585705e-05, + "loss": 3.0979, + "step": 3843 + }, + { + "epoch": 5.63, + "grad_norm": 1.50877845287323, + "learning_rate": 7.845399272149124e-05, + "loss": 3.1231, + "step": 3844 + }, + { + "epoch": 5.63, + "grad_norm": 2.6628737449645996, + "learning_rate": 7.84081932871026e-05, + "loss": 3.1347, + "step": 3845 + }, + { + "epoch": 5.63, + "grad_norm": 33.729000091552734, + "learning_rate": 7.836239860276671e-05, + "loss": 3.1319, + "step": 3846 + }, + { + "epoch": 5.64, + "grad_norm": 13.277280807495117, + "learning_rate": 7.831660867855816e-05, + "loss": 3.1439, + "step": 3847 + }, + { + "epoch": 5.64, + "grad_norm": 1.8134537935256958, + "learning_rate": 7.827082352455039e-05, + "loss": 3.1507, + "step": 3848 + }, + { + "epoch": 5.64, + "grad_norm": 2.101228952407837, + "learning_rate": 7.822504315081578e-05, + "loss": 3.1133, + "step": 3849 + }, + { + "epoch": 5.64, + "grad_norm": 2.533062696456909, + "learning_rate": 7.817926756742576e-05, + "loss": 3.1362, + "step": 3850 + }, + { + "epoch": 5.64, + "grad_norm": 4.869479656219482, + "learning_rate": 7.813349678445067e-05, + "loss": 3.1618, + "step": 3851 + }, + { + "epoch": 5.64, + "grad_norm": 2.6743147373199463, + "learning_rate": 7.808773081195972e-05, + "loss": 3.1278, + "step": 3852 + }, + { + "epoch": 5.64, + "grad_norm": 0.8585074543952942, + "learning_rate": 7.804196966002113e-05, + "loss": 3.1386, + "step": 3853 + }, + { + "epoch": 5.65, + "grad_norm": 7.510640621185303, + "learning_rate": 7.799621333870208e-05, + "loss": 3.1786, + "step": 3854 + }, + { + "epoch": 5.65, + "grad_norm": 2.670872449874878, + "learning_rate": 7.795046185806866e-05, + "loss": 3.1483, + "step": 3855 + }, + { + "epoch": 5.65, + "grad_norm": 22.928964614868164, + "learning_rate": 7.79047152281858e-05, + "loss": 3.1633, + "step": 3856 + }, + { + "epoch": 5.65, + "grad_norm": 7.530433177947998, + "learning_rate": 7.785897345911754e-05, + "loss": 3.1601, + "step": 3857 + }, + { + "epoch": 5.65, + "grad_norm": 7.938250541687012, + "learning_rate": 7.781323656092677e-05, + "loss": 3.1554, + "step": 3858 + }, + { + "epoch": 5.65, + "grad_norm": 2.2199909687042236, + "learning_rate": 7.776750454367521e-05, + "loss": 3.1512, + "step": 3859 + }, + { + "epoch": 5.65, + "grad_norm": 92.11273956298828, + "learning_rate": 7.772177741742369e-05, + "loss": 3.1275, + "step": 3860 + }, + { + "epoch": 5.66, + "grad_norm": 4.906170845031738, + "learning_rate": 7.767605519223183e-05, + "loss": 3.1245, + "step": 3861 + }, + { + "epoch": 5.66, + "grad_norm": 3.870858907699585, + "learning_rate": 7.76303378781582e-05, + "loss": 3.1105, + "step": 3862 + }, + { + "epoch": 5.66, + "grad_norm": 2.2505300045013428, + "learning_rate": 7.758462548526033e-05, + "loss": 3.1065, + "step": 3863 + }, + { + "epoch": 5.66, + "grad_norm": 2.947871208190918, + "learning_rate": 7.753891802359466e-05, + "loss": 3.1869, + "step": 3864 + }, + { + "epoch": 5.66, + "grad_norm": 3.654209613800049, + "learning_rate": 7.749321550321645e-05, + "loss": 3.1427, + "step": 3865 + }, + { + "epoch": 5.66, + "grad_norm": 3.259089469909668, + "learning_rate": 7.744751793418005e-05, + "loss": 3.1221, + "step": 3866 + }, + { + "epoch": 5.66, + "grad_norm": 2.3307595252990723, + "learning_rate": 7.740182532653853e-05, + "loss": 3.1556, + "step": 3867 + }, + { + "epoch": 5.67, + "grad_norm": 6.026763439178467, + "learning_rate": 7.735613769034399e-05, + "loss": 3.1646, + "step": 3868 + }, + { + "epoch": 5.67, + "grad_norm": 9.292561531066895, + "learning_rate": 7.731045503564743e-05, + "loss": 3.1498, + "step": 3869 + }, + { + "epoch": 5.67, + "grad_norm": 4.8578314781188965, + "learning_rate": 7.726477737249869e-05, + "loss": 3.1159, + "step": 3870 + }, + { + "epoch": 5.67, + "grad_norm": 5.385825157165527, + "learning_rate": 7.721910471094653e-05, + "loss": 3.1494, + "step": 3871 + }, + { + "epoch": 5.67, + "grad_norm": 3.9133341312408447, + "learning_rate": 7.717343706103873e-05, + "loss": 3.1402, + "step": 3872 + }, + { + "epoch": 5.67, + "grad_norm": 6.215272903442383, + "learning_rate": 7.712777443282175e-05, + "loss": 3.195, + "step": 3873 + }, + { + "epoch": 5.68, + "grad_norm": 3.6820201873779297, + "learning_rate": 7.708211683634112e-05, + "loss": 3.2287, + "step": 3874 + }, + { + "epoch": 5.68, + "grad_norm": 3.4176886081695557, + "learning_rate": 7.703646428164122e-05, + "loss": 3.1445, + "step": 3875 + }, + { + "epoch": 5.68, + "grad_norm": 8.820064544677734, + "learning_rate": 7.699081677876526e-05, + "loss": 3.2085, + "step": 3876 + }, + { + "epoch": 5.68, + "grad_norm": 17.15256118774414, + "learning_rate": 7.694517433775543e-05, + "loss": 3.1627, + "step": 3877 + }, + { + "epoch": 5.68, + "grad_norm": 1.7407811880111694, + "learning_rate": 7.689953696865271e-05, + "loss": 3.1868, + "step": 3878 + }, + { + "epoch": 5.68, + "grad_norm": 3.167860746383667, + "learning_rate": 7.685390468149707e-05, + "loss": 3.1838, + "step": 3879 + }, + { + "epoch": 5.68, + "grad_norm": 4.515937805175781, + "learning_rate": 7.680827748632728e-05, + "loss": 3.1624, + "step": 3880 + }, + { + "epoch": 5.69, + "grad_norm": 5.7442755699157715, + "learning_rate": 7.676265539318098e-05, + "loss": 3.1957, + "step": 3881 + }, + { + "epoch": 5.69, + "grad_norm": 4.945896625518799, + "learning_rate": 7.671703841209481e-05, + "loss": 3.1714, + "step": 3882 + }, + { + "epoch": 5.69, + "grad_norm": 8.167930603027344, + "learning_rate": 7.667142655310414e-05, + "loss": 3.1944, + "step": 3883 + }, + { + "epoch": 5.69, + "grad_norm": 17.595746994018555, + "learning_rate": 7.662581982624326e-05, + "loss": 3.1853, + "step": 3884 + }, + { + "epoch": 5.69, + "grad_norm": 2.751532554626465, + "learning_rate": 7.658021824154542e-05, + "loss": 3.1348, + "step": 3885 + }, + { + "epoch": 5.69, + "grad_norm": 6.661686897277832, + "learning_rate": 7.653462180904258e-05, + "loss": 3.2022, + "step": 3886 + }, + { + "epoch": 5.69, + "grad_norm": 9.880109786987305, + "learning_rate": 7.648903053876568e-05, + "loss": 3.1686, + "step": 3887 + }, + { + "epoch": 5.69, + "eval_loss": 2.989283800125122, + "eval_runtime": 1170.3187, + "eval_samples_per_second": 14.647, + "eval_steps_per_second": 0.916, + "step": 3887 + }, + { + "epoch": 5.7, + "grad_norm": 3.414788246154785, + "learning_rate": 7.644344444074453e-05, + "loss": 3.1669, + "step": 3888 + }, + { + "epoch": 5.7, + "grad_norm": 3.5142674446105957, + "learning_rate": 7.639786352500773e-05, + "loss": 3.1526, + "step": 3889 + }, + { + "epoch": 5.7, + "grad_norm": 7.385360240936279, + "learning_rate": 7.635228780158274e-05, + "loss": 3.1528, + "step": 3890 + }, + { + "epoch": 5.7, + "grad_norm": 7.439982891082764, + "learning_rate": 7.630671728049598e-05, + "loss": 3.1417, + "step": 3891 + }, + { + "epoch": 5.7, + "grad_norm": 4.842582702636719, + "learning_rate": 7.626115197177266e-05, + "loss": 3.2058, + "step": 3892 + }, + { + "epoch": 5.7, + "grad_norm": 6.158822536468506, + "learning_rate": 7.621559188543677e-05, + "loss": 3.2051, + "step": 3893 + }, + { + "epoch": 5.7, + "grad_norm": 4.262391090393066, + "learning_rate": 7.617003703151129e-05, + "loss": 3.2001, + "step": 3894 + }, + { + "epoch": 5.71, + "grad_norm": 22.497093200683594, + "learning_rate": 7.612448742001797e-05, + "loss": 3.1808, + "step": 3895 + }, + { + "epoch": 5.71, + "grad_norm": 3.1336076259613037, + "learning_rate": 7.607894306097739e-05, + "loss": 3.2113, + "step": 3896 + }, + { + "epoch": 5.71, + "grad_norm": 4.773746967315674, + "learning_rate": 7.603340396440902e-05, + "loss": 3.1906, + "step": 3897 + }, + { + "epoch": 5.71, + "grad_norm": 5.188856601715088, + "learning_rate": 7.59878701403312e-05, + "loss": 3.1663, + "step": 3898 + }, + { + "epoch": 5.71, + "grad_norm": 7.35548734664917, + "learning_rate": 7.594234159876098e-05, + "loss": 3.1467, + "step": 3899 + }, + { + "epoch": 5.71, + "grad_norm": 7.549004077911377, + "learning_rate": 7.589681834971443e-05, + "loss": 3.1928, + "step": 3900 + }, + { + "epoch": 5.72, + "grad_norm": 6.5583086013793945, + "learning_rate": 7.585130040320627e-05, + "loss": 3.2067, + "step": 3901 + }, + { + "epoch": 5.72, + "grad_norm": 14.854689598083496, + "learning_rate": 7.580578776925022e-05, + "loss": 3.2208, + "step": 3902 + }, + { + "epoch": 5.72, + "grad_norm": 11.917444229125977, + "learning_rate": 7.576028045785868e-05, + "loss": 3.1828, + "step": 3903 + }, + { + "epoch": 5.72, + "grad_norm": 6.152881145477295, + "learning_rate": 7.5714778479043e-05, + "loss": 3.204, + "step": 3904 + }, + { + "epoch": 5.72, + "grad_norm": 4.3145294189453125, + "learning_rate": 7.566928184281334e-05, + "loss": 3.2282, + "step": 3905 + }, + { + "epoch": 5.72, + "grad_norm": 3.392911434173584, + "learning_rate": 7.562379055917856e-05, + "loss": 3.1795, + "step": 3906 + }, + { + "epoch": 5.72, + "grad_norm": 5.250206470489502, + "learning_rate": 7.557830463814652e-05, + "loss": 3.1574, + "step": 3907 + }, + { + "epoch": 5.73, + "grad_norm": 11.659444808959961, + "learning_rate": 7.553282408972382e-05, + "loss": 3.1975, + "step": 3908 + }, + { + "epoch": 5.73, + "grad_norm": 6.504120826721191, + "learning_rate": 7.54873489239158e-05, + "loss": 3.2318, + "step": 3909 + }, + { + "epoch": 5.73, + "grad_norm": 3.6852314472198486, + "learning_rate": 7.544187915072676e-05, + "loss": 3.1501, + "step": 3910 + }, + { + "epoch": 5.73, + "grad_norm": 4.429998874664307, + "learning_rate": 7.539641478015976e-05, + "loss": 3.1623, + "step": 3911 + }, + { + "epoch": 5.73, + "grad_norm": 2.792065382003784, + "learning_rate": 7.535095582221656e-05, + "loss": 3.1624, + "step": 3912 + }, + { + "epoch": 5.73, + "grad_norm": 21.118961334228516, + "learning_rate": 7.530550228689794e-05, + "loss": 3.2112, + "step": 3913 + }, + { + "epoch": 5.73, + "grad_norm": 18.30959701538086, + "learning_rate": 7.526005418420332e-05, + "loss": 3.2394, + "step": 3914 + }, + { + "epoch": 5.74, + "grad_norm": 8.257102012634277, + "learning_rate": 7.521461152413092e-05, + "loss": 3.2588, + "step": 3915 + }, + { + "epoch": 5.74, + "grad_norm": 3.5304644107818604, + "learning_rate": 7.516917431667795e-05, + "loss": 3.2331, + "step": 3916 + }, + { + "epoch": 5.74, + "grad_norm": 24.953046798706055, + "learning_rate": 7.512374257184019e-05, + "loss": 3.1763, + "step": 3917 + }, + { + "epoch": 5.74, + "grad_norm": 5.283435821533203, + "learning_rate": 7.507831629961234e-05, + "loss": 3.1631, + "step": 3918 + }, + { + "epoch": 5.74, + "grad_norm": 7.136661052703857, + "learning_rate": 7.503289550998791e-05, + "loss": 3.2551, + "step": 3919 + }, + { + "epoch": 5.74, + "grad_norm": 3.8187716007232666, + "learning_rate": 7.498748021295914e-05, + "loss": 3.1214, + "step": 3920 + }, + { + "epoch": 5.74, + "grad_norm": 10.967880249023438, + "learning_rate": 7.494207041851708e-05, + "loss": 3.144, + "step": 3921 + }, + { + "epoch": 5.75, + "grad_norm": 30.177751541137695, + "learning_rate": 7.489666613665163e-05, + "loss": 3.1466, + "step": 3922 + }, + { + "epoch": 5.75, + "grad_norm": 6.267055511474609, + "learning_rate": 7.48512673773514e-05, + "loss": 3.2192, + "step": 3923 + }, + { + "epoch": 5.75, + "grad_norm": 3.4286611080169678, + "learning_rate": 7.480587415060378e-05, + "loss": 3.1519, + "step": 3924 + }, + { + "epoch": 5.75, + "grad_norm": 9.199522972106934, + "learning_rate": 7.476048646639498e-05, + "loss": 3.1738, + "step": 3925 + }, + { + "epoch": 5.75, + "grad_norm": 9.091236114501953, + "learning_rate": 7.471510433471005e-05, + "loss": 3.1704, + "step": 3926 + }, + { + "epoch": 5.75, + "grad_norm": 7.052969455718994, + "learning_rate": 7.46697277655327e-05, + "loss": 3.1835, + "step": 3927 + }, + { + "epoch": 5.76, + "grad_norm": 30.890663146972656, + "learning_rate": 7.462435676884547e-05, + "loss": 3.1883, + "step": 3928 + }, + { + "epoch": 5.76, + "grad_norm": 3.2805261611938477, + "learning_rate": 7.457899135462974e-05, + "loss": 3.2009, + "step": 3929 + }, + { + "epoch": 5.76, + "grad_norm": 11.584432601928711, + "learning_rate": 7.45336315328655e-05, + "loss": 3.2384, + "step": 3930 + }, + { + "epoch": 5.76, + "grad_norm": 6.337993144989014, + "learning_rate": 7.448827731353166e-05, + "loss": 3.1973, + "step": 3931 + }, + { + "epoch": 5.76, + "grad_norm": 13.461885452270508, + "learning_rate": 7.444292870660586e-05, + "loss": 3.2435, + "step": 3932 + }, + { + "epoch": 5.76, + "grad_norm": 9.56443977355957, + "learning_rate": 7.439758572206448e-05, + "loss": 3.1465, + "step": 3933 + }, + { + "epoch": 5.76, + "grad_norm": 7.9625115394592285, + "learning_rate": 7.43522483698826e-05, + "loss": 3.1635, + "step": 3934 + }, + { + "epoch": 5.77, + "grad_norm": 6.776467800140381, + "learning_rate": 7.430691666003424e-05, + "loss": 3.2405, + "step": 3935 + }, + { + "epoch": 5.77, + "grad_norm": 43.86566162109375, + "learning_rate": 7.426159060249204e-05, + "loss": 3.192, + "step": 3936 + }, + { + "epoch": 5.77, + "grad_norm": 12.768895149230957, + "learning_rate": 7.421627020722738e-05, + "loss": 3.1741, + "step": 3937 + }, + { + "epoch": 5.77, + "grad_norm": 6.572139263153076, + "learning_rate": 7.41709554842105e-05, + "loss": 3.1637, + "step": 3938 + }, + { + "epoch": 5.77, + "grad_norm": 5.105015277862549, + "learning_rate": 7.412564644341033e-05, + "loss": 3.2329, + "step": 3939 + }, + { + "epoch": 5.77, + "grad_norm": 4.994419574737549, + "learning_rate": 7.40803430947945e-05, + "loss": 3.2124, + "step": 3940 + }, + { + "epoch": 5.77, + "grad_norm": 13.615537643432617, + "learning_rate": 7.403504544832951e-05, + "loss": 3.2148, + "step": 3941 + }, + { + "epoch": 5.78, + "grad_norm": 6.730780124664307, + "learning_rate": 7.398975351398053e-05, + "loss": 3.2086, + "step": 3942 + }, + { + "epoch": 5.78, + "grad_norm": 8.399843215942383, + "learning_rate": 7.394446730171141e-05, + "loss": 3.2455, + "step": 3943 + }, + { + "epoch": 5.78, + "grad_norm": 7.892696857452393, + "learning_rate": 7.389918682148491e-05, + "loss": 3.263, + "step": 3944 + }, + { + "epoch": 5.78, + "grad_norm": 3.2410852909088135, + "learning_rate": 7.38539120832624e-05, + "loss": 3.2023, + "step": 3945 + }, + { + "epoch": 5.78, + "grad_norm": 43.567543029785156, + "learning_rate": 7.380864309700396e-05, + "loss": 3.2505, + "step": 3946 + }, + { + "epoch": 5.78, + "grad_norm": 8.303853988647461, + "learning_rate": 7.376337987266856e-05, + "loss": 3.2253, + "step": 3947 + }, + { + "epoch": 5.79, + "grad_norm": 5.137606620788574, + "learning_rate": 7.371812242021374e-05, + "loss": 3.2614, + "step": 3948 + }, + { + "epoch": 5.79, + "grad_norm": 7.571026802062988, + "learning_rate": 7.367287074959586e-05, + "loss": 3.3035, + "step": 3949 + }, + { + "epoch": 5.79, + "grad_norm": 5.098783016204834, + "learning_rate": 7.362762487076995e-05, + "loss": 3.24, + "step": 3950 + }, + { + "epoch": 5.79, + "grad_norm": 5.123687744140625, + "learning_rate": 7.358238479368985e-05, + "loss": 3.2333, + "step": 3951 + }, + { + "epoch": 5.79, + "grad_norm": 9.861023902893066, + "learning_rate": 7.353715052830808e-05, + "loss": 3.2665, + "step": 3952 + }, + { + "epoch": 5.79, + "grad_norm": 30.322845458984375, + "learning_rate": 7.349192208457579e-05, + "loss": 3.2337, + "step": 3953 + }, + { + "epoch": 5.79, + "grad_norm": 10.311440467834473, + "learning_rate": 7.344669947244302e-05, + "loss": 3.2307, + "step": 3954 + }, + { + "epoch": 5.8, + "grad_norm": 13.122282981872559, + "learning_rate": 7.340148270185842e-05, + "loss": 3.2456, + "step": 3955 + }, + { + "epoch": 5.8, + "grad_norm": 16.757612228393555, + "learning_rate": 7.335627178276937e-05, + "loss": 3.3208, + "step": 3956 + }, + { + "epoch": 5.8, + "grad_norm": 36.39468002319336, + "learning_rate": 7.331106672512197e-05, + "loss": 3.2313, + "step": 3957 + }, + { + "epoch": 5.8, + "grad_norm": 39.07672882080078, + "learning_rate": 7.326586753886102e-05, + "loss": 3.3304, + "step": 3958 + }, + { + "epoch": 5.8, + "grad_norm": 67.7435302734375, + "learning_rate": 7.322067423393002e-05, + "loss": 3.3623, + "step": 3959 + }, + { + "epoch": 5.8, + "grad_norm": 102.08399963378906, + "learning_rate": 7.317548682027127e-05, + "loss": 3.3498, + "step": 3960 + }, + { + "epoch": 5.8, + "grad_norm": 68.23622131347656, + "learning_rate": 7.313030530782564e-05, + "loss": 3.4284, + "step": 3961 + }, + { + "epoch": 5.81, + "grad_norm": 54.879310607910156, + "learning_rate": 7.308512970653274e-05, + "loss": 3.4101, + "step": 3962 + }, + { + "epoch": 5.81, + "grad_norm": 154.8038787841797, + "learning_rate": 7.303996002633096e-05, + "loss": 3.5803, + "step": 3963 + }, + { + "epoch": 5.81, + "grad_norm": 228.81173706054688, + "learning_rate": 7.29947962771573e-05, + "loss": 3.489, + "step": 3964 + }, + { + "epoch": 5.81, + "grad_norm": 350.7447509765625, + "learning_rate": 7.294963846894744e-05, + "loss": 3.6323, + "step": 3965 + }, + { + "epoch": 5.81, + "grad_norm": 858.3496704101562, + "learning_rate": 7.290448661163587e-05, + "loss": 4.0812, + "step": 3966 + }, + { + "epoch": 5.81, + "grad_norm": 730.7702026367188, + "learning_rate": 7.285934071515568e-05, + "loss": 4.0206, + "step": 3967 + }, + { + "epoch": 5.81, + "grad_norm": 658.6021118164062, + "learning_rate": 7.281420078943855e-05, + "loss": 4.3312, + "step": 3968 + }, + { + "epoch": 5.82, + "grad_norm": 475.8223876953125, + "learning_rate": 7.276906684441514e-05, + "loss": 4.5484, + "step": 3969 + }, + { + "epoch": 5.82, + "grad_norm": 309.2330322265625, + "learning_rate": 7.272393889001454e-05, + "loss": 4.6918, + "step": 3970 + }, + { + "epoch": 5.82, + "grad_norm": 195.97572326660156, + "learning_rate": 7.267881693616452e-05, + "loss": 4.7661, + "step": 3971 + }, + { + "epoch": 5.82, + "grad_norm": 134.8620147705078, + "learning_rate": 7.263370099279172e-05, + "loss": 4.6478, + "step": 3972 + }, + { + "epoch": 5.82, + "grad_norm": 108.79113006591797, + "learning_rate": 7.25885910698213e-05, + "loss": 4.6325, + "step": 3973 + }, + { + "epoch": 5.82, + "grad_norm": 149.6044921875, + "learning_rate": 7.254348717717714e-05, + "loss": 4.8721, + "step": 3974 + }, + { + "epoch": 5.83, + "grad_norm": 117.6059341430664, + "learning_rate": 7.249838932478177e-05, + "loss": 4.8513, + "step": 3975 + }, + { + "epoch": 5.83, + "grad_norm": 56.039024353027344, + "learning_rate": 7.245329752255647e-05, + "loss": 4.8064, + "step": 3976 + }, + { + "epoch": 5.83, + "grad_norm": 100.5042953491211, + "learning_rate": 7.24082117804211e-05, + "loss": 4.8077, + "step": 3977 + }, + { + "epoch": 5.83, + "grad_norm": 101.55089569091797, + "learning_rate": 7.236313210829419e-05, + "loss": 4.9091, + "step": 3978 + }, + { + "epoch": 5.83, + "grad_norm": 54.18492889404297, + "learning_rate": 7.231805851609306e-05, + "loss": 4.9624, + "step": 3979 + }, + { + "epoch": 5.83, + "grad_norm": 90.85334014892578, + "learning_rate": 7.227299101373352e-05, + "loss": 4.8436, + "step": 3980 + }, + { + "epoch": 5.83, + "grad_norm": 108.68850708007812, + "learning_rate": 7.222792961113012e-05, + "loss": 4.7235, + "step": 3981 + }, + { + "epoch": 5.84, + "grad_norm": 46.6330451965332, + "learning_rate": 7.21828743181961e-05, + "loss": 4.7477, + "step": 3982 + }, + { + "epoch": 5.84, + "grad_norm": 116.49347686767578, + "learning_rate": 7.213782514484333e-05, + "loss": 4.6439, + "step": 3983 + }, + { + "epoch": 5.84, + "grad_norm": 46.60487747192383, + "learning_rate": 7.209278210098224e-05, + "loss": 4.6174, + "step": 3984 + }, + { + "epoch": 5.84, + "grad_norm": 74.72257995605469, + "learning_rate": 7.204774519652211e-05, + "loss": 4.6026, + "step": 3985 + }, + { + "epoch": 5.84, + "grad_norm": 117.54887390136719, + "learning_rate": 7.200271444137072e-05, + "loss": 4.3403, + "step": 3986 + }, + { + "epoch": 5.84, + "grad_norm": 66.1172866821289, + "learning_rate": 7.195768984543448e-05, + "loss": 4.2326, + "step": 3987 + }, + { + "epoch": 5.84, + "grad_norm": 23.685178756713867, + "learning_rate": 7.191267141861858e-05, + "loss": 4.2885, + "step": 3988 + }, + { + "epoch": 5.85, + "grad_norm": 33.490779876708984, + "learning_rate": 7.186765917082674e-05, + "loss": 4.2881, + "step": 3989 + }, + { + "epoch": 5.85, + "grad_norm": 49.65608215332031, + "learning_rate": 7.18226531119613e-05, + "loss": 4.0779, + "step": 3990 + }, + { + "epoch": 5.85, + "grad_norm": 21.95450782775879, + "learning_rate": 7.177765325192338e-05, + "loss": 4.0829, + "step": 3991 + }, + { + "epoch": 5.85, + "grad_norm": 36.09683609008789, + "learning_rate": 7.173265960061261e-05, + "loss": 4.1138, + "step": 3992 + }, + { + "epoch": 5.85, + "grad_norm": 135.8228302001953, + "learning_rate": 7.168767216792724e-05, + "loss": 4.0337, + "step": 3993 + }, + { + "epoch": 5.85, + "grad_norm": 42.24320602416992, + "learning_rate": 7.164269096376432e-05, + "loss": 4.0038, + "step": 3994 + }, + { + "epoch": 5.85, + "grad_norm": 90.61548614501953, + "learning_rate": 7.15977159980193e-05, + "loss": 3.9393, + "step": 3995 + }, + { + "epoch": 5.86, + "grad_norm": 83.08573150634766, + "learning_rate": 7.155274728058645e-05, + "loss": 4.085, + "step": 3996 + }, + { + "epoch": 5.86, + "grad_norm": 51.404388427734375, + "learning_rate": 7.150778482135851e-05, + "loss": 4.025, + "step": 3997 + }, + { + "epoch": 5.86, + "grad_norm": 133.36476135253906, + "learning_rate": 7.146282863022699e-05, + "loss": 4.0341, + "step": 3998 + }, + { + "epoch": 5.86, + "grad_norm": 68.87841033935547, + "learning_rate": 7.141787871708194e-05, + "loss": 4.1013, + "step": 3999 + }, + { + "epoch": 5.86, + "grad_norm": 83.18514251708984, + "learning_rate": 7.137293509181197e-05, + "loss": 4.0575, + "step": 4000 + }, + { + "epoch": 5.86, + "grad_norm": 63.65642166137695, + "learning_rate": 7.132799776430446e-05, + "loss": 4.0338, + "step": 4001 + }, + { + "epoch": 5.87, + "grad_norm": 84.51449584960938, + "learning_rate": 7.128306674444531e-05, + "loss": 4.0289, + "step": 4002 + }, + { + "epoch": 5.87, + "grad_norm": 1112.17431640625, + "learning_rate": 7.123814204211899e-05, + "loss": 3.9574, + "step": 4003 + }, + { + "epoch": 5.87, + "grad_norm": 113.361572265625, + "learning_rate": 7.11932236672087e-05, + "loss": 3.9683, + "step": 4004 + }, + { + "epoch": 5.87, + "grad_norm": 94.9935073852539, + "learning_rate": 7.114831162959614e-05, + "loss": 3.867, + "step": 4005 + }, + { + "epoch": 5.87, + "grad_norm": 99.41561126708984, + "learning_rate": 7.110340593916164e-05, + "loss": 4.0003, + "step": 4006 + }, + { + "epoch": 5.87, + "grad_norm": 68.38880157470703, + "learning_rate": 7.105850660578422e-05, + "loss": 3.8946, + "step": 4007 + }, + { + "epoch": 5.87, + "grad_norm": 48.9935188293457, + "learning_rate": 7.101361363934136e-05, + "loss": 3.9, + "step": 4008 + }, + { + "epoch": 5.88, + "grad_norm": 75.62947082519531, + "learning_rate": 7.096872704970923e-05, + "loss": 3.9334, + "step": 4009 + }, + { + "epoch": 5.88, + "grad_norm": 259.18511962890625, + "learning_rate": 7.092384684676262e-05, + "loss": 3.9539, + "step": 4010 + }, + { + "epoch": 5.88, + "grad_norm": 60.03557586669922, + "learning_rate": 7.087897304037483e-05, + "loss": 3.9378, + "step": 4011 + }, + { + "epoch": 5.88, + "grad_norm": 60.427616119384766, + "learning_rate": 7.083410564041779e-05, + "loss": 3.922, + "step": 4012 + }, + { + "epoch": 5.88, + "grad_norm": 86.16722869873047, + "learning_rate": 7.078924465676208e-05, + "loss": 3.9123, + "step": 4013 + }, + { + "epoch": 5.88, + "grad_norm": 92.25879669189453, + "learning_rate": 7.07443900992768e-05, + "loss": 3.9653, + "step": 4014 + }, + { + "epoch": 5.88, + "grad_norm": 63.97455596923828, + "learning_rate": 7.069954197782958e-05, + "loss": 3.9709, + "step": 4015 + }, + { + "epoch": 5.89, + "grad_norm": 85.49566650390625, + "learning_rate": 7.065470030228679e-05, + "loss": 3.9614, + "step": 4016 + }, + { + "epoch": 5.89, + "grad_norm": 200.93997192382812, + "learning_rate": 7.060986508251329e-05, + "loss": 3.9715, + "step": 4017 + }, + { + "epoch": 5.89, + "grad_norm": 47.623600006103516, + "learning_rate": 7.056503632837247e-05, + "loss": 3.9969, + "step": 4018 + }, + { + "epoch": 5.89, + "grad_norm": 57.71437454223633, + "learning_rate": 7.052021404972641e-05, + "loss": 4.0892, + "step": 4019 + }, + { + "epoch": 5.89, + "grad_norm": 101.89254760742188, + "learning_rate": 7.047539825643572e-05, + "loss": 4.1731, + "step": 4020 + }, + { + "epoch": 5.89, + "grad_norm": 56.065486907958984, + "learning_rate": 7.043058895835953e-05, + "loss": 4.1648, + "step": 4021 + }, + { + "epoch": 5.89, + "grad_norm": 98.78327178955078, + "learning_rate": 7.038578616535556e-05, + "loss": 4.3178, + "step": 4022 + }, + { + "epoch": 5.9, + "grad_norm": 383.0876770019531, + "learning_rate": 7.034098988728024e-05, + "loss": 4.3457, + "step": 4023 + }, + { + "epoch": 5.9, + "grad_norm": 162.98101806640625, + "learning_rate": 7.029620013398837e-05, + "loss": 4.3242, + "step": 4024 + }, + { + "epoch": 5.9, + "grad_norm": 149.046142578125, + "learning_rate": 7.025141691533334e-05, + "loss": 4.4494, + "step": 4025 + }, + { + "epoch": 5.9, + "grad_norm": 135.7252655029297, + "learning_rate": 7.020664024116726e-05, + "loss": 4.487, + "step": 4026 + }, + { + "epoch": 5.9, + "grad_norm": 103.65221405029297, + "learning_rate": 7.016187012134068e-05, + "loss": 4.544, + "step": 4027 + }, + { + "epoch": 5.9, + "grad_norm": 95.76521301269531, + "learning_rate": 7.011710656570267e-05, + "loss": 4.5937, + "step": 4028 + }, + { + "epoch": 5.91, + "grad_norm": 287.5489196777344, + "learning_rate": 7.007234958410098e-05, + "loss": 4.6397, + "step": 4029 + }, + { + "epoch": 5.91, + "grad_norm": 55.86167526245117, + "learning_rate": 7.002759918638183e-05, + "loss": 4.6566, + "step": 4030 + }, + { + "epoch": 5.91, + "grad_norm": 50.3218879699707, + "learning_rate": 6.998285538238993e-05, + "loss": 4.6248, + "step": 4031 + }, + { + "epoch": 5.91, + "grad_norm": 414.169921875, + "learning_rate": 6.993811818196872e-05, + "loss": 4.6006, + "step": 4032 + }, + { + "epoch": 5.91, + "grad_norm": 112.764404296875, + "learning_rate": 6.989338759496006e-05, + "loss": 4.6415, + "step": 4033 + }, + { + "epoch": 5.91, + "grad_norm": 53.906734466552734, + "learning_rate": 6.984866363120432e-05, + "loss": 4.6018, + "step": 4034 + }, + { + "epoch": 5.91, + "grad_norm": 45.21807098388672, + "learning_rate": 6.980394630054054e-05, + "loss": 4.5761, + "step": 4035 + }, + { + "epoch": 5.92, + "grad_norm": 123.97179412841797, + "learning_rate": 6.975923561280624e-05, + "loss": 4.4143, + "step": 4036 + }, + { + "epoch": 5.92, + "grad_norm": 39.991905212402344, + "learning_rate": 6.971453157783739e-05, + "loss": 4.3971, + "step": 4037 + }, + { + "epoch": 5.92, + "grad_norm": 245.3745574951172, + "learning_rate": 6.966983420546866e-05, + "loss": 4.3387, + "step": 4038 + }, + { + "epoch": 5.92, + "grad_norm": 535.9085083007812, + "learning_rate": 6.962514350553313e-05, + "loss": 4.2919, + "step": 4039 + }, + { + "epoch": 5.92, + "grad_norm": 31.66384506225586, + "learning_rate": 6.958045948786245e-05, + "loss": 4.2878, + "step": 4040 + }, + { + "epoch": 5.92, + "grad_norm": 104.09449768066406, + "learning_rate": 6.953578216228685e-05, + "loss": 4.2645, + "step": 4041 + }, + { + "epoch": 5.92, + "grad_norm": 73.07312774658203, + "learning_rate": 6.9491111538635e-05, + "loss": 4.2232, + "step": 4042 + }, + { + "epoch": 5.93, + "grad_norm": 55.163421630859375, + "learning_rate": 6.944644762673413e-05, + "loss": 4.2499, + "step": 4043 + }, + { + "epoch": 5.93, + "grad_norm": 97.98957061767578, + "learning_rate": 6.940179043641005e-05, + "loss": 4.2195, + "step": 4044 + }, + { + "epoch": 5.93, + "grad_norm": 35.54196548461914, + "learning_rate": 6.9357139977487e-05, + "loss": 4.2314, + "step": 4045 + }, + { + "epoch": 5.93, + "grad_norm": 36.159202575683594, + "learning_rate": 6.931249625978782e-05, + "loss": 4.1943, + "step": 4046 + }, + { + "epoch": 5.93, + "grad_norm": 119.87337493896484, + "learning_rate": 6.926785929313376e-05, + "loss": 4.1162, + "step": 4047 + }, + { + "epoch": 5.93, + "grad_norm": 7.6323442459106445, + "learning_rate": 6.922322908734474e-05, + "loss": 4.1434, + "step": 4048 + }, + { + "epoch": 5.94, + "grad_norm": 13.205349922180176, + "learning_rate": 6.917860565223905e-05, + "loss": 4.1026, + "step": 4049 + }, + { + "epoch": 5.94, + "grad_norm": 29.02250099182129, + "learning_rate": 6.913398899763352e-05, + "loss": 4.0863, + "step": 4050 + }, + { + "epoch": 5.94, + "grad_norm": 18.842418670654297, + "learning_rate": 6.908937913334362e-05, + "loss": 4.1056, + "step": 4051 + }, + { + "epoch": 5.94, + "grad_norm": 26.442873001098633, + "learning_rate": 6.904477606918313e-05, + "loss": 4.1195, + "step": 4052 + }, + { + "epoch": 5.94, + "grad_norm": 146.73545837402344, + "learning_rate": 6.900017981496443e-05, + "loss": 4.0448, + "step": 4053 + }, + { + "epoch": 5.94, + "grad_norm": 18.860145568847656, + "learning_rate": 6.895559038049846e-05, + "loss": 4.0507, + "step": 4054 + }, + { + "epoch": 5.94, + "grad_norm": 7.4111504554748535, + "learning_rate": 6.891100777559454e-05, + "loss": 3.9955, + "step": 4055 + }, + { + "epoch": 5.95, + "grad_norm": 2.2253613471984863, + "learning_rate": 6.886643201006052e-05, + "loss": 3.9959, + "step": 4056 + }, + { + "epoch": 5.95, + "eval_loss": 3.7561120986938477, + "eval_runtime": 1174.8961, + "eval_samples_per_second": 14.59, + "eval_steps_per_second": 0.912, + "step": 4056 + }, + { + "epoch": 5.95, + "grad_norm": 2.182352066040039, + "learning_rate": 6.882186309370287e-05, + "loss": 3.9472, + "step": 4057 + }, + { + "epoch": 5.95, + "grad_norm": 1.8885817527770996, + "learning_rate": 6.877730103632638e-05, + "loss": 3.9685, + "step": 4058 + }, + { + "epoch": 5.95, + "grad_norm": 1.6018338203430176, + "learning_rate": 6.87327458477344e-05, + "loss": 3.8923, + "step": 4059 + }, + { + "epoch": 5.95, + "grad_norm": 1.4974128007888794, + "learning_rate": 6.868819753772879e-05, + "loss": 3.8199, + "step": 4060 + }, + { + "epoch": 5.95, + "grad_norm": 1.9076157808303833, + "learning_rate": 6.86436561161099e-05, + "loss": 3.8143, + "step": 4061 + }, + { + "epoch": 5.95, + "grad_norm": 1.2935985326766968, + "learning_rate": 6.859912159267648e-05, + "loss": 3.8232, + "step": 4062 + }, + { + "epoch": 5.96, + "grad_norm": 1.5463148355484009, + "learning_rate": 6.85545939772259e-05, + "loss": 3.7682, + "step": 4063 + }, + { + "epoch": 5.96, + "grad_norm": 1.2979236841201782, + "learning_rate": 6.851007327955393e-05, + "loss": 3.812, + "step": 4064 + }, + { + "epoch": 5.96, + "grad_norm": 1.235023856163025, + "learning_rate": 6.846555950945475e-05, + "loss": 3.6794, + "step": 4065 + }, + { + "epoch": 5.96, + "grad_norm": 1.587446689605713, + "learning_rate": 6.842105267672117e-05, + "loss": 3.6736, + "step": 4066 + }, + { + "epoch": 5.96, + "grad_norm": 1.6601265668869019, + "learning_rate": 6.837655279114439e-05, + "loss": 3.7298, + "step": 4067 + }, + { + "epoch": 5.96, + "grad_norm": 1.1815412044525146, + "learning_rate": 6.833205986251404e-05, + "loss": 3.6491, + "step": 4068 + }, + { + "epoch": 5.96, + "grad_norm": 2.2124860286712646, + "learning_rate": 6.828757390061828e-05, + "loss": 3.6432, + "step": 4069 + }, + { + "epoch": 5.97, + "grad_norm": 1.2953498363494873, + "learning_rate": 6.824309491524377e-05, + "loss": 3.6124, + "step": 4070 + }, + { + "epoch": 5.97, + "grad_norm": 1.5017386674880981, + "learning_rate": 6.819862291617555e-05, + "loss": 3.6377, + "step": 4071 + }, + { + "epoch": 5.97, + "grad_norm": 2.5354573726654053, + "learning_rate": 6.815415791319712e-05, + "loss": 3.6211, + "step": 4072 + }, + { + "epoch": 5.97, + "grad_norm": 2.2560806274414062, + "learning_rate": 6.810969991609055e-05, + "loss": 3.5855, + "step": 4073 + }, + { + "epoch": 5.97, + "grad_norm": 2.3098514080047607, + "learning_rate": 6.806524893463631e-05, + "loss": 3.5944, + "step": 4074 + }, + { + "epoch": 5.97, + "grad_norm": 3.807551860809326, + "learning_rate": 6.802080497861326e-05, + "loss": 3.582, + "step": 4075 + }, + { + "epoch": 5.98, + "grad_norm": 0.8520615100860596, + "learning_rate": 6.797636805779878e-05, + "loss": 3.5581, + "step": 4076 + }, + { + "epoch": 5.98, + "grad_norm": 2.77345871925354, + "learning_rate": 6.793193818196876e-05, + "loss": 3.5948, + "step": 4077 + }, + { + "epoch": 5.98, + "grad_norm": 1.2732380628585815, + "learning_rate": 6.788751536089739e-05, + "loss": 3.5607, + "step": 4078 + }, + { + "epoch": 5.98, + "grad_norm": 4.257369518280029, + "learning_rate": 6.784309960435744e-05, + "loss": 3.588, + "step": 4079 + }, + { + "epoch": 5.98, + "grad_norm": 4.027552604675293, + "learning_rate": 6.77986909221201e-05, + "loss": 3.5961, + "step": 4080 + }, + { + "epoch": 5.98, + "grad_norm": 1.3005938529968262, + "learning_rate": 6.77542893239549e-05, + "loss": 3.52, + "step": 4081 + }, + { + "epoch": 5.98, + "grad_norm": 3.389254570007324, + "learning_rate": 6.770989481963001e-05, + "loss": 3.5313, + "step": 4082 + }, + { + "epoch": 5.99, + "grad_norm": 1.3640910387039185, + "learning_rate": 6.766550741891182e-05, + "loss": 3.5291, + "step": 4083 + }, + { + "epoch": 5.99, + "grad_norm": 5.111477375030518, + "learning_rate": 6.76211271315653e-05, + "loss": 3.5717, + "step": 4084 + }, + { + "epoch": 5.99, + "grad_norm": 5.650928974151611, + "learning_rate": 6.757675396735386e-05, + "loss": 3.5892, + "step": 4085 + }, + { + "epoch": 5.99, + "grad_norm": 2.029646158218384, + "learning_rate": 6.753238793603923e-05, + "loss": 3.5179, + "step": 4086 + }, + { + "epoch": 5.99, + "grad_norm": 5.816488742828369, + "learning_rate": 6.748802904738163e-05, + "loss": 3.5582, + "step": 4087 + }, + { + "epoch": 5.99, + "grad_norm": 7.348140716552734, + "learning_rate": 6.74436773111398e-05, + "loss": 3.5371, + "step": 4088 + }, + { + "epoch": 5.99, + "grad_norm": 6.118741512298584, + "learning_rate": 6.739933273707078e-05, + "loss": 3.5298, + "step": 4089 + }, + { + "epoch": 6.0, + "grad_norm": 0.9806273579597473, + "learning_rate": 6.735499533493003e-05, + "loss": 3.5438, + "step": 4090 + }, + { + "epoch": 6.0, + "grad_norm": 9.882466316223145, + "learning_rate": 6.731066511447159e-05, + "loss": 3.5713, + "step": 4091 + }, + { + "epoch": 6.0, + "grad_norm": 14.419783592224121, + "learning_rate": 6.726634208544773e-05, + "loss": 3.569, + "step": 4092 + }, + { + "epoch": 6.0, + "grad_norm": 11.746726036071777, + "learning_rate": 6.722202625760923e-05, + "loss": 3.6069, + "step": 4093 + }, + { + "epoch": 6.0, + "grad_norm": 5.577601432800293, + "learning_rate": 6.717771764070527e-05, + "loss": 3.5287, + "step": 4094 + }, + { + "epoch": 6.0, + "grad_norm": 5.144962310791016, + "learning_rate": 6.713341624448351e-05, + "loss": 3.5013, + "step": 4095 + }, + { + "epoch": 6.0, + "grad_norm": 8.980729103088379, + "learning_rate": 6.708912207868987e-05, + "loss": 3.5225, + "step": 4096 + }, + { + "epoch": 6.01, + "grad_norm": 9.171212196350098, + "learning_rate": 6.70448351530688e-05, + "loss": 3.537, + "step": 4097 + }, + { + "epoch": 6.01, + "grad_norm": 6.954055309295654, + "learning_rate": 6.70005554773632e-05, + "loss": 3.5512, + "step": 4098 + }, + { + "epoch": 6.01, + "grad_norm": 25.35405731201172, + "learning_rate": 6.695628306131422e-05, + "loss": 3.525, + "step": 4099 + }, + { + "epoch": 6.01, + "grad_norm": 7.4213948249816895, + "learning_rate": 6.691201791466148e-05, + "loss": 3.5411, + "step": 4100 + }, + { + "epoch": 6.01, + "grad_norm": 10.75389289855957, + "learning_rate": 6.68677600471431e-05, + "loss": 3.5158, + "step": 4101 + }, + { + "epoch": 6.01, + "grad_norm": 9.7072114944458, + "learning_rate": 6.682350946849546e-05, + "loss": 3.5728, + "step": 4102 + }, + { + "epoch": 6.02, + "grad_norm": 4.010187149047852, + "learning_rate": 6.677926618845337e-05, + "loss": 3.4962, + "step": 4103 + }, + { + "epoch": 6.0, + "grad_norm": 5.2085371017456055, + "learning_rate": 6.673503021675011e-05, + "loss": 3.5387, + "step": 4104 + }, + { + "epoch": 6.0, + "grad_norm": 7.690248012542725, + "learning_rate": 6.669080156311729e-05, + "loss": 3.5012, + "step": 4105 + }, + { + "epoch": 6.0, + "grad_norm": 7.818577766418457, + "learning_rate": 6.664658023728485e-05, + "loss": 3.5594, + "step": 4106 + }, + { + "epoch": 6.01, + "grad_norm": 4.1020355224609375, + "learning_rate": 6.660236624898124e-05, + "loss": 3.4641, + "step": 4107 + }, + { + "epoch": 6.01, + "grad_norm": 3.3743233680725098, + "learning_rate": 6.655815960793328e-05, + "loss": 3.5123, + "step": 4108 + }, + { + "epoch": 6.01, + "grad_norm": 5.35082483291626, + "learning_rate": 6.651396032386603e-05, + "loss": 3.5047, + "step": 4109 + }, + { + "epoch": 6.01, + "grad_norm": 2.8813364505767822, + "learning_rate": 6.646976840650312e-05, + "loss": 3.4699, + "step": 4110 + }, + { + "epoch": 6.01, + "grad_norm": 3.898869037628174, + "learning_rate": 6.642558386556648e-05, + "loss": 3.4723, + "step": 4111 + }, + { + "epoch": 6.01, + "grad_norm": 3.897127389907837, + "learning_rate": 6.638140671077633e-05, + "loss": 3.4887, + "step": 4112 + }, + { + "epoch": 6.01, + "grad_norm": 0.8849645256996155, + "learning_rate": 6.633723695185141e-05, + "loss": 3.4564, + "step": 4113 + }, + { + "epoch": 6.02, + "grad_norm": 4.971652507781982, + "learning_rate": 6.62930745985088e-05, + "loss": 3.5218, + "step": 4114 + }, + { + "epoch": 6.02, + "grad_norm": 4.998004913330078, + "learning_rate": 6.624891966046385e-05, + "loss": 3.5211, + "step": 4115 + }, + { + "epoch": 6.02, + "grad_norm": 1.827019214630127, + "learning_rate": 6.620477214743042e-05, + "loss": 3.4959, + "step": 4116 + }, + { + "epoch": 6.02, + "grad_norm": 4.872849464416504, + "learning_rate": 6.616063206912063e-05, + "loss": 3.4849, + "step": 4117 + }, + { + "epoch": 6.02, + "grad_norm": 5.961864948272705, + "learning_rate": 6.611649943524498e-05, + "loss": 3.4685, + "step": 4118 + }, + { + "epoch": 6.02, + "grad_norm": 3.4859068393707275, + "learning_rate": 6.607237425551236e-05, + "loss": 3.4841, + "step": 4119 + }, + { + "epoch": 6.03, + "grad_norm": 4.598318099975586, + "learning_rate": 6.602825653963006e-05, + "loss": 3.5138, + "step": 4120 + }, + { + "epoch": 6.03, + "grad_norm": 8.418424606323242, + "learning_rate": 6.598414629730367e-05, + "loss": 3.4861, + "step": 4121 + }, + { + "epoch": 6.03, + "grad_norm": 6.949596881866455, + "learning_rate": 6.594004353823708e-05, + "loss": 3.4545, + "step": 4122 + }, + { + "epoch": 6.03, + "grad_norm": 1.4345308542251587, + "learning_rate": 6.589594827213268e-05, + "loss": 3.4615, + "step": 4123 + }, + { + "epoch": 6.03, + "grad_norm": 5.833983898162842, + "learning_rate": 6.585186050869112e-05, + "loss": 3.4469, + "step": 4124 + }, + { + "epoch": 6.03, + "grad_norm": 8.324283599853516, + "learning_rate": 6.580778025761136e-05, + "loss": 3.5428, + "step": 4125 + }, + { + "epoch": 6.03, + "grad_norm": 7.164588451385498, + "learning_rate": 6.576370752859085e-05, + "loss": 3.44, + "step": 4126 + }, + { + "epoch": 6.04, + "grad_norm": 4.524144172668457, + "learning_rate": 6.57196423313252e-05, + "loss": 3.4798, + "step": 4127 + }, + { + "epoch": 6.04, + "grad_norm": 3.6542019844055176, + "learning_rate": 6.56755846755085e-05, + "loss": 3.4431, + "step": 4128 + }, + { + "epoch": 6.04, + "grad_norm": 6.415909290313721, + "learning_rate": 6.563153457083315e-05, + "loss": 3.4509, + "step": 4129 + }, + { + "epoch": 6.04, + "grad_norm": 4.0182623863220215, + "learning_rate": 6.558749202698988e-05, + "loss": 3.4182, + "step": 4130 + }, + { + "epoch": 6.04, + "grad_norm": 22.318021774291992, + "learning_rate": 6.55434570536677e-05, + "loss": 3.5704, + "step": 4131 + }, + { + "epoch": 6.04, + "grad_norm": 6.184370040893555, + "learning_rate": 6.54994296605541e-05, + "loss": 3.3746, + "step": 4132 + }, + { + "epoch": 6.04, + "grad_norm": 7.076589584350586, + "learning_rate": 6.545540985733473e-05, + "loss": 3.4426, + "step": 4133 + }, + { + "epoch": 6.05, + "grad_norm": 3.5989532470703125, + "learning_rate": 6.541139765369368e-05, + "loss": 3.4139, + "step": 4134 + }, + { + "epoch": 6.05, + "grad_norm": 4.95648717880249, + "learning_rate": 6.536739305931338e-05, + "loss": 3.392, + "step": 4135 + }, + { + "epoch": 6.05, + "grad_norm": 8.156216621398926, + "learning_rate": 6.53233960838745e-05, + "loss": 3.4219, + "step": 4136 + }, + { + "epoch": 6.05, + "grad_norm": 6.330258846282959, + "learning_rate": 6.527940673705605e-05, + "loss": 3.4077, + "step": 4137 + }, + { + "epoch": 6.05, + "grad_norm": 1.3029074668884277, + "learning_rate": 6.523542502853547e-05, + "loss": 3.422, + "step": 4138 + }, + { + "epoch": 6.05, + "grad_norm": 5.435974597930908, + "learning_rate": 6.519145096798842e-05, + "loss": 3.4704, + "step": 4139 + }, + { + "epoch": 6.05, + "grad_norm": 5.678780555725098, + "learning_rate": 6.514748456508885e-05, + "loss": 3.419, + "step": 4140 + }, + { + "epoch": 6.06, + "grad_norm": 3.2500386238098145, + "learning_rate": 6.510352582950909e-05, + "loss": 3.4207, + "step": 4141 + }, + { + "epoch": 6.06, + "grad_norm": 4.012285232543945, + "learning_rate": 6.505957477091981e-05, + "loss": 3.3755, + "step": 4142 + }, + { + "epoch": 6.06, + "grad_norm": 6.388498783111572, + "learning_rate": 6.501563139898991e-05, + "loss": 3.4086, + "step": 4143 + }, + { + "epoch": 6.06, + "grad_norm": 3.951239824295044, + "learning_rate": 6.497169572338661e-05, + "loss": 3.3432, + "step": 4144 + }, + { + "epoch": 6.06, + "grad_norm": 4.527097702026367, + "learning_rate": 6.492776775377556e-05, + "loss": 3.3961, + "step": 4145 + }, + { + "epoch": 6.06, + "grad_norm": 5.017838954925537, + "learning_rate": 6.488384749982053e-05, + "loss": 3.3483, + "step": 4146 + }, + { + "epoch": 6.07, + "grad_norm": 4.438302516937256, + "learning_rate": 6.483993497118371e-05, + "loss": 3.456, + "step": 4147 + }, + { + "epoch": 6.07, + "grad_norm": 0.9947657585144043, + "learning_rate": 6.479603017752559e-05, + "loss": 3.3646, + "step": 4148 + }, + { + "epoch": 6.07, + "grad_norm": 2.749354600906372, + "learning_rate": 6.47521331285049e-05, + "loss": 3.4993, + "step": 4149 + }, + { + "epoch": 6.07, + "grad_norm": 2.3780393600463867, + "learning_rate": 6.470824383377867e-05, + "loss": 3.3783, + "step": 4150 + }, + { + "epoch": 6.07, + "grad_norm": 2.840714454650879, + "learning_rate": 6.46643623030023e-05, + "loss": 3.3901, + "step": 4151 + }, + { + "epoch": 6.07, + "grad_norm": 2.8188705444335938, + "learning_rate": 6.462048854582944e-05, + "loss": 3.3313, + "step": 4152 + }, + { + "epoch": 6.07, + "grad_norm": 0.6622484922409058, + "learning_rate": 6.457662257191195e-05, + "loss": 3.3428, + "step": 4153 + }, + { + "epoch": 6.08, + "grad_norm": 2.970710515975952, + "learning_rate": 6.453276439090012e-05, + "loss": 3.3343, + "step": 4154 + }, + { + "epoch": 6.08, + "grad_norm": 2.88561749458313, + "learning_rate": 6.448891401244244e-05, + "loss": 3.3591, + "step": 4155 + }, + { + "epoch": 6.08, + "grad_norm": 2.3622708320617676, + "learning_rate": 6.444507144618566e-05, + "loss": 3.39, + "step": 4156 + }, + { + "epoch": 6.08, + "grad_norm": 1.4716769456863403, + "learning_rate": 6.440123670177492e-05, + "loss": 3.4068, + "step": 4157 + }, + { + "epoch": 6.08, + "grad_norm": 10.4777193069458, + "learning_rate": 6.435740978885353e-05, + "loss": 3.352, + "step": 4158 + }, + { + "epoch": 6.08, + "grad_norm": 6.331012725830078, + "learning_rate": 6.43135907170631e-05, + "loss": 3.3506, + "step": 4159 + }, + { + "epoch": 6.08, + "grad_norm": 4.355317115783691, + "learning_rate": 6.426977949604358e-05, + "loss": 3.3904, + "step": 4160 + }, + { + "epoch": 6.09, + "grad_norm": 1.3198426961898804, + "learning_rate": 6.422597613543309e-05, + "loss": 3.3279, + "step": 4161 + }, + { + "epoch": 6.09, + "grad_norm": 3.2531580924987793, + "learning_rate": 6.41821806448681e-05, + "loss": 3.427, + "step": 4162 + }, + { + "epoch": 6.09, + "grad_norm": 1.780542254447937, + "learning_rate": 6.413839303398337e-05, + "loss": 3.382, + "step": 4163 + }, + { + "epoch": 6.09, + "grad_norm": 4.514572620391846, + "learning_rate": 6.40946133124118e-05, + "loss": 3.3181, + "step": 4164 + }, + { + "epoch": 6.09, + "grad_norm": 4.696477890014648, + "learning_rate": 6.405084148978471e-05, + "loss": 3.3693, + "step": 4165 + }, + { + "epoch": 6.09, + "grad_norm": 1.9454889297485352, + "learning_rate": 6.400707757573152e-05, + "loss": 3.3362, + "step": 4166 + }, + { + "epoch": 6.1, + "grad_norm": 4.424253463745117, + "learning_rate": 6.396332157988007e-05, + "loss": 3.3042, + "step": 4167 + }, + { + "epoch": 6.1, + "grad_norm": 6.185297012329102, + "learning_rate": 6.39195735118564e-05, + "loss": 3.3837, + "step": 4168 + }, + { + "epoch": 6.1, + "grad_norm": 4.592702865600586, + "learning_rate": 6.387583338128471e-05, + "loss": 3.35, + "step": 4169 + }, + { + "epoch": 6.1, + "grad_norm": 1.6287086009979248, + "learning_rate": 6.38321011977876e-05, + "loss": 3.3304, + "step": 4170 + }, + { + "epoch": 6.1, + "grad_norm": 4.277772903442383, + "learning_rate": 6.378837697098586e-05, + "loss": 3.321, + "step": 4171 + }, + { + "epoch": 6.1, + "grad_norm": 2.2683122158050537, + "learning_rate": 6.374466071049847e-05, + "loss": 3.2938, + "step": 4172 + }, + { + "epoch": 6.1, + "grad_norm": 2.8517701625823975, + "learning_rate": 6.370095242594279e-05, + "loss": 3.338, + "step": 4173 + }, + { + "epoch": 6.11, + "grad_norm": 3.4475059509277344, + "learning_rate": 6.365725212693428e-05, + "loss": 3.3719, + "step": 4174 + }, + { + "epoch": 6.11, + "grad_norm": 1.0694806575775146, + "learning_rate": 6.361355982308675e-05, + "loss": 3.3108, + "step": 4175 + }, + { + "epoch": 6.11, + "grad_norm": 4.983922958374023, + "learning_rate": 6.356987552401222e-05, + "loss": 3.326, + "step": 4176 + }, + { + "epoch": 6.11, + "grad_norm": 5.746216773986816, + "learning_rate": 6.352619923932093e-05, + "loss": 3.3831, + "step": 4177 + }, + { + "epoch": 6.11, + "grad_norm": 2.469836473464966, + "learning_rate": 6.348253097862134e-05, + "loss": 3.3593, + "step": 4178 + }, + { + "epoch": 6.11, + "grad_norm": 3.8957202434539795, + "learning_rate": 6.343887075152024e-05, + "loss": 3.3594, + "step": 4179 + }, + { + "epoch": 6.11, + "grad_norm": 5.375302791595459, + "learning_rate": 6.339521856762254e-05, + "loss": 3.3357, + "step": 4180 + }, + { + "epoch": 6.12, + "grad_norm": 4.205276966094971, + "learning_rate": 6.335157443653142e-05, + "loss": 3.3987, + "step": 4181 + }, + { + "epoch": 6.12, + "grad_norm": 1.7369492053985596, + "learning_rate": 6.330793836784836e-05, + "loss": 3.3606, + "step": 4182 + }, + { + "epoch": 6.12, + "grad_norm": 2.639575481414795, + "learning_rate": 6.326431037117296e-05, + "loss": 3.4109, + "step": 4183 + }, + { + "epoch": 6.12, + "grad_norm": 0.6227443218231201, + "learning_rate": 6.322069045610305e-05, + "loss": 3.3122, + "step": 4184 + }, + { + "epoch": 6.12, + "grad_norm": 2.4166953563690186, + "learning_rate": 6.317707863223477e-05, + "loss": 3.339, + "step": 4185 + }, + { + "epoch": 6.12, + "grad_norm": 1.3330849409103394, + "learning_rate": 6.313347490916243e-05, + "loss": 3.2775, + "step": 4186 + }, + { + "epoch": 6.12, + "grad_norm": 3.1274631023406982, + "learning_rate": 6.308987929647852e-05, + "loss": 3.2931, + "step": 4187 + }, + { + "epoch": 6.13, + "grad_norm": 2.9381067752838135, + "learning_rate": 6.30462918037738e-05, + "loss": 3.347, + "step": 4188 + }, + { + "epoch": 6.13, + "grad_norm": 0.6513715982437134, + "learning_rate": 6.300271244063726e-05, + "loss": 3.3084, + "step": 4189 + }, + { + "epoch": 6.13, + "grad_norm": 1.9384881258010864, + "learning_rate": 6.295914121665602e-05, + "loss": 3.3508, + "step": 4190 + }, + { + "epoch": 6.13, + "grad_norm": 1.33826744556427, + "learning_rate": 6.291557814141547e-05, + "loss": 3.3385, + "step": 4191 + }, + { + "epoch": 6.13, + "grad_norm": 1.159730076789856, + "learning_rate": 6.287202322449922e-05, + "loss": 3.3025, + "step": 4192 + }, + { + "epoch": 6.13, + "grad_norm": 15.805403709411621, + "learning_rate": 6.282847647548904e-05, + "loss": 3.3027, + "step": 4193 + }, + { + "epoch": 6.14, + "grad_norm": 4.59617280960083, + "learning_rate": 6.278493790396489e-05, + "loss": 3.3538, + "step": 4194 + }, + { + "epoch": 6.14, + "grad_norm": 2.9616920948028564, + "learning_rate": 6.274140751950501e-05, + "loss": 3.3287, + "step": 4195 + }, + { + "epoch": 6.14, + "grad_norm": 2.299812078475952, + "learning_rate": 6.269788533168581e-05, + "loss": 3.3451, + "step": 4196 + }, + { + "epoch": 6.14, + "grad_norm": 2.9177210330963135, + "learning_rate": 6.26543713500818e-05, + "loss": 3.3583, + "step": 4197 + }, + { + "epoch": 6.14, + "grad_norm": 0.9003336429595947, + "learning_rate": 6.261086558426586e-05, + "loss": 3.2672, + "step": 4198 + }, + { + "epoch": 6.14, + "grad_norm": 6.0708184242248535, + "learning_rate": 6.25673680438089e-05, + "loss": 3.2929, + "step": 4199 + }, + { + "epoch": 6.14, + "grad_norm": 6.111046314239502, + "learning_rate": 6.25238787382801e-05, + "loss": 3.3031, + "step": 4200 + }, + { + "epoch": 6.15, + "grad_norm": 4.454471588134766, + "learning_rate": 6.248039767724683e-05, + "loss": 3.3293, + "step": 4201 + }, + { + "epoch": 6.15, + "grad_norm": 0.9689366221427917, + "learning_rate": 6.243692487027464e-05, + "loss": 3.2844, + "step": 4202 + }, + { + "epoch": 6.15, + "grad_norm": 4.885942459106445, + "learning_rate": 6.239346032692721e-05, + "loss": 3.3346, + "step": 4203 + }, + { + "epoch": 6.15, + "grad_norm": 3.6569879055023193, + "learning_rate": 6.23500040567665e-05, + "loss": 3.2936, + "step": 4204 + }, + { + "epoch": 6.15, + "grad_norm": 1.7212425470352173, + "learning_rate": 6.23065560693526e-05, + "loss": 3.3124, + "step": 4205 + }, + { + "epoch": 6.15, + "grad_norm": 2.9631359577178955, + "learning_rate": 6.226311637424371e-05, + "loss": 3.3716, + "step": 4206 + }, + { + "epoch": 6.15, + "grad_norm": 1.4877088069915771, + "learning_rate": 6.221968498099636e-05, + "loss": 3.2881, + "step": 4207 + }, + { + "epoch": 6.16, + "grad_norm": 5.8026533126831055, + "learning_rate": 6.21762618991651e-05, + "loss": 3.2705, + "step": 4208 + }, + { + "epoch": 6.16, + "grad_norm": 6.365312576293945, + "learning_rate": 6.213284713830271e-05, + "loss": 3.3174, + "step": 4209 + }, + { + "epoch": 6.16, + "grad_norm": 5.353916645050049, + "learning_rate": 6.208944070796022e-05, + "loss": 3.3226, + "step": 4210 + }, + { + "epoch": 6.16, + "grad_norm": 1.4095226526260376, + "learning_rate": 6.204604261768669e-05, + "loss": 3.3037, + "step": 4211 + }, + { + "epoch": 6.16, + "grad_norm": 2.51399827003479, + "learning_rate": 6.200265287702942e-05, + "loss": 3.3118, + "step": 4212 + }, + { + "epoch": 6.16, + "grad_norm": 1.4615494012832642, + "learning_rate": 6.19592714955339e-05, + "loss": 3.2963, + "step": 4213 + }, + { + "epoch": 6.16, + "grad_norm": 2.4555823802948, + "learning_rate": 6.191589848274368e-05, + "loss": 3.275, + "step": 4214 + }, + { + "epoch": 6.17, + "grad_norm": 3.085233211517334, + "learning_rate": 6.187253384820061e-05, + "loss": 3.2726, + "step": 4215 + }, + { + "epoch": 6.17, + "grad_norm": 0.8324146270751953, + "learning_rate": 6.182917760144453e-05, + "loss": 3.3158, + "step": 4216 + }, + { + "epoch": 6.17, + "grad_norm": 1.7136584520339966, + "learning_rate": 6.178582975201359e-05, + "loss": 3.2768, + "step": 4217 + }, + { + "epoch": 6.17, + "grad_norm": 1.937159538269043, + "learning_rate": 6.1742490309444e-05, + "loss": 3.3338, + "step": 4218 + }, + { + "epoch": 6.17, + "grad_norm": 1.2840466499328613, + "learning_rate": 6.169915928327012e-05, + "loss": 3.2964, + "step": 4219 + }, + { + "epoch": 6.17, + "grad_norm": 2.5718352794647217, + "learning_rate": 6.165583668302457e-05, + "loss": 3.287, + "step": 4220 + }, + { + "epoch": 6.18, + "grad_norm": 4.60259485244751, + "learning_rate": 6.161252251823795e-05, + "loss": 3.2494, + "step": 4221 + }, + { + "epoch": 6.18, + "grad_norm": 1.160286545753479, + "learning_rate": 6.15692167984391e-05, + "loss": 3.3453, + "step": 4222 + }, + { + "epoch": 6.18, + "grad_norm": 2.802889823913574, + "learning_rate": 6.152591953315505e-05, + "loss": 3.3133, + "step": 4223 + }, + { + "epoch": 6.18, + "grad_norm": 1.775070071220398, + "learning_rate": 6.148263073191082e-05, + "loss": 3.2936, + "step": 4224 + }, + { + "epoch": 6.18, + "grad_norm": 1.9516983032226562, + "learning_rate": 6.14393504042297e-05, + "loss": 3.3066, + "step": 4225 + }, + { + "epoch": 6.18, + "eval_loss": 3.107570171356201, + "eval_runtime": 1177.648, + "eval_samples_per_second": 14.556, + "eval_steps_per_second": 0.91, + "step": 4225 + }, + { + "epoch": 6.18, + "grad_norm": 1.7126461267471313, + "learning_rate": 6.13960785596331e-05, + "loss": 3.2474, + "step": 4226 + }, + { + "epoch": 6.18, + "grad_norm": 1.2491967678070068, + "learning_rate": 6.135281520764049e-05, + "loss": 3.3243, + "step": 4227 + }, + { + "epoch": 6.19, + "grad_norm": 1.4934719800949097, + "learning_rate": 6.130956035776951e-05, + "loss": 3.3423, + "step": 4228 + }, + { + "epoch": 6.19, + "grad_norm": 1.9269001483917236, + "learning_rate": 6.126631401953598e-05, + "loss": 3.2633, + "step": 4229 + }, + { + "epoch": 6.19, + "grad_norm": 1.3831140995025635, + "learning_rate": 6.12230762024538e-05, + "loss": 3.286, + "step": 4230 + }, + { + "epoch": 6.19, + "grad_norm": 2.75412917137146, + "learning_rate": 6.117984691603493e-05, + "loss": 3.2788, + "step": 4231 + }, + { + "epoch": 6.19, + "grad_norm": 3.342359781265259, + "learning_rate": 6.113662616978961e-05, + "loss": 3.2737, + "step": 4232 + }, + { + "epoch": 6.19, + "grad_norm": 0.9600224494934082, + "learning_rate": 6.109341397322611e-05, + "loss": 3.2899, + "step": 4233 + }, + { + "epoch": 6.19, + "grad_norm": 2.350534200668335, + "learning_rate": 6.105021033585074e-05, + "loss": 3.2768, + "step": 4234 + }, + { + "epoch": 6.2, + "grad_norm": 1.1939469575881958, + "learning_rate": 6.10070152671681e-05, + "loss": 3.2706, + "step": 4235 + }, + { + "epoch": 6.2, + "grad_norm": 4.221333980560303, + "learning_rate": 6.096382877668079e-05, + "loss": 3.2853, + "step": 4236 + }, + { + "epoch": 6.2, + "grad_norm": 4.897353649139404, + "learning_rate": 6.092065087388953e-05, + "loss": 3.3084, + "step": 4237 + }, + { + "epoch": 6.2, + "grad_norm": 2.3200554847717285, + "learning_rate": 6.087748156829316e-05, + "loss": 3.2422, + "step": 4238 + }, + { + "epoch": 6.2, + "grad_norm": 4.042237758636475, + "learning_rate": 6.08343208693887e-05, + "loss": 3.2441, + "step": 4239 + }, + { + "epoch": 6.2, + "grad_norm": 4.729053974151611, + "learning_rate": 6.079116878667115e-05, + "loss": 3.2911, + "step": 4240 + }, + { + "epoch": 6.2, + "grad_norm": 3.419166088104248, + "learning_rate": 6.074802532963368e-05, + "loss": 3.2984, + "step": 4241 + }, + { + "epoch": 6.21, + "grad_norm": 2.1573331356048584, + "learning_rate": 6.0704890507767597e-05, + "loss": 3.2494, + "step": 4242 + }, + { + "epoch": 6.21, + "grad_norm": 3.604609489440918, + "learning_rate": 6.0661764330562276e-05, + "loss": 3.2961, + "step": 4243 + }, + { + "epoch": 6.21, + "grad_norm": 1.5325900316238403, + "learning_rate": 6.061864680750513e-05, + "loss": 3.2401, + "step": 4244 + }, + { + "epoch": 6.21, + "grad_norm": 3.4049322605133057, + "learning_rate": 6.0575537948081793e-05, + "loss": 3.292, + "step": 4245 + }, + { + "epoch": 6.21, + "grad_norm": 4.517602920532227, + "learning_rate": 6.053243776177591e-05, + "loss": 3.2811, + "step": 4246 + }, + { + "epoch": 6.21, + "grad_norm": 2.2732179164886475, + "learning_rate": 6.0489346258069177e-05, + "loss": 3.2883, + "step": 4247 + }, + { + "epoch": 6.22, + "grad_norm": 5.137287139892578, + "learning_rate": 6.044626344644151e-05, + "loss": 3.2465, + "step": 4248 + }, + { + "epoch": 6.22, + "grad_norm": 5.510692596435547, + "learning_rate": 6.040318933637081e-05, + "loss": 3.3609, + "step": 4249 + }, + { + "epoch": 6.22, + "grad_norm": 4.384718894958496, + "learning_rate": 6.036012393733306e-05, + "loss": 3.2768, + "step": 4250 + }, + { + "epoch": 6.22, + "grad_norm": 0.5632131695747375, + "learning_rate": 6.031706725880243e-05, + "loss": 3.2637, + "step": 4251 + }, + { + "epoch": 6.22, + "grad_norm": 4.083331108093262, + "learning_rate": 6.027401931025103e-05, + "loss": 3.2579, + "step": 4252 + }, + { + "epoch": 6.22, + "grad_norm": 4.434952735900879, + "learning_rate": 6.0230980101149136e-05, + "loss": 3.2995, + "step": 4253 + }, + { + "epoch": 6.22, + "grad_norm": 1.9891043901443481, + "learning_rate": 6.0187949640965146e-05, + "loss": 3.2652, + "step": 4254 + }, + { + "epoch": 6.23, + "grad_norm": 3.985672950744629, + "learning_rate": 6.014492793916539e-05, + "loss": 3.256, + "step": 4255 + }, + { + "epoch": 6.23, + "grad_norm": 5.051290988922119, + "learning_rate": 6.010191500521439e-05, + "loss": 3.2773, + "step": 4256 + }, + { + "epoch": 6.23, + "grad_norm": 3.2940456867218018, + "learning_rate": 6.005891084857473e-05, + "loss": 3.3033, + "step": 4257 + }, + { + "epoch": 6.23, + "grad_norm": 2.565002202987671, + "learning_rate": 6.001591547870701e-05, + "loss": 3.2838, + "step": 4258 + }, + { + "epoch": 6.23, + "grad_norm": 3.5721094608306885, + "learning_rate": 5.997292890506989e-05, + "loss": 3.2499, + "step": 4259 + }, + { + "epoch": 6.23, + "grad_norm": 1.7328767776489258, + "learning_rate": 5.992995113712019e-05, + "loss": 3.2925, + "step": 4260 + }, + { + "epoch": 6.23, + "grad_norm": 3.9454033374786377, + "learning_rate": 5.9886982184312725e-05, + "loss": 3.2744, + "step": 4261 + }, + { + "epoch": 6.24, + "grad_norm": 4.74951171875, + "learning_rate": 5.984402205610031e-05, + "loss": 3.2632, + "step": 4262 + }, + { + "epoch": 6.24, + "grad_norm": 2.6633002758026123, + "learning_rate": 5.980107076193393e-05, + "loss": 3.2585, + "step": 4263 + }, + { + "epoch": 6.24, + "grad_norm": 2.6646337509155273, + "learning_rate": 5.975812831126261e-05, + "loss": 3.3006, + "step": 4264 + }, + { + "epoch": 6.24, + "grad_norm": 3.7945520877838135, + "learning_rate": 5.971519471353335e-05, + "loss": 3.2882, + "step": 4265 + }, + { + "epoch": 6.24, + "grad_norm": 1.7113827466964722, + "learning_rate": 5.9672269978191266e-05, + "loss": 3.2554, + "step": 4266 + }, + { + "epoch": 6.24, + "grad_norm": 3.288426399230957, + "learning_rate": 5.9629354114679556e-05, + "loss": 3.329, + "step": 4267 + }, + { + "epoch": 6.25, + "grad_norm": 5.059687614440918, + "learning_rate": 5.958644713243937e-05, + "loss": 3.2254, + "step": 4268 + }, + { + "epoch": 6.25, + "grad_norm": 2.444850206375122, + "learning_rate": 5.9543549040909954e-05, + "loss": 3.2804, + "step": 4269 + }, + { + "epoch": 6.25, + "grad_norm": 2.6846628189086914, + "learning_rate": 5.9500659849528664e-05, + "loss": 3.2151, + "step": 4270 + }, + { + "epoch": 6.25, + "grad_norm": 3.8976359367370605, + "learning_rate": 5.945777956773077e-05, + "loss": 3.3193, + "step": 4271 + }, + { + "epoch": 6.25, + "grad_norm": 2.4183270931243896, + "learning_rate": 5.941490820494965e-05, + "loss": 3.261, + "step": 4272 + }, + { + "epoch": 6.25, + "grad_norm": 3.022622585296631, + "learning_rate": 5.937204577061677e-05, + "loss": 3.2569, + "step": 4273 + }, + { + "epoch": 6.25, + "grad_norm": 3.768277168273926, + "learning_rate": 5.932919227416156e-05, + "loss": 3.2582, + "step": 4274 + }, + { + "epoch": 6.26, + "grad_norm": 2.1014842987060547, + "learning_rate": 5.9286347725011424e-05, + "loss": 3.236, + "step": 4275 + }, + { + "epoch": 6.26, + "grad_norm": 2.9817419052124023, + "learning_rate": 5.9243512132591985e-05, + "loss": 3.2667, + "step": 4276 + }, + { + "epoch": 6.26, + "grad_norm": 4.514947891235352, + "learning_rate": 5.9200685506326734e-05, + "loss": 3.3089, + "step": 4277 + }, + { + "epoch": 6.26, + "grad_norm": 2.3111038208007812, + "learning_rate": 5.915786785563722e-05, + "loss": 3.253, + "step": 4278 + }, + { + "epoch": 6.26, + "grad_norm": 2.7890591621398926, + "learning_rate": 5.911505918994309e-05, + "loss": 3.2643, + "step": 4279 + }, + { + "epoch": 6.26, + "grad_norm": 3.4461700916290283, + "learning_rate": 5.907225951866195e-05, + "loss": 3.2625, + "step": 4280 + }, + { + "epoch": 6.26, + "grad_norm": 2.0314648151397705, + "learning_rate": 5.902946885120939e-05, + "loss": 3.2712, + "step": 4281 + }, + { + "epoch": 6.27, + "grad_norm": 2.992061138153076, + "learning_rate": 5.8986687196999135e-05, + "loss": 3.2365, + "step": 4282 + }, + { + "epoch": 6.27, + "grad_norm": 4.312194347381592, + "learning_rate": 5.8943914565442856e-05, + "loss": 3.2867, + "step": 4283 + }, + { + "epoch": 6.27, + "grad_norm": 2.6370973587036133, + "learning_rate": 5.8901150965950194e-05, + "loss": 3.239, + "step": 4284 + }, + { + "epoch": 6.27, + "grad_norm": 3.0366334915161133, + "learning_rate": 5.885839640792892e-05, + "loss": 3.2651, + "step": 4285 + }, + { + "epoch": 6.27, + "grad_norm": 4.975083827972412, + "learning_rate": 5.881565090078469e-05, + "loss": 3.2896, + "step": 4286 + }, + { + "epoch": 6.27, + "grad_norm": 2.4393575191497803, + "learning_rate": 5.8772914453921276e-05, + "loss": 3.2273, + "step": 4287 + }, + { + "epoch": 6.27, + "grad_norm": 2.5122718811035156, + "learning_rate": 5.873018707674036e-05, + "loss": 3.2617, + "step": 4288 + }, + { + "epoch": 6.28, + "grad_norm": 3.912677526473999, + "learning_rate": 5.868746877864173e-05, + "loss": 3.3112, + "step": 4289 + }, + { + "epoch": 6.28, + "grad_norm": 1.8555430173873901, + "learning_rate": 5.864475956902312e-05, + "loss": 3.322, + "step": 4290 + }, + { + "epoch": 6.28, + "grad_norm": 11.267170906066895, + "learning_rate": 5.860205945728021e-05, + "loss": 3.2759, + "step": 4291 + }, + { + "epoch": 6.28, + "grad_norm": 7.718640327453613, + "learning_rate": 5.855936845280682e-05, + "loss": 3.2407, + "step": 4292 + }, + { + "epoch": 6.28, + "grad_norm": 8.495302200317383, + "learning_rate": 5.851668656499465e-05, + "loss": 3.3493, + "step": 4293 + }, + { + "epoch": 6.28, + "grad_norm": 4.235414981842041, + "learning_rate": 5.8474013803233394e-05, + "loss": 3.2929, + "step": 4294 + }, + { + "epoch": 6.29, + "grad_norm": 2.0188095569610596, + "learning_rate": 5.843135017691085e-05, + "loss": 3.2134, + "step": 4295 + }, + { + "epoch": 6.29, + "grad_norm": 4.253573417663574, + "learning_rate": 5.8388695695412666e-05, + "loss": 3.2903, + "step": 4296 + }, + { + "epoch": 6.29, + "grad_norm": 2.978238582611084, + "learning_rate": 5.834605036812255e-05, + "loss": 3.2783, + "step": 4297 + }, + { + "epoch": 6.29, + "grad_norm": 1.3776922225952148, + "learning_rate": 5.830341420442221e-05, + "loss": 3.2773, + "step": 4298 + }, + { + "epoch": 6.29, + "grad_norm": 2.41166353225708, + "learning_rate": 5.82607872136913e-05, + "loss": 3.2214, + "step": 4299 + }, + { + "epoch": 6.29, + "grad_norm": 0.7983728647232056, + "learning_rate": 5.821816940530749e-05, + "loss": 3.3103, + "step": 4300 + }, + { + "epoch": 6.29, + "grad_norm": 1.7378346920013428, + "learning_rate": 5.8175560788646414e-05, + "loss": 3.2548, + "step": 4301 + }, + { + "epoch": 6.3, + "grad_norm": 0.7058059573173523, + "learning_rate": 5.813296137308164e-05, + "loss": 3.2865, + "step": 4302 + }, + { + "epoch": 6.3, + "grad_norm": 2.132636070251465, + "learning_rate": 5.809037116798479e-05, + "loss": 3.2757, + "step": 4303 + }, + { + "epoch": 6.3, + "grad_norm": 1.359208345413208, + "learning_rate": 5.804779018272544e-05, + "loss": 3.2997, + "step": 4304 + }, + { + "epoch": 6.3, + "grad_norm": 2.7147774696350098, + "learning_rate": 5.800521842667107e-05, + "loss": 3.2789, + "step": 4305 + }, + { + "epoch": 6.3, + "grad_norm": 1.9574331045150757, + "learning_rate": 5.7962655909187226e-05, + "loss": 3.2634, + "step": 4306 + }, + { + "epoch": 6.3, + "grad_norm": 1.1457573175430298, + "learning_rate": 5.792010263963735e-05, + "loss": 3.2802, + "step": 4307 + }, + { + "epoch": 6.3, + "grad_norm": 0.9981921315193176, + "learning_rate": 5.7877558627382886e-05, + "loss": 3.3431, + "step": 4308 + }, + { + "epoch": 6.31, + "grad_norm": 1.708836317062378, + "learning_rate": 5.783502388178328e-05, + "loss": 3.2667, + "step": 4309 + }, + { + "epoch": 6.31, + "grad_norm": 1.0355262756347656, + "learning_rate": 5.7792498412195764e-05, + "loss": 3.2917, + "step": 4310 + }, + { + "epoch": 6.31, + "grad_norm": 3.3587546348571777, + "learning_rate": 5.774998222797577e-05, + "loss": 3.2507, + "step": 4311 + }, + { + "epoch": 6.31, + "grad_norm": 3.278871774673462, + "learning_rate": 5.770747533847658e-05, + "loss": 3.2427, + "step": 4312 + }, + { + "epoch": 6.31, + "grad_norm": 0.9090314507484436, + "learning_rate": 5.766497775304931e-05, + "loss": 3.2401, + "step": 4313 + }, + { + "epoch": 6.31, + "grad_norm": 3.0170810222625732, + "learning_rate": 5.7622489481043254e-05, + "loss": 3.2708, + "step": 4314 + }, + { + "epoch": 6.31, + "grad_norm": 3.1690783500671387, + "learning_rate": 5.758001053180555e-05, + "loss": 3.2354, + "step": 4315 + }, + { + "epoch": 6.32, + "grad_norm": 0.4450177550315857, + "learning_rate": 5.753754091468115e-05, + "loss": 3.2435, + "step": 4316 + }, + { + "epoch": 6.32, + "grad_norm": 3.6252622604370117, + "learning_rate": 5.749508063901328e-05, + "loss": 3.2594, + "step": 4317 + }, + { + "epoch": 6.32, + "grad_norm": 5.237750053405762, + "learning_rate": 5.745262971414276e-05, + "loss": 3.2698, + "step": 4318 + }, + { + "epoch": 6.32, + "grad_norm": 1.0414494276046753, + "learning_rate": 5.741018814940853e-05, + "loss": 3.2509, + "step": 4319 + }, + { + "epoch": 6.32, + "grad_norm": 2.6525533199310303, + "learning_rate": 5.736775595414756e-05, + "loss": 3.2609, + "step": 4320 + }, + { + "epoch": 6.32, + "grad_norm": 3.908628225326538, + "learning_rate": 5.7325333137694525e-05, + "loss": 3.2479, + "step": 4321 + }, + { + "epoch": 6.33, + "grad_norm": 1.5807950496673584, + "learning_rate": 5.728291970938218e-05, + "loss": 3.2861, + "step": 4322 + }, + { + "epoch": 6.33, + "grad_norm": 3.6872596740722656, + "learning_rate": 5.72405156785413e-05, + "loss": 3.2316, + "step": 4323 + }, + { + "epoch": 6.33, + "grad_norm": 4.621198654174805, + "learning_rate": 5.719812105450035e-05, + "loss": 3.245, + "step": 4324 + }, + { + "epoch": 6.33, + "grad_norm": 1.5106016397476196, + "learning_rate": 5.715573584658589e-05, + "loss": 3.215, + "step": 4325 + }, + { + "epoch": 6.33, + "grad_norm": 3.5040335655212402, + "learning_rate": 5.711336006412251e-05, + "loss": 3.2626, + "step": 4326 + }, + { + "epoch": 6.33, + "grad_norm": 4.461084842681885, + "learning_rate": 5.707099371643245e-05, + "loss": 3.2804, + "step": 4327 + }, + { + "epoch": 6.33, + "grad_norm": 2.81062912940979, + "learning_rate": 5.7028636812836035e-05, + "loss": 3.2476, + "step": 4328 + }, + { + "epoch": 6.34, + "grad_norm": 2.1609275341033936, + "learning_rate": 5.6986289362651624e-05, + "loss": 3.2922, + "step": 4329 + }, + { + "epoch": 6.34, + "grad_norm": 39.97700881958008, + "learning_rate": 5.694395137519525e-05, + "loss": 3.262, + "step": 4330 + }, + { + "epoch": 6.34, + "grad_norm": 4.451782703399658, + "learning_rate": 5.690162285978102e-05, + "loss": 3.3011, + "step": 4331 + }, + { + "epoch": 6.34, + "grad_norm": 1.8545382022857666, + "learning_rate": 5.6859303825720935e-05, + "loss": 3.2684, + "step": 4332 + }, + { + "epoch": 6.34, + "grad_norm": 3.6993982791900635, + "learning_rate": 5.6816994282324875e-05, + "loss": 3.2664, + "step": 4333 + }, + { + "epoch": 6.34, + "grad_norm": 4.563552379608154, + "learning_rate": 5.67746942389007e-05, + "loss": 3.2004, + "step": 4334 + }, + { + "epoch": 6.34, + "grad_norm": 3.6278369426727295, + "learning_rate": 5.6732403704754086e-05, + "loss": 3.1965, + "step": 4335 + }, + { + "epoch": 6.35, + "grad_norm": 1.3190866708755493, + "learning_rate": 5.66901226891887e-05, + "loss": 3.2505, + "step": 4336 + }, + { + "epoch": 6.35, + "grad_norm": 3.1047582626342773, + "learning_rate": 5.664785120150607e-05, + "loss": 3.3744, + "step": 4337 + }, + { + "epoch": 6.35, + "grad_norm": 1.4010859727859497, + "learning_rate": 5.660558925100565e-05, + "loss": 3.2555, + "step": 4338 + }, + { + "epoch": 6.35, + "grad_norm": 2.454122304916382, + "learning_rate": 5.656333684698476e-05, + "loss": 3.2401, + "step": 4339 + }, + { + "epoch": 6.35, + "grad_norm": 2.7071664333343506, + "learning_rate": 5.652109399873873e-05, + "loss": 3.2436, + "step": 4340 + }, + { + "epoch": 6.35, + "grad_norm": 0.4050593674182892, + "learning_rate": 5.647886071556054e-05, + "loss": 3.2843, + "step": 4341 + }, + { + "epoch": 6.35, + "grad_norm": 3.2467408180236816, + "learning_rate": 5.6436637006741375e-05, + "loss": 3.2526, + "step": 4342 + }, + { + "epoch": 6.36, + "grad_norm": 3.083259344100952, + "learning_rate": 5.639442288157015e-05, + "loss": 3.2539, + "step": 4343 + }, + { + "epoch": 6.36, + "grad_norm": 1.2270174026489258, + "learning_rate": 5.635221834933359e-05, + "loss": 3.2435, + "step": 4344 + }, + { + "epoch": 6.36, + "grad_norm": 1.6782840490341187, + "learning_rate": 5.63100234193165e-05, + "loss": 3.2838, + "step": 4345 + }, + { + "epoch": 6.36, + "grad_norm": 1.8353314399719238, + "learning_rate": 5.6267838100801516e-05, + "loss": 3.2793, + "step": 4346 + }, + { + "epoch": 6.36, + "grad_norm": 1.642630934715271, + "learning_rate": 5.622566240306899e-05, + "loss": 3.261, + "step": 4347 + }, + { + "epoch": 6.36, + "grad_norm": 1.191734790802002, + "learning_rate": 5.618349633539741e-05, + "loss": 3.2822, + "step": 4348 + }, + { + "epoch": 6.37, + "grad_norm": 0.7267574071884155, + "learning_rate": 5.614133990706304e-05, + "loss": 3.2482, + "step": 4349 + }, + { + "epoch": 6.37, + "grad_norm": 0.9115816950798035, + "learning_rate": 5.6099193127339864e-05, + "loss": 3.2347, + "step": 4350 + }, + { + "epoch": 6.37, + "grad_norm": 1.4531807899475098, + "learning_rate": 5.6057056005500085e-05, + "loss": 3.2186, + "step": 4351 + }, + { + "epoch": 6.37, + "grad_norm": 1.0633835792541504, + "learning_rate": 5.601492855081345e-05, + "loss": 3.2526, + "step": 4352 + }, + { + "epoch": 6.37, + "grad_norm": 0.716886579990387, + "learning_rate": 5.5972810772547724e-05, + "loss": 3.2448, + "step": 4353 + }, + { + "epoch": 6.37, + "grad_norm": 0.9413835406303406, + "learning_rate": 5.5930702679968636e-05, + "loss": 3.2125, + "step": 4354 + }, + { + "epoch": 6.37, + "grad_norm": 1.2338863611221313, + "learning_rate": 5.58886042823396e-05, + "loss": 3.2228, + "step": 4355 + }, + { + "epoch": 6.38, + "grad_norm": 2.2373838424682617, + "learning_rate": 5.584651558892193e-05, + "loss": 3.2777, + "step": 4356 + }, + { + "epoch": 6.38, + "grad_norm": 1.3574821949005127, + "learning_rate": 5.580443660897503e-05, + "loss": 3.2371, + "step": 4357 + }, + { + "epoch": 6.38, + "grad_norm": 1.0699307918548584, + "learning_rate": 5.576236735175584e-05, + "loss": 3.2889, + "step": 4358 + }, + { + "epoch": 6.38, + "grad_norm": 0.737332284450531, + "learning_rate": 5.572030782651938e-05, + "loss": 3.2533, + "step": 4359 + }, + { + "epoch": 6.38, + "grad_norm": 1.1307657957077026, + "learning_rate": 5.567825804251843e-05, + "loss": 3.2832, + "step": 4360 + }, + { + "epoch": 6.38, + "grad_norm": 0.6265879273414612, + "learning_rate": 5.563621800900369e-05, + "loss": 3.2663, + "step": 4361 + }, + { + "epoch": 6.38, + "grad_norm": 2.2958106994628906, + "learning_rate": 5.559418773522367e-05, + "loss": 3.2426, + "step": 4362 + }, + { + "epoch": 6.39, + "grad_norm": 1.0793524980545044, + "learning_rate": 5.555216723042476e-05, + "loss": 3.2336, + "step": 4363 + }, + { + "epoch": 6.39, + "grad_norm": 1.8671746253967285, + "learning_rate": 5.551015650385119e-05, + "loss": 3.2449, + "step": 4364 + }, + { + "epoch": 6.39, + "grad_norm": 1.5960359573364258, + "learning_rate": 5.5468155564745e-05, + "loss": 3.2135, + "step": 4365 + }, + { + "epoch": 6.39, + "grad_norm": 2.3406100273132324, + "learning_rate": 5.542616442234618e-05, + "loss": 3.2034, + "step": 4366 + }, + { + "epoch": 6.39, + "grad_norm": 1.8202327489852905, + "learning_rate": 5.538418308589246e-05, + "loss": 3.2608, + "step": 4367 + }, + { + "epoch": 6.39, + "grad_norm": 1.2679892778396606, + "learning_rate": 5.534221156461945e-05, + "loss": 3.2336, + "step": 4368 + }, + { + "epoch": 6.4, + "grad_norm": 1.0765639543533325, + "learning_rate": 5.530024986776061e-05, + "loss": 3.2034, + "step": 4369 + }, + { + "epoch": 6.4, + "grad_norm": 2.065032720565796, + "learning_rate": 5.5258298004547246e-05, + "loss": 3.2323, + "step": 4370 + }, + { + "epoch": 6.4, + "grad_norm": 1.7677899599075317, + "learning_rate": 5.521635598420847e-05, + "loss": 3.2416, + "step": 4371 + }, + { + "epoch": 6.4, + "grad_norm": 1.2142525911331177, + "learning_rate": 5.5174423815971254e-05, + "loss": 3.2783, + "step": 4372 + }, + { + "epoch": 6.4, + "grad_norm": 1.2158989906311035, + "learning_rate": 5.513250150906039e-05, + "loss": 3.2203, + "step": 4373 + }, + { + "epoch": 6.4, + "grad_norm": 2.0104503631591797, + "learning_rate": 5.509058907269854e-05, + "loss": 3.2627, + "step": 4374 + }, + { + "epoch": 6.4, + "grad_norm": 1.794409155845642, + "learning_rate": 5.504868651610604e-05, + "loss": 3.2247, + "step": 4375 + }, + { + "epoch": 6.41, + "grad_norm": 1.241929531097412, + "learning_rate": 5.5006793848501314e-05, + "loss": 3.2352, + "step": 4376 + }, + { + "epoch": 6.41, + "grad_norm": 1.3682582378387451, + "learning_rate": 5.4964911079100446e-05, + "loss": 3.2834, + "step": 4377 + }, + { + "epoch": 6.41, + "grad_norm": 1.42366623878479, + "learning_rate": 5.492303821711724e-05, + "loss": 3.2356, + "step": 4378 + }, + { + "epoch": 6.41, + "grad_norm": 0.9967645406723022, + "learning_rate": 5.488117527176357e-05, + "loss": 3.2623, + "step": 4379 + }, + { + "epoch": 6.41, + "grad_norm": 1.6339219808578491, + "learning_rate": 5.4839322252249016e-05, + "loss": 3.2162, + "step": 4380 + }, + { + "epoch": 6.41, + "grad_norm": 0.9269053936004639, + "learning_rate": 5.479747916778088e-05, + "loss": 3.2236, + "step": 4381 + }, + { + "epoch": 6.41, + "grad_norm": 3.181241989135742, + "learning_rate": 5.4755646027564336e-05, + "loss": 3.2464, + "step": 4382 + }, + { + "epoch": 6.42, + "grad_norm": 3.343838930130005, + "learning_rate": 5.4713822840802544e-05, + "loss": 3.227, + "step": 4383 + }, + { + "epoch": 6.42, + "grad_norm": 0.8503923416137695, + "learning_rate": 5.467200961669619e-05, + "loss": 3.2975, + "step": 4384 + }, + { + "epoch": 6.42, + "grad_norm": 2.970435380935669, + "learning_rate": 5.4630206364443936e-05, + "loss": 3.2325, + "step": 4385 + }, + { + "epoch": 6.42, + "grad_norm": 3.7299089431762695, + "learning_rate": 5.4588413093242244e-05, + "loss": 3.1893, + "step": 4386 + }, + { + "epoch": 6.42, + "grad_norm": 1.3558520078659058, + "learning_rate": 5.454662981228532e-05, + "loss": 3.2106, + "step": 4387 + }, + { + "epoch": 6.42, + "grad_norm": 3.116187572479248, + "learning_rate": 5.450485653076524e-05, + "loss": 3.2292, + "step": 4388 + }, + { + "epoch": 6.42, + "grad_norm": 3.9433794021606445, + "learning_rate": 5.446309325787181e-05, + "loss": 3.2971, + "step": 4389 + }, + { + "epoch": 6.43, + "grad_norm": 2.001657247543335, + "learning_rate": 5.44213400027927e-05, + "loss": 3.2795, + "step": 4390 + }, + { + "epoch": 6.43, + "grad_norm": 2.557114601135254, + "learning_rate": 5.4379596774713334e-05, + "loss": 3.2177, + "step": 4391 + }, + { + "epoch": 6.43, + "grad_norm": 3.371882677078247, + "learning_rate": 5.4337863582816936e-05, + "loss": 3.2478, + "step": 4392 + }, + { + "epoch": 6.43, + "grad_norm": 1.5067156553268433, + "learning_rate": 5.4296140436284546e-05, + "loss": 3.2362, + "step": 4393 + }, + { + "epoch": 6.43, + "grad_norm": 2.2881274223327637, + "learning_rate": 5.425442734429495e-05, + "loss": 3.2567, + "step": 4394 + }, + { + "epoch": 6.43, + "eval_loss": 3.067939043045044, + "eval_runtime": 1166.4557, + "eval_samples_per_second": 14.696, + "eval_steps_per_second": 0.919, + "step": 4394 + }, + { + "epoch": 6.43, + "grad_norm": 2.7140753269195557, + "learning_rate": 5.4212724316024776e-05, + "loss": 3.1976, + "step": 4395 + }, + { + "epoch": 6.44, + "grad_norm": 1.5862021446228027, + "learning_rate": 5.4171031360648406e-05, + "loss": 3.2628, + "step": 4396 + }, + { + "epoch": 6.44, + "grad_norm": 1.6570472717285156, + "learning_rate": 5.4129348487337995e-05, + "loss": 3.2399, + "step": 4397 + }, + { + "epoch": 6.44, + "grad_norm": 1.0379539728164673, + "learning_rate": 5.408767570526352e-05, + "loss": 3.2495, + "step": 4398 + }, + { + "epoch": 6.44, + "grad_norm": 0.9085617065429688, + "learning_rate": 5.404601302359269e-05, + "loss": 3.2073, + "step": 4399 + }, + { + "epoch": 6.44, + "grad_norm": 1.3575804233551025, + "learning_rate": 5.400436045149102e-05, + "loss": 3.2191, + "step": 4400 + }, + { + "epoch": 6.44, + "grad_norm": 1.4507752656936646, + "learning_rate": 5.39627179981218e-05, + "loss": 3.2083, + "step": 4401 + }, + { + "epoch": 6.44, + "grad_norm": 1.4040168523788452, + "learning_rate": 5.392108567264609e-05, + "loss": 3.2492, + "step": 4402 + }, + { + "epoch": 6.45, + "grad_norm": 1.684368371963501, + "learning_rate": 5.387946348422274e-05, + "loss": 3.2433, + "step": 4403 + }, + { + "epoch": 6.45, + "grad_norm": 1.438843846321106, + "learning_rate": 5.383785144200832e-05, + "loss": 3.2335, + "step": 4404 + }, + { + "epoch": 6.45, + "grad_norm": 1.311325192451477, + "learning_rate": 5.379624955515723e-05, + "loss": 3.2418, + "step": 4405 + }, + { + "epoch": 6.45, + "grad_norm": 1.0702279806137085, + "learning_rate": 5.375465783282163e-05, + "loss": 3.2534, + "step": 4406 + }, + { + "epoch": 6.45, + "grad_norm": 1.4737073183059692, + "learning_rate": 5.371307628415131e-05, + "loss": 3.2167, + "step": 4407 + }, + { + "epoch": 6.45, + "grad_norm": 1.0920416116714478, + "learning_rate": 5.367150491829408e-05, + "loss": 3.2418, + "step": 4408 + }, + { + "epoch": 6.45, + "grad_norm": 2.5086584091186523, + "learning_rate": 5.362994374439525e-05, + "loss": 3.2499, + "step": 4409 + }, + { + "epoch": 6.46, + "grad_norm": 1.8857651948928833, + "learning_rate": 5.3588392771598e-05, + "loss": 3.2232, + "step": 4410 + }, + { + "epoch": 6.46, + "grad_norm": 1.4724599123001099, + "learning_rate": 5.354685200904337e-05, + "loss": 3.2534, + "step": 4411 + }, + { + "epoch": 6.46, + "grad_norm": 1.5440292358398438, + "learning_rate": 5.350532146586996e-05, + "loss": 3.2061, + "step": 4412 + }, + { + "epoch": 6.46, + "grad_norm": 1.186196208000183, + "learning_rate": 5.346380115121419e-05, + "loss": 3.2526, + "step": 4413 + }, + { + "epoch": 6.46, + "grad_norm": 1.4283154010772705, + "learning_rate": 5.342229107421037e-05, + "loss": 3.2498, + "step": 4414 + }, + { + "epoch": 6.46, + "grad_norm": 1.7490146160125732, + "learning_rate": 5.3380791243990325e-05, + "loss": 3.2403, + "step": 4415 + }, + { + "epoch": 6.46, + "grad_norm": 1.5179111957550049, + "learning_rate": 5.333930166968374e-05, + "loss": 3.2239, + "step": 4416 + }, + { + "epoch": 6.47, + "grad_norm": 1.3260568380355835, + "learning_rate": 5.329782236041816e-05, + "loss": 3.2357, + "step": 4417 + }, + { + "epoch": 6.47, + "grad_norm": 0.9309088587760925, + "learning_rate": 5.325635332531864e-05, + "loss": 3.2212, + "step": 4418 + }, + { + "epoch": 6.47, + "grad_norm": 1.8849228620529175, + "learning_rate": 5.321489457350812e-05, + "loss": 3.2158, + "step": 4419 + }, + { + "epoch": 6.47, + "grad_norm": 2.031489610671997, + "learning_rate": 5.3173446114107265e-05, + "loss": 3.229, + "step": 4420 + }, + { + "epoch": 6.47, + "grad_norm": 1.7095611095428467, + "learning_rate": 5.313200795623443e-05, + "loss": 3.2323, + "step": 4421 + }, + { + "epoch": 6.47, + "grad_norm": 2.060396671295166, + "learning_rate": 5.309058010900576e-05, + "loss": 3.2519, + "step": 4422 + }, + { + "epoch": 6.48, + "grad_norm": 0.750653088092804, + "learning_rate": 5.3049162581535094e-05, + "loss": 3.2344, + "step": 4423 + }, + { + "epoch": 6.48, + "grad_norm": 1.6587752103805542, + "learning_rate": 5.300775538293401e-05, + "loss": 3.2506, + "step": 4424 + }, + { + "epoch": 6.48, + "grad_norm": 0.8414655923843384, + "learning_rate": 5.296635852231182e-05, + "loss": 3.2065, + "step": 4425 + }, + { + "epoch": 6.48, + "grad_norm": 0.4840925335884094, + "learning_rate": 5.292497200877553e-05, + "loss": 3.2054, + "step": 4426 + }, + { + "epoch": 6.48, + "grad_norm": 0.6098539233207703, + "learning_rate": 5.288359585142993e-05, + "loss": 3.2024, + "step": 4427 + }, + { + "epoch": 6.48, + "grad_norm": 0.46651723980903625, + "learning_rate": 5.2842230059377486e-05, + "loss": 3.2161, + "step": 4428 + }, + { + "epoch": 6.48, + "grad_norm": 0.508057177066803, + "learning_rate": 5.280087464171838e-05, + "loss": 3.1871, + "step": 4429 + }, + { + "epoch": 6.49, + "grad_norm": 0.5408399701118469, + "learning_rate": 5.275952960755055e-05, + "loss": 3.2624, + "step": 4430 + }, + { + "epoch": 6.49, + "grad_norm": 1.0045307874679565, + "learning_rate": 5.271819496596964e-05, + "loss": 3.2303, + "step": 4431 + }, + { + "epoch": 6.49, + "grad_norm": 0.6911265850067139, + "learning_rate": 5.2676870726068906e-05, + "loss": 3.2087, + "step": 4432 + }, + { + "epoch": 6.49, + "grad_norm": 0.6002917289733887, + "learning_rate": 5.2635556896939486e-05, + "loss": 3.1978, + "step": 4433 + }, + { + "epoch": 6.49, + "grad_norm": 0.9630332589149475, + "learning_rate": 5.259425348767018e-05, + "loss": 3.2238, + "step": 4434 + }, + { + "epoch": 6.49, + "grad_norm": 0.7534378170967102, + "learning_rate": 5.2552960507347325e-05, + "loss": 3.2308, + "step": 4435 + }, + { + "epoch": 6.49, + "grad_norm": 0.9733850359916687, + "learning_rate": 5.251167796505523e-05, + "loss": 3.2392, + "step": 4436 + }, + { + "epoch": 6.5, + "grad_norm": 1.1052712202072144, + "learning_rate": 5.247040586987575e-05, + "loss": 3.2257, + "step": 4437 + }, + { + "epoch": 6.5, + "grad_norm": 0.8000559210777283, + "learning_rate": 5.2429144230888384e-05, + "loss": 3.2837, + "step": 4438 + }, + { + "epoch": 6.5, + "grad_norm": 0.7555025219917297, + "learning_rate": 5.238789305717053e-05, + "loss": 3.2463, + "step": 4439 + }, + { + "epoch": 6.5, + "grad_norm": 1.0577791929244995, + "learning_rate": 5.2346652357797146e-05, + "loss": 3.2328, + "step": 4440 + }, + { + "epoch": 6.5, + "grad_norm": 0.6421762704849243, + "learning_rate": 5.230542214184081e-05, + "loss": 3.2186, + "step": 4441 + }, + { + "epoch": 6.5, + "grad_norm": 1.1943224668502808, + "learning_rate": 5.226420241837204e-05, + "loss": 3.2487, + "step": 4442 + }, + { + "epoch": 6.5, + "grad_norm": 0.6367546916007996, + "learning_rate": 5.22229931964588e-05, + "loss": 3.2938, + "step": 4443 + }, + { + "epoch": 6.51, + "grad_norm": 1.3476903438568115, + "learning_rate": 5.2181794485166824e-05, + "loss": 3.2098, + "step": 4444 + }, + { + "epoch": 6.51, + "grad_norm": 1.686813235282898, + "learning_rate": 5.214060629355968e-05, + "loss": 3.2456, + "step": 4445 + }, + { + "epoch": 6.51, + "grad_norm": 0.49779945611953735, + "learning_rate": 5.209942863069837e-05, + "loss": 3.2537, + "step": 4446 + }, + { + "epoch": 6.51, + "grad_norm": 1.0947519540786743, + "learning_rate": 5.20582615056417e-05, + "loss": 3.3028, + "step": 4447 + }, + { + "epoch": 6.51, + "grad_norm": 0.6501612663269043, + "learning_rate": 5.201710492744629e-05, + "loss": 3.2317, + "step": 4448 + }, + { + "epoch": 6.51, + "grad_norm": 0.4535963535308838, + "learning_rate": 5.19759589051662e-05, + "loss": 3.23, + "step": 4449 + }, + { + "epoch": 6.52, + "grad_norm": 0.5250782370567322, + "learning_rate": 5.193482344785327e-05, + "loss": 3.2435, + "step": 4450 + }, + { + "epoch": 6.52, + "grad_norm": 0.400921493768692, + "learning_rate": 5.1893698564557124e-05, + "loss": 3.2188, + "step": 4451 + }, + { + "epoch": 6.52, + "grad_norm": 0.902998149394989, + "learning_rate": 5.1852584264324866e-05, + "loss": 3.2495, + "step": 4452 + }, + { + "epoch": 6.52, + "grad_norm": 0.7799638509750366, + "learning_rate": 5.1811480556201375e-05, + "loss": 3.234, + "step": 4453 + }, + { + "epoch": 6.52, + "grad_norm": 1.8399096727371216, + "learning_rate": 5.1770387449229276e-05, + "loss": 3.3015, + "step": 4454 + }, + { + "epoch": 6.52, + "grad_norm": 0.94755619764328, + "learning_rate": 5.172930495244869e-05, + "loss": 3.2394, + "step": 4455 + }, + { + "epoch": 6.52, + "grad_norm": 2.66386079788208, + "learning_rate": 5.168823307489751e-05, + "loss": 3.2519, + "step": 4456 + }, + { + "epoch": 6.53, + "grad_norm": 1.893293857574463, + "learning_rate": 5.1647171825611284e-05, + "loss": 3.1926, + "step": 4457 + }, + { + "epoch": 6.53, + "grad_norm": 0.6333227753639221, + "learning_rate": 5.160612121362322e-05, + "loss": 3.2169, + "step": 4458 + }, + { + "epoch": 6.53, + "grad_norm": 1.2479146718978882, + "learning_rate": 5.156508124796415e-05, + "loss": 3.2308, + "step": 4459 + }, + { + "epoch": 6.53, + "grad_norm": 0.3292587399482727, + "learning_rate": 5.152405193766263e-05, + "loss": 3.2359, + "step": 4460 + }, + { + "epoch": 6.53, + "grad_norm": 1.253524899482727, + "learning_rate": 5.1483033291744797e-05, + "loss": 3.2176, + "step": 4461 + }, + { + "epoch": 6.53, + "grad_norm": 0.4894808232784271, + "learning_rate": 5.14420253192345e-05, + "loss": 3.2433, + "step": 4462 + }, + { + "epoch": 6.53, + "grad_norm": 1.1630624532699585, + "learning_rate": 5.140102802915321e-05, + "loss": 3.213, + "step": 4463 + }, + { + "epoch": 6.54, + "grad_norm": 0.8647369742393494, + "learning_rate": 5.136004143052007e-05, + "loss": 3.2736, + "step": 4464 + }, + { + "epoch": 6.54, + "grad_norm": 0.9852306842803955, + "learning_rate": 5.1319065532351865e-05, + "loss": 3.2649, + "step": 4465 + }, + { + "epoch": 6.54, + "grad_norm": 0.7359192371368408, + "learning_rate": 5.1278100343662916e-05, + "loss": 3.2311, + "step": 4466 + }, + { + "epoch": 6.54, + "grad_norm": 0.38752880692481995, + "learning_rate": 5.123714587346543e-05, + "loss": 3.2308, + "step": 4467 + }, + { + "epoch": 6.54, + "grad_norm": 1.2246233224868774, + "learning_rate": 5.119620213076909e-05, + "loss": 3.1778, + "step": 4468 + }, + { + "epoch": 6.54, + "grad_norm": 0.7400122284889221, + "learning_rate": 5.115526912458113e-05, + "loss": 3.2127, + "step": 4469 + }, + { + "epoch": 6.55, + "grad_norm": 0.9233433604240417, + "learning_rate": 5.111434686390666e-05, + "loss": 3.184, + "step": 4470 + }, + { + "epoch": 6.55, + "grad_norm": 1.7230321168899536, + "learning_rate": 5.107343535774831e-05, + "loss": 3.2116, + "step": 4471 + }, + { + "epoch": 6.55, + "grad_norm": 1.6479278802871704, + "learning_rate": 5.1032534615106195e-05, + "loss": 3.2049, + "step": 4472 + }, + { + "epoch": 6.55, + "grad_norm": 2.626024007797241, + "learning_rate": 5.0991644644978356e-05, + "loss": 3.2434, + "step": 4473 + }, + { + "epoch": 6.55, + "grad_norm": 3.1591696739196777, + "learning_rate": 5.095076545636028e-05, + "loss": 3.2116, + "step": 4474 + }, + { + "epoch": 6.55, + "grad_norm": 0.8667658567428589, + "learning_rate": 5.0909897058245025e-05, + "loss": 3.1994, + "step": 4475 + }, + { + "epoch": 6.55, + "grad_norm": 3.626915454864502, + "learning_rate": 5.08690394596235e-05, + "loss": 3.221, + "step": 4476 + }, + { + "epoch": 6.56, + "grad_norm": 4.862388610839844, + "learning_rate": 5.0828192669484e-05, + "loss": 3.2172, + "step": 4477 + }, + { + "epoch": 6.56, + "grad_norm": 2.2208385467529297, + "learning_rate": 5.078735669681257e-05, + "loss": 3.2153, + "step": 4478 + }, + { + "epoch": 6.56, + "grad_norm": 2.200000286102295, + "learning_rate": 5.074653155059286e-05, + "loss": 3.2971, + "step": 4479 + }, + { + "epoch": 6.56, + "grad_norm": 2.9845163822174072, + "learning_rate": 5.0705717239806125e-05, + "loss": 3.2522, + "step": 4480 + }, + { + "epoch": 6.56, + "grad_norm": 1.4984869956970215, + "learning_rate": 5.066491377343126e-05, + "loss": 3.2009, + "step": 4481 + }, + { + "epoch": 6.56, + "grad_norm": 2.6040592193603516, + "learning_rate": 5.0624121160444716e-05, + "loss": 3.1945, + "step": 4482 + }, + { + "epoch": 6.56, + "grad_norm": 3.3462085723876953, + "learning_rate": 5.0583339409820616e-05, + "loss": 3.1927, + "step": 4483 + }, + { + "epoch": 6.57, + "grad_norm": 1.2761520147323608, + "learning_rate": 5.0542568530530676e-05, + "loss": 3.1981, + "step": 4484 + }, + { + "epoch": 6.57, + "grad_norm": 2.611973285675049, + "learning_rate": 5.05018085315442e-05, + "loss": 3.1832, + "step": 4485 + }, + { + "epoch": 6.57, + "grad_norm": 3.3917155265808105, + "learning_rate": 5.046105942182815e-05, + "loss": 3.2319, + "step": 4486 + }, + { + "epoch": 6.57, + "grad_norm": 2.2328946590423584, + "learning_rate": 5.042032121034701e-05, + "loss": 3.2443, + "step": 4487 + }, + { + "epoch": 6.57, + "grad_norm": 2.1951651573181152, + "learning_rate": 5.037959390606294e-05, + "loss": 3.2073, + "step": 4488 + }, + { + "epoch": 6.57, + "grad_norm": 3.5785024166107178, + "learning_rate": 5.033887751793567e-05, + "loss": 3.264, + "step": 4489 + }, + { + "epoch": 6.57, + "grad_norm": 1.865713119506836, + "learning_rate": 5.0298172054922534e-05, + "loss": 3.2455, + "step": 4490 + }, + { + "epoch": 6.58, + "grad_norm": 2.268017292022705, + "learning_rate": 5.0257477525978446e-05, + "loss": 3.2089, + "step": 4491 + }, + { + "epoch": 6.58, + "grad_norm": 2.626833200454712, + "learning_rate": 5.021679394005596e-05, + "loss": 3.2474, + "step": 4492 + }, + { + "epoch": 6.58, + "grad_norm": 1.4748424291610718, + "learning_rate": 5.0176121306105164e-05, + "loss": 3.2171, + "step": 4493 + }, + { + "epoch": 6.58, + "grad_norm": 1.906065583229065, + "learning_rate": 5.0135459633073776e-05, + "loss": 3.2625, + "step": 4494 + }, + { + "epoch": 6.58, + "grad_norm": 2.797215700149536, + "learning_rate": 5.00948089299071e-05, + "loss": 3.2316, + "step": 4495 + }, + { + "epoch": 6.58, + "grad_norm": 0.824268639087677, + "learning_rate": 5.0054169205548016e-05, + "loss": 3.211, + "step": 4496 + }, + { + "epoch": 6.59, + "grad_norm": 2.206386089324951, + "learning_rate": 5.001354046893697e-05, + "loss": 3.2278, + "step": 4497 + }, + { + "epoch": 6.59, + "grad_norm": 2.08886981010437, + "learning_rate": 4.997292272901203e-05, + "loss": 3.2376, + "step": 4498 + }, + { + "epoch": 6.59, + "grad_norm": 0.7808924913406372, + "learning_rate": 4.993231599470887e-05, + "loss": 3.2411, + "step": 4499 + }, + { + "epoch": 6.59, + "grad_norm": 1.8568205833435059, + "learning_rate": 4.9891720274960566e-05, + "loss": 3.2718, + "step": 4500 + }, + { + "epoch": 6.59, + "grad_norm": 1.4308000802993774, + "learning_rate": 4.985113557869803e-05, + "loss": 3.2299, + "step": 4501 + }, + { + "epoch": 6.59, + "grad_norm": 1.8029470443725586, + "learning_rate": 4.981056191484962e-05, + "loss": 3.2458, + "step": 4502 + }, + { + "epoch": 6.59, + "grad_norm": 1.7437525987625122, + "learning_rate": 4.9769999292341194e-05, + "loss": 3.2255, + "step": 4503 + }, + { + "epoch": 6.6, + "grad_norm": 1.1221601963043213, + "learning_rate": 4.9729447720096265e-05, + "loss": 3.1949, + "step": 4504 + }, + { + "epoch": 6.6, + "grad_norm": 1.0662968158721924, + "learning_rate": 4.968890720703602e-05, + "loss": 3.1963, + "step": 4505 + }, + { + "epoch": 6.6, + "grad_norm": 1.716160535812378, + "learning_rate": 4.9648377762078966e-05, + "loss": 3.2037, + "step": 4506 + }, + { + "epoch": 6.6, + "grad_norm": 1.2884470224380493, + "learning_rate": 4.960785939414134e-05, + "loss": 3.2482, + "step": 4507 + }, + { + "epoch": 6.6, + "grad_norm": 2.104020595550537, + "learning_rate": 4.956735211213701e-05, + "loss": 3.2097, + "step": 4508 + }, + { + "epoch": 6.6, + "grad_norm": 2.3521175384521484, + "learning_rate": 4.9526855924977175e-05, + "loss": 3.2618, + "step": 4509 + }, + { + "epoch": 6.6, + "grad_norm": 1.780421257019043, + "learning_rate": 4.9486370841570803e-05, + "loss": 3.2177, + "step": 4510 + }, + { + "epoch": 6.61, + "grad_norm": 1.345507264137268, + "learning_rate": 4.94458968708243e-05, + "loss": 3.1907, + "step": 4511 + }, + { + "epoch": 6.61, + "grad_norm": 1.3987131118774414, + "learning_rate": 4.9405434021641705e-05, + "loss": 3.2361, + "step": 4512 + }, + { + "epoch": 6.61, + "grad_norm": 1.295686960220337, + "learning_rate": 4.9364982302924545e-05, + "loss": 3.2055, + "step": 4513 + }, + { + "epoch": 6.61, + "grad_norm": 1.0267897844314575, + "learning_rate": 4.9324541723571936e-05, + "loss": 3.2089, + "step": 4514 + }, + { + "epoch": 6.61, + "grad_norm": 1.2543052434921265, + "learning_rate": 4.9284112292480535e-05, + "loss": 3.2271, + "step": 4515 + }, + { + "epoch": 6.61, + "grad_norm": 1.0035556554794312, + "learning_rate": 4.924369401854455e-05, + "loss": 3.2686, + "step": 4516 + }, + { + "epoch": 6.61, + "grad_norm": 1.4459877014160156, + "learning_rate": 4.920328691065573e-05, + "loss": 3.2149, + "step": 4517 + }, + { + "epoch": 6.62, + "grad_norm": 1.3623418807983398, + "learning_rate": 4.916289097770337e-05, + "loss": 3.2456, + "step": 4518 + }, + { + "epoch": 6.62, + "grad_norm": 1.6200556755065918, + "learning_rate": 4.9122506228574286e-05, + "loss": 3.1883, + "step": 4519 + }, + { + "epoch": 6.62, + "grad_norm": 1.0919500589370728, + "learning_rate": 4.908213267215287e-05, + "loss": 3.1899, + "step": 4520 + }, + { + "epoch": 6.62, + "grad_norm": 1.162226676940918, + "learning_rate": 4.904177031732103e-05, + "loss": 3.2402, + "step": 4521 + }, + { + "epoch": 6.62, + "grad_norm": 1.1378471851348877, + "learning_rate": 4.900141917295822e-05, + "loss": 3.3377, + "step": 4522 + }, + { + "epoch": 6.62, + "grad_norm": 1.183754563331604, + "learning_rate": 4.896107924794141e-05, + "loss": 3.2368, + "step": 4523 + }, + { + "epoch": 6.63, + "grad_norm": 1.1688711643218994, + "learning_rate": 4.892075055114511e-05, + "loss": 3.2358, + "step": 4524 + }, + { + "epoch": 6.63, + "grad_norm": 0.9402367472648621, + "learning_rate": 4.888043309144138e-05, + "loss": 3.1947, + "step": 4525 + }, + { + "epoch": 6.63, + "grad_norm": 0.9889912009239197, + "learning_rate": 4.8840126877699787e-05, + "loss": 3.2409, + "step": 4526 + }, + { + "epoch": 6.63, + "grad_norm": 1.3844680786132812, + "learning_rate": 4.879983191878742e-05, + "loss": 3.2658, + "step": 4527 + }, + { + "epoch": 6.63, + "grad_norm": 0.8452149033546448, + "learning_rate": 4.875954822356893e-05, + "loss": 3.1865, + "step": 4528 + }, + { + "epoch": 6.63, + "grad_norm": 0.9102594256401062, + "learning_rate": 4.871927580090636e-05, + "loss": 3.2226, + "step": 4529 + }, + { + "epoch": 6.63, + "grad_norm": 0.7820479869842529, + "learning_rate": 4.867901465965949e-05, + "loss": 3.2125, + "step": 4530 + }, + { + "epoch": 6.64, + "grad_norm": 1.0344125032424927, + "learning_rate": 4.863876480868549e-05, + "loss": 3.1863, + "step": 4531 + }, + { + "epoch": 6.64, + "grad_norm": 1.1749505996704102, + "learning_rate": 4.8598526256838947e-05, + "loss": 3.2641, + "step": 4532 + }, + { + "epoch": 6.64, + "grad_norm": 1.117828130722046, + "learning_rate": 4.855829901297223e-05, + "loss": 3.1992, + "step": 4533 + }, + { + "epoch": 6.64, + "grad_norm": 1.255987286567688, + "learning_rate": 4.8518083085934954e-05, + "loss": 3.1951, + "step": 4534 + }, + { + "epoch": 6.64, + "grad_norm": 0.6159536838531494, + "learning_rate": 4.847787848457433e-05, + "loss": 3.2087, + "step": 4535 + }, + { + "epoch": 6.64, + "grad_norm": 1.0251574516296387, + "learning_rate": 4.843768521773523e-05, + "loss": 3.202, + "step": 4536 + }, + { + "epoch": 6.64, + "grad_norm": 1.4189834594726562, + "learning_rate": 4.8397503294259795e-05, + "loss": 3.2329, + "step": 4537 + }, + { + "epoch": 6.65, + "grad_norm": 1.0889549255371094, + "learning_rate": 4.835733272298777e-05, + "loss": 3.2087, + "step": 4538 + }, + { + "epoch": 6.65, + "grad_norm": 1.1167737245559692, + "learning_rate": 4.831717351275653e-05, + "loss": 3.2098, + "step": 4539 + }, + { + "epoch": 6.65, + "grad_norm": 1.141223430633545, + "learning_rate": 4.827702567240072e-05, + "loss": 3.2151, + "step": 4540 + }, + { + "epoch": 6.65, + "grad_norm": 1.1248409748077393, + "learning_rate": 4.823688921075258e-05, + "loss": 3.2147, + "step": 4541 + }, + { + "epoch": 6.65, + "grad_norm": 1.471689224243164, + "learning_rate": 4.8196764136642e-05, + "loss": 3.1923, + "step": 4542 + }, + { + "epoch": 6.65, + "grad_norm": 1.6416572332382202, + "learning_rate": 4.81566504588961e-05, + "loss": 3.2319, + "step": 4543 + }, + { + "epoch": 6.65, + "grad_norm": 1.1048762798309326, + "learning_rate": 4.811654818633966e-05, + "loss": 3.2002, + "step": 4544 + }, + { + "epoch": 6.66, + "grad_norm": 1.3798350095748901, + "learning_rate": 4.807645732779491e-05, + "loss": 3.1903, + "step": 4545 + }, + { + "epoch": 6.66, + "grad_norm": 1.5549540519714355, + "learning_rate": 4.8036377892081565e-05, + "loss": 3.181, + "step": 4546 + }, + { + "epoch": 6.66, + "grad_norm": 0.42195671796798706, + "learning_rate": 4.7996309888016844e-05, + "loss": 3.1872, + "step": 4547 + }, + { + "epoch": 6.66, + "grad_norm": 1.59599769115448, + "learning_rate": 4.795625332441545e-05, + "loss": 3.2153, + "step": 4548 + }, + { + "epoch": 6.66, + "grad_norm": 1.7859076261520386, + "learning_rate": 4.7916208210089516e-05, + "loss": 3.246, + "step": 4549 + }, + { + "epoch": 6.66, + "grad_norm": 1.5909998416900635, + "learning_rate": 4.787617455384874e-05, + "loss": 3.1916, + "step": 4550 + }, + { + "epoch": 6.67, + "grad_norm": 2.1160545349121094, + "learning_rate": 4.783615236450024e-05, + "loss": 3.1881, + "step": 4551 + }, + { + "epoch": 6.67, + "grad_norm": 0.9319918751716614, + "learning_rate": 4.779614165084864e-05, + "loss": 3.2286, + "step": 4552 + }, + { + "epoch": 6.67, + "grad_norm": 2.0976734161376953, + "learning_rate": 4.775614242169606e-05, + "loss": 3.2103, + "step": 4553 + }, + { + "epoch": 6.67, + "grad_norm": 1.695967197418213, + "learning_rate": 4.7716154685841944e-05, + "loss": 3.1638, + "step": 4554 + }, + { + "epoch": 6.67, + "grad_norm": 2.1299004554748535, + "learning_rate": 4.767617845208344e-05, + "loss": 3.21, + "step": 4555 + }, + { + "epoch": 6.67, + "grad_norm": 1.7198915481567383, + "learning_rate": 4.763621372921506e-05, + "loss": 3.2789, + "step": 4556 + }, + { + "epoch": 6.67, + "grad_norm": 0.6392155885696411, + "learning_rate": 4.759626052602866e-05, + "loss": 3.2074, + "step": 4557 + }, + { + "epoch": 6.68, + "grad_norm": 1.7603440284729004, + "learning_rate": 4.7556318851313784e-05, + "loss": 3.1682, + "step": 4558 + }, + { + "epoch": 6.68, + "grad_norm": 0.6075698733329773, + "learning_rate": 4.7516388713857344e-05, + "loss": 3.2082, + "step": 4559 + }, + { + "epoch": 6.68, + "grad_norm": 1.1561111211776733, + "learning_rate": 4.7476470122443576e-05, + "loss": 3.2178, + "step": 4560 + }, + { + "epoch": 6.68, + "grad_norm": 0.6579719185829163, + "learning_rate": 4.743656308585441e-05, + "loss": 3.2283, + "step": 4561 + }, + { + "epoch": 6.68, + "grad_norm": 1.3691036701202393, + "learning_rate": 4.739666761286915e-05, + "loss": 3.2209, + "step": 4562 + }, + { + "epoch": 6.68, + "grad_norm": 0.4905560314655304, + "learning_rate": 4.735678371226441e-05, + "loss": 3.1764, + "step": 4563 + }, + { + "epoch": 6.68, + "eval_loss": 3.0459163188934326, + "eval_runtime": 1166.1122, + "eval_samples_per_second": 14.7, + "eval_steps_per_second": 0.919, + "step": 4563 + }, + { + "epoch": 6.68, + "grad_norm": 0.34124380350112915, + "learning_rate": 4.731691139281449e-05, + "loss": 3.2197, + "step": 4564 + }, + { + "epoch": 6.69, + "grad_norm": 0.7039169073104858, + "learning_rate": 4.7277050663291035e-05, + "loss": 3.2013, + "step": 4565 + }, + { + "epoch": 6.69, + "grad_norm": 1.3341939449310303, + "learning_rate": 4.723720153246303e-05, + "loss": 3.2084, + "step": 4566 + }, + { + "epoch": 6.69, + "grad_norm": 0.4564155638217926, + "learning_rate": 4.719736400909716e-05, + "loss": 3.1865, + "step": 4567 + }, + { + "epoch": 6.69, + "grad_norm": 0.9328960180282593, + "learning_rate": 4.715753810195729e-05, + "loss": 3.2035, + "step": 4568 + }, + { + "epoch": 6.69, + "grad_norm": 0.6796619892120361, + "learning_rate": 4.7117723819804884e-05, + "loss": 3.2024, + "step": 4569 + }, + { + "epoch": 6.69, + "grad_norm": 0.4047408699989319, + "learning_rate": 4.7077921171398895e-05, + "loss": 3.1935, + "step": 4570 + }, + { + "epoch": 6.7, + "grad_norm": 0.6333791017532349, + "learning_rate": 4.7038130165495555e-05, + "loss": 3.1794, + "step": 4571 + }, + { + "epoch": 6.7, + "grad_norm": 0.6958187818527222, + "learning_rate": 4.6998350810848605e-05, + "loss": 3.2401, + "step": 4572 + }, + { + "epoch": 6.7, + "grad_norm": 1.0862935781478882, + "learning_rate": 4.6958583116209344e-05, + "loss": 3.2027, + "step": 4573 + }, + { + "epoch": 6.7, + "grad_norm": 0.5394841432571411, + "learning_rate": 4.691882709032629e-05, + "loss": 3.2092, + "step": 4574 + }, + { + "epoch": 6.7, + "grad_norm": 1.7290090322494507, + "learning_rate": 4.687908274194555e-05, + "loss": 3.2199, + "step": 4575 + }, + { + "epoch": 6.7, + "grad_norm": 1.2097941637039185, + "learning_rate": 4.683935007981061e-05, + "loss": 3.2319, + "step": 4576 + }, + { + "epoch": 6.7, + "grad_norm": 1.8092429637908936, + "learning_rate": 4.6799629112662366e-05, + "loss": 3.1968, + "step": 4577 + }, + { + "epoch": 6.71, + "grad_norm": 1.7616616487503052, + "learning_rate": 4.675991984923921e-05, + "loss": 3.2206, + "step": 4578 + }, + { + "epoch": 6.71, + "grad_norm": 0.6763386726379395, + "learning_rate": 4.672022229827687e-05, + "loss": 3.1857, + "step": 4579 + }, + { + "epoch": 6.71, + "grad_norm": 1.4232481718063354, + "learning_rate": 4.668053646850856e-05, + "loss": 3.184, + "step": 4580 + }, + { + "epoch": 6.71, + "grad_norm": 1.7722804546356201, + "learning_rate": 4.664086236866491e-05, + "loss": 3.225, + "step": 4581 + }, + { + "epoch": 6.71, + "grad_norm": 1.0724995136260986, + "learning_rate": 4.660120000747393e-05, + "loss": 3.1924, + "step": 4582 + }, + { + "epoch": 6.71, + "grad_norm": 0.3242776393890381, + "learning_rate": 4.656154939366109e-05, + "loss": 3.2068, + "step": 4583 + }, + { + "epoch": 6.71, + "grad_norm": 1.4377954006195068, + "learning_rate": 4.652191053594924e-05, + "loss": 3.193, + "step": 4584 + }, + { + "epoch": 6.72, + "grad_norm": 4.462525844573975, + "learning_rate": 4.64822834430587e-05, + "loss": 3.275, + "step": 4585 + }, + { + "epoch": 6.72, + "grad_norm": 1.4812723398208618, + "learning_rate": 4.644266812370714e-05, + "loss": 3.2099, + "step": 4586 + }, + { + "epoch": 6.72, + "grad_norm": 0.6171543002128601, + "learning_rate": 4.640306458660966e-05, + "loss": 3.178, + "step": 4587 + }, + { + "epoch": 6.72, + "grad_norm": 1.2330536842346191, + "learning_rate": 4.636347284047877e-05, + "loss": 3.2013, + "step": 4588 + }, + { + "epoch": 6.72, + "grad_norm": 0.5005393624305725, + "learning_rate": 4.6323892894024414e-05, + "loss": 3.2114, + "step": 4589 + }, + { + "epoch": 6.72, + "grad_norm": 1.613756775856018, + "learning_rate": 4.628432475595393e-05, + "loss": 3.1852, + "step": 4590 + }, + { + "epoch": 6.72, + "grad_norm": 0.8560778498649597, + "learning_rate": 4.6244768434971944e-05, + "loss": 3.2414, + "step": 4591 + }, + { + "epoch": 6.73, + "grad_norm": 1.844510555267334, + "learning_rate": 4.620522393978068e-05, + "loss": 3.1791, + "step": 4592 + }, + { + "epoch": 6.73, + "grad_norm": 1.6803126335144043, + "learning_rate": 4.616569127907968e-05, + "loss": 3.2298, + "step": 4593 + }, + { + "epoch": 6.73, + "grad_norm": 1.2029790878295898, + "learning_rate": 4.6126170461565746e-05, + "loss": 3.2044, + "step": 4594 + }, + { + "epoch": 6.73, + "grad_norm": 0.8809365630149841, + "learning_rate": 4.608666149593328e-05, + "loss": 3.1935, + "step": 4595 + }, + { + "epoch": 6.73, + "grad_norm": 0.6382874846458435, + "learning_rate": 4.604716439087403e-05, + "loss": 3.1965, + "step": 4596 + }, + { + "epoch": 6.73, + "grad_norm": 1.0096112489700317, + "learning_rate": 4.600767915507695e-05, + "loss": 3.1924, + "step": 4597 + }, + { + "epoch": 6.74, + "grad_norm": 0.8474308252334595, + "learning_rate": 4.596820579722867e-05, + "loss": 3.1994, + "step": 4598 + }, + { + "epoch": 6.74, + "grad_norm": 1.0024635791778564, + "learning_rate": 4.592874432601304e-05, + "loss": 3.1834, + "step": 4599 + }, + { + "epoch": 6.74, + "grad_norm": 0.9615989923477173, + "learning_rate": 4.588929475011125e-05, + "loss": 3.1855, + "step": 4600 + }, + { + "epoch": 6.74, + "grad_norm": 0.4216245114803314, + "learning_rate": 4.584985707820198e-05, + "loss": 3.2381, + "step": 4601 + }, + { + "epoch": 6.74, + "grad_norm": 0.5493209362030029, + "learning_rate": 4.581043131896126e-05, + "loss": 3.1567, + "step": 4602 + }, + { + "epoch": 6.74, + "grad_norm": 0.6222708225250244, + "learning_rate": 4.5771017481062475e-05, + "loss": 3.2149, + "step": 4603 + }, + { + "epoch": 6.74, + "grad_norm": 0.5318418741226196, + "learning_rate": 4.573161557317642e-05, + "loss": 3.2158, + "step": 4604 + }, + { + "epoch": 6.75, + "grad_norm": 0.7692480087280273, + "learning_rate": 4.5692225603971264e-05, + "loss": 3.2096, + "step": 4605 + }, + { + "epoch": 6.75, + "grad_norm": 0.5248206257820129, + "learning_rate": 4.56528475821125e-05, + "loss": 3.1933, + "step": 4606 + }, + { + "epoch": 6.75, + "grad_norm": 1.0218887329101562, + "learning_rate": 4.561348151626306e-05, + "loss": 3.1819, + "step": 4607 + }, + { + "epoch": 6.75, + "grad_norm": 2.8102357387542725, + "learning_rate": 4.55741274150832e-05, + "loss": 3.2286, + "step": 4608 + }, + { + "epoch": 6.75, + "grad_norm": 0.5493364334106445, + "learning_rate": 4.5534785287230575e-05, + "loss": 3.1981, + "step": 4609 + }, + { + "epoch": 6.75, + "grad_norm": 0.438005656003952, + "learning_rate": 4.5495455141360165e-05, + "loss": 3.1831, + "step": 4610 + }, + { + "epoch": 6.75, + "grad_norm": 1.3687691688537598, + "learning_rate": 4.5456136986124355e-05, + "loss": 3.1993, + "step": 4611 + }, + { + "epoch": 6.76, + "grad_norm": 0.7376412749290466, + "learning_rate": 4.5416830830172876e-05, + "loss": 3.2093, + "step": 4612 + }, + { + "epoch": 6.76, + "grad_norm": 0.710468590259552, + "learning_rate": 4.53775366821528e-05, + "loss": 3.1834, + "step": 4613 + }, + { + "epoch": 6.76, + "grad_norm": 0.6461208462715149, + "learning_rate": 4.53382545507086e-05, + "loss": 3.2279, + "step": 4614 + }, + { + "epoch": 6.76, + "grad_norm": 0.4084911644458771, + "learning_rate": 4.5298984444482075e-05, + "loss": 3.1921, + "step": 4615 + }, + { + "epoch": 6.76, + "grad_norm": 0.5108605027198792, + "learning_rate": 4.5259726372112376e-05, + "loss": 3.2015, + "step": 4616 + }, + { + "epoch": 6.76, + "grad_norm": 0.36462754011154175, + "learning_rate": 4.522048034223603e-05, + "loss": 3.2099, + "step": 4617 + }, + { + "epoch": 6.76, + "grad_norm": 0.5964351892471313, + "learning_rate": 4.518124636348688e-05, + "loss": 3.223, + "step": 4618 + }, + { + "epoch": 6.77, + "grad_norm": 0.532785177230835, + "learning_rate": 4.514202444449615e-05, + "loss": 3.1739, + "step": 4619 + }, + { + "epoch": 6.77, + "grad_norm": 0.7253513336181641, + "learning_rate": 4.510281459389242e-05, + "loss": 3.202, + "step": 4620 + }, + { + "epoch": 6.77, + "grad_norm": 1.3560603857040405, + "learning_rate": 4.506361682030156e-05, + "loss": 3.2175, + "step": 4621 + }, + { + "epoch": 6.77, + "grad_norm": 0.6363286375999451, + "learning_rate": 4.502443113234688e-05, + "loss": 3.2485, + "step": 4622 + }, + { + "epoch": 6.77, + "grad_norm": 1.2093783617019653, + "learning_rate": 4.498525753864884e-05, + "loss": 3.2012, + "step": 4623 + }, + { + "epoch": 6.77, + "grad_norm": 1.1001776456832886, + "learning_rate": 4.4946096047825515e-05, + "loss": 3.1938, + "step": 4624 + }, + { + "epoch": 6.78, + "grad_norm": 0.4927830398082733, + "learning_rate": 4.490694666849208e-05, + "loss": 3.234, + "step": 4625 + }, + { + "epoch": 6.78, + "grad_norm": 1.6655173301696777, + "learning_rate": 4.486780940926112e-05, + "loss": 3.1661, + "step": 4626 + }, + { + "epoch": 6.78, + "grad_norm": 1.3249220848083496, + "learning_rate": 4.482868427874268e-05, + "loss": 3.2283, + "step": 4627 + }, + { + "epoch": 6.78, + "grad_norm": 2.2860875129699707, + "learning_rate": 4.478957128554392e-05, + "loss": 3.2115, + "step": 4628 + }, + { + "epoch": 6.78, + "grad_norm": 2.1470510959625244, + "learning_rate": 4.4750470438269446e-05, + "loss": 3.2151, + "step": 4629 + }, + { + "epoch": 6.78, + "grad_norm": 1.3575111627578735, + "learning_rate": 4.4711381745521275e-05, + "loss": 3.1938, + "step": 4630 + }, + { + "epoch": 6.78, + "grad_norm": 1.7453752756118774, + "learning_rate": 4.467230521589856e-05, + "loss": 3.1969, + "step": 4631 + }, + { + "epoch": 6.79, + "grad_norm": 1.7523752450942993, + "learning_rate": 4.463324085799786e-05, + "loss": 3.2266, + "step": 4632 + }, + { + "epoch": 6.79, + "grad_norm": 1.433594822883606, + "learning_rate": 4.4594188680413206e-05, + "loss": 3.1962, + "step": 4633 + }, + { + "epoch": 6.79, + "grad_norm": 1.0330843925476074, + "learning_rate": 4.4555148691735685e-05, + "loss": 3.2325, + "step": 4634 + }, + { + "epoch": 6.79, + "grad_norm": 0.9396421313285828, + "learning_rate": 4.451612090055388e-05, + "loss": 3.2274, + "step": 4635 + }, + { + "epoch": 6.79, + "grad_norm": 1.1527962684631348, + "learning_rate": 4.447710531545364e-05, + "loss": 3.2075, + "step": 4636 + }, + { + "epoch": 6.79, + "grad_norm": 0.8900932669639587, + "learning_rate": 4.443810194501813e-05, + "loss": 3.1952, + "step": 4637 + }, + { + "epoch": 6.79, + "grad_norm": 1.3143221139907837, + "learning_rate": 4.4399110797827835e-05, + "loss": 3.2086, + "step": 4638 + }, + { + "epoch": 6.8, + "grad_norm": 1.1299476623535156, + "learning_rate": 4.4360131882460555e-05, + "loss": 3.2, + "step": 4639 + }, + { + "epoch": 6.8, + "grad_norm": 0.8985263109207153, + "learning_rate": 4.432116520749137e-05, + "loss": 3.15, + "step": 4640 + }, + { + "epoch": 6.8, + "grad_norm": 1.0726135969161987, + "learning_rate": 4.4282210781492684e-05, + "loss": 3.2016, + "step": 4641 + }, + { + "epoch": 6.8, + "grad_norm": 0.556115984916687, + "learning_rate": 4.424326861303424e-05, + "loss": 3.2035, + "step": 4642 + }, + { + "epoch": 6.8, + "grad_norm": 1.1372523307800293, + "learning_rate": 4.420433871068302e-05, + "loss": 3.1998, + "step": 4643 + }, + { + "epoch": 6.8, + "grad_norm": 0.9902729988098145, + "learning_rate": 4.416542108300338e-05, + "loss": 3.1661, + "step": 4644 + }, + { + "epoch": 6.8, + "grad_norm": 0.6972740292549133, + "learning_rate": 4.412651573855689e-05, + "loss": 3.191, + "step": 4645 + }, + { + "epoch": 6.81, + "grad_norm": 0.7196264863014221, + "learning_rate": 4.408762268590251e-05, + "loss": 3.1772, + "step": 4646 + }, + { + "epoch": 6.81, + "grad_norm": 0.6130596995353699, + "learning_rate": 4.404874193359646e-05, + "loss": 3.1859, + "step": 4647 + }, + { + "epoch": 6.81, + "grad_norm": 0.8541386723518372, + "learning_rate": 4.4009873490192135e-05, + "loss": 3.1891, + "step": 4648 + }, + { + "epoch": 6.81, + "grad_norm": 0.4068596065044403, + "learning_rate": 4.397101736424046e-05, + "loss": 3.2386, + "step": 4649 + }, + { + "epoch": 6.81, + "grad_norm": 0.35903048515319824, + "learning_rate": 4.3932173564289506e-05, + "loss": 3.2046, + "step": 4650 + }, + { + "epoch": 6.81, + "grad_norm": 1.0451300144195557, + "learning_rate": 4.389334209888455e-05, + "loss": 3.2677, + "step": 4651 + }, + { + "epoch": 6.82, + "grad_norm": 0.3824341595172882, + "learning_rate": 4.385452297656836e-05, + "loss": 3.1955, + "step": 4652 + }, + { + "epoch": 6.82, + "grad_norm": 0.9694157838821411, + "learning_rate": 4.381571620588088e-05, + "loss": 3.2113, + "step": 4653 + }, + { + "epoch": 6.82, + "grad_norm": 0.6231606602668762, + "learning_rate": 4.3776921795359216e-05, + "loss": 3.1968, + "step": 4654 + }, + { + "epoch": 6.82, + "grad_norm": 0.39795351028442383, + "learning_rate": 4.373813975353802e-05, + "loss": 3.1859, + "step": 4655 + }, + { + "epoch": 6.82, + "grad_norm": 0.3494223654270172, + "learning_rate": 4.3699370088949066e-05, + "loss": 3.2194, + "step": 4656 + }, + { + "epoch": 6.82, + "grad_norm": 0.7372296452522278, + "learning_rate": 4.366061281012129e-05, + "loss": 3.2062, + "step": 4657 + }, + { + "epoch": 6.82, + "grad_norm": 1.144978642463684, + "learning_rate": 4.362186792558122e-05, + "loss": 3.1948, + "step": 4658 + }, + { + "epoch": 6.83, + "grad_norm": 0.5863081812858582, + "learning_rate": 4.358313544385233e-05, + "loss": 3.1734, + "step": 4659 + }, + { + "epoch": 6.83, + "grad_norm": 1.334794282913208, + "learning_rate": 4.354441537345553e-05, + "loss": 3.187, + "step": 4660 + }, + { + "epoch": 6.83, + "grad_norm": 0.4816846251487732, + "learning_rate": 4.350570772290906e-05, + "loss": 3.1916, + "step": 4661 + }, + { + "epoch": 6.83, + "grad_norm": 0.8258687853813171, + "learning_rate": 4.3467012500728245e-05, + "loss": 3.2171, + "step": 4662 + }, + { + "epoch": 6.83, + "grad_norm": 0.9091712236404419, + "learning_rate": 4.342832971542578e-05, + "loss": 3.1695, + "step": 4663 + }, + { + "epoch": 6.83, + "grad_norm": 1.3142671585083008, + "learning_rate": 4.3389659375511725e-05, + "loss": 3.1681, + "step": 4664 + }, + { + "epoch": 6.83, + "grad_norm": 2.6121108531951904, + "learning_rate": 4.335100148949318e-05, + "loss": 3.2161, + "step": 4665 + }, + { + "epoch": 6.84, + "grad_norm": 0.6220512986183167, + "learning_rate": 4.3312356065874635e-05, + "loss": 3.1868, + "step": 4666 + }, + { + "epoch": 6.84, + "grad_norm": 1.4593570232391357, + "learning_rate": 4.327372311315792e-05, + "loss": 3.1968, + "step": 4667 + }, + { + "epoch": 6.84, + "grad_norm": 1.4904062747955322, + "learning_rate": 4.323510263984192e-05, + "loss": 3.2013, + "step": 4668 + }, + { + "epoch": 6.84, + "grad_norm": 0.8087517023086548, + "learning_rate": 4.3196494654422916e-05, + "loss": 3.2106, + "step": 4669 + }, + { + "epoch": 6.84, + "grad_norm": 0.5217124223709106, + "learning_rate": 4.315789916539441e-05, + "loss": 3.2188, + "step": 4670 + }, + { + "epoch": 6.84, + "grad_norm": 0.7983100414276123, + "learning_rate": 4.311931618124715e-05, + "loss": 3.217, + "step": 4671 + }, + { + "epoch": 6.84, + "grad_norm": 0.49825963377952576, + "learning_rate": 4.308074571046913e-05, + "loss": 3.2089, + "step": 4672 + }, + { + "epoch": 6.85, + "grad_norm": 0.9408275485038757, + "learning_rate": 4.30421877615456e-05, + "loss": 3.1924, + "step": 4673 + }, + { + "epoch": 6.85, + "grad_norm": 0.6962942481040955, + "learning_rate": 4.300364234295905e-05, + "loss": 3.2022, + "step": 4674 + }, + { + "epoch": 6.85, + "grad_norm": 0.5625227093696594, + "learning_rate": 4.296510946318921e-05, + "loss": 3.2056, + "step": 4675 + }, + { + "epoch": 6.85, + "grad_norm": 0.6208386421203613, + "learning_rate": 4.292658913071306e-05, + "loss": 3.1973, + "step": 4676 + }, + { + "epoch": 6.85, + "grad_norm": 0.5512813925743103, + "learning_rate": 4.2888081354004805e-05, + "loss": 3.2122, + "step": 4677 + }, + { + "epoch": 6.85, + "grad_norm": 1.9432053565979004, + "learning_rate": 4.2849586141535945e-05, + "loss": 3.2239, + "step": 4678 + }, + { + "epoch": 6.86, + "grad_norm": 4.5366291999816895, + "learning_rate": 4.281110350177506e-05, + "loss": 3.1618, + "step": 4679 + }, + { + "epoch": 6.86, + "grad_norm": 0.9245926737785339, + "learning_rate": 4.277263344318817e-05, + "loss": 3.2309, + "step": 4680 + }, + { + "epoch": 6.86, + "grad_norm": 0.3581990897655487, + "learning_rate": 4.2734175974238444e-05, + "loss": 3.1872, + "step": 4681 + }, + { + "epoch": 6.86, + "grad_norm": 0.9905257821083069, + "learning_rate": 4.269573110338615e-05, + "loss": 3.2077, + "step": 4682 + }, + { + "epoch": 6.86, + "grad_norm": 1.2098913192749023, + "learning_rate": 4.265729883908902e-05, + "loss": 3.2249, + "step": 4683 + }, + { + "epoch": 6.86, + "grad_norm": 0.711704671382904, + "learning_rate": 4.2618879189801885e-05, + "loss": 3.1625, + "step": 4684 + }, + { + "epoch": 6.86, + "grad_norm": 2.870743989944458, + "learning_rate": 4.25804721639767e-05, + "loss": 3.1953, + "step": 4685 + }, + { + "epoch": 6.87, + "grad_norm": 1.7720706462860107, + "learning_rate": 4.254207777006287e-05, + "loss": 3.1671, + "step": 4686 + }, + { + "epoch": 6.87, + "grad_norm": 0.8603296875953674, + "learning_rate": 4.25036960165069e-05, + "loss": 3.166, + "step": 4687 + }, + { + "epoch": 6.87, + "grad_norm": 2.4739768505096436, + "learning_rate": 4.246532691175241e-05, + "loss": 3.1946, + "step": 4688 + }, + { + "epoch": 6.87, + "grad_norm": 2.0245025157928467, + "learning_rate": 4.242697046424046e-05, + "loss": 3.2304, + "step": 4689 + }, + { + "epoch": 6.87, + "grad_norm": 0.7027145028114319, + "learning_rate": 4.2388626682409194e-05, + "loss": 3.2526, + "step": 4690 + }, + { + "epoch": 6.87, + "grad_norm": 2.3779094219207764, + "learning_rate": 4.23502955746939e-05, + "loss": 3.2621, + "step": 4691 + }, + { + "epoch": 6.87, + "grad_norm": 1.9404739141464233, + "learning_rate": 4.23119771495273e-05, + "loss": 3.1891, + "step": 4692 + }, + { + "epoch": 6.88, + "grad_norm": 0.6578683257102966, + "learning_rate": 4.227367141533909e-05, + "loss": 3.2023, + "step": 4693 + }, + { + "epoch": 6.88, + "grad_norm": 5.048638820648193, + "learning_rate": 4.2235378380556255e-05, + "loss": 3.136, + "step": 4694 + }, + { + "epoch": 6.88, + "grad_norm": 2.4565985202789307, + "learning_rate": 4.2197098053603136e-05, + "loss": 3.1868, + "step": 4695 + }, + { + "epoch": 6.88, + "grad_norm": 0.770892858505249, + "learning_rate": 4.215883044290103e-05, + "loss": 3.1455, + "step": 4696 + }, + { + "epoch": 6.88, + "grad_norm": 2.0807790756225586, + "learning_rate": 4.2120575556868594e-05, + "loss": 3.1894, + "step": 4697 + }, + { + "epoch": 6.88, + "grad_norm": 2.7063095569610596, + "learning_rate": 4.208233340392164e-05, + "loss": 3.1982, + "step": 4698 + }, + { + "epoch": 6.89, + "grad_norm": 1.1912204027175903, + "learning_rate": 4.2044103992473195e-05, + "loss": 3.1818, + "step": 4699 + }, + { + "epoch": 6.89, + "grad_norm": 2.9936110973358154, + "learning_rate": 4.2005887330933457e-05, + "loss": 3.2109, + "step": 4700 + }, + { + "epoch": 6.89, + "grad_norm": 3.9401204586029053, + "learning_rate": 4.196768342770986e-05, + "loss": 3.2246, + "step": 4701 + }, + { + "epoch": 6.89, + "grad_norm": 2.341156482696533, + "learning_rate": 4.1929492291206983e-05, + "loss": 3.1819, + "step": 4702 + }, + { + "epoch": 6.89, + "grad_norm": 0.7375439405441284, + "learning_rate": 4.1891313929826635e-05, + "loss": 3.2047, + "step": 4703 + }, + { + "epoch": 6.89, + "grad_norm": 1.9721698760986328, + "learning_rate": 4.185314835196781e-05, + "loss": 3.2131, + "step": 4704 + }, + { + "epoch": 6.89, + "grad_norm": 0.8218457102775574, + "learning_rate": 4.181499556602665e-05, + "loss": 3.1891, + "step": 4705 + }, + { + "epoch": 6.9, + "grad_norm": 1.2695006132125854, + "learning_rate": 4.177685558039652e-05, + "loss": 3.1774, + "step": 4706 + }, + { + "epoch": 6.9, + "grad_norm": 1.4132906198501587, + "learning_rate": 4.1738728403467974e-05, + "loss": 3.1856, + "step": 4707 + }, + { + "epoch": 6.9, + "grad_norm": 0.5927780866622925, + "learning_rate": 4.170061404362874e-05, + "loss": 3.2209, + "step": 4708 + }, + { + "epoch": 6.9, + "grad_norm": 1.482879877090454, + "learning_rate": 4.166251250926371e-05, + "loss": 3.1621, + "step": 4709 + }, + { + "epoch": 6.9, + "grad_norm": 0.4204716384410858, + "learning_rate": 4.1624423808754956e-05, + "loss": 3.171, + "step": 4710 + }, + { + "epoch": 6.9, + "grad_norm": 1.5936245918273926, + "learning_rate": 4.158634795048175e-05, + "loss": 3.1746, + "step": 4711 + }, + { + "epoch": 6.9, + "grad_norm": 1.1268365383148193, + "learning_rate": 4.154828494282056e-05, + "loss": 3.1932, + "step": 4712 + }, + { + "epoch": 6.91, + "grad_norm": 1.2134582996368408, + "learning_rate": 4.151023479414489e-05, + "loss": 3.2064, + "step": 4713 + }, + { + "epoch": 6.91, + "grad_norm": 1.2098625898361206, + "learning_rate": 4.147219751282562e-05, + "loss": 3.2025, + "step": 4714 + }, + { + "epoch": 6.91, + "grad_norm": 4.699478626251221, + "learning_rate": 4.143417310723069e-05, + "loss": 3.1962, + "step": 4715 + }, + { + "epoch": 6.91, + "grad_norm": 1.845900297164917, + "learning_rate": 4.139616158572512e-05, + "loss": 3.1635, + "step": 4716 + }, + { + "epoch": 6.91, + "grad_norm": 1.1032308340072632, + "learning_rate": 4.1358162956671295e-05, + "loss": 3.1781, + "step": 4717 + }, + { + "epoch": 6.91, + "grad_norm": 1.5497393608093262, + "learning_rate": 4.1320177228428646e-05, + "loss": 3.1825, + "step": 4718 + }, + { + "epoch": 6.91, + "grad_norm": 2.2472078800201416, + "learning_rate": 4.128220440935372e-05, + "loss": 3.1866, + "step": 4719 + }, + { + "epoch": 6.92, + "grad_norm": 92.44196319580078, + "learning_rate": 4.124424450780029e-05, + "loss": 3.2326, + "step": 4720 + }, + { + "epoch": 6.92, + "grad_norm": 1.094028115272522, + "learning_rate": 4.120629753211936e-05, + "loss": 3.2104, + "step": 4721 + }, + { + "epoch": 6.92, + "grad_norm": 0.9184134602546692, + "learning_rate": 4.116836349065892e-05, + "loss": 3.2233, + "step": 4722 + }, + { + "epoch": 6.92, + "grad_norm": 1.4922984838485718, + "learning_rate": 4.1130442391764214e-05, + "loss": 3.1623, + "step": 4723 + }, + { + "epoch": 6.92, + "grad_norm": 0.9635801911354065, + "learning_rate": 4.109253424377772e-05, + "loss": 3.1788, + "step": 4724 + }, + { + "epoch": 6.92, + "grad_norm": 1.1866681575775146, + "learning_rate": 4.105463905503888e-05, + "loss": 3.2056, + "step": 4725 + }, + { + "epoch": 6.93, + "grad_norm": 0.8471019268035889, + "learning_rate": 4.101675683388442e-05, + "loss": 3.1978, + "step": 4726 + }, + { + "epoch": 6.93, + "grad_norm": 0.5931624174118042, + "learning_rate": 4.0978887588648155e-05, + "loss": 3.2222, + "step": 4727 + }, + { + "epoch": 6.93, + "grad_norm": 0.9211606383323669, + "learning_rate": 4.094103132766109e-05, + "loss": 3.1993, + "step": 4728 + }, + { + "epoch": 6.93, + "grad_norm": 0.7005136013031006, + "learning_rate": 4.090318805925133e-05, + "loss": 3.2065, + "step": 4729 + }, + { + "epoch": 6.93, + "grad_norm": 0.9550567269325256, + "learning_rate": 4.086535779174415e-05, + "loss": 3.1617, + "step": 4730 + }, + { + "epoch": 6.93, + "grad_norm": 0.9668344259262085, + "learning_rate": 4.082754053346195e-05, + "loss": 3.2148, + "step": 4731 + }, + { + "epoch": 6.93, + "grad_norm": 0.6807712912559509, + "learning_rate": 4.078973629272428e-05, + "loss": 3.1848, + "step": 4732 + }, + { + "epoch": 6.93, + "eval_loss": 3.0342421531677246, + "eval_runtime": 1171.9875, + "eval_samples_per_second": 14.626, + "eval_steps_per_second": 0.915, + "step": 4732 + }, + { + "epoch": 6.94, + "grad_norm": 1.207279920578003, + "learning_rate": 4.075194507784782e-05, + "loss": 3.1595, + "step": 4733 + }, + { + "epoch": 6.94, + "grad_norm": 1.3139106035232544, + "learning_rate": 4.071416689714637e-05, + "loss": 3.1985, + "step": 4734 + }, + { + "epoch": 6.94, + "grad_norm": 1.053685188293457, + "learning_rate": 4.0676401758930894e-05, + "loss": 3.2048, + "step": 4735 + }, + { + "epoch": 6.94, + "grad_norm": 1.043835997581482, + "learning_rate": 4.063864967150945e-05, + "loss": 3.1763, + "step": 4736 + }, + { + "epoch": 6.94, + "grad_norm": 0.4672507047653198, + "learning_rate": 4.060091064318727e-05, + "loss": 3.1722, + "step": 4737 + }, + { + "epoch": 6.94, + "grad_norm": 1.6286187171936035, + "learning_rate": 4.0563184682266676e-05, + "loss": 3.2824, + "step": 4738 + }, + { + "epoch": 6.94, + "grad_norm": 0.8605678081512451, + "learning_rate": 4.052547179704711e-05, + "loss": 3.198, + "step": 4739 + }, + { + "epoch": 6.95, + "grad_norm": 1.1343073844909668, + "learning_rate": 4.048777199582518e-05, + "loss": 3.1726, + "step": 4740 + }, + { + "epoch": 6.95, + "grad_norm": 0.6983288526535034, + "learning_rate": 4.0450085286894565e-05, + "loss": 3.2139, + "step": 4741 + }, + { + "epoch": 6.95, + "grad_norm": 0.8509522080421448, + "learning_rate": 4.041241167854612e-05, + "loss": 3.176, + "step": 4742 + }, + { + "epoch": 6.95, + "grad_norm": 0.8853251934051514, + "learning_rate": 4.037475117906775e-05, + "loss": 3.1622, + "step": 4743 + }, + { + "epoch": 6.95, + "grad_norm": 1.2624071836471558, + "learning_rate": 4.033710379674458e-05, + "loss": 3.1633, + "step": 4744 + }, + { + "epoch": 6.95, + "grad_norm": 1.0235035419464111, + "learning_rate": 4.029946953985865e-05, + "loss": 3.1794, + "step": 4745 + }, + { + "epoch": 6.95, + "grad_norm": 1.0335735082626343, + "learning_rate": 4.0261848416689394e-05, + "loss": 3.2398, + "step": 4746 + }, + { + "epoch": 6.96, + "grad_norm": 1.2208091020584106, + "learning_rate": 4.022424043551312e-05, + "loss": 3.2424, + "step": 4747 + }, + { + "epoch": 6.96, + "grad_norm": 0.8216298818588257, + "learning_rate": 4.0186645604603304e-05, + "loss": 3.1948, + "step": 4748 + }, + { + "epoch": 6.96, + "grad_norm": 1.2754603624343872, + "learning_rate": 4.014906393223067e-05, + "loss": 3.1608, + "step": 4749 + }, + { + "epoch": 6.96, + "grad_norm": 1.283899188041687, + "learning_rate": 4.011149542666284e-05, + "loss": 3.2403, + "step": 4750 + }, + { + "epoch": 6.96, + "grad_norm": 0.569480836391449, + "learning_rate": 4.007394009616462e-05, + "loss": 3.2395, + "step": 4751 + }, + { + "epoch": 6.96, + "grad_norm": 0.7259752750396729, + "learning_rate": 4.003639794899804e-05, + "loss": 3.1733, + "step": 4752 + }, + { + "epoch": 6.97, + "grad_norm": 0.3225840628147125, + "learning_rate": 3.9998868993422026e-05, + "loss": 3.2326, + "step": 4753 + }, + { + "epoch": 6.97, + "grad_norm": 1.8098729848861694, + "learning_rate": 3.996135323769269e-05, + "loss": 3.2116, + "step": 4754 + }, + { + "epoch": 6.97, + "grad_norm": 0.893907904624939, + "learning_rate": 3.9923850690063345e-05, + "loss": 3.1768, + "step": 4755 + }, + { + "epoch": 6.97, + "grad_norm": 0.552627444267273, + "learning_rate": 3.988636135878421e-05, + "loss": 3.2011, + "step": 4756 + }, + { + "epoch": 6.97, + "grad_norm": 0.809914231300354, + "learning_rate": 3.984888525210266e-05, + "loss": 3.1777, + "step": 4757 + }, + { + "epoch": 6.97, + "grad_norm": 0.4486781060695648, + "learning_rate": 3.981142237826332e-05, + "loss": 3.193, + "step": 4758 + }, + { + "epoch": 6.97, + "grad_norm": 0.7444935441017151, + "learning_rate": 3.977397274550766e-05, + "loss": 3.1806, + "step": 4759 + }, + { + "epoch": 6.98, + "grad_norm": 0.49755245447158813, + "learning_rate": 3.973653636207437e-05, + "loss": 3.1676, + "step": 4760 + }, + { + "epoch": 6.98, + "grad_norm": 1.1105209589004517, + "learning_rate": 3.9699113236199226e-05, + "loss": 3.1885, + "step": 4761 + }, + { + "epoch": 6.98, + "grad_norm": 0.8656307458877563, + "learning_rate": 3.9661703376115034e-05, + "loss": 3.2385, + "step": 4762 + }, + { + "epoch": 6.98, + "grad_norm": 1.250411033630371, + "learning_rate": 3.962430679005175e-05, + "loss": 3.1652, + "step": 4763 + }, + { + "epoch": 6.98, + "grad_norm": 1.7875831127166748, + "learning_rate": 3.958692348623634e-05, + "loss": 3.1795, + "step": 4764 + }, + { + "epoch": 6.98, + "grad_norm": 0.575053870677948, + "learning_rate": 3.954955347289289e-05, + "loss": 3.2002, + "step": 4765 + }, + { + "epoch": 6.98, + "grad_norm": 1.1401422023773193, + "learning_rate": 3.9512196758242556e-05, + "loss": 3.1345, + "step": 4766 + }, + { + "epoch": 6.99, + "grad_norm": 0.6520401835441589, + "learning_rate": 3.947485335050357e-05, + "loss": 3.2271, + "step": 4767 + }, + { + "epoch": 6.99, + "grad_norm": 0.7057487964630127, + "learning_rate": 3.9437523257891215e-05, + "loss": 3.2535, + "step": 4768 + }, + { + "epoch": 6.99, + "grad_norm": 1.187171220779419, + "learning_rate": 3.940020648861792e-05, + "loss": 3.153, + "step": 4769 + }, + { + "epoch": 6.99, + "grad_norm": 0.44475510716438293, + "learning_rate": 3.9362903050892996e-05, + "loss": 3.1748, + "step": 4770 + }, + { + "epoch": 6.99, + "grad_norm": 2.0689454078674316, + "learning_rate": 3.9325612952923065e-05, + "loss": 3.17, + "step": 4771 + }, + { + "epoch": 6.99, + "grad_norm": 1.3041399717330933, + "learning_rate": 3.92883362029117e-05, + "loss": 3.185, + "step": 4772 + }, + { + "epoch": 6.99, + "grad_norm": 0.5959802865982056, + "learning_rate": 3.925107280905943e-05, + "loss": 3.1533, + "step": 4773 + }, + { + "epoch": 7.0, + "grad_norm": 1.501555323600769, + "learning_rate": 3.9213822779564056e-05, + "loss": 3.1333, + "step": 4774 + }, + { + "epoch": 7.0, + "grad_norm": 0.4824700951576233, + "learning_rate": 3.917658612262033e-05, + "loss": 3.173, + "step": 4775 + }, + { + "epoch": 7.0, + "grad_norm": 0.9984436631202698, + "learning_rate": 3.913936284641997e-05, + "loss": 3.1509, + "step": 4776 + }, + { + "epoch": 7.0, + "grad_norm": 0.47394973039627075, + "learning_rate": 3.910215295915195e-05, + "loss": 3.1744, + "step": 4777 + }, + { + "epoch": 7.0, + "grad_norm": 1.570947527885437, + "learning_rate": 3.906495646900219e-05, + "loss": 3.1831, + "step": 4778 + }, + { + "epoch": 7.0, + "grad_norm": 0.9516702890396118, + "learning_rate": 3.902777338415358e-05, + "loss": 3.1569, + "step": 4779 + }, + { + "epoch": 7.01, + "grad_norm": 2.1789183616638184, + "learning_rate": 3.899060371278627e-05, + "loss": 3.1952, + "step": 4780 + }, + { + "epoch": 7.01, + "grad_norm": 3.014054536819458, + "learning_rate": 3.895344746307723e-05, + "loss": 3.2197, + "step": 4781 + }, + { + "epoch": 7.01, + "grad_norm": 1.0081833600997925, + "learning_rate": 3.8916304643200604e-05, + "loss": 3.1867, + "step": 4782 + }, + { + "epoch": 7.01, + "grad_norm": 1.2014774084091187, + "learning_rate": 3.887917526132767e-05, + "loss": 3.2003, + "step": 4783 + }, + { + "epoch": 7.01, + "grad_norm": 2.214982748031616, + "learning_rate": 3.8842059325626536e-05, + "loss": 3.1835, + "step": 4784 + }, + { + "epoch": 7.01, + "grad_norm": 1.1680030822753906, + "learning_rate": 3.880495684426243e-05, + "loss": 3.2276, + "step": 4785 + }, + { + "epoch": 7.01, + "grad_norm": 1.6248522996902466, + "learning_rate": 3.8767867825397796e-05, + "loss": 3.1552, + "step": 4786 + }, + { + "epoch": 7.0, + "grad_norm": 1.5934488773345947, + "learning_rate": 3.873079227719184e-05, + "loss": 3.1989, + "step": 4787 + }, + { + "epoch": 7.0, + "grad_norm": 1.3276259899139404, + "learning_rate": 3.869373020780094e-05, + "loss": 3.2109, + "step": 4788 + }, + { + "epoch": 7.0, + "grad_norm": 1.9052053689956665, + "learning_rate": 3.865668162537861e-05, + "loss": 3.1952, + "step": 4789 + }, + { + "epoch": 7.0, + "grad_norm": 1.7343652248382568, + "learning_rate": 3.8619646538075184e-05, + "loss": 3.2074, + "step": 4790 + }, + { + "epoch": 7.01, + "grad_norm": 1.1672378778457642, + "learning_rate": 3.858262495403816e-05, + "loss": 3.1704, + "step": 4791 + }, + { + "epoch": 7.01, + "grad_norm": 1.6275311708450317, + "learning_rate": 3.854561688141205e-05, + "loss": 3.1723, + "step": 4792 + }, + { + "epoch": 7.01, + "grad_norm": 2.027491569519043, + "learning_rate": 3.8508622328338376e-05, + "loss": 3.2521, + "step": 4793 + }, + { + "epoch": 7.01, + "grad_norm": 1.3763444423675537, + "learning_rate": 3.8471641302955684e-05, + "loss": 3.1991, + "step": 4794 + }, + { + "epoch": 7.01, + "grad_norm": 1.7834948301315308, + "learning_rate": 3.843467381339957e-05, + "loss": 3.1685, + "step": 4795 + }, + { + "epoch": 7.01, + "grad_norm": 1.548448085784912, + "learning_rate": 3.839771986780263e-05, + "loss": 3.1715, + "step": 4796 + }, + { + "epoch": 7.02, + "grad_norm": 2.0860743522644043, + "learning_rate": 3.836077947429446e-05, + "loss": 3.2339, + "step": 4797 + }, + { + "epoch": 7.02, + "grad_norm": 2.655271053314209, + "learning_rate": 3.8323852641001733e-05, + "loss": 3.1881, + "step": 4798 + }, + { + "epoch": 7.02, + "grad_norm": 1.268930435180664, + "learning_rate": 3.828693937604809e-05, + "loss": 3.1635, + "step": 4799 + }, + { + "epoch": 7.02, + "grad_norm": 1.7928105592727661, + "learning_rate": 3.8250039687554195e-05, + "loss": 3.1764, + "step": 4800 + }, + { + "epoch": 7.02, + "grad_norm": 2.381535768508911, + "learning_rate": 3.821315358363775e-05, + "loss": 3.1892, + "step": 4801 + }, + { + "epoch": 7.02, + "grad_norm": 1.9187071323394775, + "learning_rate": 3.817628107241343e-05, + "loss": 3.1783, + "step": 4802 + }, + { + "epoch": 7.02, + "grad_norm": 0.999586284160614, + "learning_rate": 3.813942216199299e-05, + "loss": 3.1682, + "step": 4803 + }, + { + "epoch": 7.03, + "grad_norm": 1.993181824684143, + "learning_rate": 3.810257686048503e-05, + "loss": 3.2129, + "step": 4804 + }, + { + "epoch": 7.03, + "grad_norm": 1.1408189535140991, + "learning_rate": 3.806574517599537e-05, + "loss": 3.2541, + "step": 4805 + }, + { + "epoch": 7.03, + "grad_norm": 1.8819730281829834, + "learning_rate": 3.802892711662676e-05, + "loss": 3.1628, + "step": 4806 + }, + { + "epoch": 7.03, + "grad_norm": 2.9541916847229004, + "learning_rate": 3.799212269047878e-05, + "loss": 3.224, + "step": 4807 + }, + { + "epoch": 7.03, + "grad_norm": 1.8541984558105469, + "learning_rate": 3.795533190564831e-05, + "loss": 3.1792, + "step": 4808 + }, + { + "epoch": 7.03, + "grad_norm": 0.8892889022827148, + "learning_rate": 3.791855477022903e-05, + "loss": 3.2253, + "step": 4809 + }, + { + "epoch": 7.03, + "grad_norm": 1.7917122840881348, + "learning_rate": 3.788179129231159e-05, + "loss": 3.1897, + "step": 4810 + }, + { + "epoch": 7.04, + "grad_norm": 1.6101915836334229, + "learning_rate": 3.78450414799838e-05, + "loss": 3.1593, + "step": 4811 + }, + { + "epoch": 7.04, + "grad_norm": 0.8278429508209229, + "learning_rate": 3.780830534133037e-05, + "loss": 3.1712, + "step": 4812 + }, + { + "epoch": 7.04, + "grad_norm": 1.5922514200210571, + "learning_rate": 3.777158288443291e-05, + "loss": 3.1608, + "step": 4813 + }, + { + "epoch": 7.04, + "grad_norm": 0.9403831362724304, + "learning_rate": 3.7734874117370244e-05, + "loss": 3.2142, + "step": 4814 + }, + { + "epoch": 7.04, + "grad_norm": 1.2470545768737793, + "learning_rate": 3.769817904821796e-05, + "loss": 3.2069, + "step": 4815 + }, + { + "epoch": 7.04, + "grad_norm": 0.7299978733062744, + "learning_rate": 3.766149768504876e-05, + "loss": 3.1938, + "step": 4816 + }, + { + "epoch": 7.04, + "grad_norm": 1.2859150171279907, + "learning_rate": 3.762483003593229e-05, + "loss": 3.1506, + "step": 4817 + }, + { + "epoch": 7.05, + "grad_norm": 0.8429161906242371, + "learning_rate": 3.7588176108935194e-05, + "loss": 3.1859, + "step": 4818 + }, + { + "epoch": 7.05, + "grad_norm": 1.0546910762786865, + "learning_rate": 3.75515359121211e-05, + "loss": 3.1802, + "step": 4819 + }, + { + "epoch": 7.05, + "grad_norm": 1.477935791015625, + "learning_rate": 3.7514909453550575e-05, + "loss": 3.19, + "step": 4820 + }, + { + "epoch": 7.05, + "grad_norm": 0.5975956916809082, + "learning_rate": 3.7478296741281215e-05, + "loss": 3.192, + "step": 4821 + }, + { + "epoch": 7.05, + "grad_norm": 0.9366810321807861, + "learning_rate": 3.744169778336758e-05, + "loss": 3.1711, + "step": 4822 + }, + { + "epoch": 7.05, + "grad_norm": 0.5407786965370178, + "learning_rate": 3.740511258786118e-05, + "loss": 3.2555, + "step": 4823 + }, + { + "epoch": 7.06, + "grad_norm": 0.5257308483123779, + "learning_rate": 3.736854116281052e-05, + "loss": 3.1868, + "step": 4824 + }, + { + "epoch": 7.06, + "grad_norm": 0.5943123698234558, + "learning_rate": 3.7331983516261073e-05, + "loss": 3.1969, + "step": 4825 + }, + { + "epoch": 7.06, + "grad_norm": 0.31985798478126526, + "learning_rate": 3.729543965625526e-05, + "loss": 3.2147, + "step": 4826 + }, + { + "epoch": 7.06, + "grad_norm": 0.932587206363678, + "learning_rate": 3.7258909590832506e-05, + "loss": 3.1519, + "step": 4827 + }, + { + "epoch": 7.06, + "grad_norm": 1.0996321439743042, + "learning_rate": 3.722239332802916e-05, + "loss": 3.2083, + "step": 4828 + }, + { + "epoch": 7.06, + "grad_norm": 0.7307466864585876, + "learning_rate": 3.718589087587857e-05, + "loss": 3.1716, + "step": 4829 + }, + { + "epoch": 7.06, + "grad_norm": 0.39040249586105347, + "learning_rate": 3.7149402242411023e-05, + "loss": 3.1738, + "step": 4830 + }, + { + "epoch": 7.07, + "grad_norm": 0.8237361907958984, + "learning_rate": 3.711292743565378e-05, + "loss": 3.186, + "step": 4831 + }, + { + "epoch": 7.07, + "grad_norm": 0.6313247084617615, + "learning_rate": 3.707646646363104e-05, + "loss": 3.186, + "step": 4832 + }, + { + "epoch": 7.07, + "grad_norm": 1.1303566694259644, + "learning_rate": 3.7040019334363996e-05, + "loss": 3.2012, + "step": 4833 + }, + { + "epoch": 7.07, + "grad_norm": 1.1949049234390259, + "learning_rate": 3.7003586055870755e-05, + "loss": 3.2074, + "step": 4834 + }, + { + "epoch": 7.07, + "grad_norm": 0.9406141638755798, + "learning_rate": 3.696716663616641e-05, + "loss": 3.1975, + "step": 4835 + }, + { + "epoch": 7.07, + "grad_norm": 0.7907782793045044, + "learning_rate": 3.6930761083262955e-05, + "loss": 3.1521, + "step": 4836 + }, + { + "epoch": 7.07, + "grad_norm": 0.45947736501693726, + "learning_rate": 3.689436940516944e-05, + "loss": 3.1649, + "step": 4837 + }, + { + "epoch": 7.08, + "grad_norm": 1.1769413948059082, + "learning_rate": 3.6857991609891675e-05, + "loss": 3.1585, + "step": 4838 + }, + { + "epoch": 7.08, + "grad_norm": 0.3579048812389374, + "learning_rate": 3.6821627705432614e-05, + "loss": 3.2008, + "step": 4839 + }, + { + "epoch": 7.08, + "grad_norm": 1.2127845287322998, + "learning_rate": 3.678527769979211e-05, + "loss": 3.194, + "step": 4840 + }, + { + "epoch": 7.08, + "grad_norm": 0.7427309155464172, + "learning_rate": 3.6748941600966815e-05, + "loss": 3.171, + "step": 4841 + }, + { + "epoch": 7.08, + "grad_norm": 1.3657749891281128, + "learning_rate": 3.671261941695045e-05, + "loss": 3.2067, + "step": 4842 + }, + { + "epoch": 7.08, + "grad_norm": 0.7372667193412781, + "learning_rate": 3.6676311155733746e-05, + "loss": 3.1599, + "step": 4843 + }, + { + "epoch": 7.09, + "grad_norm": 1.225062370300293, + "learning_rate": 3.664001682530418e-05, + "loss": 3.1744, + "step": 4844 + }, + { + "epoch": 7.09, + "grad_norm": 1.3079476356506348, + "learning_rate": 3.6603736433646264e-05, + "loss": 3.149, + "step": 4845 + }, + { + "epoch": 7.09, + "grad_norm": 0.3697514533996582, + "learning_rate": 3.656746998874153e-05, + "loss": 3.1942, + "step": 4846 + }, + { + "epoch": 7.09, + "grad_norm": 2.1522505283355713, + "learning_rate": 3.653121749856826e-05, + "loss": 3.1853, + "step": 4847 + }, + { + "epoch": 7.09, + "grad_norm": 0.901470959186554, + "learning_rate": 3.649497897110179e-05, + "loss": 3.1828, + "step": 4848 + }, + { + "epoch": 7.09, + "grad_norm": 0.8554129004478455, + "learning_rate": 3.645875441431435e-05, + "loss": 3.224, + "step": 4849 + }, + { + "epoch": 7.09, + "grad_norm": 1.8987047672271729, + "learning_rate": 3.642254383617511e-05, + "loss": 3.1925, + "step": 4850 + }, + { + "epoch": 7.1, + "grad_norm": 0.45335274934768677, + "learning_rate": 3.6386347244650155e-05, + "loss": 3.1787, + "step": 4851 + }, + { + "epoch": 7.1, + "grad_norm": 0.8638046979904175, + "learning_rate": 3.635016464770249e-05, + "loss": 3.1878, + "step": 4852 + }, + { + "epoch": 7.1, + "grad_norm": 0.6001507639884949, + "learning_rate": 3.6313996053292034e-05, + "loss": 3.1502, + "step": 4853 + }, + { + "epoch": 7.1, + "grad_norm": 0.9555931687355042, + "learning_rate": 3.627784146937563e-05, + "loss": 3.2091, + "step": 4854 + }, + { + "epoch": 7.1, + "grad_norm": 0.5545978546142578, + "learning_rate": 3.624170090390706e-05, + "loss": 3.2264, + "step": 4855 + }, + { + "epoch": 7.1, + "grad_norm": 0.2554129362106323, + "learning_rate": 3.620557436483701e-05, + "loss": 3.1756, + "step": 4856 + }, + { + "epoch": 7.1, + "grad_norm": 0.648738443851471, + "learning_rate": 3.6169461860113074e-05, + "loss": 3.1885, + "step": 4857 + }, + { + "epoch": 7.11, + "grad_norm": 0.39746615290641785, + "learning_rate": 3.613336339767974e-05, + "loss": 3.1645, + "step": 4858 + }, + { + "epoch": 7.11, + "grad_norm": 0.6752750277519226, + "learning_rate": 3.6097278985478454e-05, + "loss": 3.1669, + "step": 4859 + }, + { + "epoch": 7.11, + "grad_norm": 0.4994068443775177, + "learning_rate": 3.606120863144753e-05, + "loss": 3.2155, + "step": 4860 + }, + { + "epoch": 7.11, + "grad_norm": 0.9360056519508362, + "learning_rate": 3.6025152343522205e-05, + "loss": 3.198, + "step": 4861 + }, + { + "epoch": 7.11, + "grad_norm": 0.3693661093711853, + "learning_rate": 3.598911012963463e-05, + "loss": 3.1723, + "step": 4862 + }, + { + "epoch": 7.11, + "grad_norm": 0.7912516593933105, + "learning_rate": 3.5953081997713876e-05, + "loss": 3.1946, + "step": 4863 + }, + { + "epoch": 7.11, + "grad_norm": 0.5820804834365845, + "learning_rate": 3.591706795568579e-05, + "loss": 3.1551, + "step": 4864 + }, + { + "epoch": 7.12, + "grad_norm": 0.41594260931015015, + "learning_rate": 3.588106801147333e-05, + "loss": 3.2372, + "step": 4865 + }, + { + "epoch": 7.12, + "grad_norm": 1.0947422981262207, + "learning_rate": 3.5845082172996224e-05, + "loss": 3.1926, + "step": 4866 + }, + { + "epoch": 7.12, + "grad_norm": 0.4444226324558258, + "learning_rate": 3.580911044817104e-05, + "loss": 3.2154, + "step": 4867 + }, + { + "epoch": 7.12, + "grad_norm": 0.8094576001167297, + "learning_rate": 3.577315284491138e-05, + "loss": 3.1802, + "step": 4868 + }, + { + "epoch": 7.12, + "grad_norm": 0.6356644034385681, + "learning_rate": 3.573720937112772e-05, + "loss": 3.1484, + "step": 4869 + }, + { + "epoch": 7.12, + "grad_norm": 0.9300285577774048, + "learning_rate": 3.570128003472725e-05, + "loss": 3.1834, + "step": 4870 + }, + { + "epoch": 7.13, + "grad_norm": 0.419990211725235, + "learning_rate": 3.5665364843614336e-05, + "loss": 3.1857, + "step": 4871 + }, + { + "epoch": 7.13, + "grad_norm": 1.047782063484192, + "learning_rate": 3.562946380568997e-05, + "loss": 3.2032, + "step": 4872 + }, + { + "epoch": 7.13, + "grad_norm": 0.564288318157196, + "learning_rate": 3.5593576928852134e-05, + "loss": 3.1501, + "step": 4873 + }, + { + "epoch": 7.13, + "grad_norm": 0.6687262654304504, + "learning_rate": 3.555770422099581e-05, + "loss": 3.1657, + "step": 4874 + }, + { + "epoch": 7.13, + "grad_norm": 0.4760758876800537, + "learning_rate": 3.5521845690012655e-05, + "loss": 3.1708, + "step": 4875 + }, + { + "epoch": 7.13, + "grad_norm": 0.6651167273521423, + "learning_rate": 3.5486001343791295e-05, + "loss": 3.1546, + "step": 4876 + }, + { + "epoch": 7.13, + "grad_norm": 0.554828941822052, + "learning_rate": 3.5450171190217364e-05, + "loss": 3.1451, + "step": 4877 + }, + { + "epoch": 7.14, + "grad_norm": 1.176149606704712, + "learning_rate": 3.5414355237173136e-05, + "loss": 3.2025, + "step": 4878 + }, + { + "epoch": 7.14, + "grad_norm": 0.4831768572330475, + "learning_rate": 3.537855349253788e-05, + "loss": 3.2084, + "step": 4879 + }, + { + "epoch": 7.14, + "grad_norm": 0.946924090385437, + "learning_rate": 3.5342765964187864e-05, + "loss": 3.2116, + "step": 4880 + }, + { + "epoch": 7.14, + "grad_norm": 0.4359959661960602, + "learning_rate": 3.530699265999599e-05, + "loss": 3.1784, + "step": 4881 + }, + { + "epoch": 7.14, + "grad_norm": 0.9432086944580078, + "learning_rate": 3.5271233587832164e-05, + "loss": 3.2123, + "step": 4882 + }, + { + "epoch": 7.14, + "grad_norm": 0.4890124797821045, + "learning_rate": 3.523548875556316e-05, + "loss": 3.2041, + "step": 4883 + }, + { + "epoch": 7.14, + "grad_norm": 0.6941928267478943, + "learning_rate": 3.51997581710526e-05, + "loss": 3.2032, + "step": 4884 + }, + { + "epoch": 7.15, + "grad_norm": 0.5665786862373352, + "learning_rate": 3.5164041842160966e-05, + "loss": 3.193, + "step": 4885 + }, + { + "epoch": 7.15, + "grad_norm": 0.7066811919212341, + "learning_rate": 3.5128339776745625e-05, + "loss": 3.1766, + "step": 4886 + }, + { + "epoch": 7.15, + "grad_norm": 0.6680645942687988, + "learning_rate": 3.5092651982660774e-05, + "loss": 3.2185, + "step": 4887 + }, + { + "epoch": 7.15, + "grad_norm": 0.7763155698776245, + "learning_rate": 3.5056978467757506e-05, + "loss": 3.18, + "step": 4888 + }, + { + "epoch": 7.15, + "grad_norm": 0.8374306559562683, + "learning_rate": 3.502131923988373e-05, + "loss": 3.2049, + "step": 4889 + }, + { + "epoch": 7.15, + "grad_norm": 1.1140564680099487, + "learning_rate": 3.4985674306884274e-05, + "loss": 3.195, + "step": 4890 + }, + { + "epoch": 7.15, + "grad_norm": 0.9668914675712585, + "learning_rate": 3.495004367660075e-05, + "loss": 3.1662, + "step": 4891 + }, + { + "epoch": 7.16, + "grad_norm": 0.8925195336341858, + "learning_rate": 3.491442735687167e-05, + "loss": 3.198, + "step": 4892 + }, + { + "epoch": 7.16, + "grad_norm": 1.2694224119186401, + "learning_rate": 3.4878825355532394e-05, + "loss": 3.1972, + "step": 4893 + }, + { + "epoch": 7.16, + "grad_norm": 0.7892391085624695, + "learning_rate": 3.4843237680415156e-05, + "loss": 3.2424, + "step": 4894 + }, + { + "epoch": 7.16, + "grad_norm": 0.8622798323631287, + "learning_rate": 3.480766433934889e-05, + "loss": 3.2112, + "step": 4895 + }, + { + "epoch": 7.16, + "grad_norm": 0.8695278167724609, + "learning_rate": 3.477210534015961e-05, + "loss": 3.2158, + "step": 4896 + }, + { + "epoch": 7.16, + "grad_norm": 0.8544620871543884, + "learning_rate": 3.473656069067005e-05, + "loss": 3.2501, + "step": 4897 + }, + { + "epoch": 7.17, + "grad_norm": 0.6005237102508545, + "learning_rate": 3.470103039869968e-05, + "loss": 3.205, + "step": 4898 + }, + { + "epoch": 7.17, + "grad_norm": 0.796326220035553, + "learning_rate": 3.466551447206506e-05, + "loss": 3.1567, + "step": 4899 + }, + { + "epoch": 7.17, + "grad_norm": 0.40793824195861816, + "learning_rate": 3.463001291857943e-05, + "loss": 3.1772, + "step": 4900 + }, + { + "epoch": 7.17, + "grad_norm": 0.804324209690094, + "learning_rate": 3.459452574605281e-05, + "loss": 3.181, + "step": 4901 + }, + { + "epoch": 7.17, + "eval_loss": 3.0279037952423096, + "eval_runtime": 1163.8402, + "eval_samples_per_second": 14.729, + "eval_steps_per_second": 0.921, + "step": 4901 + }, + { + "epoch": 7.17, + "grad_norm": 0.47186848521232605, + "learning_rate": 3.455905296229222e-05, + "loss": 3.1301, + "step": 4902 + }, + { + "epoch": 7.17, + "grad_norm": 0.7227790951728821, + "learning_rate": 3.452359457510145e-05, + "loss": 3.1657, + "step": 4903 + }, + { + "epoch": 7.17, + "grad_norm": 0.9041692614555359, + "learning_rate": 3.4488150592281e-05, + "loss": 3.2102, + "step": 4904 + }, + { + "epoch": 7.18, + "grad_norm": 1.273749589920044, + "learning_rate": 3.4452721021628445e-05, + "loss": 3.2073, + "step": 4905 + }, + { + "epoch": 7.18, + "grad_norm": 1.3366678953170776, + "learning_rate": 3.441730587093795e-05, + "loss": 3.1913, + "step": 4906 + }, + { + "epoch": 7.18, + "grad_norm": 0.45730334520339966, + "learning_rate": 3.438190514800063e-05, + "loss": 3.2091, + "step": 4907 + }, + { + "epoch": 7.18, + "grad_norm": 1.0588980913162231, + "learning_rate": 3.4346518860604485e-05, + "loss": 3.2037, + "step": 4908 + }, + { + "epoch": 7.18, + "grad_norm": 0.6257373094558716, + "learning_rate": 3.4311147016534164e-05, + "loss": 3.1525, + "step": 4909 + }, + { + "epoch": 7.18, + "grad_norm": 0.7281100749969482, + "learning_rate": 3.4275789623571234e-05, + "loss": 3.2051, + "step": 4910 + }, + { + "epoch": 7.18, + "grad_norm": 0.8004705905914307, + "learning_rate": 3.42404466894942e-05, + "loss": 3.201, + "step": 4911 + }, + { + "epoch": 7.19, + "grad_norm": 0.34791865944862366, + "learning_rate": 3.4205118222078145e-05, + "loss": 3.1746, + "step": 4912 + }, + { + "epoch": 7.19, + "grad_norm": 1.0412063598632812, + "learning_rate": 3.416980422909516e-05, + "loss": 3.2195, + "step": 4913 + }, + { + "epoch": 7.19, + "grad_norm": 0.489105224609375, + "learning_rate": 3.413450471831407e-05, + "loss": 3.1446, + "step": 4914 + }, + { + "epoch": 7.19, + "grad_norm": 0.49872738122940063, + "learning_rate": 3.4099219697500514e-05, + "loss": 3.128, + "step": 4915 + }, + { + "epoch": 7.19, + "grad_norm": 0.7999401092529297, + "learning_rate": 3.4063949174416984e-05, + "loss": 3.1953, + "step": 4916 + }, + { + "epoch": 7.19, + "grad_norm": 0.4927428364753723, + "learning_rate": 3.402869315682275e-05, + "loss": 3.1533, + "step": 4917 + }, + { + "epoch": 7.19, + "grad_norm": 0.29562923312187195, + "learning_rate": 3.3993451652473916e-05, + "loss": 3.1752, + "step": 4918 + }, + { + "epoch": 7.2, + "grad_norm": 0.9475629329681396, + "learning_rate": 3.395822466912335e-05, + "loss": 3.1526, + "step": 4919 + }, + { + "epoch": 7.2, + "grad_norm": 0.9323421120643616, + "learning_rate": 3.3923012214520763e-05, + "loss": 3.1457, + "step": 4920 + }, + { + "epoch": 7.2, + "grad_norm": 0.30518239736557007, + "learning_rate": 3.388781429641267e-05, + "loss": 3.1892, + "step": 4921 + }, + { + "epoch": 7.2, + "grad_norm": 0.4029679000377655, + "learning_rate": 3.385263092254236e-05, + "loss": 3.1918, + "step": 4922 + }, + { + "epoch": 7.2, + "grad_norm": 0.7984765768051147, + "learning_rate": 3.381746210064995e-05, + "loss": 3.2006, + "step": 4923 + }, + { + "epoch": 7.2, + "grad_norm": 0.9310452342033386, + "learning_rate": 3.378230783847234e-05, + "loss": 3.1526, + "step": 4924 + }, + { + "epoch": 7.21, + "grad_norm": 0.5400900840759277, + "learning_rate": 3.374716814374323e-05, + "loss": 3.2063, + "step": 4925 + }, + { + "epoch": 7.21, + "grad_norm": 0.7315930128097534, + "learning_rate": 3.371204302419313e-05, + "loss": 3.1709, + "step": 4926 + }, + { + "epoch": 7.21, + "grad_norm": 0.8185734152793884, + "learning_rate": 3.367693248754933e-05, + "loss": 3.1659, + "step": 4927 + }, + { + "epoch": 7.21, + "grad_norm": 0.4014480710029602, + "learning_rate": 3.364183654153592e-05, + "loss": 3.1291, + "step": 4928 + }, + { + "epoch": 7.21, + "grad_norm": 0.45573729276657104, + "learning_rate": 3.360675519387371e-05, + "loss": 3.1648, + "step": 4929 + }, + { + "epoch": 7.21, + "grad_norm": 0.49166086316108704, + "learning_rate": 3.357168845228043e-05, + "loss": 3.168, + "step": 4930 + }, + { + "epoch": 7.21, + "grad_norm": 0.37833172082901, + "learning_rate": 3.353663632447054e-05, + "loss": 3.1675, + "step": 4931 + }, + { + "epoch": 7.22, + "grad_norm": 0.9355205297470093, + "learning_rate": 3.350159881815518e-05, + "loss": 3.1798, + "step": 4932 + }, + { + "epoch": 7.22, + "grad_norm": 0.888450562953949, + "learning_rate": 3.346657594104246e-05, + "loss": 3.1609, + "step": 4933 + }, + { + "epoch": 7.22, + "grad_norm": 0.39355748891830444, + "learning_rate": 3.343156770083717e-05, + "loss": 3.1912, + "step": 4934 + }, + { + "epoch": 7.22, + "grad_norm": 0.4100874960422516, + "learning_rate": 3.339657410524081e-05, + "loss": 3.1741, + "step": 4935 + }, + { + "epoch": 7.22, + "grad_norm": 0.6284832954406738, + "learning_rate": 3.336159516195182e-05, + "loss": 3.1456, + "step": 4936 + }, + { + "epoch": 7.22, + "grad_norm": 9.06452465057373, + "learning_rate": 3.3326630878665334e-05, + "loss": 3.1432, + "step": 4937 + }, + { + "epoch": 7.22, + "grad_norm": 1.4369940757751465, + "learning_rate": 3.329168126307319e-05, + "loss": 3.1878, + "step": 4938 + }, + { + "epoch": 7.23, + "grad_norm": 0.6629584431648254, + "learning_rate": 3.3256746322864127e-05, + "loss": 3.1782, + "step": 4939 + }, + { + "epoch": 7.23, + "grad_norm": 1.5383516550064087, + "learning_rate": 3.322182606572357e-05, + "loss": 3.1567, + "step": 4940 + }, + { + "epoch": 7.23, + "grad_norm": 1.498374581336975, + "learning_rate": 3.318692049933375e-05, + "loss": 3.2173, + "step": 4941 + }, + { + "epoch": 7.23, + "grad_norm": 0.5786330699920654, + "learning_rate": 3.315202963137366e-05, + "loss": 3.1327, + "step": 4942 + }, + { + "epoch": 7.23, + "grad_norm": 1.2023847103118896, + "learning_rate": 3.311715346951906e-05, + "loss": 3.2942, + "step": 4943 + }, + { + "epoch": 7.23, + "grad_norm": 0.6449021697044373, + "learning_rate": 3.3082292021442454e-05, + "loss": 3.1564, + "step": 4944 + }, + { + "epoch": 7.24, + "grad_norm": 0.36229845881462097, + "learning_rate": 3.304744529481315e-05, + "loss": 3.1787, + "step": 4945 + }, + { + "epoch": 7.24, + "grad_norm": 0.9449754953384399, + "learning_rate": 3.301261329729719e-05, + "loss": 3.1928, + "step": 4946 + }, + { + "epoch": 7.24, + "grad_norm": 0.9095726609230042, + "learning_rate": 3.297779603655736e-05, + "loss": 3.2106, + "step": 4947 + }, + { + "epoch": 7.24, + "grad_norm": 0.346010684967041, + "learning_rate": 3.2942993520253254e-05, + "loss": 3.1908, + "step": 4948 + }, + { + "epoch": 7.24, + "grad_norm": 0.7208975553512573, + "learning_rate": 3.2908205756041176e-05, + "loss": 3.1851, + "step": 4949 + }, + { + "epoch": 7.24, + "grad_norm": 0.5966201424598694, + "learning_rate": 3.2873432751574194e-05, + "loss": 3.2213, + "step": 4950 + }, + { + "epoch": 7.24, + "grad_norm": 0.3596455156803131, + "learning_rate": 3.2838674514502154e-05, + "loss": 3.2082, + "step": 4951 + }, + { + "epoch": 7.25, + "grad_norm": 0.8812998533248901, + "learning_rate": 3.2803931052471615e-05, + "loss": 3.1496, + "step": 4952 + }, + { + "epoch": 7.25, + "grad_norm": 0.7055163979530334, + "learning_rate": 3.276920237312593e-05, + "loss": 3.1859, + "step": 4953 + }, + { + "epoch": 7.25, + "grad_norm": 0.7300843596458435, + "learning_rate": 3.273448848410515e-05, + "loss": 3.1804, + "step": 4954 + }, + { + "epoch": 7.25, + "grad_norm": 1.0114784240722656, + "learning_rate": 3.269978939304612e-05, + "loss": 3.1612, + "step": 4955 + }, + { + "epoch": 7.25, + "grad_norm": 0.46303462982177734, + "learning_rate": 3.266510510758238e-05, + "loss": 3.1675, + "step": 4956 + }, + { + "epoch": 7.25, + "grad_norm": 0.722417950630188, + "learning_rate": 3.263043563534428e-05, + "loss": 3.1732, + "step": 4957 + }, + { + "epoch": 7.25, + "grad_norm": 0.6245961785316467, + "learning_rate": 3.2595780983958834e-05, + "loss": 3.1481, + "step": 4958 + }, + { + "epoch": 7.26, + "grad_norm": 0.31019535660743713, + "learning_rate": 3.256114116104986e-05, + "loss": 3.1808, + "step": 4959 + }, + { + "epoch": 7.26, + "grad_norm": 1.2723808288574219, + "learning_rate": 3.25265161742379e-05, + "loss": 3.1502, + "step": 4960 + }, + { + "epoch": 7.26, + "grad_norm": 0.6151775121688843, + "learning_rate": 3.249190603114013e-05, + "loss": 3.1393, + "step": 4961 + }, + { + "epoch": 7.26, + "grad_norm": 1.0856354236602783, + "learning_rate": 3.245731073937068e-05, + "loss": 3.1964, + "step": 4962 + }, + { + "epoch": 7.26, + "grad_norm": 0.665896475315094, + "learning_rate": 3.2422730306540186e-05, + "loss": 3.1715, + "step": 4963 + }, + { + "epoch": 7.26, + "grad_norm": 1.0534874200820923, + "learning_rate": 3.23881647402561e-05, + "loss": 3.1952, + "step": 4964 + }, + { + "epoch": 7.26, + "grad_norm": 1.2023156881332397, + "learning_rate": 3.235361404812274e-05, + "loss": 3.1622, + "step": 4965 + }, + { + "epoch": 7.27, + "grad_norm": 0.9104660749435425, + "learning_rate": 3.23190782377409e-05, + "loss": 3.1731, + "step": 4966 + }, + { + "epoch": 7.27, + "grad_norm": 1.6818856000900269, + "learning_rate": 3.228455731670823e-05, + "loss": 3.2543, + "step": 4967 + }, + { + "epoch": 7.27, + "grad_norm": 0.40944576263427734, + "learning_rate": 3.225005129261922e-05, + "loss": 3.1609, + "step": 4968 + }, + { + "epoch": 7.27, + "grad_norm": 2.257155656814575, + "learning_rate": 3.221556017306484e-05, + "loss": 3.2051, + "step": 4969 + }, + { + "epoch": 7.27, + "grad_norm": 1.4875580072402954, + "learning_rate": 3.218108396563293e-05, + "loss": 3.1654, + "step": 4970 + }, + { + "epoch": 7.27, + "grad_norm": 0.7136067152023315, + "learning_rate": 3.21466226779081e-05, + "loss": 3.1851, + "step": 4971 + }, + { + "epoch": 7.28, + "grad_norm": 1.1615363359451294, + "learning_rate": 3.2112176317471535e-05, + "loss": 3.1378, + "step": 4972 + }, + { + "epoch": 7.28, + "grad_norm": 1.23421311378479, + "learning_rate": 3.20777448919012e-05, + "loss": 3.1976, + "step": 4973 + }, + { + "epoch": 7.28, + "grad_norm": 0.3943505883216858, + "learning_rate": 3.204332840877179e-05, + "loss": 3.1514, + "step": 4974 + }, + { + "epoch": 7.28, + "grad_norm": 1.666935682296753, + "learning_rate": 3.200892687565471e-05, + "loss": 3.1276, + "step": 4975 + }, + { + "epoch": 7.28, + "grad_norm": 0.877071738243103, + "learning_rate": 3.1974540300118064e-05, + "loss": 3.1572, + "step": 4976 + }, + { + "epoch": 7.28, + "grad_norm": 1.395138144493103, + "learning_rate": 3.1940168689726666e-05, + "loss": 3.1789, + "step": 4977 + }, + { + "epoch": 7.28, + "grad_norm": 1.2696112394332886, + "learning_rate": 3.190581205204204e-05, + "loss": 3.1909, + "step": 4978 + }, + { + "epoch": 7.29, + "grad_norm": 0.9251769185066223, + "learning_rate": 3.1871470394622404e-05, + "loss": 3.1486, + "step": 4979 + }, + { + "epoch": 7.29, + "grad_norm": 0.9306465983390808, + "learning_rate": 3.183714372502272e-05, + "loss": 3.1897, + "step": 4980 + }, + { + "epoch": 7.29, + "grad_norm": 0.6166656613349915, + "learning_rate": 3.1802832050794595e-05, + "loss": 3.1902, + "step": 4981 + }, + { + "epoch": 7.29, + "grad_norm": 1.0859153270721436, + "learning_rate": 3.1768535379486395e-05, + "loss": 3.2179, + "step": 4982 + }, + { + "epoch": 7.29, + "grad_norm": 0.5373363494873047, + "learning_rate": 3.173425371864314e-05, + "loss": 3.1732, + "step": 4983 + }, + { + "epoch": 7.29, + "grad_norm": 0.6277896761894226, + "learning_rate": 3.169998707580657e-05, + "loss": 3.1735, + "step": 4984 + }, + { + "epoch": 7.29, + "grad_norm": 0.8959996104240417, + "learning_rate": 3.166573545851514e-05, + "loss": 3.1425, + "step": 4985 + }, + { + "epoch": 7.3, + "grad_norm": 0.5353897213935852, + "learning_rate": 3.16314988743039e-05, + "loss": 3.1447, + "step": 4986 + }, + { + "epoch": 7.3, + "grad_norm": 0.8923834562301636, + "learning_rate": 3.1597277330704756e-05, + "loss": 3.1684, + "step": 4987 + }, + { + "epoch": 7.3, + "grad_norm": 0.456402063369751, + "learning_rate": 3.1563070835246213e-05, + "loss": 3.1917, + "step": 4988 + }, + { + "epoch": 7.3, + "grad_norm": 0.744957685470581, + "learning_rate": 3.152887939545337e-05, + "loss": 3.205, + "step": 4989 + }, + { + "epoch": 7.3, + "grad_norm": 0.7469781041145325, + "learning_rate": 3.149470301884823e-05, + "loss": 3.1544, + "step": 4990 + }, + { + "epoch": 7.3, + "grad_norm": 0.48788437247276306, + "learning_rate": 3.146054171294935e-05, + "loss": 3.1809, + "step": 4991 + }, + { + "epoch": 7.3, + "grad_norm": 1.1828556060791016, + "learning_rate": 3.14263954852719e-05, + "loss": 3.1552, + "step": 4992 + }, + { + "epoch": 7.31, + "grad_norm": 0.35930806398391724, + "learning_rate": 3.1392264343327914e-05, + "loss": 3.1549, + "step": 4993 + }, + { + "epoch": 7.31, + "grad_norm": 1.010923981666565, + "learning_rate": 3.135814829462602e-05, + "loss": 3.181, + "step": 4994 + }, + { + "epoch": 7.31, + "grad_norm": 1.8422837257385254, + "learning_rate": 3.132404734667142e-05, + "loss": 3.1774, + "step": 4995 + }, + { + "epoch": 7.31, + "grad_norm": 0.7072595357894897, + "learning_rate": 3.1289961506966214e-05, + "loss": 3.1996, + "step": 4996 + }, + { + "epoch": 7.31, + "grad_norm": 0.8350168466567993, + "learning_rate": 3.125589078300898e-05, + "loss": 3.2288, + "step": 4997 + }, + { + "epoch": 7.31, + "grad_norm": 0.7736479043960571, + "learning_rate": 3.122183518229503e-05, + "loss": 3.1702, + "step": 4998 + }, + { + "epoch": 7.32, + "grad_norm": 0.6932786107063293, + "learning_rate": 3.1187794712316474e-05, + "loss": 3.1662, + "step": 4999 + }, + { + "epoch": 7.32, + "grad_norm": 0.5773895382881165, + "learning_rate": 3.11537693805619e-05, + "loss": 3.1859, + "step": 5000 + }, + { + "epoch": 7.32, + "grad_norm": 0.8364612460136414, + "learning_rate": 3.111975919451664e-05, + "loss": 3.1769, + "step": 5001 + }, + { + "epoch": 7.32, + "grad_norm": 0.3345867097377777, + "learning_rate": 3.10857641616628e-05, + "loss": 3.1939, + "step": 5002 + }, + { + "epoch": 7.32, + "grad_norm": 0.6873171925544739, + "learning_rate": 3.105178428947898e-05, + "loss": 3.2176, + "step": 5003 + }, + { + "epoch": 7.32, + "grad_norm": 0.5719138383865356, + "learning_rate": 3.10178195854405e-05, + "loss": 3.174, + "step": 5004 + }, + { + "epoch": 7.32, + "grad_norm": 0.9275087714195251, + "learning_rate": 3.0983870057019484e-05, + "loss": 3.2174, + "step": 5005 + }, + { + "epoch": 7.33, + "grad_norm": 1.1784770488739014, + "learning_rate": 3.09499357116845e-05, + "loss": 3.1778, + "step": 5006 + }, + { + "epoch": 7.33, + "grad_norm": 0.7836307287216187, + "learning_rate": 3.0916016556900904e-05, + "loss": 3.1627, + "step": 5007 + }, + { + "epoch": 7.33, + "grad_norm": 0.482113242149353, + "learning_rate": 3.088211260013068e-05, + "loss": 3.1841, + "step": 5008 + }, + { + "epoch": 7.33, + "grad_norm": 0.48355069756507874, + "learning_rate": 3.084822384883249e-05, + "loss": 3.1691, + "step": 5009 + }, + { + "epoch": 7.33, + "grad_norm": 0.4077546000480652, + "learning_rate": 3.0814350310461615e-05, + "loss": 3.1581, + "step": 5010 + }, + { + "epoch": 7.33, + "grad_norm": 0.7942917943000793, + "learning_rate": 3.078049199247e-05, + "loss": 3.1964, + "step": 5011 + }, + { + "epoch": 7.33, + "grad_norm": 0.38686656951904297, + "learning_rate": 3.0746648902306264e-05, + "loss": 3.1606, + "step": 5012 + }, + { + "epoch": 7.34, + "grad_norm": 1.3776711225509644, + "learning_rate": 3.071282104741565e-05, + "loss": 3.2248, + "step": 5013 + }, + { + "epoch": 7.34, + "grad_norm": 0.2463265061378479, + "learning_rate": 3.067900843524008e-05, + "loss": 3.1848, + "step": 5014 + }, + { + "epoch": 7.34, + "grad_norm": 0.5669382214546204, + "learning_rate": 3.064521107321806e-05, + "loss": 3.1501, + "step": 5015 + }, + { + "epoch": 7.34, + "grad_norm": 0.40979495644569397, + "learning_rate": 3.061142896878486e-05, + "loss": 3.1762, + "step": 5016 + }, + { + "epoch": 7.34, + "grad_norm": 0.5121455192565918, + "learning_rate": 3.05776621293722e-05, + "loss": 3.1893, + "step": 5017 + }, + { + "epoch": 7.34, + "grad_norm": 0.3571344017982483, + "learning_rate": 3.054391056240865e-05, + "loss": 3.1726, + "step": 5018 + }, + { + "epoch": 7.34, + "grad_norm": 0.3279051184654236, + "learning_rate": 3.0510174275319325e-05, + "loss": 3.1238, + "step": 5019 + }, + { + "epoch": 7.35, + "grad_norm": 0.5023763179779053, + "learning_rate": 3.0476453275525906e-05, + "loss": 3.1893, + "step": 5020 + }, + { + "epoch": 7.35, + "grad_norm": 0.45298734307289124, + "learning_rate": 3.0442747570446872e-05, + "loss": 3.192, + "step": 5021 + }, + { + "epoch": 7.35, + "grad_norm": 0.580197274684906, + "learning_rate": 3.0409057167497258e-05, + "loss": 3.1934, + "step": 5022 + }, + { + "epoch": 7.35, + "grad_norm": 0.4648624360561371, + "learning_rate": 3.037538207408862e-05, + "loss": 3.1733, + "step": 5023 + }, + { + "epoch": 7.35, + "grad_norm": 0.8324884176254272, + "learning_rate": 3.034172229762935e-05, + "loss": 3.1584, + "step": 5024 + }, + { + "epoch": 7.35, + "grad_norm": 0.6418886184692383, + "learning_rate": 3.0308077845524384e-05, + "loss": 3.1809, + "step": 5025 + }, + { + "epoch": 7.36, + "grad_norm": 0.7795079946517944, + "learning_rate": 3.0274448725175176e-05, + "loss": 3.1515, + "step": 5026 + }, + { + "epoch": 7.36, + "grad_norm": 0.5563576221466064, + "learning_rate": 3.024083494397999e-05, + "loss": 3.1772, + "step": 5027 + }, + { + "epoch": 7.36, + "grad_norm": 0.42387935519218445, + "learning_rate": 3.0207236509333647e-05, + "loss": 3.1866, + "step": 5028 + }, + { + "epoch": 7.36, + "grad_norm": 0.5205991864204407, + "learning_rate": 3.017365342862748e-05, + "loss": 3.1517, + "step": 5029 + }, + { + "epoch": 7.36, + "grad_norm": 0.5210472941398621, + "learning_rate": 3.0140085709249667e-05, + "loss": 3.172, + "step": 5030 + }, + { + "epoch": 7.36, + "grad_norm": 0.5895211100578308, + "learning_rate": 3.010653335858479e-05, + "loss": 3.239, + "step": 5031 + }, + { + "epoch": 7.36, + "grad_norm": 0.38003700971603394, + "learning_rate": 3.007299638401416e-05, + "loss": 3.1581, + "step": 5032 + }, + { + "epoch": 7.37, + "grad_norm": 0.3639942705631256, + "learning_rate": 3.0039474792915702e-05, + "loss": 3.2157, + "step": 5033 + }, + { + "epoch": 7.37, + "grad_norm": 1.5683830976486206, + "learning_rate": 3.0005968592663925e-05, + "loss": 3.1839, + "step": 5034 + }, + { + "epoch": 7.37, + "grad_norm": 0.5200375914573669, + "learning_rate": 2.9972477790629994e-05, + "loss": 3.1836, + "step": 5035 + }, + { + "epoch": 7.37, + "grad_norm": 0.7715541124343872, + "learning_rate": 2.9939002394181637e-05, + "loss": 3.1778, + "step": 5036 + }, + { + "epoch": 7.37, + "grad_norm": 1.0058480501174927, + "learning_rate": 2.9905542410683218e-05, + "loss": 3.1782, + "step": 5037 + }, + { + "epoch": 7.37, + "grad_norm": 0.6472846269607544, + "learning_rate": 2.987209784749573e-05, + "loss": 3.1847, + "step": 5038 + }, + { + "epoch": 7.37, + "grad_norm": 1.0543522834777832, + "learning_rate": 2.9838668711976724e-05, + "loss": 3.1228, + "step": 5039 + }, + { + "epoch": 7.38, + "grad_norm": 0.9698706865310669, + "learning_rate": 2.9805255011480415e-05, + "loss": 3.163, + "step": 5040 + }, + { + "epoch": 7.38, + "grad_norm": 0.7419912219047546, + "learning_rate": 2.9771856753357573e-05, + "loss": 3.1658, + "step": 5041 + }, + { + "epoch": 7.38, + "grad_norm": 0.8493951559066772, + "learning_rate": 2.9738473944955603e-05, + "loss": 3.1264, + "step": 5042 + }, + { + "epoch": 7.38, + "grad_norm": 0.9127371311187744, + "learning_rate": 2.970510659361848e-05, + "loss": 3.2144, + "step": 5043 + }, + { + "epoch": 7.38, + "grad_norm": 0.5002891421318054, + "learning_rate": 2.967175470668683e-05, + "loss": 3.1733, + "step": 5044 + }, + { + "epoch": 7.38, + "grad_norm": 0.7260929346084595, + "learning_rate": 2.9638418291497817e-05, + "loss": 3.1292, + "step": 5045 + }, + { + "epoch": 7.39, + "grad_norm": 0.8466237783432007, + "learning_rate": 2.9605097355385227e-05, + "loss": 3.194, + "step": 5046 + }, + { + "epoch": 7.39, + "grad_norm": 0.29703596234321594, + "learning_rate": 2.9571791905679468e-05, + "loss": 3.1485, + "step": 5047 + }, + { + "epoch": 7.39, + "grad_norm": 0.8655198216438293, + "learning_rate": 2.9538501949707487e-05, + "loss": 3.1638, + "step": 5048 + }, + { + "epoch": 7.39, + "grad_norm": 0.5336217880249023, + "learning_rate": 2.9505227494792874e-05, + "loss": 3.1671, + "step": 5049 + }, + { + "epoch": 7.39, + "grad_norm": 0.2816140949726105, + "learning_rate": 2.947196854825581e-05, + "loss": 3.1982, + "step": 5050 + }, + { + "epoch": 7.39, + "grad_norm": 0.6799694299697876, + "learning_rate": 2.9438725117412947e-05, + "loss": 3.1824, + "step": 5051 + }, + { + "epoch": 7.39, + "grad_norm": 0.7424922585487366, + "learning_rate": 2.9405497209577703e-05, + "loss": 3.1787, + "step": 5052 + }, + { + "epoch": 7.4, + "grad_norm": 0.5237893462181091, + "learning_rate": 2.9372284832059993e-05, + "loss": 3.1705, + "step": 5053 + }, + { + "epoch": 7.4, + "grad_norm": 0.7378460764884949, + "learning_rate": 2.933908799216625e-05, + "loss": 3.1769, + "step": 5054 + }, + { + "epoch": 7.4, + "grad_norm": 0.9311485886573792, + "learning_rate": 2.9305906697199615e-05, + "loss": 3.1574, + "step": 5055 + }, + { + "epoch": 7.4, + "grad_norm": 0.41424453258514404, + "learning_rate": 2.9272740954459776e-05, + "loss": 3.132, + "step": 5056 + }, + { + "epoch": 7.4, + "grad_norm": 0.4336400330066681, + "learning_rate": 2.92395907712429e-05, + "loss": 3.1724, + "step": 5057 + }, + { + "epoch": 7.4, + "grad_norm": 0.902377188205719, + "learning_rate": 2.920645615484181e-05, + "loss": 3.1471, + "step": 5058 + }, + { + "epoch": 7.4, + "grad_norm": 0.7323518991470337, + "learning_rate": 2.9173337112545985e-05, + "loss": 3.2012, + "step": 5059 + }, + { + "epoch": 7.41, + "grad_norm": 0.6311755776405334, + "learning_rate": 2.9140233651641314e-05, + "loss": 3.1655, + "step": 5060 + }, + { + "epoch": 7.41, + "grad_norm": 1.4856607913970947, + "learning_rate": 2.9107145779410317e-05, + "loss": 3.1768, + "step": 5061 + }, + { + "epoch": 7.41, + "grad_norm": 0.5299568176269531, + "learning_rate": 2.9074073503132217e-05, + "loss": 3.1368, + "step": 5062 + }, + { + "epoch": 7.41, + "grad_norm": 0.5073440074920654, + "learning_rate": 2.904101683008259e-05, + "loss": 3.1784, + "step": 5063 + }, + { + "epoch": 7.41, + "grad_norm": 0.7405803799629211, + "learning_rate": 2.9007975767533714e-05, + "loss": 3.145, + "step": 5064 + }, + { + "epoch": 7.41, + "grad_norm": 0.39113080501556396, + "learning_rate": 2.897495032275439e-05, + "loss": 3.148, + "step": 5065 + }, + { + "epoch": 7.41, + "grad_norm": 0.8356419801712036, + "learning_rate": 2.8941940503010022e-05, + "loss": 3.1611, + "step": 5066 + }, + { + "epoch": 7.42, + "grad_norm": 0.6192629337310791, + "learning_rate": 2.8908946315562524e-05, + "loss": 3.1742, + "step": 5067 + }, + { + "epoch": 7.42, + "grad_norm": 1.1198054552078247, + "learning_rate": 2.8875967767670408e-05, + "loss": 3.1789, + "step": 5068 + }, + { + "epoch": 7.42, + "grad_norm": 0.4452805519104004, + "learning_rate": 2.884300486658873e-05, + "loss": 3.1533, + "step": 5069 + }, + { + "epoch": 7.42, + "grad_norm": 0.8051960468292236, + "learning_rate": 2.881005761956911e-05, + "loss": 3.1688, + "step": 5070 + }, + { + "epoch": 7.42, + "eval_loss": 3.0202746391296387, + "eval_runtime": 1162.3223, + "eval_samples_per_second": 14.748, + "eval_steps_per_second": 0.922, + "step": 5070 + }, + { + "epoch": 7.42, + "grad_norm": 1.1692814826965332, + "learning_rate": 2.8777126033859726e-05, + "loss": 3.1559, + "step": 5071 + }, + { + "epoch": 7.42, + "grad_norm": 0.3186073303222656, + "learning_rate": 2.8744210116705296e-05, + "loss": 3.1507, + "step": 5072 + }, + { + "epoch": 7.43, + "grad_norm": 0.6718719005584717, + "learning_rate": 2.8711309875347102e-05, + "loss": 3.1667, + "step": 5073 + }, + { + "epoch": 7.43, + "grad_norm": 0.7967109680175781, + "learning_rate": 2.8678425317022973e-05, + "loss": 3.1415, + "step": 5074 + }, + { + "epoch": 7.43, + "grad_norm": 0.36231231689453125, + "learning_rate": 2.8645556448967315e-05, + "loss": 3.1264, + "step": 5075 + }, + { + "epoch": 7.43, + "grad_norm": 0.5855808258056641, + "learning_rate": 2.8612703278411025e-05, + "loss": 3.1217, + "step": 5076 + }, + { + "epoch": 7.43, + "grad_norm": 0.6386885046958923, + "learning_rate": 2.8579865812581608e-05, + "loss": 3.1476, + "step": 5077 + }, + { + "epoch": 7.43, + "grad_norm": 0.4413103759288788, + "learning_rate": 2.8547044058703076e-05, + "loss": 3.2046, + "step": 5078 + }, + { + "epoch": 7.43, + "grad_norm": 0.7146703004837036, + "learning_rate": 2.851423802399599e-05, + "loss": 3.2516, + "step": 5079 + }, + { + "epoch": 7.44, + "grad_norm": 0.7587980031967163, + "learning_rate": 2.8481447715677466e-05, + "loss": 3.2025, + "step": 5080 + }, + { + "epoch": 7.44, + "grad_norm": 1.2787212133407593, + "learning_rate": 2.8448673140961147e-05, + "loss": 3.1572, + "step": 5081 + }, + { + "epoch": 7.44, + "grad_norm": 0.5361132621765137, + "learning_rate": 2.8415914307057245e-05, + "loss": 3.1564, + "step": 5082 + }, + { + "epoch": 7.44, + "grad_norm": 1.6003608703613281, + "learning_rate": 2.8383171221172412e-05, + "loss": 3.1482, + "step": 5083 + }, + { + "epoch": 7.44, + "grad_norm": 0.9171953201293945, + "learning_rate": 2.8350443890510003e-05, + "loss": 3.1886, + "step": 5084 + }, + { + "epoch": 7.44, + "grad_norm": 0.39613091945648193, + "learning_rate": 2.8317732322269754e-05, + "loss": 3.1506, + "step": 5085 + }, + { + "epoch": 7.44, + "grad_norm": 1.17970609664917, + "learning_rate": 2.8285036523647944e-05, + "loss": 3.1526, + "step": 5086 + }, + { + "epoch": 7.45, + "grad_norm": 0.781102180480957, + "learning_rate": 2.8252356501837562e-05, + "loss": 3.1663, + "step": 5087 + }, + { + "epoch": 7.45, + "grad_norm": 0.49025315046310425, + "learning_rate": 2.8219692264027886e-05, + "loss": 3.1647, + "step": 5088 + }, + { + "epoch": 7.45, + "grad_norm": 2.0356221199035645, + "learning_rate": 2.8187043817404833e-05, + "loss": 3.2115, + "step": 5089 + }, + { + "epoch": 7.45, + "grad_norm": 1.2565196752548218, + "learning_rate": 2.815441116915093e-05, + "loss": 3.2334, + "step": 5090 + }, + { + "epoch": 7.45, + "grad_norm": 0.6307234168052673, + "learning_rate": 2.812179432644506e-05, + "loss": 3.1614, + "step": 5091 + }, + { + "epoch": 7.45, + "grad_norm": 0.8804566264152527, + "learning_rate": 2.808919329646269e-05, + "loss": 3.1703, + "step": 5092 + }, + { + "epoch": 7.45, + "grad_norm": 0.5766822695732117, + "learning_rate": 2.8056608086375934e-05, + "loss": 3.1264, + "step": 5093 + }, + { + "epoch": 7.46, + "grad_norm": 0.9480114579200745, + "learning_rate": 2.8024038703353227e-05, + "loss": 3.1382, + "step": 5094 + }, + { + "epoch": 7.46, + "grad_norm": 0.5312039852142334, + "learning_rate": 2.7991485154559617e-05, + "loss": 3.1555, + "step": 5095 + }, + { + "epoch": 7.46, + "grad_norm": 0.5020516514778137, + "learning_rate": 2.7958947447156746e-05, + "loss": 3.1825, + "step": 5096 + }, + { + "epoch": 7.46, + "grad_norm": 0.8643108010292053, + "learning_rate": 2.7926425588302606e-05, + "loss": 3.1903, + "step": 5097 + }, + { + "epoch": 7.46, + "grad_norm": 0.38518649339675903, + "learning_rate": 2.789391958515183e-05, + "loss": 3.1832, + "step": 5098 + }, + { + "epoch": 7.46, + "grad_norm": 0.9875998497009277, + "learning_rate": 2.7861429444855514e-05, + "loss": 3.1695, + "step": 5099 + }, + { + "epoch": 7.47, + "grad_norm": 0.5301055312156677, + "learning_rate": 2.7828955174561257e-05, + "loss": 3.1645, + "step": 5100 + }, + { + "epoch": 7.47, + "grad_norm": 0.7682579755783081, + "learning_rate": 2.7796496781413206e-05, + "loss": 3.1384, + "step": 5101 + }, + { + "epoch": 7.47, + "grad_norm": 1.0809067487716675, + "learning_rate": 2.7764054272551965e-05, + "loss": 3.1427, + "step": 5102 + }, + { + "epoch": 7.47, + "grad_norm": 0.6575438380241394, + "learning_rate": 2.773162765511469e-05, + "loss": 3.1146, + "step": 5103 + }, + { + "epoch": 7.47, + "grad_norm": 1.6334797143936157, + "learning_rate": 2.7699216936234994e-05, + "loss": 3.1582, + "step": 5104 + }, + { + "epoch": 7.47, + "grad_norm": 1.1224489212036133, + "learning_rate": 2.7666822123043046e-05, + "loss": 3.1531, + "step": 5105 + }, + { + "epoch": 7.47, + "grad_norm": 0.6989713311195374, + "learning_rate": 2.7634443222665475e-05, + "loss": 3.1587, + "step": 5106 + }, + { + "epoch": 7.48, + "grad_norm": 1.6318511962890625, + "learning_rate": 2.760208024222546e-05, + "loss": 3.1676, + "step": 5107 + }, + { + "epoch": 7.48, + "grad_norm": 1.0403834581375122, + "learning_rate": 2.7569733188842528e-05, + "loss": 3.2076, + "step": 5108 + }, + { + "epoch": 7.48, + "grad_norm": 0.9188293814659119, + "learning_rate": 2.7537402069632922e-05, + "loss": 3.1641, + "step": 5109 + }, + { + "epoch": 7.48, + "grad_norm": 1.1109488010406494, + "learning_rate": 2.7505086891709277e-05, + "loss": 3.1396, + "step": 5110 + }, + { + "epoch": 7.48, + "grad_norm": 0.40556856989860535, + "learning_rate": 2.747278766218061e-05, + "loss": 3.1475, + "step": 5111 + }, + { + "epoch": 7.48, + "grad_norm": 1.6644717454910278, + "learning_rate": 2.744050438815263e-05, + "loss": 3.1237, + "step": 5112 + }, + { + "epoch": 7.48, + "grad_norm": 1.2431093454360962, + "learning_rate": 2.740823707672744e-05, + "loss": 3.1342, + "step": 5113 + }, + { + "epoch": 7.49, + "grad_norm": 0.5421004295349121, + "learning_rate": 2.7375985735003552e-05, + "loss": 3.153, + "step": 5114 + }, + { + "epoch": 7.49, + "grad_norm": 1.7382439374923706, + "learning_rate": 2.7343750370076125e-05, + "loss": 3.1935, + "step": 5115 + }, + { + "epoch": 7.49, + "grad_norm": 0.757526695728302, + "learning_rate": 2.7311530989036726e-05, + "loss": 3.173, + "step": 5116 + }, + { + "epoch": 7.49, + "grad_norm": 1.094909906387329, + "learning_rate": 2.7279327598973303e-05, + "loss": 3.1654, + "step": 5117 + }, + { + "epoch": 7.49, + "grad_norm": 1.3784464597702026, + "learning_rate": 2.7247140206970534e-05, + "loss": 3.1394, + "step": 5118 + }, + { + "epoch": 7.49, + "grad_norm": 0.33335521817207336, + "learning_rate": 2.721496882010931e-05, + "loss": 3.1698, + "step": 5119 + }, + { + "epoch": 7.49, + "grad_norm": 0.5880882740020752, + "learning_rate": 2.7182813445467138e-05, + "loss": 3.1336, + "step": 5120 + }, + { + "epoch": 7.5, + "grad_norm": 1.3899345397949219, + "learning_rate": 2.7150674090118054e-05, + "loss": 3.2018, + "step": 5121 + }, + { + "epoch": 7.5, + "grad_norm": 0.4972773790359497, + "learning_rate": 2.7118550761132444e-05, + "loss": 3.174, + "step": 5122 + }, + { + "epoch": 7.5, + "grad_norm": 1.0646159648895264, + "learning_rate": 2.7086443465577195e-05, + "loss": 3.16, + "step": 5123 + }, + { + "epoch": 7.5, + "grad_norm": 0.5726083517074585, + "learning_rate": 2.705435221051581e-05, + "loss": 3.1459, + "step": 5124 + }, + { + "epoch": 7.5, + "grad_norm": 0.5148174166679382, + "learning_rate": 2.7022277003008035e-05, + "loss": 3.1702, + "step": 5125 + }, + { + "epoch": 7.5, + "grad_norm": 0.5842214226722717, + "learning_rate": 2.6990217850110222e-05, + "loss": 3.189, + "step": 5126 + }, + { + "epoch": 7.51, + "grad_norm": 0.450296550989151, + "learning_rate": 2.6958174758875253e-05, + "loss": 3.1581, + "step": 5127 + }, + { + "epoch": 7.51, + "grad_norm": 0.6981111168861389, + "learning_rate": 2.6926147736352302e-05, + "loss": 3.1709, + "step": 5128 + }, + { + "epoch": 7.51, + "grad_norm": 0.6360636949539185, + "learning_rate": 2.689413678958713e-05, + "loss": 3.163, + "step": 5129 + }, + { + "epoch": 7.51, + "grad_norm": 0.6991342306137085, + "learning_rate": 2.686214192562193e-05, + "loss": 3.1673, + "step": 5130 + }, + { + "epoch": 7.51, + "grad_norm": 0.4301992952823639, + "learning_rate": 2.6830163151495358e-05, + "loss": 3.15, + "step": 5131 + }, + { + "epoch": 7.51, + "grad_norm": 0.6416985988616943, + "learning_rate": 2.679820047424253e-05, + "loss": 3.1364, + "step": 5132 + }, + { + "epoch": 7.51, + "grad_norm": 0.6713948845863342, + "learning_rate": 2.6766253900895023e-05, + "loss": 3.1668, + "step": 5133 + }, + { + "epoch": 7.52, + "grad_norm": 0.4859183728694916, + "learning_rate": 2.6734323438480858e-05, + "loss": 3.153, + "step": 5134 + }, + { + "epoch": 7.52, + "grad_norm": 0.5950589179992676, + "learning_rate": 2.6702409094024538e-05, + "loss": 3.164, + "step": 5135 + }, + { + "epoch": 7.52, + "grad_norm": 0.4921453297138214, + "learning_rate": 2.6670510874547006e-05, + "loss": 3.1513, + "step": 5136 + }, + { + "epoch": 7.52, + "grad_norm": 0.3888862133026123, + "learning_rate": 2.6638628787065633e-05, + "loss": 3.1621, + "step": 5137 + }, + { + "epoch": 7.52, + "grad_norm": 0.43584099411964417, + "learning_rate": 2.660676283859429e-05, + "loss": 3.1327, + "step": 5138 + }, + { + "epoch": 7.52, + "grad_norm": 0.659436047077179, + "learning_rate": 2.6574913036143266e-05, + "loss": 3.1782, + "step": 5139 + }, + { + "epoch": 7.52, + "grad_norm": 0.33366990089416504, + "learning_rate": 2.6543079386719295e-05, + "loss": 3.1657, + "step": 5140 + }, + { + "epoch": 7.53, + "grad_norm": 0.6816606521606445, + "learning_rate": 2.651126189732561e-05, + "loss": 3.145, + "step": 5141 + }, + { + "epoch": 7.53, + "grad_norm": 0.5122085213661194, + "learning_rate": 2.6479460574961746e-05, + "loss": 3.1146, + "step": 5142 + }, + { + "epoch": 7.53, + "grad_norm": 0.7051358819007874, + "learning_rate": 2.6447675426623864e-05, + "loss": 3.1565, + "step": 5143 + }, + { + "epoch": 7.53, + "grad_norm": 0.3721584379673004, + "learning_rate": 2.6415906459304495e-05, + "loss": 3.165, + "step": 5144 + }, + { + "epoch": 7.53, + "grad_norm": 0.30766353011131287, + "learning_rate": 2.638415367999252e-05, + "loss": 3.1973, + "step": 5145 + }, + { + "epoch": 7.53, + "grad_norm": 0.41517648100852966, + "learning_rate": 2.63524170956734e-05, + "loss": 3.1766, + "step": 5146 + }, + { + "epoch": 7.53, + "grad_norm": 0.433137983083725, + "learning_rate": 2.6320696713328986e-05, + "loss": 3.2217, + "step": 5147 + }, + { + "epoch": 7.54, + "grad_norm": 0.3852587044239044, + "learning_rate": 2.6288992539937463e-05, + "loss": 3.1605, + "step": 5148 + }, + { + "epoch": 7.54, + "grad_norm": 0.5037500858306885, + "learning_rate": 2.625730458247362e-05, + "loss": 3.1981, + "step": 5149 + }, + { + "epoch": 7.54, + "grad_norm": 0.4805777966976166, + "learning_rate": 2.62256328479086e-05, + "loss": 3.1315, + "step": 5150 + }, + { + "epoch": 7.54, + "grad_norm": 0.7443678975105286, + "learning_rate": 2.6193977343209874e-05, + "loss": 3.173, + "step": 5151 + }, + { + "epoch": 7.54, + "grad_norm": 0.6087154746055603, + "learning_rate": 2.6162338075341564e-05, + "loss": 3.1921, + "step": 5152 + }, + { + "epoch": 7.54, + "grad_norm": 2.0148138999938965, + "learning_rate": 2.613071505126402e-05, + "loss": 3.1314, + "step": 5153 + }, + { + "epoch": 7.55, + "grad_norm": 0.5105882883071899, + "learning_rate": 2.6099108277934103e-05, + "loss": 3.1305, + "step": 5154 + }, + { + "epoch": 7.55, + "grad_norm": 0.4445195496082306, + "learning_rate": 2.6067517762305116e-05, + "loss": 3.1308, + "step": 5155 + }, + { + "epoch": 7.55, + "grad_norm": 0.3402569890022278, + "learning_rate": 2.6035943511326733e-05, + "loss": 3.1384, + "step": 5156 + }, + { + "epoch": 7.55, + "grad_norm": 0.42875614762306213, + "learning_rate": 2.6004385531945107e-05, + "loss": 3.1476, + "step": 5157 + }, + { + "epoch": 7.55, + "grad_norm": 0.46508556604385376, + "learning_rate": 2.5972843831102755e-05, + "loss": 3.1542, + "step": 5158 + }, + { + "epoch": 7.55, + "grad_norm": 0.7432370185852051, + "learning_rate": 2.594131841573866e-05, + "loss": 3.1507, + "step": 5159 + }, + { + "epoch": 7.55, + "grad_norm": 0.5330306887626648, + "learning_rate": 2.590980929278819e-05, + "loss": 3.1661, + "step": 5160 + }, + { + "epoch": 7.56, + "grad_norm": 0.37541455030441284, + "learning_rate": 2.587831646918315e-05, + "loss": 3.1454, + "step": 5161 + }, + { + "epoch": 7.56, + "grad_norm": 0.6398892402648926, + "learning_rate": 2.584683995185174e-05, + "loss": 3.1901, + "step": 5162 + }, + { + "epoch": 7.56, + "grad_norm": 0.9277903437614441, + "learning_rate": 2.5815379747718594e-05, + "loss": 3.1594, + "step": 5163 + }, + { + "epoch": 7.56, + "grad_norm": 0.8719270825386047, + "learning_rate": 2.5783935863704732e-05, + "loss": 3.1646, + "step": 5164 + }, + { + "epoch": 7.56, + "grad_norm": 0.811159074306488, + "learning_rate": 2.5752508306727608e-05, + "loss": 3.2009, + "step": 5165 + }, + { + "epoch": 7.56, + "grad_norm": 0.834298849105835, + "learning_rate": 2.5721097083701084e-05, + "loss": 3.1747, + "step": 5166 + }, + { + "epoch": 7.56, + "grad_norm": 1.0362366437911987, + "learning_rate": 2.5689702201535394e-05, + "loss": 3.1428, + "step": 5167 + }, + { + "epoch": 7.57, + "grad_norm": 0.8834418058395386, + "learning_rate": 2.5658323667137218e-05, + "loss": 3.1559, + "step": 5168 + }, + { + "epoch": 7.57, + "grad_norm": 0.4252467155456543, + "learning_rate": 2.5626961487409616e-05, + "loss": 3.1528, + "step": 5169 + }, + { + "epoch": 7.57, + "grad_norm": 1.2788466215133667, + "learning_rate": 2.559561566925206e-05, + "loss": 3.1789, + "step": 5170 + }, + { + "epoch": 7.57, + "grad_norm": 0.736139714717865, + "learning_rate": 2.556428621956043e-05, + "loss": 3.1242, + "step": 5171 + }, + { + "epoch": 7.57, + "grad_norm": 0.38194677233695984, + "learning_rate": 2.553297314522698e-05, + "loss": 3.1737, + "step": 5172 + }, + { + "epoch": 7.57, + "grad_norm": 0.5845624804496765, + "learning_rate": 2.5501676453140378e-05, + "loss": 3.1726, + "step": 5173 + }, + { + "epoch": 7.58, + "grad_norm": 0.7914636731147766, + "learning_rate": 2.5470396150185692e-05, + "loss": 3.1579, + "step": 5174 + }, + { + "epoch": 7.58, + "grad_norm": 0.39859166741371155, + "learning_rate": 2.5439132243244412e-05, + "loss": 3.1319, + "step": 5175 + }, + { + "epoch": 7.58, + "grad_norm": 1.092436671257019, + "learning_rate": 2.540788473919433e-05, + "loss": 3.1286, + "step": 5176 + }, + { + "epoch": 7.58, + "grad_norm": 0.6279719471931458, + "learning_rate": 2.5376653644909675e-05, + "loss": 3.1328, + "step": 5177 + }, + { + "epoch": 7.58, + "grad_norm": 1.2582124471664429, + "learning_rate": 2.534543896726117e-05, + "loss": 3.1486, + "step": 5178 + }, + { + "epoch": 7.58, + "grad_norm": 0.6606663465499878, + "learning_rate": 2.5314240713115756e-05, + "loss": 3.1386, + "step": 5179 + }, + { + "epoch": 7.58, + "grad_norm": 0.5657997727394104, + "learning_rate": 2.528305888933683e-05, + "loss": 3.1366, + "step": 5180 + }, + { + "epoch": 7.59, + "grad_norm": 0.7459825873374939, + "learning_rate": 2.5251893502784275e-05, + "loss": 3.1625, + "step": 5181 + }, + { + "epoch": 7.59, + "grad_norm": 0.6678188443183899, + "learning_rate": 2.522074456031418e-05, + "loss": 3.1981, + "step": 5182 + }, + { + "epoch": 7.59, + "grad_norm": 0.7135474681854248, + "learning_rate": 2.5189612068779112e-05, + "loss": 3.1958, + "step": 5183 + }, + { + "epoch": 7.59, + "grad_norm": 0.5656193494796753, + "learning_rate": 2.5158496035028077e-05, + "loss": 3.1218, + "step": 5184 + }, + { + "epoch": 7.59, + "grad_norm": 0.5109965205192566, + "learning_rate": 2.5127396465906327e-05, + "loss": 3.1465, + "step": 5185 + }, + { + "epoch": 7.59, + "grad_norm": 0.5121713876724243, + "learning_rate": 2.5096313368255585e-05, + "loss": 3.1401, + "step": 5186 + }, + { + "epoch": 7.59, + "grad_norm": 0.38621145486831665, + "learning_rate": 2.506524674891392e-05, + "loss": 3.1519, + "step": 5187 + }, + { + "epoch": 7.6, + "grad_norm": 0.557099461555481, + "learning_rate": 2.5034196614715778e-05, + "loss": 3.1632, + "step": 5188 + }, + { + "epoch": 7.6, + "grad_norm": 0.42920613288879395, + "learning_rate": 2.500316297249198e-05, + "loss": 3.1272, + "step": 5189 + }, + { + "epoch": 7.6, + "grad_norm": 0.765871524810791, + "learning_rate": 2.4972145829069714e-05, + "loss": 3.1568, + "step": 5190 + }, + { + "epoch": 7.6, + "grad_norm": 0.4879997670650482, + "learning_rate": 2.4941145191272563e-05, + "loss": 3.174, + "step": 5191 + }, + { + "epoch": 7.6, + "grad_norm": 0.5009108781814575, + "learning_rate": 2.4910161065920445e-05, + "loss": 3.1467, + "step": 5192 + }, + { + "epoch": 7.6, + "grad_norm": 0.6382935643196106, + "learning_rate": 2.4879193459829665e-05, + "loss": 3.1552, + "step": 5193 + }, + { + "epoch": 7.6, + "grad_norm": 0.57361900806427, + "learning_rate": 2.484824237981288e-05, + "loss": 3.1511, + "step": 5194 + }, + { + "epoch": 7.61, + "grad_norm": 0.2623114287853241, + "learning_rate": 2.4817307832679126e-05, + "loss": 3.1664, + "step": 5195 + }, + { + "epoch": 7.61, + "grad_norm": 0.5569726228713989, + "learning_rate": 2.4786389825233802e-05, + "loss": 3.1431, + "step": 5196 + }, + { + "epoch": 7.61, + "grad_norm": 0.5350569486618042, + "learning_rate": 2.4755488364278656e-05, + "loss": 3.173, + "step": 5197 + }, + { + "epoch": 7.61, + "grad_norm": 0.38636651635169983, + "learning_rate": 2.4724603456611805e-05, + "loss": 3.1996, + "step": 5198 + }, + { + "epoch": 7.61, + "grad_norm": 0.6508245468139648, + "learning_rate": 2.4693735109027716e-05, + "loss": 3.1693, + "step": 5199 + }, + { + "epoch": 7.61, + "grad_norm": 0.7632821798324585, + "learning_rate": 2.4662883328317222e-05, + "loss": 3.1948, + "step": 5200 + }, + { + "epoch": 7.62, + "grad_norm": 0.5081233978271484, + "learning_rate": 2.4632048121267547e-05, + "loss": 3.1938, + "step": 5201 + }, + { + "epoch": 7.62, + "grad_norm": 0.7682433128356934, + "learning_rate": 2.4601229494662125e-05, + "loss": 3.1553, + "step": 5202 + }, + { + "epoch": 7.62, + "grad_norm": 1.0782290697097778, + "learning_rate": 2.4570427455280943e-05, + "loss": 3.1649, + "step": 5203 + }, + { + "epoch": 7.62, + "grad_norm": 0.6353168487548828, + "learning_rate": 2.4539642009900253e-05, + "loss": 3.1517, + "step": 5204 + }, + { + "epoch": 7.62, + "grad_norm": 0.7350844144821167, + "learning_rate": 2.4508873165292535e-05, + "loss": 3.1375, + "step": 5205 + }, + { + "epoch": 7.62, + "grad_norm": 0.9207132458686829, + "learning_rate": 2.4478120928226846e-05, + "loss": 3.1478, + "step": 5206 + }, + { + "epoch": 7.62, + "grad_norm": 0.7946307063102722, + "learning_rate": 2.444738530546845e-05, + "loss": 3.1392, + "step": 5207 + }, + { + "epoch": 7.63, + "grad_norm": 0.9071816802024841, + "learning_rate": 2.441666630377889e-05, + "loss": 3.1275, + "step": 5208 + }, + { + "epoch": 7.63, + "grad_norm": 1.2288057804107666, + "learning_rate": 2.438596392991628e-05, + "loss": 3.1333, + "step": 5209 + }, + { + "epoch": 7.63, + "grad_norm": 1.0466946363449097, + "learning_rate": 2.435527819063482e-05, + "loss": 3.143, + "step": 5210 + }, + { + "epoch": 7.63, + "grad_norm": 0.5097655057907104, + "learning_rate": 2.432460909268518e-05, + "loss": 3.2069, + "step": 5211 + }, + { + "epoch": 7.63, + "grad_norm": 0.8036677241325378, + "learning_rate": 2.4293956642814453e-05, + "loss": 3.1592, + "step": 5212 + }, + { + "epoch": 7.63, + "grad_norm": 3.4330554008483887, + "learning_rate": 2.4263320847765868e-05, + "loss": 3.1551, + "step": 5213 + }, + { + "epoch": 7.63, + "grad_norm": 0.9436863660812378, + "learning_rate": 2.4232701714279115e-05, + "loss": 3.1353, + "step": 5214 + }, + { + "epoch": 7.64, + "grad_norm": 0.30926942825317383, + "learning_rate": 2.420209924909026e-05, + "loss": 3.1577, + "step": 5215 + }, + { + "epoch": 7.64, + "grad_norm": 0.7689909338951111, + "learning_rate": 2.417151345893157e-05, + "loss": 3.1326, + "step": 5216 + }, + { + "epoch": 7.64, + "grad_norm": 0.8262006640434265, + "learning_rate": 2.4140944350531712e-05, + "loss": 3.1372, + "step": 5217 + }, + { + "epoch": 7.64, + "grad_norm": 0.35750913619995117, + "learning_rate": 2.411039193061577e-05, + "loss": 3.1526, + "step": 5218 + }, + { + "epoch": 7.64, + "grad_norm": 0.7210152745246887, + "learning_rate": 2.407985620590498e-05, + "loss": 3.1702, + "step": 5219 + }, + { + "epoch": 7.64, + "grad_norm": 0.37686386704444885, + "learning_rate": 2.404933718311704e-05, + "loss": 3.1609, + "step": 5220 + }, + { + "epoch": 7.64, + "grad_norm": 0.38531693816185, + "learning_rate": 2.401883486896591e-05, + "loss": 3.16, + "step": 5221 + }, + { + "epoch": 7.65, + "grad_norm": 3.7467172145843506, + "learning_rate": 2.3988349270161913e-05, + "loss": 3.1974, + "step": 5222 + }, + { + "epoch": 7.65, + "grad_norm": 0.4558284878730774, + "learning_rate": 2.3957880393411667e-05, + "loss": 3.1724, + "step": 5223 + }, + { + "epoch": 7.65, + "grad_norm": 0.5120177268981934, + "learning_rate": 2.3927428245418127e-05, + "loss": 3.1466, + "step": 5224 + }, + { + "epoch": 7.65, + "grad_norm": 0.47643178701400757, + "learning_rate": 2.3896992832880548e-05, + "loss": 3.1508, + "step": 5225 + }, + { + "epoch": 7.65, + "grad_norm": 0.333384245634079, + "learning_rate": 2.3866574162494538e-05, + "loss": 3.1346, + "step": 5226 + }, + { + "epoch": 7.65, + "grad_norm": 0.5989437699317932, + "learning_rate": 2.383617224095198e-05, + "loss": 3.1354, + "step": 5227 + }, + { + "epoch": 7.66, + "grad_norm": 0.35983359813690186, + "learning_rate": 2.38057870749411e-05, + "loss": 3.1599, + "step": 5228 + }, + { + "epoch": 7.66, + "grad_norm": 0.3712514042854309, + "learning_rate": 2.3775418671146443e-05, + "loss": 3.1584, + "step": 5229 + }, + { + "epoch": 7.66, + "grad_norm": 0.3716100752353668, + "learning_rate": 2.3745067036248836e-05, + "loss": 3.1766, + "step": 5230 + }, + { + "epoch": 7.66, + "grad_norm": 0.27831578254699707, + "learning_rate": 2.3714732176925446e-05, + "loss": 3.1285, + "step": 5231 + }, + { + "epoch": 7.66, + "grad_norm": 0.47703278064727783, + "learning_rate": 2.368441409984976e-05, + "loss": 3.1467, + "step": 5232 + }, + { + "epoch": 7.66, + "grad_norm": 0.577670156955719, + "learning_rate": 2.3654112811691477e-05, + "loss": 3.1222, + "step": 5233 + }, + { + "epoch": 7.66, + "grad_norm": 0.44678911566734314, + "learning_rate": 2.3623828319116748e-05, + "loss": 3.1783, + "step": 5234 + }, + { + "epoch": 7.67, + "grad_norm": 0.9070417881011963, + "learning_rate": 2.359356062878797e-05, + "loss": 3.1324, + "step": 5235 + }, + { + "epoch": 7.67, + "grad_norm": 0.5070209503173828, + "learning_rate": 2.356330974736375e-05, + "loss": 3.129, + "step": 5236 + }, + { + "epoch": 7.67, + "grad_norm": 0.6462606191635132, + "learning_rate": 2.353307568149915e-05, + "loss": 3.1474, + "step": 5237 + }, + { + "epoch": 7.67, + "grad_norm": 0.34412530064582825, + "learning_rate": 2.3502858437845476e-05, + "loss": 3.1597, + "step": 5238 + }, + { + "epoch": 7.67, + "grad_norm": 0.4765468239784241, + "learning_rate": 2.347265802305022e-05, + "loss": 3.1474, + "step": 5239 + }, + { + "epoch": 7.67, + "eval_loss": 3.013108015060425, + "eval_runtime": 1170.7492, + "eval_samples_per_second": 14.642, + "eval_steps_per_second": 0.916, + "step": 5239 + }, + { + "epoch": 7.67, + "grad_norm": 0.5959687232971191, + "learning_rate": 2.3442474443757366e-05, + "loss": 3.1172, + "step": 5240 + }, + { + "epoch": 7.67, + "grad_norm": 0.27279970049858093, + "learning_rate": 2.3412307706607108e-05, + "loss": 3.1737, + "step": 5241 + }, + { + "epoch": 7.68, + "grad_norm": 0.5665635466575623, + "learning_rate": 2.3382157818235816e-05, + "loss": 3.1308, + "step": 5242 + }, + { + "epoch": 7.68, + "grad_norm": 0.3132018446922302, + "learning_rate": 2.3352024785276404e-05, + "loss": 3.1657, + "step": 5243 + }, + { + "epoch": 7.68, + "grad_norm": 0.622239351272583, + "learning_rate": 2.3321908614357824e-05, + "loss": 3.1689, + "step": 5244 + }, + { + "epoch": 7.68, + "grad_norm": 0.5310043692588806, + "learning_rate": 2.3291809312105462e-05, + "loss": 3.2016, + "step": 5245 + }, + { + "epoch": 7.68, + "grad_norm": 0.3280445337295532, + "learning_rate": 2.3261726885141023e-05, + "loss": 3.1654, + "step": 5246 + }, + { + "epoch": 7.68, + "grad_norm": 0.4684401750564575, + "learning_rate": 2.3231661340082366e-05, + "loss": 3.1932, + "step": 5247 + }, + { + "epoch": 7.68, + "grad_norm": 0.5372167825698853, + "learning_rate": 2.3201612683543704e-05, + "loss": 3.2027, + "step": 5248 + }, + { + "epoch": 7.69, + "grad_norm": 0.865881621837616, + "learning_rate": 2.3171580922135627e-05, + "loss": 3.1653, + "step": 5249 + }, + { + "epoch": 7.69, + "grad_norm": 0.650731086730957, + "learning_rate": 2.3141566062464847e-05, + "loss": 3.1495, + "step": 5250 + }, + { + "epoch": 7.69, + "grad_norm": 1.0147230625152588, + "learning_rate": 2.311156811113444e-05, + "loss": 3.1422, + "step": 5251 + }, + { + "epoch": 7.69, + "grad_norm": 0.34079644083976746, + "learning_rate": 2.308158707474377e-05, + "loss": 3.125, + "step": 5252 + }, + { + "epoch": 7.69, + "grad_norm": 0.8186140656471252, + "learning_rate": 2.3051622959888464e-05, + "loss": 3.1991, + "step": 5253 + }, + { + "epoch": 7.69, + "grad_norm": 0.8478550314903259, + "learning_rate": 2.3021675773160424e-05, + "loss": 3.1535, + "step": 5254 + }, + { + "epoch": 7.7, + "grad_norm": 0.6345941424369812, + "learning_rate": 2.299174552114781e-05, + "loss": 3.1704, + "step": 5255 + }, + { + "epoch": 7.7, + "grad_norm": 0.9661188721656799, + "learning_rate": 2.296183221043511e-05, + "loss": 3.1492, + "step": 5256 + }, + { + "epoch": 7.7, + "grad_norm": 0.4356244206428528, + "learning_rate": 2.2931935847603037e-05, + "loss": 3.159, + "step": 5257 + }, + { + "epoch": 7.7, + "grad_norm": 0.6363608837127686, + "learning_rate": 2.290205643922859e-05, + "loss": 3.1579, + "step": 5258 + }, + { + "epoch": 7.7, + "grad_norm": 1.1210684776306152, + "learning_rate": 2.2872193991885026e-05, + "loss": 3.1412, + "step": 5259 + }, + { + "epoch": 7.7, + "grad_norm": 0.38606759905815125, + "learning_rate": 2.2842348512141908e-05, + "loss": 3.1999, + "step": 5260 + }, + { + "epoch": 7.7, + "grad_norm": 1.1856557130813599, + "learning_rate": 2.281252000656503e-05, + "loss": 3.1595, + "step": 5261 + }, + { + "epoch": 7.71, + "grad_norm": 1.063137412071228, + "learning_rate": 2.2782708481716463e-05, + "loss": 3.1713, + "step": 5262 + }, + { + "epoch": 7.71, + "grad_norm": 0.6546114087104797, + "learning_rate": 2.275291394415454e-05, + "loss": 3.1581, + "step": 5263 + }, + { + "epoch": 7.71, + "grad_norm": 1.1830710172653198, + "learning_rate": 2.2723136400433865e-05, + "loss": 3.1447, + "step": 5264 + }, + { + "epoch": 7.71, + "grad_norm": 0.6315742135047913, + "learning_rate": 2.269337585710529e-05, + "loss": 3.1487, + "step": 5265 + }, + { + "epoch": 7.71, + "grad_norm": 0.8535947799682617, + "learning_rate": 2.2663632320715976e-05, + "loss": 3.1631, + "step": 5266 + }, + { + "epoch": 7.71, + "grad_norm": 1.0756810903549194, + "learning_rate": 2.2633905797809197e-05, + "loss": 3.1389, + "step": 5267 + }, + { + "epoch": 7.71, + "grad_norm": 0.7916353344917297, + "learning_rate": 2.2604196294924694e-05, + "loss": 3.13, + "step": 5268 + }, + { + "epoch": 7.72, + "grad_norm": 0.46646764874458313, + "learning_rate": 2.257450381859836e-05, + "loss": 3.1633, + "step": 5269 + }, + { + "epoch": 7.72, + "grad_norm": 0.6788039803504944, + "learning_rate": 2.254482837536224e-05, + "loss": 3.1178, + "step": 5270 + }, + { + "epoch": 7.72, + "grad_norm": 0.5274389982223511, + "learning_rate": 2.251516997174482e-05, + "loss": 3.1824, + "step": 5271 + }, + { + "epoch": 7.72, + "grad_norm": 0.5712849497795105, + "learning_rate": 2.2485528614270757e-05, + "loss": 3.1324, + "step": 5272 + }, + { + "epoch": 7.72, + "grad_norm": 0.39241471886634827, + "learning_rate": 2.2455904309460886e-05, + "loss": 3.1527, + "step": 5273 + }, + { + "epoch": 7.72, + "grad_norm": 0.3917776346206665, + "learning_rate": 2.242629706383237e-05, + "loss": 3.1554, + "step": 5274 + }, + { + "epoch": 7.73, + "grad_norm": 0.44407832622528076, + "learning_rate": 2.2396706883898676e-05, + "loss": 3.1729, + "step": 5275 + }, + { + "epoch": 7.73, + "grad_norm": 0.6070168614387512, + "learning_rate": 2.236713377616936e-05, + "loss": 3.1218, + "step": 5276 + }, + { + "epoch": 7.73, + "grad_norm": 0.43469035625457764, + "learning_rate": 2.2337577747150353e-05, + "loss": 3.1484, + "step": 5277 + }, + { + "epoch": 7.73, + "grad_norm": 0.3631550967693329, + "learning_rate": 2.2308038803343755e-05, + "loss": 3.1192, + "step": 5278 + }, + { + "epoch": 7.73, + "grad_norm": 0.45660415291786194, + "learning_rate": 2.227851695124796e-05, + "loss": 3.1704, + "step": 5279 + }, + { + "epoch": 7.73, + "grad_norm": 0.3175928592681885, + "learning_rate": 2.224901219735758e-05, + "loss": 3.1534, + "step": 5280 + }, + { + "epoch": 7.73, + "grad_norm": 0.6207453608512878, + "learning_rate": 2.221952454816344e-05, + "loss": 3.191, + "step": 5281 + }, + { + "epoch": 7.74, + "grad_norm": 0.5500271320343018, + "learning_rate": 2.219005401015264e-05, + "loss": 3.1639, + "step": 5282 + }, + { + "epoch": 7.74, + "grad_norm": 0.5747382044792175, + "learning_rate": 2.21606005898085e-05, + "loss": 3.132, + "step": 5283 + }, + { + "epoch": 7.74, + "grad_norm": 0.26599442958831787, + "learning_rate": 2.2131164293610572e-05, + "loss": 3.1491, + "step": 5284 + }, + { + "epoch": 7.74, + "grad_norm": 0.46995365619659424, + "learning_rate": 2.2101745128034656e-05, + "loss": 3.1383, + "step": 5285 + }, + { + "epoch": 7.74, + "grad_norm": 0.23462985455989838, + "learning_rate": 2.207234309955275e-05, + "loss": 3.1604, + "step": 5286 + }, + { + "epoch": 7.74, + "grad_norm": 0.45813092589378357, + "learning_rate": 2.2042958214633115e-05, + "loss": 3.1589, + "step": 5287 + }, + { + "epoch": 7.74, + "grad_norm": 0.48195019364356995, + "learning_rate": 2.2013590479740232e-05, + "loss": 3.1074, + "step": 5288 + }, + { + "epoch": 7.75, + "grad_norm": 0.9091559052467346, + "learning_rate": 2.1984239901334802e-05, + "loss": 3.1518, + "step": 5289 + }, + { + "epoch": 7.75, + "grad_norm": 0.9648109078407288, + "learning_rate": 2.195490648587375e-05, + "loss": 3.2077, + "step": 5290 + }, + { + "epoch": 7.75, + "grad_norm": 0.6132333874702454, + "learning_rate": 2.1925590239810235e-05, + "loss": 3.1481, + "step": 5291 + }, + { + "epoch": 7.75, + "grad_norm": 0.4865446984767914, + "learning_rate": 2.1896291169593643e-05, + "loss": 3.1402, + "step": 5292 + }, + { + "epoch": 7.75, + "grad_norm": 0.561164140701294, + "learning_rate": 2.186700928166956e-05, + "loss": 3.1597, + "step": 5293 + }, + { + "epoch": 7.75, + "grad_norm": 0.46440842747688293, + "learning_rate": 2.1837744582479814e-05, + "loss": 3.1493, + "step": 5294 + }, + { + "epoch": 7.75, + "grad_norm": 0.5474674701690674, + "learning_rate": 2.1808497078462442e-05, + "loss": 3.1622, + "step": 5295 + }, + { + "epoch": 7.76, + "grad_norm": 0.304200679063797, + "learning_rate": 2.1779266776051688e-05, + "loss": 3.1756, + "step": 5296 + }, + { + "epoch": 7.76, + "grad_norm": 0.534535825252533, + "learning_rate": 2.175005368167804e-05, + "loss": 3.2078, + "step": 5297 + }, + { + "epoch": 7.76, + "grad_norm": 0.7717536687850952, + "learning_rate": 2.17208578017682e-05, + "loss": 3.1603, + "step": 5298 + }, + { + "epoch": 7.76, + "grad_norm": 0.8810925483703613, + "learning_rate": 2.169167914274498e-05, + "loss": 3.1335, + "step": 5299 + }, + { + "epoch": 7.76, + "grad_norm": 0.9111099243164062, + "learning_rate": 2.166251771102761e-05, + "loss": 3.1247, + "step": 5300 + }, + { + "epoch": 7.76, + "grad_norm": 0.5138559937477112, + "learning_rate": 2.163337351303132e-05, + "loss": 3.183, + "step": 5301 + }, + { + "epoch": 7.77, + "grad_norm": 0.5875454545021057, + "learning_rate": 2.1604246555167638e-05, + "loss": 3.1424, + "step": 5302 + }, + { + "epoch": 7.77, + "grad_norm": 0.35727164149284363, + "learning_rate": 2.157513684384439e-05, + "loss": 3.1641, + "step": 5303 + }, + { + "epoch": 7.77, + "grad_norm": 0.720236599445343, + "learning_rate": 2.1546044385465426e-05, + "loss": 3.1459, + "step": 5304 + }, + { + "epoch": 7.77, + "grad_norm": 0.49461638927459717, + "learning_rate": 2.151696918643089e-05, + "loss": 3.1613, + "step": 5305 + }, + { + "epoch": 7.77, + "grad_norm": 0.5594753623008728, + "learning_rate": 2.1487911253137217e-05, + "loss": 3.1251, + "step": 5306 + }, + { + "epoch": 7.77, + "grad_norm": 1.0345263481140137, + "learning_rate": 2.1458870591976877e-05, + "loss": 3.1507, + "step": 5307 + }, + { + "epoch": 7.77, + "grad_norm": 0.31530436873435974, + "learning_rate": 2.1429847209338628e-05, + "loss": 3.1615, + "step": 5308 + }, + { + "epoch": 7.78, + "grad_norm": 0.3019442558288574, + "learning_rate": 2.1400841111607483e-05, + "loss": 3.1626, + "step": 5309 + }, + { + "epoch": 7.78, + "grad_norm": 0.40825822949409485, + "learning_rate": 2.137185230516453e-05, + "loss": 3.1289, + "step": 5310 + }, + { + "epoch": 7.78, + "grad_norm": 0.4319787323474884, + "learning_rate": 2.134288079638712e-05, + "loss": 3.1428, + "step": 5311 + }, + { + "epoch": 7.78, + "grad_norm": 0.5881616473197937, + "learning_rate": 2.131392659164879e-05, + "loss": 3.1675, + "step": 5312 + }, + { + "epoch": 7.78, + "grad_norm": 0.4166703224182129, + "learning_rate": 2.128498969731929e-05, + "loss": 3.1611, + "step": 5313 + }, + { + "epoch": 7.78, + "grad_norm": 0.7921746373176575, + "learning_rate": 2.125607011976454e-05, + "loss": 3.1281, + "step": 5314 + }, + { + "epoch": 7.78, + "grad_norm": 0.454723060131073, + "learning_rate": 2.1227167865346632e-05, + "loss": 3.1568, + "step": 5315 + }, + { + "epoch": 7.79, + "grad_norm": 0.49115943908691406, + "learning_rate": 2.1198282940423887e-05, + "loss": 3.1556, + "step": 5316 + }, + { + "epoch": 7.79, + "grad_norm": 0.4243347644805908, + "learning_rate": 2.1169415351350786e-05, + "loss": 3.1701, + "step": 5317 + }, + { + "epoch": 7.79, + "grad_norm": 0.9867798089981079, + "learning_rate": 2.1140565104478004e-05, + "loss": 3.1729, + "step": 5318 + }, + { + "epoch": 7.79, + "grad_norm": 1.0699418783187866, + "learning_rate": 2.1111732206152424e-05, + "loss": 3.1711, + "step": 5319 + }, + { + "epoch": 7.79, + "grad_norm": 0.7602143287658691, + "learning_rate": 2.1082916662717055e-05, + "loss": 3.1824, + "step": 5320 + }, + { + "epoch": 7.79, + "grad_norm": 0.7598912715911865, + "learning_rate": 2.1054118480511154e-05, + "loss": 3.1232, + "step": 5321 + }, + { + "epoch": 7.79, + "grad_norm": 0.7849280834197998, + "learning_rate": 2.1025337665870104e-05, + "loss": 3.1169, + "step": 5322 + }, + { + "epoch": 7.8, + "grad_norm": 0.5915687084197998, + "learning_rate": 2.0996574225125532e-05, + "loss": 3.1269, + "step": 5323 + }, + { + "epoch": 7.8, + "grad_norm": 0.9598039984703064, + "learning_rate": 2.096782816460513e-05, + "loss": 3.1504, + "step": 5324 + }, + { + "epoch": 7.8, + "grad_norm": 0.6661960482597351, + "learning_rate": 2.0939099490632896e-05, + "loss": 3.1284, + "step": 5325 + }, + { + "epoch": 7.8, + "grad_norm": 0.4720488488674164, + "learning_rate": 2.0910388209528964e-05, + "loss": 3.1584, + "step": 5326 + }, + { + "epoch": 7.8, + "grad_norm": 0.6807810664176941, + "learning_rate": 2.0881694327609536e-05, + "loss": 3.1586, + "step": 5327 + }, + { + "epoch": 7.8, + "grad_norm": 0.7132989764213562, + "learning_rate": 2.085301785118716e-05, + "loss": 3.1526, + "step": 5328 + }, + { + "epoch": 7.81, + "grad_norm": 0.4347662031650543, + "learning_rate": 2.0824358786570453e-05, + "loss": 3.1645, + "step": 5329 + }, + { + "epoch": 7.81, + "grad_norm": 1.1308764219284058, + "learning_rate": 2.0795717140064163e-05, + "loss": 3.162, + "step": 5330 + }, + { + "epoch": 7.81, + "grad_norm": 0.759826123714447, + "learning_rate": 2.076709291796931e-05, + "loss": 3.184, + "step": 5331 + }, + { + "epoch": 7.81, + "grad_norm": 0.6993570923805237, + "learning_rate": 2.0738486126583056e-05, + "loss": 3.1636, + "step": 5332 + }, + { + "epoch": 7.81, + "grad_norm": 0.5368728637695312, + "learning_rate": 2.0709896772198612e-05, + "loss": 3.1551, + "step": 5333 + }, + { + "epoch": 7.81, + "grad_norm": 0.5664252042770386, + "learning_rate": 2.0681324861105556e-05, + "loss": 3.1349, + "step": 5334 + }, + { + "epoch": 7.81, + "grad_norm": 0.6612221598625183, + "learning_rate": 2.0652770399589427e-05, + "loss": 3.1495, + "step": 5335 + }, + { + "epoch": 7.82, + "grad_norm": 0.48035261034965515, + "learning_rate": 2.0624233393932024e-05, + "loss": 3.1624, + "step": 5336 + }, + { + "epoch": 7.82, + "grad_norm": 0.36596471071243286, + "learning_rate": 2.059571385041138e-05, + "loss": 3.1674, + "step": 5337 + }, + { + "epoch": 7.82, + "grad_norm": 0.6980510950088501, + "learning_rate": 2.056721177530151e-05, + "loss": 3.1126, + "step": 5338 + }, + { + "epoch": 7.82, + "grad_norm": 0.567153811454773, + "learning_rate": 2.0538727174872684e-05, + "loss": 3.1449, + "step": 5339 + }, + { + "epoch": 7.82, + "grad_norm": 0.6353252530097961, + "learning_rate": 2.0510260055391418e-05, + "loss": 3.1646, + "step": 5340 + }, + { + "epoch": 7.82, + "grad_norm": 0.6695141792297363, + "learning_rate": 2.0481810423120196e-05, + "loss": 3.1426, + "step": 5341 + }, + { + "epoch": 7.82, + "grad_norm": 1.0574151277542114, + "learning_rate": 2.045337828431775e-05, + "loss": 3.1435, + "step": 5342 + }, + { + "epoch": 7.83, + "grad_norm": 0.3314790725708008, + "learning_rate": 2.0424963645239027e-05, + "loss": 3.1263, + "step": 5343 + }, + { + "epoch": 7.83, + "grad_norm": 0.708838939666748, + "learning_rate": 2.0396566512134996e-05, + "loss": 3.178, + "step": 5344 + }, + { + "epoch": 7.83, + "grad_norm": 0.670356810092926, + "learning_rate": 2.036818689125284e-05, + "loss": 3.1153, + "step": 5345 + }, + { + "epoch": 7.83, + "grad_norm": 0.6451178193092346, + "learning_rate": 2.0339824788835916e-05, + "loss": 3.1816, + "step": 5346 + }, + { + "epoch": 7.83, + "grad_norm": 0.5722768306732178, + "learning_rate": 2.031148021112367e-05, + "loss": 3.1584, + "step": 5347 + }, + { + "epoch": 7.83, + "grad_norm": 0.8647962808609009, + "learning_rate": 2.028315316435173e-05, + "loss": 3.118, + "step": 5348 + }, + { + "epoch": 7.83, + "grad_norm": 0.5145653486251831, + "learning_rate": 2.0254843654751852e-05, + "loss": 3.1143, + "step": 5349 + }, + { + "epoch": 7.84, + "grad_norm": 0.7415019273757935, + "learning_rate": 2.0226551688551954e-05, + "loss": 3.1961, + "step": 5350 + }, + { + "epoch": 7.84, + "grad_norm": 0.6742720603942871, + "learning_rate": 2.0198277271976052e-05, + "loss": 3.107, + "step": 5351 + }, + { + "epoch": 7.84, + "grad_norm": 0.5243716835975647, + "learning_rate": 2.017002041124435e-05, + "loss": 3.1568, + "step": 5352 + }, + { + "epoch": 7.84, + "grad_norm": 0.6386037468910217, + "learning_rate": 2.0141781112573154e-05, + "loss": 3.1501, + "step": 5353 + }, + { + "epoch": 7.84, + "grad_norm": 0.9167563915252686, + "learning_rate": 2.011355938217496e-05, + "loss": 3.1591, + "step": 5354 + }, + { + "epoch": 7.84, + "grad_norm": 0.3091890811920166, + "learning_rate": 2.0085355226258285e-05, + "loss": 3.1714, + "step": 5355 + }, + { + "epoch": 7.85, + "grad_norm": 0.6145365834236145, + "learning_rate": 2.00571686510279e-05, + "loss": 3.1247, + "step": 5356 + }, + { + "epoch": 7.85, + "grad_norm": 0.3485010862350464, + "learning_rate": 2.00289996626847e-05, + "loss": 3.1696, + "step": 5357 + }, + { + "epoch": 7.85, + "grad_norm": 0.5586562156677246, + "learning_rate": 2.0000848267425575e-05, + "loss": 3.1249, + "step": 5358 + }, + { + "epoch": 7.85, + "grad_norm": 0.45884886384010315, + "learning_rate": 1.9972714471443733e-05, + "loss": 3.1264, + "step": 5359 + }, + { + "epoch": 7.85, + "grad_norm": 0.3801732361316681, + "learning_rate": 1.9944598280928408e-05, + "loss": 3.1699, + "step": 5360 + }, + { + "epoch": 7.85, + "grad_norm": 0.5741209983825684, + "learning_rate": 1.9916499702064894e-05, + "loss": 3.1625, + "step": 5361 + }, + { + "epoch": 7.85, + "grad_norm": 0.23692893981933594, + "learning_rate": 1.9888418741034787e-05, + "loss": 3.1282, + "step": 5362 + }, + { + "epoch": 7.86, + "grad_norm": 0.23494169116020203, + "learning_rate": 1.9860355404015696e-05, + "loss": 3.1757, + "step": 5363 + }, + { + "epoch": 7.86, + "grad_norm": 0.5047478675842285, + "learning_rate": 1.9832309697181294e-05, + "loss": 3.1371, + "step": 5364 + }, + { + "epoch": 7.86, + "grad_norm": 0.39867815375328064, + "learning_rate": 1.9804281626701525e-05, + "loss": 3.1522, + "step": 5365 + }, + { + "epoch": 7.86, + "grad_norm": 0.4054205119609833, + "learning_rate": 1.977627119874237e-05, + "loss": 3.1453, + "step": 5366 + }, + { + "epoch": 7.86, + "grad_norm": 0.37046530842781067, + "learning_rate": 1.9748278419465872e-05, + "loss": 3.1295, + "step": 5367 + }, + { + "epoch": 7.86, + "grad_norm": 0.4440865218639374, + "learning_rate": 1.972030329503035e-05, + "loss": 3.1436, + "step": 5368 + }, + { + "epoch": 7.86, + "grad_norm": 0.25685054063796997, + "learning_rate": 1.9692345831590064e-05, + "loss": 3.147, + "step": 5369 + }, + { + "epoch": 7.87, + "grad_norm": 0.423784077167511, + "learning_rate": 1.966440603529549e-05, + "loss": 3.161, + "step": 5370 + }, + { + "epoch": 7.87, + "grad_norm": 0.407570481300354, + "learning_rate": 1.963648391229321e-05, + "loss": 3.1711, + "step": 5371 + }, + { + "epoch": 7.87, + "grad_norm": 1.7502979040145874, + "learning_rate": 1.9608579468725895e-05, + "loss": 3.1717, + "step": 5372 + }, + { + "epoch": 7.87, + "grad_norm": 0.37750712037086487, + "learning_rate": 1.958069271073234e-05, + "loss": 3.1518, + "step": 5373 + }, + { + "epoch": 7.87, + "grad_norm": 0.44295188784599304, + "learning_rate": 1.9552823644447437e-05, + "loss": 3.1532, + "step": 5374 + }, + { + "epoch": 7.87, + "grad_norm": 0.6612734794616699, + "learning_rate": 1.9524972276002207e-05, + "loss": 3.1507, + "step": 5375 + }, + { + "epoch": 7.88, + "grad_norm": 0.5701128244400024, + "learning_rate": 1.9497138611523748e-05, + "loss": 3.1726, + "step": 5376 + }, + { + "epoch": 7.88, + "grad_norm": 0.48842257261276245, + "learning_rate": 1.9469322657135293e-05, + "loss": 3.1591, + "step": 5377 + }, + { + "epoch": 7.88, + "grad_norm": 0.40046364068984985, + "learning_rate": 1.944152441895616e-05, + "loss": 3.1994, + "step": 5378 + }, + { + "epoch": 7.88, + "grad_norm": 0.62181556224823, + "learning_rate": 1.9413743903101776e-05, + "loss": 3.1446, + "step": 5379 + }, + { + "epoch": 7.88, + "grad_norm": 0.7460499405860901, + "learning_rate": 1.9385981115683672e-05, + "loss": 3.1688, + "step": 5380 + }, + { + "epoch": 7.88, + "grad_norm": 0.2939889132976532, + "learning_rate": 1.9358236062809464e-05, + "loss": 3.1436, + "step": 5381 + }, + { + "epoch": 7.88, + "grad_norm": 0.6719468832015991, + "learning_rate": 1.9330508750582897e-05, + "loss": 3.1482, + "step": 5382 + }, + { + "epoch": 7.89, + "grad_norm": 0.41709572076797485, + "learning_rate": 1.930279918510377e-05, + "loss": 3.1564, + "step": 5383 + }, + { + "epoch": 7.89, + "grad_norm": 0.5399413704872131, + "learning_rate": 1.9275107372468014e-05, + "loss": 3.1684, + "step": 5384 + }, + { + "epoch": 7.89, + "grad_norm": 0.6927580237388611, + "learning_rate": 1.9247433318767648e-05, + "loss": 3.1473, + "step": 5385 + }, + { + "epoch": 7.89, + "grad_norm": 0.5529474020004272, + "learning_rate": 1.921977703009077e-05, + "loss": 3.1235, + "step": 5386 + }, + { + "epoch": 7.89, + "grad_norm": 1.1675437688827515, + "learning_rate": 1.9192138512521584e-05, + "loss": 3.1587, + "step": 5387 + }, + { + "epoch": 7.89, + "grad_norm": 0.3985271751880646, + "learning_rate": 1.916451777214039e-05, + "loss": 3.1479, + "step": 5388 + }, + { + "epoch": 7.89, + "grad_norm": 0.8106251358985901, + "learning_rate": 1.913691481502351e-05, + "loss": 3.1356, + "step": 5389 + }, + { + "epoch": 7.9, + "grad_norm": 0.767872154712677, + "learning_rate": 1.910932964724347e-05, + "loss": 3.1462, + "step": 5390 + }, + { + "epoch": 7.9, + "grad_norm": 0.41784894466400146, + "learning_rate": 1.9081762274868832e-05, + "loss": 3.1637, + "step": 5391 + }, + { + "epoch": 7.9, + "grad_norm": 0.8937169313430786, + "learning_rate": 1.9054212703964147e-05, + "loss": 3.1743, + "step": 5392 + }, + { + "epoch": 7.9, + "grad_norm": 3.9727253913879395, + "learning_rate": 1.9026680940590224e-05, + "loss": 3.1128, + "step": 5393 + }, + { + "epoch": 7.9, + "grad_norm": 0.4264329969882965, + "learning_rate": 1.8999166990803864e-05, + "loss": 3.1802, + "step": 5394 + }, + { + "epoch": 7.9, + "grad_norm": 0.4454759657382965, + "learning_rate": 1.8971670860657887e-05, + "loss": 3.1421, + "step": 5395 + }, + { + "epoch": 7.9, + "grad_norm": 0.8566815257072449, + "learning_rate": 1.8944192556201266e-05, + "loss": 3.1259, + "step": 5396 + }, + { + "epoch": 7.91, + "grad_norm": 0.42065227031707764, + "learning_rate": 1.8916732083479138e-05, + "loss": 3.1286, + "step": 5397 + }, + { + "epoch": 7.91, + "grad_norm": 1.0363951921463013, + "learning_rate": 1.888928944853251e-05, + "loss": 3.1475, + "step": 5398 + }, + { + "epoch": 7.91, + "grad_norm": 0.8358655571937561, + "learning_rate": 1.886186465739861e-05, + "loss": 3.1292, + "step": 5399 + }, + { + "epoch": 7.91, + "grad_norm": 0.5414959788322449, + "learning_rate": 1.8834457716110778e-05, + "loss": 3.1235, + "step": 5400 + }, + { + "epoch": 7.91, + "grad_norm": 0.6389888525009155, + "learning_rate": 1.8807068630698267e-05, + "loss": 3.1118, + "step": 5401 + }, + { + "epoch": 7.91, + "grad_norm": 0.5987576246261597, + "learning_rate": 1.8779697407186526e-05, + "loss": 3.1534, + "step": 5402 + }, + { + "epoch": 7.92, + "grad_norm": 0.9241575598716736, + "learning_rate": 1.8752344051597036e-05, + "loss": 3.207, + "step": 5403 + }, + { + "epoch": 7.92, + "grad_norm": 0.5730569958686829, + "learning_rate": 1.8725008569947365e-05, + "loss": 3.1383, + "step": 5404 + }, + { + "epoch": 7.92, + "grad_norm": 0.43397852778434753, + "learning_rate": 1.8697690968251126e-05, + "loss": 3.1723, + "step": 5405 + }, + { + "epoch": 7.92, + "grad_norm": 0.4378036558628082, + "learning_rate": 1.8670391252518017e-05, + "loss": 3.1829, + "step": 5406 + }, + { + "epoch": 7.92, + "grad_norm": 0.8141224980354309, + "learning_rate": 1.864310942875377e-05, + "loss": 3.1639, + "step": 5407 + }, + { + "epoch": 7.92, + "grad_norm": 0.25414714217185974, + "learning_rate": 1.8615845502960228e-05, + "loss": 3.1672, + "step": 5408 + }, + { + "epoch": 7.92, + "eval_loss": 3.008017063140869, + "eval_runtime": 1168.0951, + "eval_samples_per_second": 14.675, + "eval_steps_per_second": 0.918, + "step": 5408 + }, + { + "epoch": 7.92, + "grad_norm": 0.5402032136917114, + "learning_rate": 1.858859948113526e-05, + "loss": 3.1674, + "step": 5409 + }, + { + "epoch": 7.93, + "grad_norm": 0.5880429148674011, + "learning_rate": 1.8561371369272818e-05, + "loss": 3.1517, + "step": 5410 + }, + { + "epoch": 7.93, + "grad_norm": 0.44701412320137024, + "learning_rate": 1.8534161173362897e-05, + "loss": 3.1729, + "step": 5411 + }, + { + "epoch": 7.93, + "grad_norm": 0.36186301708221436, + "learning_rate": 1.8506968899391554e-05, + "loss": 3.1521, + "step": 5412 + }, + { + "epoch": 7.93, + "grad_norm": 0.35009482502937317, + "learning_rate": 1.847979455334091e-05, + "loss": 3.1199, + "step": 5413 + }, + { + "epoch": 7.93, + "grad_norm": 0.34212368726730347, + "learning_rate": 1.8452638141189138e-05, + "loss": 3.1637, + "step": 5414 + }, + { + "epoch": 7.93, + "grad_norm": 0.35525619983673096, + "learning_rate": 1.8425499668910472e-05, + "loss": 3.1613, + "step": 5415 + }, + { + "epoch": 7.93, + "grad_norm": 0.9575212597846985, + "learning_rate": 1.8398379142475196e-05, + "loss": 3.1801, + "step": 5416 + }, + { + "epoch": 7.94, + "grad_norm": 0.2845044434070587, + "learning_rate": 1.837127656784966e-05, + "loss": 3.1833, + "step": 5417 + }, + { + "epoch": 7.94, + "grad_norm": 0.29446759819984436, + "learning_rate": 1.834419195099617e-05, + "loss": 3.1385, + "step": 5418 + }, + { + "epoch": 7.94, + "grad_norm": 0.3464059829711914, + "learning_rate": 1.831712529787325e-05, + "loss": 3.1476, + "step": 5419 + }, + { + "epoch": 7.94, + "grad_norm": 0.24105912446975708, + "learning_rate": 1.829007661443538e-05, + "loss": 3.1684, + "step": 5420 + }, + { + "epoch": 7.94, + "grad_norm": 0.4570894241333008, + "learning_rate": 1.8263045906633002e-05, + "loss": 3.1824, + "step": 5421 + }, + { + "epoch": 7.94, + "grad_norm": 0.5457885265350342, + "learning_rate": 1.8236033180412793e-05, + "loss": 3.1701, + "step": 5422 + }, + { + "epoch": 7.94, + "grad_norm": 0.29805809259414673, + "learning_rate": 1.8209038441717306e-05, + "loss": 3.1578, + "step": 5423 + }, + { + "epoch": 7.95, + "grad_norm": 0.47300076484680176, + "learning_rate": 1.8182061696485207e-05, + "loss": 3.1372, + "step": 5424 + }, + { + "epoch": 7.95, + "grad_norm": 0.7307349443435669, + "learning_rate": 1.815510295065126e-05, + "loss": 3.1615, + "step": 5425 + }, + { + "epoch": 7.95, + "grad_norm": 0.28895077109336853, + "learning_rate": 1.8128162210146136e-05, + "loss": 3.1576, + "step": 5426 + }, + { + "epoch": 7.95, + "grad_norm": 0.4990285634994507, + "learning_rate": 1.8101239480896626e-05, + "loss": 3.1346, + "step": 5427 + }, + { + "epoch": 7.95, + "grad_norm": 0.42847058176994324, + "learning_rate": 1.8074334768825606e-05, + "loss": 3.1791, + "step": 5428 + }, + { + "epoch": 7.95, + "grad_norm": 0.3061131536960602, + "learning_rate": 1.804744807985188e-05, + "loss": 3.1364, + "step": 5429 + }, + { + "epoch": 7.96, + "grad_norm": 0.4757770001888275, + "learning_rate": 1.8020579419890326e-05, + "loss": 3.1907, + "step": 5430 + }, + { + "epoch": 7.96, + "grad_norm": 0.4148907959461212, + "learning_rate": 1.799372879485196e-05, + "loss": 3.2014, + "step": 5431 + }, + { + "epoch": 7.96, + "grad_norm": 0.30733799934387207, + "learning_rate": 1.7966896210643634e-05, + "loss": 3.1103, + "step": 5432 + }, + { + "epoch": 7.96, + "grad_norm": 0.40864160656929016, + "learning_rate": 1.794008167316836e-05, + "loss": 3.1728, + "step": 5433 + }, + { + "epoch": 7.96, + "grad_norm": 0.5808351039886475, + "learning_rate": 1.7913285188325235e-05, + "loss": 3.1226, + "step": 5434 + }, + { + "epoch": 7.96, + "grad_norm": 0.5464041829109192, + "learning_rate": 1.7886506762009225e-05, + "loss": 3.1737, + "step": 5435 + }, + { + "epoch": 7.96, + "grad_norm": 0.587399959564209, + "learning_rate": 1.7859746400111422e-05, + "loss": 3.1581, + "step": 5436 + }, + { + "epoch": 7.97, + "grad_norm": 0.36822310090065, + "learning_rate": 1.7833004108518925e-05, + "loss": 3.1762, + "step": 5437 + }, + { + "epoch": 7.97, + "grad_norm": 2.8975210189819336, + "learning_rate": 1.7806279893114875e-05, + "loss": 3.1847, + "step": 5438 + }, + { + "epoch": 7.97, + "grad_norm": 0.3770262897014618, + "learning_rate": 1.77795737597784e-05, + "loss": 3.1737, + "step": 5439 + }, + { + "epoch": 7.97, + "grad_norm": 0.4018249213695526, + "learning_rate": 1.7752885714384694e-05, + "loss": 3.1248, + "step": 5440 + }, + { + "epoch": 7.97, + "grad_norm": 0.26656123995780945, + "learning_rate": 1.772621576280492e-05, + "loss": 3.1343, + "step": 5441 + }, + { + "epoch": 7.97, + "grad_norm": 0.29161426424980164, + "learning_rate": 1.7699563910906314e-05, + "loss": 3.1363, + "step": 5442 + }, + { + "epoch": 7.97, + "grad_norm": 0.4624899625778198, + "learning_rate": 1.767293016455209e-05, + "loss": 3.1913, + "step": 5443 + }, + { + "epoch": 7.98, + "grad_norm": 0.529693067073822, + "learning_rate": 1.7646314529601505e-05, + "loss": 3.1301, + "step": 5444 + }, + { + "epoch": 7.98, + "grad_norm": 0.25641757249832153, + "learning_rate": 1.7619717011909842e-05, + "loss": 3.1114, + "step": 5445 + }, + { + "epoch": 7.98, + "grad_norm": 0.26415735483169556, + "learning_rate": 1.7593137617328282e-05, + "loss": 3.1425, + "step": 5446 + }, + { + "epoch": 7.98, + "grad_norm": 0.6229835748672485, + "learning_rate": 1.7566576351704224e-05, + "loss": 3.1519, + "step": 5447 + }, + { + "epoch": 7.98, + "grad_norm": 0.6144125461578369, + "learning_rate": 1.7540033220880936e-05, + "loss": 3.1652, + "step": 5448 + }, + { + "epoch": 7.98, + "grad_norm": 0.30449098348617554, + "learning_rate": 1.7513508230697673e-05, + "loss": 3.1811, + "step": 5449 + }, + { + "epoch": 7.98, + "grad_norm": 0.4084562361240387, + "learning_rate": 1.748700138698982e-05, + "loss": 3.185, + "step": 5450 + }, + { + "epoch": 7.99, + "grad_norm": 0.6092720627784729, + "learning_rate": 1.7460512695588704e-05, + "loss": 3.1706, + "step": 5451 + }, + { + "epoch": 7.99, + "grad_norm": 0.5944820046424866, + "learning_rate": 1.7434042162321597e-05, + "loss": 3.165, + "step": 5452 + }, + { + "epoch": 7.99, + "grad_norm": 0.3383338749408722, + "learning_rate": 1.7407589793011912e-05, + "loss": 3.171, + "step": 5453 + }, + { + "epoch": 7.99, + "grad_norm": 0.7481070160865784, + "learning_rate": 1.738115559347898e-05, + "loss": 3.1244, + "step": 5454 + }, + { + "epoch": 7.99, + "grad_norm": 0.6213645339012146, + "learning_rate": 1.735473956953808e-05, + "loss": 3.198, + "step": 5455 + }, + { + "epoch": 7.99, + "grad_norm": 0.5867860913276672, + "learning_rate": 1.7328341727000664e-05, + "loss": 3.1474, + "step": 5456 + }, + { + "epoch": 8.0, + "grad_norm": 0.4407004714012146, + "learning_rate": 1.7301962071674004e-05, + "loss": 3.1451, + "step": 5457 + }, + { + "epoch": 8.0, + "grad_norm": 0.6691040992736816, + "learning_rate": 1.7275600609361443e-05, + "loss": 3.1377, + "step": 5458 + }, + { + "epoch": 8.0, + "grad_norm": 0.63309645652771, + "learning_rate": 1.72492573458624e-05, + "loss": 3.1297, + "step": 5459 + }, + { + "epoch": 8.0, + "grad_norm": 0.481118768453598, + "learning_rate": 1.722293228697216e-05, + "loss": 3.2125, + "step": 5460 + }, + { + "epoch": 8.0, + "grad_norm": 0.5907300710678101, + "learning_rate": 1.7196625438482027e-05, + "loss": 3.1457, + "step": 5461 + }, + { + "epoch": 8.0, + "grad_norm": 0.7083212733268738, + "learning_rate": 1.7170336806179433e-05, + "loss": 3.1235, + "step": 5462 + }, + { + "epoch": 8.0, + "grad_norm": 0.9391943216323853, + "learning_rate": 1.7144066395847615e-05, + "loss": 3.1399, + "step": 5463 + }, + { + "epoch": 8.01, + "grad_norm": 0.5589471459388733, + "learning_rate": 1.7117814213265893e-05, + "loss": 3.129, + "step": 5464 + }, + { + "epoch": 8.01, + "grad_norm": 0.7485381960868835, + "learning_rate": 1.7091580264209627e-05, + "loss": 3.1755, + "step": 5465 + }, + { + "epoch": 8.01, + "grad_norm": 0.8377997279167175, + "learning_rate": 1.706536455445007e-05, + "loss": 3.159, + "step": 5466 + }, + { + "epoch": 8.01, + "grad_norm": 0.469113290309906, + "learning_rate": 1.7039167089754492e-05, + "loss": 3.1715, + "step": 5467 + }, + { + "epoch": 8.01, + "grad_norm": 0.610246479511261, + "learning_rate": 1.701298787588619e-05, + "loss": 3.1326, + "step": 5468 + }, + { + "epoch": 8.01, + "grad_norm": 0.643669068813324, + "learning_rate": 1.698682691860439e-05, + "loss": 3.1593, + "step": 5469 + }, + { + "epoch": 8.01, + "grad_norm": 0.5371613502502441, + "learning_rate": 1.6960684223664337e-05, + "loss": 3.1264, + "step": 5470 + }, + { + "epoch": 8.0, + "grad_norm": 0.4291907250881195, + "learning_rate": 1.6934559796817263e-05, + "loss": 3.1593, + "step": 5471 + }, + { + "epoch": 8.0, + "grad_norm": 0.6830047965049744, + "learning_rate": 1.690845364381034e-05, + "loss": 3.1404, + "step": 5472 + }, + { + "epoch": 8.0, + "grad_norm": 0.419023722410202, + "learning_rate": 1.6882365770386766e-05, + "loss": 3.1497, + "step": 5473 + }, + { + "epoch": 8.01, + "grad_norm": 0.6376973986625671, + "learning_rate": 1.685629618228569e-05, + "loss": 3.1803, + "step": 5474 + }, + { + "epoch": 8.01, + "grad_norm": 0.814529299736023, + "learning_rate": 1.6830244885242252e-05, + "loss": 3.1458, + "step": 5475 + }, + { + "epoch": 8.01, + "grad_norm": 0.5563649535179138, + "learning_rate": 1.6804211884987563e-05, + "loss": 3.1326, + "step": 5476 + }, + { + "epoch": 8.01, + "grad_norm": 0.8076274394989014, + "learning_rate": 1.677819718724871e-05, + "loss": 3.1377, + "step": 5477 + }, + { + "epoch": 8.01, + "grad_norm": 0.39461883902549744, + "learning_rate": 1.675220079774874e-05, + "loss": 3.1151, + "step": 5478 + }, + { + "epoch": 8.01, + "grad_norm": 0.4176673889160156, + "learning_rate": 1.6726222722206718e-05, + "loss": 3.1486, + "step": 5479 + }, + { + "epoch": 8.01, + "grad_norm": 0.4115258455276489, + "learning_rate": 1.6700262966337578e-05, + "loss": 3.1341, + "step": 5480 + }, + { + "epoch": 8.02, + "grad_norm": 0.7755107879638672, + "learning_rate": 1.6674321535852343e-05, + "loss": 3.1913, + "step": 5481 + }, + { + "epoch": 8.02, + "grad_norm": 0.29228803515434265, + "learning_rate": 1.664839843645799e-05, + "loss": 3.1166, + "step": 5482 + }, + { + "epoch": 8.02, + "grad_norm": 0.7680675983428955, + "learning_rate": 1.662249367385732e-05, + "loss": 3.1529, + "step": 5483 + }, + { + "epoch": 8.02, + "grad_norm": 1.2723925113677979, + "learning_rate": 1.6596607253749308e-05, + "loss": 3.148, + "step": 5484 + }, + { + "epoch": 8.02, + "grad_norm": 0.5549447536468506, + "learning_rate": 1.6570739181828775e-05, + "loss": 3.1678, + "step": 5485 + }, + { + "epoch": 8.02, + "grad_norm": 0.6857723593711853, + "learning_rate": 1.654488946378645e-05, + "loss": 3.2432, + "step": 5486 + }, + { + "epoch": 8.02, + "grad_norm": 0.9772923588752747, + "learning_rate": 1.651905810530918e-05, + "loss": 3.1304, + "step": 5487 + }, + { + "epoch": 8.03, + "grad_norm": 0.33374878764152527, + "learning_rate": 1.649324511207969e-05, + "loss": 3.1321, + "step": 5488 + }, + { + "epoch": 8.03, + "grad_norm": 0.5361008644104004, + "learning_rate": 1.6467450489776582e-05, + "loss": 3.1434, + "step": 5489 + }, + { + "epoch": 8.03, + "grad_norm": 0.5644283890724182, + "learning_rate": 1.644167424407461e-05, + "loss": 3.1415, + "step": 5490 + }, + { + "epoch": 8.03, + "grad_norm": 0.3192185163497925, + "learning_rate": 1.641591638064429e-05, + "loss": 3.1669, + "step": 5491 + }, + { + "epoch": 8.03, + "grad_norm": 2.5619218349456787, + "learning_rate": 1.639017690515222e-05, + "loss": 3.1329, + "step": 5492 + }, + { + "epoch": 8.03, + "grad_norm": 0.5575476288795471, + "learning_rate": 1.6364455823260906e-05, + "loss": 3.1527, + "step": 5493 + }, + { + "epoch": 8.03, + "grad_norm": 0.6022101044654846, + "learning_rate": 1.6338753140628792e-05, + "loss": 3.1372, + "step": 5494 + }, + { + "epoch": 8.04, + "grad_norm": 0.5677317380905151, + "learning_rate": 1.631306886291033e-05, + "loss": 3.1356, + "step": 5495 + }, + { + "epoch": 8.04, + "grad_norm": 0.7927194833755493, + "learning_rate": 1.6287402995755864e-05, + "loss": 3.1394, + "step": 5496 + }, + { + "epoch": 8.04, + "grad_norm": 0.3976461887359619, + "learning_rate": 1.6261755544811728e-05, + "loss": 3.129, + "step": 5497 + }, + { + "epoch": 8.04, + "grad_norm": 0.6121003031730652, + "learning_rate": 1.6236126515720184e-05, + "loss": 3.1544, + "step": 5498 + }, + { + "epoch": 8.04, + "grad_norm": 0.4400195777416229, + "learning_rate": 1.621051591411944e-05, + "loss": 3.1182, + "step": 5499 + }, + { + "epoch": 8.04, + "grad_norm": 0.4340709149837494, + "learning_rate": 1.6184923745643667e-05, + "loss": 3.1473, + "step": 5500 + }, + { + "epoch": 8.05, + "grad_norm": 0.8094406723976135, + "learning_rate": 1.615935001592297e-05, + "loss": 3.1462, + "step": 5501 + }, + { + "epoch": 8.05, + "grad_norm": 0.40194961428642273, + "learning_rate": 1.6133794730583384e-05, + "loss": 3.1298, + "step": 5502 + }, + { + "epoch": 8.05, + "grad_norm": 0.44503289461135864, + "learning_rate": 1.6108257895246925e-05, + "loss": 3.1624, + "step": 5503 + }, + { + "epoch": 8.05, + "grad_norm": 0.23914110660552979, + "learning_rate": 1.6082739515531497e-05, + "loss": 3.1345, + "step": 5504 + }, + { + "epoch": 8.05, + "grad_norm": 0.37540099024772644, + "learning_rate": 1.6057239597051e-05, + "loss": 3.1487, + "step": 5505 + }, + { + "epoch": 8.05, + "grad_norm": 0.31591150164604187, + "learning_rate": 1.603175814541522e-05, + "loss": 3.1514, + "step": 5506 + }, + { + "epoch": 8.05, + "grad_norm": 0.5472672581672668, + "learning_rate": 1.6006295166229924e-05, + "loss": 3.1564, + "step": 5507 + }, + { + "epoch": 8.06, + "grad_norm": 0.42764073610305786, + "learning_rate": 1.598085066509679e-05, + "loss": 3.1504, + "step": 5508 + }, + { + "epoch": 8.06, + "grad_norm": 0.45837774872779846, + "learning_rate": 1.5955424647613438e-05, + "loss": 3.1613, + "step": 5509 + }, + { + "epoch": 8.06, + "grad_norm": 0.3727029263973236, + "learning_rate": 1.593001711937343e-05, + "loss": 3.1321, + "step": 5510 + }, + { + "epoch": 8.06, + "grad_norm": 0.38336122035980225, + "learning_rate": 1.5904628085966234e-05, + "loss": 3.1526, + "step": 5511 + }, + { + "epoch": 8.06, + "grad_norm": 0.33697742223739624, + "learning_rate": 1.587925755297728e-05, + "loss": 3.1447, + "step": 5512 + }, + { + "epoch": 8.06, + "grad_norm": 0.42843082547187805, + "learning_rate": 1.5853905525987932e-05, + "loss": 3.1343, + "step": 5513 + }, + { + "epoch": 8.06, + "grad_norm": 0.4407825469970703, + "learning_rate": 1.5828572010575426e-05, + "loss": 3.1385, + "step": 5514 + }, + { + "epoch": 8.07, + "grad_norm": 0.3060319125652313, + "learning_rate": 1.5803257012312956e-05, + "loss": 3.1189, + "step": 5515 + }, + { + "epoch": 8.07, + "grad_norm": 0.3679053783416748, + "learning_rate": 1.5777960536769743e-05, + "loss": 3.1458, + "step": 5516 + }, + { + "epoch": 8.07, + "grad_norm": 0.5479602813720703, + "learning_rate": 1.5752682589510747e-05, + "loss": 3.1543, + "step": 5517 + }, + { + "epoch": 8.07, + "grad_norm": 0.6712786555290222, + "learning_rate": 1.5727423176096955e-05, + "loss": 3.1429, + "step": 5518 + }, + { + "epoch": 8.07, + "grad_norm": 0.24910399317741394, + "learning_rate": 1.5702182302085343e-05, + "loss": 3.2076, + "step": 5519 + }, + { + "epoch": 8.07, + "grad_norm": 0.42461472749710083, + "learning_rate": 1.567695997302867e-05, + "loss": 3.1566, + "step": 5520 + }, + { + "epoch": 8.07, + "grad_norm": 0.2487383484840393, + "learning_rate": 1.5651756194475666e-05, + "loss": 3.1567, + "step": 5521 + }, + { + "epoch": 8.08, + "grad_norm": 0.602077841758728, + "learning_rate": 1.5626570971971068e-05, + "loss": 3.1662, + "step": 5522 + }, + { + "epoch": 8.08, + "grad_norm": 0.5867159962654114, + "learning_rate": 1.560140431105539e-05, + "loss": 3.1886, + "step": 5523 + }, + { + "epoch": 8.08, + "grad_norm": 0.2727794945240021, + "learning_rate": 1.5576256217265152e-05, + "loss": 3.127, + "step": 5524 + }, + { + "epoch": 8.08, + "grad_norm": 0.33053722977638245, + "learning_rate": 1.5551126696132756e-05, + "loss": 3.1243, + "step": 5525 + }, + { + "epoch": 8.08, + "grad_norm": 0.5441333055496216, + "learning_rate": 1.552601575318654e-05, + "loss": 3.1573, + "step": 5526 + }, + { + "epoch": 8.08, + "grad_norm": 0.4156208634376526, + "learning_rate": 1.550092339395074e-05, + "loss": 3.1993, + "step": 5527 + }, + { + "epoch": 8.09, + "grad_norm": 0.5446397066116333, + "learning_rate": 1.5475849623945505e-05, + "loss": 3.1158, + "step": 5528 + }, + { + "epoch": 8.09, + "grad_norm": 0.4556122124195099, + "learning_rate": 1.545079444868689e-05, + "loss": 3.1386, + "step": 5529 + }, + { + "epoch": 8.09, + "grad_norm": 0.6932422518730164, + "learning_rate": 1.5425757873686875e-05, + "loss": 3.1687, + "step": 5530 + }, + { + "epoch": 8.09, + "grad_norm": 0.28542694449424744, + "learning_rate": 1.5400739904453344e-05, + "loss": 3.1192, + "step": 5531 + }, + { + "epoch": 8.09, + "grad_norm": 0.3106916844844818, + "learning_rate": 1.537574054649007e-05, + "loss": 3.1941, + "step": 5532 + }, + { + "epoch": 8.09, + "grad_norm": 0.5748286247253418, + "learning_rate": 1.5350759805296744e-05, + "loss": 3.1547, + "step": 5533 + }, + { + "epoch": 8.09, + "grad_norm": 0.2282506823539734, + "learning_rate": 1.532579768636898e-05, + "loss": 3.1223, + "step": 5534 + }, + { + "epoch": 8.1, + "grad_norm": 0.36305752396583557, + "learning_rate": 1.5300854195198257e-05, + "loss": 3.1469, + "step": 5535 + }, + { + "epoch": 8.1, + "grad_norm": 0.3789515197277069, + "learning_rate": 1.527592933727198e-05, + "loss": 3.1726, + "step": 5536 + }, + { + "epoch": 8.1, + "grad_norm": 0.32144132256507874, + "learning_rate": 1.5251023118073448e-05, + "loss": 3.1429, + "step": 5537 + }, + { + "epoch": 8.1, + "grad_norm": 0.47844165563583374, + "learning_rate": 1.5226135543081876e-05, + "loss": 3.1361, + "step": 5538 + }, + { + "epoch": 8.1, + "grad_norm": 0.3552798628807068, + "learning_rate": 1.520126661777238e-05, + "loss": 3.1202, + "step": 5539 + }, + { + "epoch": 8.1, + "grad_norm": 0.24844308197498322, + "learning_rate": 1.5176416347615885e-05, + "loss": 3.1319, + "step": 5540 + }, + { + "epoch": 8.1, + "grad_norm": 0.7230424880981445, + "learning_rate": 1.515158473807935e-05, + "loss": 3.1792, + "step": 5541 + }, + { + "epoch": 8.11, + "grad_norm": 0.3808114528656006, + "learning_rate": 1.5126771794625572e-05, + "loss": 3.1422, + "step": 5542 + }, + { + "epoch": 8.11, + "grad_norm": 0.6194649934768677, + "learning_rate": 1.5101977522713152e-05, + "loss": 3.1325, + "step": 5543 + }, + { + "epoch": 8.11, + "grad_norm": 0.6643770933151245, + "learning_rate": 1.5077201927796747e-05, + "loss": 3.1341, + "step": 5544 + }, + { + "epoch": 8.11, + "grad_norm": 0.20721003413200378, + "learning_rate": 1.50524450153268e-05, + "loss": 3.1474, + "step": 5545 + }, + { + "epoch": 8.11, + "grad_norm": 0.3616236746311188, + "learning_rate": 1.5027706790749619e-05, + "loss": 3.1442, + "step": 5546 + }, + { + "epoch": 8.11, + "grad_norm": 1.006027340888977, + "learning_rate": 1.5002987259507539e-05, + "loss": 3.2174, + "step": 5547 + }, + { + "epoch": 8.12, + "grad_norm": 0.40877261757850647, + "learning_rate": 1.4978286427038601e-05, + "loss": 3.203, + "step": 5548 + }, + { + "epoch": 8.12, + "grad_norm": 0.6168388724327087, + "learning_rate": 1.4953604298776847e-05, + "loss": 3.1284, + "step": 5549 + }, + { + "epoch": 8.12, + "grad_norm": 0.846990704536438, + "learning_rate": 1.4928940880152232e-05, + "loss": 3.1267, + "step": 5550 + }, + { + "epoch": 8.12, + "grad_norm": 0.47257474064826965, + "learning_rate": 1.490429617659047e-05, + "loss": 3.1828, + "step": 5551 + }, + { + "epoch": 8.12, + "grad_norm": 0.5591023564338684, + "learning_rate": 1.4879670193513251e-05, + "loss": 3.1506, + "step": 5552 + }, + { + "epoch": 8.12, + "grad_norm": 1.040032982826233, + "learning_rate": 1.4855062936338183e-05, + "loss": 3.2159, + "step": 5553 + }, + { + "epoch": 8.12, + "grad_norm": 0.566142737865448, + "learning_rate": 1.4830474410478623e-05, + "loss": 3.1234, + "step": 5554 + }, + { + "epoch": 8.13, + "grad_norm": 0.6576405763626099, + "learning_rate": 1.4805904621343903e-05, + "loss": 3.1391, + "step": 5555 + }, + { + "epoch": 8.13, + "grad_norm": 0.622337818145752, + "learning_rate": 1.4781353574339252e-05, + "loss": 3.1823, + "step": 5556 + }, + { + "epoch": 8.13, + "grad_norm": 0.45685091614723206, + "learning_rate": 1.4756821274865695e-05, + "loss": 3.1574, + "step": 5557 + }, + { + "epoch": 8.13, + "grad_norm": 0.45965492725372314, + "learning_rate": 1.473230772832015e-05, + "loss": 3.1754, + "step": 5558 + }, + { + "epoch": 8.13, + "grad_norm": 0.5647199153900146, + "learning_rate": 1.4707812940095511e-05, + "loss": 3.1166, + "step": 5559 + }, + { + "epoch": 8.13, + "grad_norm": 0.6305474638938904, + "learning_rate": 1.4683336915580414e-05, + "loss": 3.134, + "step": 5560 + }, + { + "epoch": 8.13, + "grad_norm": 0.23273073136806488, + "learning_rate": 1.4658879660159419e-05, + "loss": 3.1207, + "step": 5561 + }, + { + "epoch": 8.14, + "grad_norm": 0.5071410536766052, + "learning_rate": 1.4634441179212965e-05, + "loss": 3.1623, + "step": 5562 + }, + { + "epoch": 8.14, + "grad_norm": 0.34845244884490967, + "learning_rate": 1.4610021478117352e-05, + "loss": 3.1172, + "step": 5563 + }, + { + "epoch": 8.14, + "grad_norm": 0.43220025300979614, + "learning_rate": 1.4585620562244773e-05, + "loss": 3.1265, + "step": 5564 + }, + { + "epoch": 8.14, + "grad_norm": 0.3767712712287903, + "learning_rate": 1.4561238436963231e-05, + "loss": 3.1457, + "step": 5565 + }, + { + "epoch": 8.14, + "grad_norm": 0.4660220742225647, + "learning_rate": 1.453687510763666e-05, + "loss": 3.1654, + "step": 5566 + }, + { + "epoch": 8.14, + "grad_norm": 0.3686300814151764, + "learning_rate": 1.4512530579624806e-05, + "loss": 3.1813, + "step": 5567 + }, + { + "epoch": 8.14, + "grad_norm": 0.40277543663978577, + "learning_rate": 1.4488204858283328e-05, + "loss": 3.1764, + "step": 5568 + }, + { + "epoch": 8.15, + "grad_norm": 0.4759199917316437, + "learning_rate": 1.446389794896369e-05, + "loss": 3.2285, + "step": 5569 + }, + { + "epoch": 8.15, + "grad_norm": 0.6583293676376343, + "learning_rate": 1.4439609857013304e-05, + "loss": 3.1806, + "step": 5570 + }, + { + "epoch": 8.15, + "grad_norm": 2.4492862224578857, + "learning_rate": 1.4415340587775295e-05, + "loss": 3.1365, + "step": 5571 + }, + { + "epoch": 8.15, + "grad_norm": 0.5203627943992615, + "learning_rate": 1.4391090146588816e-05, + "loss": 3.1313, + "step": 5572 + }, + { + "epoch": 8.15, + "grad_norm": 0.3894268870353699, + "learning_rate": 1.4366858538788796e-05, + "loss": 3.1248, + "step": 5573 + }, + { + "epoch": 8.15, + "grad_norm": 0.5935128331184387, + "learning_rate": 1.4342645769705977e-05, + "loss": 3.1253, + "step": 5574 + }, + { + "epoch": 8.16, + "grad_norm": 0.44079282879829407, + "learning_rate": 1.4318451844667058e-05, + "loss": 3.1611, + "step": 5575 + }, + { + "epoch": 8.16, + "grad_norm": 7.371267318725586, + "learning_rate": 1.429427676899454e-05, + "loss": 3.1284, + "step": 5576 + }, + { + "epoch": 8.16, + "grad_norm": 0.3392414152622223, + "learning_rate": 1.4270120548006716e-05, + "loss": 3.1202, + "step": 5577 + }, + { + "epoch": 8.16, + "eval_loss": 3.003582000732422, + "eval_runtime": 1164.277, + "eval_samples_per_second": 14.723, + "eval_steps_per_second": 0.921, + "step": 5577 + }, + { + "epoch": 8.16, + "grad_norm": 0.4595149755477905, + "learning_rate": 1.4245983187017854e-05, + "loss": 3.1606, + "step": 5578 + }, + { + "epoch": 8.16, + "grad_norm": 0.20213989913463593, + "learning_rate": 1.4221864691338015e-05, + "loss": 3.1206, + "step": 5579 + }, + { + "epoch": 8.16, + "grad_norm": 0.389874666929245, + "learning_rate": 1.419776506627305e-05, + "loss": 3.1593, + "step": 5580 + }, + { + "epoch": 8.16, + "grad_norm": 0.7267783284187317, + "learning_rate": 1.417368431712478e-05, + "loss": 3.1118, + "step": 5581 + }, + { + "epoch": 8.17, + "grad_norm": 0.2689005136489868, + "learning_rate": 1.4149622449190768e-05, + "loss": 3.1423, + "step": 5582 + }, + { + "epoch": 8.17, + "grad_norm": 0.2806105613708496, + "learning_rate": 1.4125579467764438e-05, + "loss": 3.1678, + "step": 5583 + }, + { + "epoch": 8.17, + "grad_norm": 0.22201931476593018, + "learning_rate": 1.4101555378135178e-05, + "loss": 3.1606, + "step": 5584 + }, + { + "epoch": 8.17, + "grad_norm": 0.5508745908737183, + "learning_rate": 1.407755018558804e-05, + "loss": 3.1305, + "step": 5585 + }, + { + "epoch": 8.17, + "grad_norm": 0.5338426232337952, + "learning_rate": 1.4053563895404032e-05, + "loss": 3.1826, + "step": 5586 + }, + { + "epoch": 8.17, + "grad_norm": 0.45947423577308655, + "learning_rate": 1.4029596512859989e-05, + "loss": 3.1468, + "step": 5587 + }, + { + "epoch": 8.17, + "grad_norm": 0.2771443724632263, + "learning_rate": 1.4005648043228558e-05, + "loss": 3.1542, + "step": 5588 + }, + { + "epoch": 8.18, + "grad_norm": 0.372646689414978, + "learning_rate": 1.398171849177825e-05, + "loss": 3.1321, + "step": 5589 + }, + { + "epoch": 8.18, + "grad_norm": 0.682151734828949, + "learning_rate": 1.3957807863773398e-05, + "loss": 3.1698, + "step": 5590 + }, + { + "epoch": 8.18, + "grad_norm": 0.39036300778388977, + "learning_rate": 1.3933916164474192e-05, + "loss": 3.129, + "step": 5591 + }, + { + "epoch": 8.18, + "grad_norm": 0.4261132478713989, + "learning_rate": 1.3910043399136651e-05, + "loss": 3.1635, + "step": 5592 + }, + { + "epoch": 8.18, + "grad_norm": 0.4003067910671234, + "learning_rate": 1.3886189573012597e-05, + "loss": 3.1571, + "step": 5593 + }, + { + "epoch": 8.18, + "grad_norm": 1.0211633443832397, + "learning_rate": 1.3862354691349732e-05, + "loss": 3.1219, + "step": 5594 + }, + { + "epoch": 8.18, + "grad_norm": 0.3579561114311218, + "learning_rate": 1.3838538759391573e-05, + "loss": 3.1424, + "step": 5595 + }, + { + "epoch": 8.19, + "grad_norm": 0.5200756788253784, + "learning_rate": 1.3814741782377461e-05, + "loss": 3.1432, + "step": 5596 + }, + { + "epoch": 8.19, + "grad_norm": 0.9823718667030334, + "learning_rate": 1.3790963765542553e-05, + "loss": 3.1739, + "step": 5597 + }, + { + "epoch": 8.19, + "grad_norm": 0.4610528349876404, + "learning_rate": 1.3767204714117876e-05, + "loss": 3.1228, + "step": 5598 + }, + { + "epoch": 8.19, + "grad_norm": 0.46500396728515625, + "learning_rate": 1.3743464633330261e-05, + "loss": 3.1569, + "step": 5599 + }, + { + "epoch": 8.19, + "grad_norm": 0.48705846071243286, + "learning_rate": 1.3719743528402362e-05, + "loss": 3.1635, + "step": 5600 + }, + { + "epoch": 8.19, + "grad_norm": 0.4838919937610626, + "learning_rate": 1.369604140455265e-05, + "loss": 3.1584, + "step": 5601 + }, + { + "epoch": 8.2, + "grad_norm": 0.46300509572029114, + "learning_rate": 1.3672358266995456e-05, + "loss": 3.1597, + "step": 5602 + }, + { + "epoch": 8.2, + "grad_norm": 0.32473286986351013, + "learning_rate": 1.3648694120940896e-05, + "loss": 3.1018, + "step": 5603 + }, + { + "epoch": 8.2, + "grad_norm": 0.7043519020080566, + "learning_rate": 1.3625048971594957e-05, + "loss": 3.1544, + "step": 5604 + }, + { + "epoch": 8.2, + "grad_norm": 0.36091065406799316, + "learning_rate": 1.3601422824159338e-05, + "loss": 3.124, + "step": 5605 + }, + { + "epoch": 8.2, + "grad_norm": 0.2936200499534607, + "learning_rate": 1.3577815683831708e-05, + "loss": 3.1584, + "step": 5606 + }, + { + "epoch": 8.2, + "grad_norm": 0.5868943333625793, + "learning_rate": 1.3554227555805465e-05, + "loss": 3.1279, + "step": 5607 + }, + { + "epoch": 8.2, + "grad_norm": 0.4498915672302246, + "learning_rate": 1.3530658445269783e-05, + "loss": 3.1589, + "step": 5608 + }, + { + "epoch": 8.21, + "grad_norm": 0.7671617269515991, + "learning_rate": 1.350710835740977e-05, + "loss": 3.1504, + "step": 5609 + }, + { + "epoch": 8.21, + "grad_norm": 3.664649486541748, + "learning_rate": 1.3483577297406303e-05, + "loss": 3.1071, + "step": 5610 + }, + { + "epoch": 8.21, + "grad_norm": 0.42762038111686707, + "learning_rate": 1.3460065270435984e-05, + "loss": 3.133, + "step": 5611 + }, + { + "epoch": 8.21, + "grad_norm": 0.3746001720428467, + "learning_rate": 1.3436572281671333e-05, + "loss": 3.1491, + "step": 5612 + }, + { + "epoch": 8.21, + "grad_norm": 0.5500866770744324, + "learning_rate": 1.3413098336280694e-05, + "loss": 3.1355, + "step": 5613 + }, + { + "epoch": 8.21, + "grad_norm": 0.5390032529830933, + "learning_rate": 1.3389643439428124e-05, + "loss": 3.1345, + "step": 5614 + }, + { + "epoch": 8.21, + "grad_norm": 0.30622246861457825, + "learning_rate": 1.3366207596273561e-05, + "loss": 3.1552, + "step": 5615 + }, + { + "epoch": 8.22, + "grad_norm": 0.3916447162628174, + "learning_rate": 1.3342790811972727e-05, + "loss": 3.1234, + "step": 5616 + }, + { + "epoch": 8.22, + "grad_norm": 0.5330905914306641, + "learning_rate": 1.3319393091677168e-05, + "loss": 3.1187, + "step": 5617 + }, + { + "epoch": 8.22, + "grad_norm": 0.700232744216919, + "learning_rate": 1.3296014440534222e-05, + "loss": 3.1759, + "step": 5618 + }, + { + "epoch": 8.22, + "grad_norm": 0.5803514122962952, + "learning_rate": 1.3272654863687028e-05, + "loss": 3.1565, + "step": 5619 + }, + { + "epoch": 8.22, + "grad_norm": 0.35048308968544006, + "learning_rate": 1.3249314366274546e-05, + "loss": 3.1275, + "step": 5620 + }, + { + "epoch": 8.22, + "grad_norm": 0.6838834285736084, + "learning_rate": 1.3225992953431532e-05, + "loss": 3.1853, + "step": 5621 + }, + { + "epoch": 8.22, + "grad_norm": 0.37435978651046753, + "learning_rate": 1.3202690630288528e-05, + "loss": 3.1035, + "step": 5622 + }, + { + "epoch": 8.23, + "grad_norm": 0.3659183979034424, + "learning_rate": 1.31794074019719e-05, + "loss": 3.1258, + "step": 5623 + }, + { + "epoch": 8.23, + "grad_norm": 0.9304698705673218, + "learning_rate": 1.3156143273603794e-05, + "loss": 3.1519, + "step": 5624 + }, + { + "epoch": 8.23, + "grad_norm": 0.29267510771751404, + "learning_rate": 1.3132898250302172e-05, + "loss": 3.1569, + "step": 5625 + }, + { + "epoch": 8.23, + "grad_norm": 0.7396655678749084, + "learning_rate": 1.3109672337180768e-05, + "loss": 3.1207, + "step": 5626 + }, + { + "epoch": 8.23, + "grad_norm": 0.7278848886489868, + "learning_rate": 1.3086465539349146e-05, + "loss": 3.1382, + "step": 5627 + }, + { + "epoch": 8.23, + "grad_norm": 0.5418538451194763, + "learning_rate": 1.3063277861912637e-05, + "loss": 3.1373, + "step": 5628 + }, + { + "epoch": 8.24, + "grad_norm": 0.7232877016067505, + "learning_rate": 1.304010930997237e-05, + "loss": 3.1655, + "step": 5629 + }, + { + "epoch": 8.24, + "grad_norm": 0.8563187122344971, + "learning_rate": 1.3016959888625269e-05, + "loss": 3.1261, + "step": 5630 + }, + { + "epoch": 8.24, + "grad_norm": 0.671583354473114, + "learning_rate": 1.299382960296407e-05, + "loss": 3.1386, + "step": 5631 + }, + { + "epoch": 8.24, + "grad_norm": 0.49029281735420227, + "learning_rate": 1.2970718458077258e-05, + "loss": 3.1581, + "step": 5632 + }, + { + "epoch": 8.24, + "grad_norm": 0.7014363408088684, + "learning_rate": 1.2947626459049146e-05, + "loss": 3.1533, + "step": 5633 + }, + { + "epoch": 8.24, + "grad_norm": 0.8424006104469299, + "learning_rate": 1.2924553610959821e-05, + "loss": 3.1507, + "step": 5634 + }, + { + "epoch": 8.24, + "grad_norm": 0.45622050762176514, + "learning_rate": 1.290149991888514e-05, + "loss": 3.155, + "step": 5635 + }, + { + "epoch": 8.25, + "grad_norm": 0.5782467126846313, + "learning_rate": 1.2878465387896788e-05, + "loss": 3.1422, + "step": 5636 + }, + { + "epoch": 8.25, + "grad_norm": 0.8892111778259277, + "learning_rate": 1.2855450023062155e-05, + "loss": 3.1708, + "step": 5637 + }, + { + "epoch": 8.25, + "grad_norm": 0.40920886397361755, + "learning_rate": 1.2832453829444534e-05, + "loss": 3.1274, + "step": 5638 + }, + { + "epoch": 8.25, + "grad_norm": 0.720196008682251, + "learning_rate": 1.2809476812102883e-05, + "loss": 3.2203, + "step": 5639 + }, + { + "epoch": 8.25, + "grad_norm": 0.7434779405593872, + "learning_rate": 1.2786518976091977e-05, + "loss": 3.1798, + "step": 5640 + }, + { + "epoch": 8.25, + "grad_norm": 0.5027379393577576, + "learning_rate": 1.2763580326462476e-05, + "loss": 3.1386, + "step": 5641 + }, + { + "epoch": 8.25, + "grad_norm": 0.6186532974243164, + "learning_rate": 1.2740660868260633e-05, + "loss": 3.1423, + "step": 5642 + }, + { + "epoch": 8.26, + "grad_norm": 0.6929596662521362, + "learning_rate": 1.2717760606528605e-05, + "loss": 3.1433, + "step": 5643 + }, + { + "epoch": 8.26, + "grad_norm": 0.3948609530925751, + "learning_rate": 1.2694879546304338e-05, + "loss": 3.1478, + "step": 5644 + }, + { + "epoch": 8.26, + "grad_norm": 0.4978497624397278, + "learning_rate": 1.2672017692621463e-05, + "loss": 3.1263, + "step": 5645 + }, + { + "epoch": 8.26, + "grad_norm": 0.6163767576217651, + "learning_rate": 1.264917505050942e-05, + "loss": 3.1549, + "step": 5646 + }, + { + "epoch": 8.26, + "grad_norm": 0.49492985010147095, + "learning_rate": 1.2626351624993515e-05, + "loss": 3.1269, + "step": 5647 + }, + { + "epoch": 8.26, + "grad_norm": 0.3817135691642761, + "learning_rate": 1.2603547421094675e-05, + "loss": 3.1334, + "step": 5648 + }, + { + "epoch": 8.27, + "grad_norm": 0.5009942650794983, + "learning_rate": 1.2580762443829707e-05, + "loss": 3.175, + "step": 5649 + }, + { + "epoch": 8.27, + "grad_norm": 0.5749830007553101, + "learning_rate": 1.2557996698211139e-05, + "loss": 3.143, + "step": 5650 + }, + { + "epoch": 8.27, + "grad_norm": 0.44454360008239746, + "learning_rate": 1.2535250189247282e-05, + "loss": 3.1268, + "step": 5651 + }, + { + "epoch": 8.27, + "grad_norm": 1.2505532503128052, + "learning_rate": 1.2512522921942226e-05, + "loss": 3.1423, + "step": 5652 + }, + { + "epoch": 8.27, + "grad_norm": 0.49076762795448303, + "learning_rate": 1.2489814901295804e-05, + "loss": 3.1443, + "step": 5653 + }, + { + "epoch": 8.27, + "grad_norm": 0.4465857148170471, + "learning_rate": 1.2467126132303642e-05, + "loss": 3.16, + "step": 5654 + }, + { + "epoch": 8.27, + "grad_norm": 0.5789979696273804, + "learning_rate": 1.2444456619957113e-05, + "loss": 3.124, + "step": 5655 + }, + { + "epoch": 8.28, + "grad_norm": 0.3660390079021454, + "learning_rate": 1.2421806369243361e-05, + "loss": 3.104, + "step": 5656 + }, + { + "epoch": 8.28, + "grad_norm": 0.7501944899559021, + "learning_rate": 1.2399175385145278e-05, + "loss": 3.1535, + "step": 5657 + }, + { + "epoch": 8.28, + "grad_norm": 0.7539718151092529, + "learning_rate": 1.2376563672641562e-05, + "loss": 3.1868, + "step": 5658 + }, + { + "epoch": 8.28, + "grad_norm": 0.3036991059780121, + "learning_rate": 1.2353971236706564e-05, + "loss": 3.1484, + "step": 5659 + }, + { + "epoch": 8.28, + "grad_norm": 0.8954304456710815, + "learning_rate": 1.233139808231053e-05, + "loss": 3.1336, + "step": 5660 + }, + { + "epoch": 8.28, + "grad_norm": 0.4805630147457123, + "learning_rate": 1.2308844214419423e-05, + "loss": 3.1618, + "step": 5661 + }, + { + "epoch": 8.28, + "grad_norm": 0.39005419611930847, + "learning_rate": 1.2286309637994864e-05, + "loss": 3.1422, + "step": 5662 + }, + { + "epoch": 8.29, + "grad_norm": 0.5603565573692322, + "learning_rate": 1.2263794357994373e-05, + "loss": 3.1942, + "step": 5663 + }, + { + "epoch": 8.29, + "grad_norm": 0.558560311794281, + "learning_rate": 1.2241298379371168e-05, + "loss": 3.1461, + "step": 5664 + }, + { + "epoch": 8.29, + "grad_norm": 0.408338725566864, + "learning_rate": 1.2218821707074136e-05, + "loss": 3.1577, + "step": 5665 + }, + { + "epoch": 8.29, + "grad_norm": 0.6621556878089905, + "learning_rate": 1.2196364346048072e-05, + "loss": 3.1458, + "step": 5666 + }, + { + "epoch": 8.29, + "grad_norm": 0.5381448268890381, + "learning_rate": 1.2173926301233429e-05, + "loss": 3.1414, + "step": 5667 + }, + { + "epoch": 8.29, + "grad_norm": 0.33626240491867065, + "learning_rate": 1.215150757756639e-05, + "loss": 3.162, + "step": 5668 + }, + { + "epoch": 8.29, + "grad_norm": 0.31144070625305176, + "learning_rate": 1.2129108179978954e-05, + "loss": 3.1176, + "step": 5669 + }, + { + "epoch": 8.3, + "grad_norm": 0.28460976481437683, + "learning_rate": 1.2106728113398869e-05, + "loss": 3.1216, + "step": 5670 + }, + { + "epoch": 8.3, + "grad_norm": 0.5316192507743835, + "learning_rate": 1.2084367382749517e-05, + "loss": 3.1335, + "step": 5671 + }, + { + "epoch": 8.3, + "grad_norm": 0.4095101058483124, + "learning_rate": 1.2062025992950187e-05, + "loss": 3.1225, + "step": 5672 + }, + { + "epoch": 8.3, + "grad_norm": 0.45470380783081055, + "learning_rate": 1.2039703948915804e-05, + "loss": 3.1579, + "step": 5673 + }, + { + "epoch": 8.3, + "grad_norm": 0.30135294795036316, + "learning_rate": 1.2017401255557037e-05, + "loss": 3.1336, + "step": 5674 + }, + { + "epoch": 8.3, + "grad_norm": 0.7919064164161682, + "learning_rate": 1.1995117917780396e-05, + "loss": 3.1425, + "step": 5675 + }, + { + "epoch": 8.31, + "grad_norm": 0.42938122153282166, + "learning_rate": 1.1972853940488015e-05, + "loss": 3.1453, + "step": 5676 + }, + { + "epoch": 8.31, + "grad_norm": 0.3541744351387024, + "learning_rate": 1.1950609328577823e-05, + "loss": 3.1628, + "step": 5677 + }, + { + "epoch": 8.31, + "grad_norm": 0.2010623663663864, + "learning_rate": 1.1928384086943533e-05, + "loss": 3.1708, + "step": 5678 + }, + { + "epoch": 8.31, + "grad_norm": 0.6417263746261597, + "learning_rate": 1.1906178220474485e-05, + "loss": 3.1305, + "step": 5679 + }, + { + "epoch": 8.31, + "grad_norm": 0.3581140637397766, + "learning_rate": 1.1883991734055844e-05, + "loss": 3.1437, + "step": 5680 + }, + { + "epoch": 8.31, + "grad_norm": 0.40174880623817444, + "learning_rate": 1.186182463256853e-05, + "loss": 3.1109, + "step": 5681 + }, + { + "epoch": 8.31, + "grad_norm": 0.6692104935646057, + "learning_rate": 1.1839676920889098e-05, + "loss": 3.1739, + "step": 5682 + }, + { + "epoch": 8.32, + "grad_norm": 0.2795259952545166, + "learning_rate": 1.1817548603889917e-05, + "loss": 3.1526, + "step": 5683 + }, + { + "epoch": 8.32, + "grad_norm": 0.21500855684280396, + "learning_rate": 1.1795439686439069e-05, + "loss": 3.1436, + "step": 5684 + }, + { + "epoch": 8.32, + "grad_norm": 0.7537773251533508, + "learning_rate": 1.1773350173400377e-05, + "loss": 3.1438, + "step": 5685 + }, + { + "epoch": 8.32, + "grad_norm": 0.520550549030304, + "learning_rate": 1.1751280069633375e-05, + "loss": 3.1659, + "step": 5686 + }, + { + "epoch": 8.32, + "grad_norm": 0.27544260025024414, + "learning_rate": 1.1729229379993345e-05, + "loss": 3.1702, + "step": 5687 + }, + { + "epoch": 8.32, + "grad_norm": 0.36824995279312134, + "learning_rate": 1.1707198109331274e-05, + "loss": 3.128, + "step": 5688 + }, + { + "epoch": 8.32, + "grad_norm": 0.268276184797287, + "learning_rate": 1.168518626249392e-05, + "loss": 3.1329, + "step": 5689 + }, + { + "epoch": 8.33, + "grad_norm": 0.43285050988197327, + "learning_rate": 1.1663193844323727e-05, + "loss": 3.1342, + "step": 5690 + }, + { + "epoch": 8.33, + "grad_norm": 0.28300830721855164, + "learning_rate": 1.1641220859658874e-05, + "loss": 3.1185, + "step": 5691 + }, + { + "epoch": 8.33, + "grad_norm": 0.33784517645835876, + "learning_rate": 1.1619267313333281e-05, + "loss": 3.1615, + "step": 5692 + }, + { + "epoch": 8.33, + "grad_norm": 0.3320251405239105, + "learning_rate": 1.1597333210176586e-05, + "loss": 3.1881, + "step": 5693 + }, + { + "epoch": 8.33, + "grad_norm": 0.3278450667858124, + "learning_rate": 1.1575418555014127e-05, + "loss": 3.1494, + "step": 5694 + }, + { + "epoch": 8.33, + "grad_norm": 0.30050769448280334, + "learning_rate": 1.1553523352667018e-05, + "loss": 3.1541, + "step": 5695 + }, + { + "epoch": 8.33, + "grad_norm": 0.26941055059432983, + "learning_rate": 1.1531647607951979e-05, + "loss": 3.1437, + "step": 5696 + }, + { + "epoch": 8.34, + "grad_norm": 0.5336388349533081, + "learning_rate": 1.1509791325681607e-05, + "loss": 3.1115, + "step": 5697 + }, + { + "epoch": 8.34, + "grad_norm": 0.30893999338150024, + "learning_rate": 1.1487954510664122e-05, + "loss": 3.1613, + "step": 5698 + }, + { + "epoch": 8.34, + "grad_norm": 0.48479652404785156, + "learning_rate": 1.1466137167703428e-05, + "loss": 3.183, + "step": 5699 + }, + { + "epoch": 8.34, + "grad_norm": 0.33634239435195923, + "learning_rate": 1.1444339301599239e-05, + "loss": 3.1524, + "step": 5700 + }, + { + "epoch": 8.34, + "grad_norm": 0.45775729417800903, + "learning_rate": 1.1422560917146962e-05, + "loss": 3.1708, + "step": 5701 + }, + { + "epoch": 8.34, + "grad_norm": 0.2623753845691681, + "learning_rate": 1.1400802019137613e-05, + "loss": 3.1317, + "step": 5702 + }, + { + "epoch": 8.35, + "grad_norm": 0.4868522584438324, + "learning_rate": 1.1379062612358082e-05, + "loss": 3.1486, + "step": 5703 + }, + { + "epoch": 8.35, + "grad_norm": 0.32729241251945496, + "learning_rate": 1.1357342701590889e-05, + "loss": 3.1077, + "step": 5704 + }, + { + "epoch": 8.35, + "grad_norm": 0.4166317582130432, + "learning_rate": 1.1335642291614201e-05, + "loss": 3.1305, + "step": 5705 + }, + { + "epoch": 8.35, + "grad_norm": 0.3061630129814148, + "learning_rate": 1.1313961387202044e-05, + "loss": 3.1469, + "step": 5706 + }, + { + "epoch": 8.35, + "grad_norm": 0.32585567235946655, + "learning_rate": 1.1292299993124012e-05, + "loss": 3.1448, + "step": 5707 + }, + { + "epoch": 8.35, + "grad_norm": 0.5888083577156067, + "learning_rate": 1.1270658114145493e-05, + "loss": 3.1815, + "step": 5708 + }, + { + "epoch": 8.35, + "grad_norm": 0.28794151544570923, + "learning_rate": 1.1249035755027559e-05, + "loss": 3.1343, + "step": 5709 + }, + { + "epoch": 8.36, + "grad_norm": 0.6074895262718201, + "learning_rate": 1.122743292052697e-05, + "loss": 3.1481, + "step": 5710 + }, + { + "epoch": 8.36, + "grad_norm": 0.6733052730560303, + "learning_rate": 1.1205849615396202e-05, + "loss": 3.1235, + "step": 5711 + }, + { + "epoch": 8.36, + "grad_norm": 0.2199082374572754, + "learning_rate": 1.1184285844383458e-05, + "loss": 3.1172, + "step": 5712 + }, + { + "epoch": 8.36, + "grad_norm": 0.32932931184768677, + "learning_rate": 1.1162741612232608e-05, + "loss": 3.1226, + "step": 5713 + }, + { + "epoch": 8.36, + "grad_norm": 0.7292448878288269, + "learning_rate": 1.114121692368324e-05, + "loss": 3.1721, + "step": 5714 + }, + { + "epoch": 8.36, + "grad_norm": 0.34164637327194214, + "learning_rate": 1.1119711783470643e-05, + "loss": 3.1505, + "step": 5715 + }, + { + "epoch": 8.36, + "grad_norm": 0.2828170657157898, + "learning_rate": 1.1098226196325811e-05, + "loss": 3.1372, + "step": 5716 + }, + { + "epoch": 8.37, + "grad_norm": 0.3506002724170685, + "learning_rate": 1.107676016697542e-05, + "loss": 3.1324, + "step": 5717 + }, + { + "epoch": 8.37, + "grad_norm": 0.5819188952445984, + "learning_rate": 1.1055313700141857e-05, + "loss": 3.1621, + "step": 5718 + }, + { + "epoch": 8.37, + "grad_norm": 0.2485204041004181, + "learning_rate": 1.1033886800543191e-05, + "loss": 3.17, + "step": 5719 + }, + { + "epoch": 8.37, + "grad_norm": 0.3673901855945587, + "learning_rate": 1.1012479472893212e-05, + "loss": 3.1383, + "step": 5720 + }, + { + "epoch": 8.37, + "grad_norm": 0.3317386507987976, + "learning_rate": 1.099109172190138e-05, + "loss": 3.1596, + "step": 5721 + }, + { + "epoch": 8.37, + "grad_norm": 0.5169317722320557, + "learning_rate": 1.096972355227286e-05, + "loss": 3.1451, + "step": 5722 + }, + { + "epoch": 8.37, + "grad_norm": 0.4397293031215668, + "learning_rate": 1.0948374968708497e-05, + "loss": 3.1589, + "step": 5723 + }, + { + "epoch": 8.38, + "grad_norm": 0.23546314239501953, + "learning_rate": 1.0927045975904848e-05, + "loss": 3.1252, + "step": 5724 + }, + { + "epoch": 8.38, + "grad_norm": 0.2676627039909363, + "learning_rate": 1.0905736578554138e-05, + "loss": 3.1651, + "step": 5725 + }, + { + "epoch": 8.38, + "grad_norm": 0.9150519371032715, + "learning_rate": 1.0884446781344303e-05, + "loss": 3.1197, + "step": 5726 + }, + { + "epoch": 8.38, + "grad_norm": 0.5215425491333008, + "learning_rate": 1.0863176588958956e-05, + "loss": 3.1501, + "step": 5727 + }, + { + "epoch": 8.38, + "grad_norm": 0.48437726497650146, + "learning_rate": 1.0841926006077385e-05, + "loss": 3.1656, + "step": 5728 + }, + { + "epoch": 8.38, + "grad_norm": 0.7072523832321167, + "learning_rate": 1.0820695037374595e-05, + "loss": 3.1917, + "step": 5729 + }, + { + "epoch": 8.39, + "grad_norm": 0.4144083857536316, + "learning_rate": 1.0799483687521216e-05, + "loss": 3.1268, + "step": 5730 + }, + { + "epoch": 8.39, + "grad_norm": 0.2927367389202118, + "learning_rate": 1.0778291961183639e-05, + "loss": 3.1329, + "step": 5731 + }, + { + "epoch": 8.39, + "grad_norm": 0.3951784372329712, + "learning_rate": 1.0757119863023924e-05, + "loss": 3.1293, + "step": 5732 + }, + { + "epoch": 8.39, + "grad_norm": 0.6036468744277954, + "learning_rate": 1.0735967397699731e-05, + "loss": 3.1348, + "step": 5733 + }, + { + "epoch": 8.39, + "grad_norm": 0.4498322606086731, + "learning_rate": 1.0714834569864474e-05, + "loss": 3.1327, + "step": 5734 + }, + { + "epoch": 8.39, + "grad_norm": 0.4379383325576782, + "learning_rate": 1.0693721384167288e-05, + "loss": 3.1598, + "step": 5735 + }, + { + "epoch": 8.39, + "grad_norm": 0.3004567325115204, + "learning_rate": 1.0672627845252881e-05, + "loss": 3.1152, + "step": 5736 + }, + { + "epoch": 8.4, + "grad_norm": 0.7020764946937561, + "learning_rate": 1.0651553957761673e-05, + "loss": 3.1358, + "step": 5737 + }, + { + "epoch": 8.4, + "grad_norm": 0.3879508078098297, + "learning_rate": 1.0630499726329846e-05, + "loss": 3.148, + "step": 5738 + }, + { + "epoch": 8.4, + "grad_norm": 0.34494709968566895, + "learning_rate": 1.0609465155589137e-05, + "loss": 3.1244, + "step": 5739 + }, + { + "epoch": 8.4, + "grad_norm": 0.45144951343536377, + "learning_rate": 1.0588450250167026e-05, + "loss": 3.139, + "step": 5740 + }, + { + "epoch": 8.4, + "grad_norm": 0.4698631167411804, + "learning_rate": 1.0567455014686645e-05, + "loss": 3.1501, + "step": 5741 + }, + { + "epoch": 8.4, + "grad_norm": 0.3781033158302307, + "learning_rate": 1.0546479453766811e-05, + "loss": 3.1047, + "step": 5742 + }, + { + "epoch": 8.4, + "grad_norm": 0.9458195567131042, + "learning_rate": 1.052552357202201e-05, + "loss": 3.1385, + "step": 5743 + }, + { + "epoch": 8.41, + "grad_norm": 0.1932523101568222, + "learning_rate": 1.0504587374062391e-05, + "loss": 3.1302, + "step": 5744 + }, + { + "epoch": 8.41, + "grad_norm": 0.541278064250946, + "learning_rate": 1.0483670864493778e-05, + "loss": 3.1655, + "step": 5745 + }, + { + "epoch": 8.41, + "grad_norm": 1.0147125720977783, + "learning_rate": 1.0462774047917656e-05, + "loss": 3.1368, + "step": 5746 + }, + { + "epoch": 8.41, + "eval_loss": 2.9998810291290283, + "eval_runtime": 1167.1096, + "eval_samples_per_second": 14.688, + "eval_steps_per_second": 0.919, + "step": 5746 + }, + { + "epoch": 8.41, + "grad_norm": 0.746507465839386, + "learning_rate": 1.0441896928931206e-05, + "loss": 3.1443, + "step": 5747 + }, + { + "epoch": 8.41, + "grad_norm": 0.41886982321739197, + "learning_rate": 1.0421039512127229e-05, + "loss": 3.1343, + "step": 5748 + }, + { + "epoch": 8.41, + "grad_norm": 1.3022066354751587, + "learning_rate": 1.0400201802094223e-05, + "loss": 3.1096, + "step": 5749 + }, + { + "epoch": 8.42, + "grad_norm": 0.7943986058235168, + "learning_rate": 1.0379383803416364e-05, + "loss": 3.1392, + "step": 5750 + }, + { + "epoch": 8.42, + "grad_norm": 0.40819865465164185, + "learning_rate": 1.0358585520673446e-05, + "loss": 3.1234, + "step": 5751 + }, + { + "epoch": 8.42, + "grad_norm": 0.6574892401695251, + "learning_rate": 1.0337806958440966e-05, + "loss": 3.1238, + "step": 5752 + }, + { + "epoch": 8.42, + "grad_norm": 0.598135232925415, + "learning_rate": 1.0317048121290074e-05, + "loss": 3.1246, + "step": 5753 + }, + { + "epoch": 8.42, + "grad_norm": 0.543494701385498, + "learning_rate": 1.0296309013787553e-05, + "loss": 3.125, + "step": 5754 + }, + { + "epoch": 8.42, + "grad_norm": 0.754518985748291, + "learning_rate": 1.0275589640495908e-05, + "loss": 3.1477, + "step": 5755 + }, + { + "epoch": 8.42, + "grad_norm": 0.27329519391059875, + "learning_rate": 1.0254890005973205e-05, + "loss": 3.1207, + "step": 5756 + }, + { + "epoch": 8.43, + "grad_norm": 0.7898842096328735, + "learning_rate": 1.0234210114773257e-05, + "loss": 3.1593, + "step": 5757 + }, + { + "epoch": 8.43, + "grad_norm": 0.5713753700256348, + "learning_rate": 1.0213549971445535e-05, + "loss": 3.1235, + "step": 5758 + }, + { + "epoch": 8.43, + "grad_norm": 0.3577013909816742, + "learning_rate": 1.0192909580535049e-05, + "loss": 3.1596, + "step": 5759 + }, + { + "epoch": 8.43, + "grad_norm": 0.6142879128456116, + "learning_rate": 1.017228894658262e-05, + "loss": 3.148, + "step": 5760 + }, + { + "epoch": 8.43, + "grad_norm": 0.71822589635849, + "learning_rate": 1.0151688074124643e-05, + "loss": 3.1361, + "step": 5761 + }, + { + "epoch": 8.43, + "grad_norm": 0.7075265049934387, + "learning_rate": 1.0131106967693116e-05, + "loss": 3.1091, + "step": 5762 + }, + { + "epoch": 8.43, + "grad_norm": 0.4524644911289215, + "learning_rate": 1.0110545631815815e-05, + "loss": 3.1273, + "step": 5763 + }, + { + "epoch": 8.44, + "grad_norm": 0.451507568359375, + "learning_rate": 1.0090004071016046e-05, + "loss": 3.141, + "step": 5764 + }, + { + "epoch": 8.44, + "grad_norm": 0.5503039956092834, + "learning_rate": 1.0069482289812816e-05, + "loss": 3.1444, + "step": 5765 + }, + { + "epoch": 8.44, + "grad_norm": 0.2659579813480377, + "learning_rate": 1.0048980292720822e-05, + "loss": 3.1474, + "step": 5766 + }, + { + "epoch": 8.44, + "grad_norm": 0.25024622678756714, + "learning_rate": 1.0028498084250325e-05, + "loss": 3.1663, + "step": 5767 + }, + { + "epoch": 8.44, + "grad_norm": 0.6892327070236206, + "learning_rate": 1.0008035668907279e-05, + "loss": 3.1352, + "step": 5768 + }, + { + "epoch": 8.44, + "grad_norm": 0.32359933853149414, + "learning_rate": 9.987593051193311e-06, + "loss": 3.1438, + "step": 5769 + }, + { + "epoch": 8.44, + "grad_norm": 0.36346298456192017, + "learning_rate": 9.967170235605616e-06, + "loss": 3.1841, + "step": 5770 + }, + { + "epoch": 8.45, + "grad_norm": 0.42581936717033386, + "learning_rate": 9.946767226637088e-06, + "loss": 3.1161, + "step": 5771 + }, + { + "epoch": 8.45, + "grad_norm": 0.5385969281196594, + "learning_rate": 9.926384028776293e-06, + "loss": 3.1526, + "step": 5772 + }, + { + "epoch": 8.45, + "grad_norm": 0.4430517256259918, + "learning_rate": 9.90602064650734e-06, + "loss": 3.179, + "step": 5773 + }, + { + "epoch": 8.45, + "grad_norm": 0.7233391404151917, + "learning_rate": 9.885677084310063e-06, + "loss": 3.1161, + "step": 5774 + }, + { + "epoch": 8.45, + "grad_norm": 0.36595597863197327, + "learning_rate": 9.865353346659911e-06, + "loss": 3.1277, + "step": 5775 + }, + { + "epoch": 8.45, + "grad_norm": 0.7899670600891113, + "learning_rate": 9.845049438027976e-06, + "loss": 3.159, + "step": 5776 + }, + { + "epoch": 8.46, + "grad_norm": 0.3608589172363281, + "learning_rate": 9.824765362880962e-06, + "loss": 3.1284, + "step": 5777 + }, + { + "epoch": 8.46, + "grad_norm": 0.4470582902431488, + "learning_rate": 9.804501125681243e-06, + "loss": 3.1371, + "step": 5778 + }, + { + "epoch": 8.46, + "grad_norm": 0.45787444710731506, + "learning_rate": 9.784256730886809e-06, + "loss": 3.1443, + "step": 5779 + }, + { + "epoch": 8.46, + "grad_norm": 0.28761786222457886, + "learning_rate": 9.7640321829513e-06, + "loss": 3.1522, + "step": 5780 + }, + { + "epoch": 8.46, + "grad_norm": 0.8798906207084656, + "learning_rate": 9.743827486323975e-06, + "loss": 3.1395, + "step": 5781 + }, + { + "epoch": 8.46, + "grad_norm": 0.5044177174568176, + "learning_rate": 9.723642645449737e-06, + "loss": 3.1485, + "step": 5782 + }, + { + "epoch": 8.46, + "grad_norm": 0.43212029337882996, + "learning_rate": 9.703477664769123e-06, + "loss": 3.1306, + "step": 5783 + }, + { + "epoch": 8.47, + "grad_norm": 0.38974976539611816, + "learning_rate": 9.683332548718248e-06, + "loss": 3.1621, + "step": 5784 + }, + { + "epoch": 8.47, + "grad_norm": 0.425968736410141, + "learning_rate": 9.66320730172896e-06, + "loss": 3.1229, + "step": 5785 + }, + { + "epoch": 8.47, + "grad_norm": 2.754948377609253, + "learning_rate": 9.64310192822867e-06, + "loss": 3.1593, + "step": 5786 + }, + { + "epoch": 8.47, + "grad_norm": 0.7210294008255005, + "learning_rate": 9.623016432640375e-06, + "loss": 3.1309, + "step": 5787 + }, + { + "epoch": 8.47, + "grad_norm": 0.17482349276542664, + "learning_rate": 9.602950819382806e-06, + "loss": 3.1226, + "step": 5788 + }, + { + "epoch": 8.47, + "grad_norm": 0.35730379819869995, + "learning_rate": 9.582905092870254e-06, + "loss": 3.155, + "step": 5789 + }, + { + "epoch": 8.47, + "grad_norm": 0.3369613587856293, + "learning_rate": 9.562879257512602e-06, + "loss": 3.135, + "step": 5790 + }, + { + "epoch": 8.48, + "grad_norm": 0.9082013368606567, + "learning_rate": 9.542873317715439e-06, + "loss": 3.1319, + "step": 5791 + }, + { + "epoch": 8.48, + "grad_norm": 0.5040355920791626, + "learning_rate": 9.522887277879956e-06, + "loss": 3.1234, + "step": 5792 + }, + { + "epoch": 8.48, + "grad_norm": 0.47945570945739746, + "learning_rate": 9.502921142402887e-06, + "loss": 3.1499, + "step": 5793 + }, + { + "epoch": 8.48, + "grad_norm": 0.9931325912475586, + "learning_rate": 9.482974915676702e-06, + "loss": 3.158, + "step": 5794 + }, + { + "epoch": 8.48, + "grad_norm": 0.5050073266029358, + "learning_rate": 9.46304860208943e-06, + "loss": 3.1161, + "step": 5795 + }, + { + "epoch": 8.48, + "grad_norm": 0.21985039114952087, + "learning_rate": 9.443142206024668e-06, + "loss": 3.1283, + "step": 5796 + }, + { + "epoch": 8.48, + "grad_norm": 0.38089579343795776, + "learning_rate": 9.423255731861791e-06, + "loss": 3.1516, + "step": 5797 + }, + { + "epoch": 8.49, + "grad_norm": 0.8046565651893616, + "learning_rate": 9.403389183975609e-06, + "loss": 3.1305, + "step": 5798 + }, + { + "epoch": 8.49, + "grad_norm": 0.7955847382545471, + "learning_rate": 9.383542566736635e-06, + "loss": 3.1691, + "step": 5799 + }, + { + "epoch": 8.49, + "grad_norm": 0.2818087339401245, + "learning_rate": 9.363715884511049e-06, + "loss": 3.1686, + "step": 5800 + }, + { + "epoch": 8.49, + "grad_norm": 0.37908995151519775, + "learning_rate": 9.343909141660535e-06, + "loss": 3.1416, + "step": 5801 + }, + { + "epoch": 8.49, + "grad_norm": 0.9841012358665466, + "learning_rate": 9.324122342542429e-06, + "loss": 3.1679, + "step": 5802 + }, + { + "epoch": 8.49, + "grad_norm": 0.7049186825752258, + "learning_rate": 9.304355491509776e-06, + "loss": 3.1367, + "step": 5803 + }, + { + "epoch": 8.5, + "grad_norm": 0.4633147120475769, + "learning_rate": 9.284608592911082e-06, + "loss": 3.1225, + "step": 5804 + }, + { + "epoch": 8.5, + "grad_norm": 0.8353789448738098, + "learning_rate": 9.26488165109054e-06, + "loss": 3.1142, + "step": 5805 + }, + { + "epoch": 8.5, + "grad_norm": 0.740578830242157, + "learning_rate": 9.245174670387958e-06, + "loss": 3.1183, + "step": 5806 + }, + { + "epoch": 8.5, + "grad_norm": 0.40835264325141907, + "learning_rate": 9.225487655138742e-06, + "loss": 3.1315, + "step": 5807 + }, + { + "epoch": 8.5, + "grad_norm": 0.6107511520385742, + "learning_rate": 9.205820609673887e-06, + "loss": 3.1433, + "step": 5808 + }, + { + "epoch": 8.5, + "grad_norm": 0.8635002970695496, + "learning_rate": 9.18617353832003e-06, + "loss": 3.1264, + "step": 5809 + }, + { + "epoch": 8.5, + "grad_norm": 0.37021952867507935, + "learning_rate": 9.166546445399394e-06, + "loss": 3.1448, + "step": 5810 + }, + { + "epoch": 8.51, + "grad_norm": 0.7059236168861389, + "learning_rate": 9.146939335229788e-06, + "loss": 3.1168, + "step": 5811 + }, + { + "epoch": 8.51, + "grad_norm": 0.4614633619785309, + "learning_rate": 9.127352212124662e-06, + "loss": 3.1413, + "step": 5812 + }, + { + "epoch": 8.51, + "grad_norm": 0.4562102258205414, + "learning_rate": 9.107785080393061e-06, + "loss": 3.1227, + "step": 5813 + }, + { + "epoch": 8.51, + "grad_norm": 0.5066782832145691, + "learning_rate": 9.088237944339594e-06, + "loss": 3.1707, + "step": 5814 + }, + { + "epoch": 8.51, + "grad_norm": 0.47414615750312805, + "learning_rate": 9.068710808264524e-06, + "loss": 3.1667, + "step": 5815 + }, + { + "epoch": 8.51, + "grad_norm": 0.6087870001792908, + "learning_rate": 9.04920367646368e-06, + "loss": 3.1496, + "step": 5816 + }, + { + "epoch": 8.51, + "grad_norm": 0.5062857270240784, + "learning_rate": 9.029716553228517e-06, + "loss": 3.1633, + "step": 5817 + }, + { + "epoch": 8.52, + "grad_norm": 0.28521206974983215, + "learning_rate": 9.01024944284602e-06, + "loss": 3.1468, + "step": 5818 + }, + { + "epoch": 8.52, + "grad_norm": 0.46825653314590454, + "learning_rate": 8.99080234959888e-06, + "loss": 3.1665, + "step": 5819 + }, + { + "epoch": 8.52, + "grad_norm": 0.682184636592865, + "learning_rate": 8.971375277765326e-06, + "loss": 3.1446, + "step": 5820 + }, + { + "epoch": 8.52, + "grad_norm": 0.2446981519460678, + "learning_rate": 8.951968231619124e-06, + "loss": 3.189, + "step": 5821 + }, + { + "epoch": 8.52, + "grad_norm": 0.23873744904994965, + "learning_rate": 8.932581215429748e-06, + "loss": 3.1293, + "step": 5822 + }, + { + "epoch": 8.52, + "grad_norm": 0.33204108476638794, + "learning_rate": 8.913214233462231e-06, + "loss": 3.13, + "step": 5823 + }, + { + "epoch": 8.52, + "grad_norm": 0.32620128989219666, + "learning_rate": 8.893867289977097e-06, + "loss": 3.1604, + "step": 5824 + }, + { + "epoch": 8.53, + "grad_norm": 0.6817585825920105, + "learning_rate": 8.874540389230612e-06, + "loss": 3.1272, + "step": 5825 + }, + { + "epoch": 8.53, + "grad_norm": 0.2696726322174072, + "learning_rate": 8.855233535474571e-06, + "loss": 3.139, + "step": 5826 + }, + { + "epoch": 8.53, + "grad_norm": 0.46153542399406433, + "learning_rate": 8.835946732956302e-06, + "loss": 3.136, + "step": 5827 + }, + { + "epoch": 8.53, + "grad_norm": 0.5768734216690063, + "learning_rate": 8.816679985918785e-06, + "loss": 3.1362, + "step": 5828 + }, + { + "epoch": 8.53, + "grad_norm": 0.5621840953826904, + "learning_rate": 8.797433298600622e-06, + "loss": 3.1742, + "step": 5829 + }, + { + "epoch": 8.53, + "grad_norm": 1.3828614950180054, + "learning_rate": 8.7782066752359e-06, + "loss": 3.1435, + "step": 5830 + }, + { + "epoch": 8.54, + "grad_norm": 0.3548658788204193, + "learning_rate": 8.75900012005436e-06, + "loss": 3.2216, + "step": 5831 + }, + { + "epoch": 8.54, + "grad_norm": 0.25544172525405884, + "learning_rate": 8.739813637281336e-06, + "loss": 3.1576, + "step": 5832 + }, + { + "epoch": 8.54, + "grad_norm": 0.6139233112335205, + "learning_rate": 8.720647231137713e-06, + "loss": 3.162, + "step": 5833 + }, + { + "epoch": 8.54, + "grad_norm": 0.3499756157398224, + "learning_rate": 8.701500905839966e-06, + "loss": 3.1198, + "step": 5834 + }, + { + "epoch": 8.54, + "grad_norm": 0.5180540084838867, + "learning_rate": 8.682374665600168e-06, + "loss": 3.1238, + "step": 5835 + }, + { + "epoch": 8.54, + "grad_norm": 0.29475992918014526, + "learning_rate": 8.663268514625966e-06, + "loss": 3.1271, + "step": 5836 + }, + { + "epoch": 8.54, + "grad_norm": 0.6238717436790466, + "learning_rate": 8.644182457120575e-06, + "loss": 3.1083, + "step": 5837 + }, + { + "epoch": 8.55, + "grad_norm": 0.24445649981498718, + "learning_rate": 8.625116497282803e-06, + "loss": 3.1107, + "step": 5838 + }, + { + "epoch": 8.55, + "grad_norm": 0.6731110215187073, + "learning_rate": 8.60607063930703e-06, + "loss": 3.1931, + "step": 5839 + }, + { + "epoch": 8.55, + "grad_norm": 0.3173396587371826, + "learning_rate": 8.587044887383221e-06, + "loss": 3.1099, + "step": 5840 + }, + { + "epoch": 8.55, + "grad_norm": 0.27860239148139954, + "learning_rate": 8.568039245696924e-06, + "loss": 3.143, + "step": 5841 + }, + { + "epoch": 8.55, + "grad_norm": 0.25917676091194153, + "learning_rate": 8.549053718429235e-06, + "loss": 3.1283, + "step": 5842 + }, + { + "epoch": 8.55, + "grad_norm": 0.3333747386932373, + "learning_rate": 8.530088309756856e-06, + "loss": 3.1291, + "step": 5843 + }, + { + "epoch": 8.55, + "grad_norm": 0.22568973898887634, + "learning_rate": 8.511143023852041e-06, + "loss": 3.1724, + "step": 5844 + }, + { + "epoch": 8.56, + "grad_norm": 0.3453158140182495, + "learning_rate": 8.492217864882623e-06, + "loss": 3.1325, + "step": 5845 + }, + { + "epoch": 8.56, + "grad_norm": 0.4224380850791931, + "learning_rate": 8.473312837012026e-06, + "loss": 3.1217, + "step": 5846 + }, + { + "epoch": 8.56, + "grad_norm": 0.3284725844860077, + "learning_rate": 8.454427944399224e-06, + "loss": 3.1243, + "step": 5847 + }, + { + "epoch": 8.56, + "grad_norm": 0.47590208053588867, + "learning_rate": 8.435563191198758e-06, + "loss": 3.1304, + "step": 5848 + }, + { + "epoch": 8.56, + "grad_norm": 0.3408375680446625, + "learning_rate": 8.416718581560746e-06, + "loss": 3.1498, + "step": 5849 + }, + { + "epoch": 8.56, + "grad_norm": 0.28439176082611084, + "learning_rate": 8.39789411963089e-06, + "loss": 3.1231, + "step": 5850 + }, + { + "epoch": 8.57, + "grad_norm": 0.36393728852272034, + "learning_rate": 8.379089809550456e-06, + "loss": 3.1541, + "step": 5851 + }, + { + "epoch": 8.57, + "grad_norm": 0.7330798506736755, + "learning_rate": 8.360305655456225e-06, + "loss": 3.112, + "step": 5852 + }, + { + "epoch": 8.57, + "grad_norm": 1.5145879983901978, + "learning_rate": 8.34154166148059e-06, + "loss": 3.1475, + "step": 5853 + }, + { + "epoch": 8.57, + "grad_norm": 0.6329509615898132, + "learning_rate": 8.322797831751561e-06, + "loss": 3.1115, + "step": 5854 + }, + { + "epoch": 8.57, + "grad_norm": 0.5651994347572327, + "learning_rate": 8.304074170392595e-06, + "loss": 3.1165, + "step": 5855 + }, + { + "epoch": 8.57, + "grad_norm": 0.5233526825904846, + "learning_rate": 8.285370681522775e-06, + "loss": 3.1514, + "step": 5856 + }, + { + "epoch": 8.57, + "grad_norm": 0.27295824885368347, + "learning_rate": 8.266687369256798e-06, + "loss": 3.0979, + "step": 5857 + }, + { + "epoch": 8.58, + "grad_norm": 0.3821253478527069, + "learning_rate": 8.248024237704821e-06, + "loss": 3.1523, + "step": 5858 + }, + { + "epoch": 8.58, + "grad_norm": 0.3980197310447693, + "learning_rate": 8.229381290972583e-06, + "loss": 3.1384, + "step": 5859 + }, + { + "epoch": 8.58, + "grad_norm": 0.3785715103149414, + "learning_rate": 8.210758533161488e-06, + "loss": 3.1334, + "step": 5860 + }, + { + "epoch": 8.58, + "grad_norm": 0.43636593222618103, + "learning_rate": 8.192155968368353e-06, + "loss": 3.174, + "step": 5861 + }, + { + "epoch": 8.58, + "grad_norm": 0.35782870650291443, + "learning_rate": 8.173573600685614e-06, + "loss": 3.1439, + "step": 5862 + }, + { + "epoch": 8.58, + "grad_norm": 0.28232669830322266, + "learning_rate": 8.155011434201333e-06, + "loss": 3.1474, + "step": 5863 + }, + { + "epoch": 8.58, + "grad_norm": 0.4299192428588867, + "learning_rate": 8.136469472998986e-06, + "loss": 3.151, + "step": 5864 + }, + { + "epoch": 8.59, + "grad_norm": 0.24973128736019135, + "learning_rate": 8.117947721157726e-06, + "loss": 3.1622, + "step": 5865 + }, + { + "epoch": 8.59, + "grad_norm": 0.3395419716835022, + "learning_rate": 8.099446182752202e-06, + "loss": 3.1173, + "step": 5866 + }, + { + "epoch": 8.59, + "grad_norm": 0.328425794839859, + "learning_rate": 8.080964861852625e-06, + "loss": 3.177, + "step": 5867 + }, + { + "epoch": 8.59, + "grad_norm": 0.509938657283783, + "learning_rate": 8.062503762524754e-06, + "loss": 3.174, + "step": 5868 + }, + { + "epoch": 8.59, + "grad_norm": 0.16413269937038422, + "learning_rate": 8.044062888829928e-06, + "loss": 3.1419, + "step": 5869 + }, + { + "epoch": 8.59, + "grad_norm": 0.6065314412117004, + "learning_rate": 8.025642244825005e-06, + "loss": 3.1232, + "step": 5870 + }, + { + "epoch": 8.59, + "grad_norm": 0.5938077569007874, + "learning_rate": 8.007241834562395e-06, + "loss": 3.1267, + "step": 5871 + }, + { + "epoch": 8.6, + "grad_norm": 0.2933906316757202, + "learning_rate": 7.988861662090085e-06, + "loss": 3.1554, + "step": 5872 + }, + { + "epoch": 8.6, + "grad_norm": 0.36898481845855713, + "learning_rate": 7.970501731451573e-06, + "loss": 3.1333, + "step": 5873 + }, + { + "epoch": 8.6, + "grad_norm": 0.5361416935920715, + "learning_rate": 7.95216204668593e-06, + "loss": 3.1383, + "step": 5874 + }, + { + "epoch": 8.6, + "grad_norm": 0.8210395574569702, + "learning_rate": 7.933842611827746e-06, + "loss": 3.1449, + "step": 5875 + }, + { + "epoch": 8.6, + "grad_norm": 0.7221469879150391, + "learning_rate": 7.915543430907202e-06, + "loss": 3.15, + "step": 5876 + }, + { + "epoch": 8.6, + "grad_norm": 0.25091636180877686, + "learning_rate": 7.897264507949986e-06, + "loss": 3.1368, + "step": 5877 + }, + { + "epoch": 8.61, + "grad_norm": 0.9012389183044434, + "learning_rate": 7.879005846977305e-06, + "loss": 3.1382, + "step": 5878 + }, + { + "epoch": 8.61, + "grad_norm": 0.8301836848258972, + "learning_rate": 7.86076745200599e-06, + "loss": 3.1148, + "step": 5879 + }, + { + "epoch": 8.61, + "grad_norm": 0.5587486624717712, + "learning_rate": 7.842549327048365e-06, + "loss": 3.1546, + "step": 5880 + }, + { + "epoch": 8.61, + "grad_norm": 0.4740186333656311, + "learning_rate": 7.824351476112234e-06, + "loss": 3.1153, + "step": 5881 + }, + { + "epoch": 8.61, + "grad_norm": 0.7473950982093811, + "learning_rate": 7.806173903201063e-06, + "loss": 3.1274, + "step": 5882 + }, + { + "epoch": 8.61, + "grad_norm": 1.0179914236068726, + "learning_rate": 7.788016612313798e-06, + "loss": 3.1127, + "step": 5883 + }, + { + "epoch": 8.61, + "grad_norm": 0.5002656579017639, + "learning_rate": 7.769879607444863e-06, + "loss": 3.1289, + "step": 5884 + }, + { + "epoch": 8.62, + "grad_norm": 0.2755470871925354, + "learning_rate": 7.751762892584358e-06, + "loss": 3.1618, + "step": 5885 + }, + { + "epoch": 8.62, + "grad_norm": 0.5702336430549622, + "learning_rate": 7.733666471717772e-06, + "loss": 3.1114, + "step": 5886 + }, + { + "epoch": 8.62, + "grad_norm": 1.010840892791748, + "learning_rate": 7.715590348826196e-06, + "loss": 3.1286, + "step": 5887 + }, + { + "epoch": 8.62, + "grad_norm": 0.9508697986602783, + "learning_rate": 7.69753452788633e-06, + "loss": 3.0807, + "step": 5888 + }, + { + "epoch": 8.62, + "grad_norm": 0.23731788992881775, + "learning_rate": 7.679499012870261e-06, + "loss": 3.1327, + "step": 5889 + }, + { + "epoch": 8.62, + "grad_norm": 0.806941032409668, + "learning_rate": 7.661483807745683e-06, + "loss": 3.1694, + "step": 5890 + }, + { + "epoch": 8.62, + "grad_norm": 0.7129907011985779, + "learning_rate": 7.643488916475882e-06, + "loss": 3.1413, + "step": 5891 + }, + { + "epoch": 8.63, + "grad_norm": 0.8230921030044556, + "learning_rate": 7.625514343019557e-06, + "loss": 3.1352, + "step": 5892 + }, + { + "epoch": 8.63, + "grad_norm": 0.6420795917510986, + "learning_rate": 7.60756009133099e-06, + "loss": 3.1604, + "step": 5893 + }, + { + "epoch": 8.63, + "grad_norm": 0.5177065134048462, + "learning_rate": 7.5896261653600535e-06, + "loss": 3.1341, + "step": 5894 + }, + { + "epoch": 8.63, + "grad_norm": 1.5158321857452393, + "learning_rate": 7.571712569052037e-06, + "loss": 3.1549, + "step": 5895 + }, + { + "epoch": 8.63, + "grad_norm": 0.8775774836540222, + "learning_rate": 7.5538193063478095e-06, + "loss": 3.1505, + "step": 5896 + }, + { + "epoch": 8.63, + "grad_norm": 0.23683175444602966, + "learning_rate": 7.535946381183812e-06, + "loss": 3.1582, + "step": 5897 + }, + { + "epoch": 8.63, + "grad_norm": 0.3066456615924835, + "learning_rate": 7.518093797491943e-06, + "loss": 3.1531, + "step": 5898 + }, + { + "epoch": 8.64, + "grad_norm": 0.45848727226257324, + "learning_rate": 7.50026155919964e-06, + "loss": 3.1185, + "step": 5899 + }, + { + "epoch": 8.64, + "grad_norm": 0.4867948591709137, + "learning_rate": 7.482449670229896e-06, + "loss": 3.1427, + "step": 5900 + }, + { + "epoch": 8.64, + "grad_norm": 0.34094056487083435, + "learning_rate": 7.464658134501201e-06, + "loss": 3.1407, + "step": 5901 + }, + { + "epoch": 8.64, + "grad_norm": 0.3147621750831604, + "learning_rate": 7.446886955927568e-06, + "loss": 3.1378, + "step": 5902 + }, + { + "epoch": 8.64, + "grad_norm": 0.40778762102127075, + "learning_rate": 7.429136138418535e-06, + "loss": 3.1767, + "step": 5903 + }, + { + "epoch": 8.64, + "grad_norm": 0.40993595123291016, + "learning_rate": 7.4114056858791805e-06, + "loss": 3.1326, + "step": 5904 + }, + { + "epoch": 8.65, + "grad_norm": 0.25892430543899536, + "learning_rate": 7.393695602210071e-06, + "loss": 3.1982, + "step": 5905 + }, + { + "epoch": 8.65, + "grad_norm": 0.4945090711116791, + "learning_rate": 7.376005891307302e-06, + "loss": 3.1257, + "step": 5906 + }, + { + "epoch": 8.65, + "grad_norm": 0.39548638463020325, + "learning_rate": 7.358336557062507e-06, + "loss": 3.1185, + "step": 5907 + }, + { + "epoch": 8.65, + "grad_norm": 0.5258082151412964, + "learning_rate": 7.340687603362828e-06, + "loss": 3.1174, + "step": 5908 + }, + { + "epoch": 8.65, + "grad_norm": 0.34763848781585693, + "learning_rate": 7.3230590340908626e-06, + "loss": 3.1277, + "step": 5909 + }, + { + "epoch": 8.65, + "grad_norm": 0.2732061743736267, + "learning_rate": 7.30545085312484e-06, + "loss": 3.1047, + "step": 5910 + }, + { + "epoch": 8.65, + "grad_norm": 0.4295293986797333, + "learning_rate": 7.2878630643384295e-06, + "loss": 3.128, + "step": 5911 + }, + { + "epoch": 8.66, + "grad_norm": 0.41177213191986084, + "learning_rate": 7.270295671600791e-06, + "loss": 3.1348, + "step": 5912 + }, + { + "epoch": 8.66, + "grad_norm": 0.24607589840888977, + "learning_rate": 7.252748678776666e-06, + "loss": 3.1461, + "step": 5913 + }, + { + "epoch": 8.66, + "grad_norm": 0.3988043963909149, + "learning_rate": 7.235222089726279e-06, + "loss": 3.1677, + "step": 5914 + }, + { + "epoch": 8.66, + "grad_norm": 0.48878470063209534, + "learning_rate": 7.217715908305333e-06, + "loss": 3.1104, + "step": 5915 + }, + { + "epoch": 8.66, + "eval_loss": 2.9968373775482178, + "eval_runtime": 1167.9535, + "eval_samples_per_second": 14.677, + "eval_steps_per_second": 0.918, + "step": 5915 + }, + { + "epoch": 8.66, + "grad_norm": 0.22077223658561707, + "learning_rate": 7.200230138365094e-06, + "loss": 3.1239, + "step": 5916 + }, + { + "epoch": 8.66, + "grad_norm": 0.3054116368293762, + "learning_rate": 7.182764783752327e-06, + "loss": 3.1718, + "step": 5917 + }, + { + "epoch": 8.66, + "grad_norm": 0.3858598470687866, + "learning_rate": 7.165319848309238e-06, + "loss": 3.1447, + "step": 5918 + }, + { + "epoch": 8.67, + "grad_norm": 0.6377255916595459, + "learning_rate": 7.1478953358736665e-06, + "loss": 3.1457, + "step": 5919 + }, + { + "epoch": 8.67, + "grad_norm": 0.4466152787208557, + "learning_rate": 7.1304912502788365e-06, + "loss": 3.1208, + "step": 5920 + }, + { + "epoch": 8.67, + "grad_norm": 0.528590738773346, + "learning_rate": 7.11310759535353e-06, + "loss": 3.1644, + "step": 5921 + }, + { + "epoch": 8.67, + "grad_norm": 0.32386285066604614, + "learning_rate": 7.095744374922087e-06, + "loss": 3.1177, + "step": 5922 + }, + { + "epoch": 8.67, + "grad_norm": 0.3173229694366455, + "learning_rate": 7.0784015928042516e-06, + "loss": 3.1442, + "step": 5923 + }, + { + "epoch": 8.67, + "grad_norm": 0.32550540566444397, + "learning_rate": 7.061079252815328e-06, + "loss": 3.1537, + "step": 5924 + }, + { + "epoch": 8.67, + "grad_norm": 0.37319254875183105, + "learning_rate": 7.043777358766124e-06, + "loss": 3.1576, + "step": 5925 + }, + { + "epoch": 8.68, + "grad_norm": 0.30680203437805176, + "learning_rate": 7.026495914462938e-06, + "loss": 3.1132, + "step": 5926 + }, + { + "epoch": 8.68, + "grad_norm": 0.37015992403030396, + "learning_rate": 7.009234923707564e-06, + "loss": 3.1623, + "step": 5927 + }, + { + "epoch": 8.68, + "grad_norm": 0.4574234187602997, + "learning_rate": 6.991994390297307e-06, + "loss": 3.0993, + "step": 5928 + }, + { + "epoch": 8.68, + "grad_norm": 0.39552009105682373, + "learning_rate": 6.974774318024979e-06, + "loss": 3.1413, + "step": 5929 + }, + { + "epoch": 8.68, + "grad_norm": 0.6035201549530029, + "learning_rate": 6.9575747106788715e-06, + "loss": 3.1387, + "step": 5930 + }, + { + "epoch": 8.68, + "grad_norm": 0.3027237355709076, + "learning_rate": 6.9403955720427905e-06, + "loss": 3.127, + "step": 5931 + }, + { + "epoch": 8.69, + "grad_norm": 0.5305870175361633, + "learning_rate": 6.923236905896024e-06, + "loss": 3.1486, + "step": 5932 + }, + { + "epoch": 8.69, + "grad_norm": 0.3524973690509796, + "learning_rate": 6.906098716013354e-06, + "loss": 3.0994, + "step": 5933 + }, + { + "epoch": 8.69, + "grad_norm": 0.2866142690181732, + "learning_rate": 6.888981006165096e-06, + "loss": 3.1436, + "step": 5934 + }, + { + "epoch": 8.69, + "grad_norm": 0.3267587721347809, + "learning_rate": 6.871883780117005e-06, + "loss": 3.1174, + "step": 5935 + }, + { + "epoch": 8.69, + "grad_norm": 0.573542594909668, + "learning_rate": 6.854807041630362e-06, + "loss": 3.144, + "step": 5936 + }, + { + "epoch": 8.69, + "grad_norm": 0.23285147547721863, + "learning_rate": 6.837750794461939e-06, + "loss": 3.1232, + "step": 5937 + }, + { + "epoch": 8.69, + "grad_norm": 0.5822926163673401, + "learning_rate": 6.820715042364001e-06, + "loss": 3.1213, + "step": 5938 + }, + { + "epoch": 8.7, + "grad_norm": 0.42947494983673096, + "learning_rate": 6.803699789084295e-06, + "loss": 3.1078, + "step": 5939 + }, + { + "epoch": 8.7, + "grad_norm": 0.4609920382499695, + "learning_rate": 6.786705038366059e-06, + "loss": 3.1454, + "step": 5940 + }, + { + "epoch": 8.7, + "grad_norm": 0.46844562888145447, + "learning_rate": 6.7697307939480255e-06, + "loss": 3.1492, + "step": 5941 + }, + { + "epoch": 8.7, + "grad_norm": 0.4904913902282715, + "learning_rate": 6.75277705956443e-06, + "loss": 3.1276, + "step": 5942 + }, + { + "epoch": 8.7, + "grad_norm": 0.4832492470741272, + "learning_rate": 6.735843838944944e-06, + "loss": 3.128, + "step": 5943 + }, + { + "epoch": 8.7, + "grad_norm": 0.898986279964447, + "learning_rate": 6.718931135814787e-06, + "loss": 3.1299, + "step": 5944 + }, + { + "epoch": 8.7, + "grad_norm": 0.9148275256156921, + "learning_rate": 6.702038953894663e-06, + "loss": 3.1266, + "step": 5945 + }, + { + "epoch": 8.71, + "grad_norm": 0.2876090407371521, + "learning_rate": 6.685167296900685e-06, + "loss": 3.1499, + "step": 5946 + }, + { + "epoch": 8.71, + "grad_norm": 0.33294424414634705, + "learning_rate": 6.668316168544542e-06, + "loss": 3.1116, + "step": 5947 + }, + { + "epoch": 8.71, + "grad_norm": 0.3966740071773529, + "learning_rate": 6.651485572533378e-06, + "loss": 3.1724, + "step": 5948 + }, + { + "epoch": 8.71, + "grad_norm": 0.2978176176548004, + "learning_rate": 6.6346755125697765e-06, + "loss": 3.1483, + "step": 5949 + }, + { + "epoch": 8.71, + "grad_norm": 0.258612722158432, + "learning_rate": 6.617885992351846e-06, + "loss": 3.1448, + "step": 5950 + }, + { + "epoch": 8.71, + "grad_norm": 0.33197665214538574, + "learning_rate": 6.6011170155732e-06, + "loss": 3.1419, + "step": 5951 + }, + { + "epoch": 8.72, + "grad_norm": 0.4256979823112488, + "learning_rate": 6.584368585922874e-06, + "loss": 3.1285, + "step": 5952 + }, + { + "epoch": 8.72, + "grad_norm": 0.26601269841194153, + "learning_rate": 6.56764070708541e-06, + "loss": 3.1451, + "step": 5953 + }, + { + "epoch": 8.72, + "grad_norm": 0.37005001306533813, + "learning_rate": 6.5509333827408315e-06, + "loss": 3.1264, + "step": 5954 + }, + { + "epoch": 8.72, + "grad_norm": 0.4265497028827667, + "learning_rate": 6.534246616564654e-06, + "loss": 3.1624, + "step": 5955 + }, + { + "epoch": 8.72, + "grad_norm": 0.283138632774353, + "learning_rate": 6.5175804122278304e-06, + "loss": 3.1231, + "step": 5956 + }, + { + "epoch": 8.72, + "grad_norm": 0.4558214843273163, + "learning_rate": 6.500934773396827e-06, + "loss": 3.14, + "step": 5957 + }, + { + "epoch": 8.72, + "grad_norm": 0.23857197165489197, + "learning_rate": 6.48430970373356e-06, + "loss": 3.1686, + "step": 5958 + }, + { + "epoch": 8.73, + "grad_norm": 0.40693220496177673, + "learning_rate": 6.467705206895458e-06, + "loss": 3.1246, + "step": 5959 + }, + { + "epoch": 8.73, + "grad_norm": 0.3224029839038849, + "learning_rate": 6.451121286535378e-06, + "loss": 3.1159, + "step": 5960 + }, + { + "epoch": 8.73, + "grad_norm": 0.36411646008491516, + "learning_rate": 6.434557946301678e-06, + "loss": 3.1206, + "step": 5961 + }, + { + "epoch": 8.73, + "grad_norm": 0.4473271369934082, + "learning_rate": 6.418015189838189e-06, + "loss": 3.1924, + "step": 5962 + }, + { + "epoch": 8.73, + "grad_norm": 0.4669060707092285, + "learning_rate": 6.401493020784199e-06, + "loss": 3.1359, + "step": 5963 + }, + { + "epoch": 8.73, + "grad_norm": 0.4600774347782135, + "learning_rate": 6.384991442774468e-06, + "loss": 3.101, + "step": 5964 + }, + { + "epoch": 8.73, + "grad_norm": 0.3088182210922241, + "learning_rate": 6.368510459439248e-06, + "loss": 3.093, + "step": 5965 + }, + { + "epoch": 8.74, + "grad_norm": 0.5628402829170227, + "learning_rate": 6.352050074404226e-06, + "loss": 3.1187, + "step": 5966 + }, + { + "epoch": 8.74, + "grad_norm": 0.2512408196926117, + "learning_rate": 6.335610291290594e-06, + "loss": 3.1644, + "step": 5967 + }, + { + "epoch": 8.74, + "grad_norm": 0.3120940327644348, + "learning_rate": 6.319191113714984e-06, + "loss": 3.1145, + "step": 5968 + }, + { + "epoch": 8.74, + "grad_norm": 0.3800273835659027, + "learning_rate": 6.3027925452895154e-06, + "loss": 3.1278, + "step": 5969 + }, + { + "epoch": 8.74, + "grad_norm": 0.28590822219848633, + "learning_rate": 6.2864145896217475e-06, + "loss": 3.1181, + "step": 5970 + }, + { + "epoch": 8.74, + "grad_norm": 0.7252241373062134, + "learning_rate": 6.270057250314731e-06, + "loss": 3.1463, + "step": 5971 + }, + { + "epoch": 8.74, + "grad_norm": 0.3182050287723541, + "learning_rate": 6.2537205309669665e-06, + "loss": 3.1443, + "step": 5972 + }, + { + "epoch": 8.75, + "grad_norm": 0.2928391098976135, + "learning_rate": 6.237404435172434e-06, + "loss": 3.1282, + "step": 5973 + }, + { + "epoch": 8.75, + "grad_norm": 0.3506748676300049, + "learning_rate": 6.221108966520561e-06, + "loss": 3.156, + "step": 5974 + }, + { + "epoch": 8.75, + "grad_norm": 0.42503002285957336, + "learning_rate": 6.204834128596215e-06, + "loss": 3.088, + "step": 5975 + }, + { + "epoch": 8.75, + "grad_norm": 0.7118878960609436, + "learning_rate": 6.1885799249797985e-06, + "loss": 3.1189, + "step": 5976 + }, + { + "epoch": 8.75, + "grad_norm": 0.6801698207855225, + "learning_rate": 6.172346359247094e-06, + "loss": 3.1899, + "step": 5977 + }, + { + "epoch": 8.75, + "grad_norm": 0.5313051342964172, + "learning_rate": 6.156133434969369e-06, + "loss": 3.1453, + "step": 5978 + }, + { + "epoch": 8.76, + "grad_norm": 0.5182479023933411, + "learning_rate": 6.139941155713402e-06, + "loss": 3.1668, + "step": 5979 + }, + { + "epoch": 8.76, + "grad_norm": 0.40206000208854675, + "learning_rate": 6.123769525041345e-06, + "loss": 3.1273, + "step": 5980 + }, + { + "epoch": 8.76, + "grad_norm": 0.7351994514465332, + "learning_rate": 6.10761854651084e-06, + "loss": 3.1369, + "step": 5981 + }, + { + "epoch": 8.76, + "grad_norm": 0.2961914837360382, + "learning_rate": 6.091488223675057e-06, + "loss": 3.1798, + "step": 5982 + }, + { + "epoch": 8.76, + "grad_norm": 0.45421791076660156, + "learning_rate": 6.075378560082501e-06, + "loss": 3.1241, + "step": 5983 + }, + { + "epoch": 8.76, + "grad_norm": 0.4831741750240326, + "learning_rate": 6.059289559277181e-06, + "loss": 3.1234, + "step": 5984 + }, + { + "epoch": 8.76, + "grad_norm": 0.36750924587249756, + "learning_rate": 6.043221224798634e-06, + "loss": 3.1557, + "step": 5985 + }, + { + "epoch": 8.77, + "grad_norm": 0.3191670775413513, + "learning_rate": 6.027173560181731e-06, + "loss": 3.1495, + "step": 5986 + }, + { + "epoch": 8.77, + "grad_norm": 0.47878777980804443, + "learning_rate": 6.011146568956871e-06, + "loss": 3.1158, + "step": 5987 + }, + { + "epoch": 8.77, + "grad_norm": 0.5439360737800598, + "learning_rate": 5.995140254649878e-06, + "loss": 3.128, + "step": 5988 + }, + { + "epoch": 8.77, + "grad_norm": 0.37722069025039673, + "learning_rate": 5.979154620782035e-06, + "loss": 3.1135, + "step": 5989 + }, + { + "epoch": 8.77, + "grad_norm": 0.27893126010894775, + "learning_rate": 5.963189670870073e-06, + "loss": 3.1115, + "step": 5990 + }, + { + "epoch": 8.77, + "grad_norm": 0.3667724132537842, + "learning_rate": 5.947245408426172e-06, + "loss": 3.1513, + "step": 5991 + }, + { + "epoch": 8.77, + "grad_norm": 0.34727010130882263, + "learning_rate": 5.931321836957982e-06, + "loss": 3.1473, + "step": 5992 + }, + { + "epoch": 8.78, + "grad_norm": 0.5472817420959473, + "learning_rate": 5.915418959968555e-06, + "loss": 3.1246, + "step": 5993 + }, + { + "epoch": 8.78, + "grad_norm": 0.46271124482154846, + "learning_rate": 5.899536780956438e-06, + "loss": 3.1131, + "step": 5994 + }, + { + "epoch": 8.78, + "grad_norm": 0.39179834723472595, + "learning_rate": 5.8836753034155925e-06, + "loss": 3.1154, + "step": 5995 + }, + { + "epoch": 8.78, + "grad_norm": 0.5964497923851013, + "learning_rate": 5.867834530835437e-06, + "loss": 3.1636, + "step": 5996 + }, + { + "epoch": 8.78, + "grad_norm": 0.3770958483219147, + "learning_rate": 5.85201446670085e-06, + "loss": 3.1155, + "step": 5997 + }, + { + "epoch": 8.78, + "grad_norm": 0.6412743330001831, + "learning_rate": 5.8362151144921165e-06, + "loss": 3.1764, + "step": 5998 + }, + { + "epoch": 8.78, + "grad_norm": 0.39634811878204346, + "learning_rate": 5.82043647768502e-06, + "loss": 3.134, + "step": 5999 + }, + { + "epoch": 8.79, + "grad_norm": 0.2597801089286804, + "learning_rate": 5.8046785597507096e-06, + "loss": 3.1321, + "step": 6000 + }, + { + "epoch": 8.79, + "grad_norm": 0.4642222821712494, + "learning_rate": 5.7889413641558535e-06, + "loss": 3.1118, + "step": 6001 + }, + { + "epoch": 8.79, + "grad_norm": 0.8907680511474609, + "learning_rate": 5.77322489436255e-06, + "loss": 3.1598, + "step": 6002 + }, + { + "epoch": 8.79, + "grad_norm": 0.6435474753379822, + "learning_rate": 5.757529153828245e-06, + "loss": 3.1429, + "step": 6003 + }, + { + "epoch": 8.79, + "grad_norm": 6.3963494300842285, + "learning_rate": 5.741854146005965e-06, + "loss": 3.2169, + "step": 6004 + }, + { + "epoch": 8.79, + "grad_norm": 0.403425931930542, + "learning_rate": 5.726199874344096e-06, + "loss": 3.134, + "step": 6005 + }, + { + "epoch": 8.8, + "grad_norm": 0.4647369980812073, + "learning_rate": 5.710566342286428e-06, + "loss": 3.1384, + "step": 6006 + }, + { + "epoch": 8.8, + "grad_norm": 0.27036553621292114, + "learning_rate": 5.6949535532722864e-06, + "loss": 3.1547, + "step": 6007 + }, + { + "epoch": 8.8, + "grad_norm": 0.27433207631111145, + "learning_rate": 5.6793615107363695e-06, + "loss": 3.1312, + "step": 6008 + }, + { + "epoch": 8.8, + "grad_norm": 0.4025347828865051, + "learning_rate": 5.6637902181087774e-06, + "loss": 3.1355, + "step": 6009 + }, + { + "epoch": 8.8, + "grad_norm": 0.4300908148288727, + "learning_rate": 5.64823967881516e-06, + "loss": 3.1247, + "step": 6010 + }, + { + "epoch": 8.8, + "grad_norm": 0.42426836490631104, + "learning_rate": 5.632709896276478e-06, + "loss": 3.1543, + "step": 6011 + }, + { + "epoch": 8.8, + "grad_norm": 0.43459025025367737, + "learning_rate": 5.617200873909178e-06, + "loss": 3.109, + "step": 6012 + }, + { + "epoch": 8.81, + "grad_norm": 0.42334458231925964, + "learning_rate": 5.601712615125199e-06, + "loss": 3.1084, + "step": 6013 + }, + { + "epoch": 8.81, + "grad_norm": 0.33407434821128845, + "learning_rate": 5.58624512333179e-06, + "loss": 3.1399, + "step": 6014 + }, + { + "epoch": 8.81, + "grad_norm": 0.29522469639778137, + "learning_rate": 5.570798401931721e-06, + "loss": 3.126, + "step": 6015 + }, + { + "epoch": 8.81, + "grad_norm": 0.3701851963996887, + "learning_rate": 5.5553724543231825e-06, + "loss": 3.1287, + "step": 6016 + }, + { + "epoch": 8.81, + "grad_norm": 0.4674670994281769, + "learning_rate": 5.539967283899749e-06, + "loss": 3.1133, + "step": 6017 + }, + { + "epoch": 8.81, + "grad_norm": 0.8055880665779114, + "learning_rate": 5.5245828940504675e-06, + "loss": 3.1813, + "step": 6018 + }, + { + "epoch": 8.81, + "grad_norm": 0.18001839518547058, + "learning_rate": 5.5092192881598286e-06, + "loss": 3.158, + "step": 6019 + }, + { + "epoch": 8.82, + "grad_norm": 0.4075320065021515, + "learning_rate": 5.4938764696076726e-06, + "loss": 3.1312, + "step": 6020 + }, + { + "epoch": 8.82, + "grad_norm": 0.7519275546073914, + "learning_rate": 5.478554441769357e-06, + "loss": 3.1851, + "step": 6021 + }, + { + "epoch": 8.82, + "grad_norm": 0.7654166221618652, + "learning_rate": 5.463253208015595e-06, + "loss": 3.101, + "step": 6022 + }, + { + "epoch": 8.82, + "grad_norm": 0.5612555742263794, + "learning_rate": 5.447972771712584e-06, + "loss": 3.1467, + "step": 6023 + }, + { + "epoch": 8.82, + "grad_norm": 0.3048601448535919, + "learning_rate": 5.4327131362218895e-06, + "loss": 3.1308, + "step": 6024 + }, + { + "epoch": 8.82, + "grad_norm": 0.3827259838581085, + "learning_rate": 5.4174743049005495e-06, + "loss": 3.1142, + "step": 6025 + }, + { + "epoch": 8.82, + "grad_norm": 0.5008516311645508, + "learning_rate": 5.402256281101004e-06, + "loss": 3.1305, + "step": 6026 + }, + { + "epoch": 8.83, + "grad_norm": 0.5051065683364868, + "learning_rate": 5.387059068171107e-06, + "loss": 3.1348, + "step": 6027 + }, + { + "epoch": 8.83, + "grad_norm": 0.6256129145622253, + "learning_rate": 5.3718826694541425e-06, + "loss": 3.1257, + "step": 6028 + }, + { + "epoch": 8.83, + "grad_norm": 0.838629961013794, + "learning_rate": 5.356727088288826e-06, + "loss": 3.1441, + "step": 6029 + }, + { + "epoch": 8.83, + "grad_norm": 0.42572328448295593, + "learning_rate": 5.341592328009282e-06, + "loss": 3.1358, + "step": 6030 + }, + { + "epoch": 8.83, + "grad_norm": 0.5678095817565918, + "learning_rate": 5.326478391945045e-06, + "loss": 3.1257, + "step": 6031 + }, + { + "epoch": 8.83, + "grad_norm": 0.608961284160614, + "learning_rate": 5.311385283421089e-06, + "loss": 3.1949, + "step": 6032 + }, + { + "epoch": 8.84, + "grad_norm": 0.48100271821022034, + "learning_rate": 5.296313005757813e-06, + "loss": 3.1098, + "step": 6033 + }, + { + "epoch": 8.84, + "grad_norm": 0.370830774307251, + "learning_rate": 5.281261562270978e-06, + "loss": 3.1347, + "step": 6034 + }, + { + "epoch": 8.84, + "grad_norm": 0.3011663258075714, + "learning_rate": 5.266230956271834e-06, + "loss": 3.158, + "step": 6035 + }, + { + "epoch": 8.84, + "grad_norm": 0.573705792427063, + "learning_rate": 5.251221191067013e-06, + "loss": 3.165, + "step": 6036 + }, + { + "epoch": 8.84, + "grad_norm": 0.9574636816978455, + "learning_rate": 5.236232269958519e-06, + "loss": 3.1486, + "step": 6037 + }, + { + "epoch": 8.84, + "grad_norm": 0.2764081656932831, + "learning_rate": 5.221264196243869e-06, + "loss": 3.1264, + "step": 6038 + }, + { + "epoch": 8.84, + "grad_norm": 0.3352329432964325, + "learning_rate": 5.206316973215919e-06, + "loss": 3.105, + "step": 6039 + }, + { + "epoch": 8.85, + "grad_norm": 0.5301186442375183, + "learning_rate": 5.191390604162938e-06, + "loss": 3.1347, + "step": 6040 + }, + { + "epoch": 8.85, + "grad_norm": 0.44588613510131836, + "learning_rate": 5.176485092368655e-06, + "loss": 3.143, + "step": 6041 + }, + { + "epoch": 8.85, + "grad_norm": 0.5561344027519226, + "learning_rate": 5.161600441112191e-06, + "loss": 3.1433, + "step": 6042 + }, + { + "epoch": 8.85, + "grad_norm": 0.3855009973049164, + "learning_rate": 5.146736653668016e-06, + "loss": 3.1325, + "step": 6043 + }, + { + "epoch": 8.85, + "grad_norm": 0.4017171263694763, + "learning_rate": 5.131893733306137e-06, + "loss": 3.1241, + "step": 6044 + }, + { + "epoch": 8.85, + "grad_norm": 0.6462271213531494, + "learning_rate": 5.1170716832918405e-06, + "loss": 3.1079, + "step": 6045 + }, + { + "epoch": 8.85, + "grad_norm": 0.6944392323493958, + "learning_rate": 5.102270506885898e-06, + "loss": 3.1106, + "step": 6046 + }, + { + "epoch": 8.86, + "grad_norm": 0.43211668729782104, + "learning_rate": 5.087490207344458e-06, + "loss": 3.1481, + "step": 6047 + }, + { + "epoch": 8.86, + "grad_norm": 0.48566940426826477, + "learning_rate": 5.072730787919111e-06, + "loss": 3.1552, + "step": 6048 + }, + { + "epoch": 8.86, + "grad_norm": 0.3225969672203064, + "learning_rate": 5.057992251856802e-06, + "loss": 3.1173, + "step": 6049 + }, + { + "epoch": 8.86, + "grad_norm": 0.6626749634742737, + "learning_rate": 5.043274602399939e-06, + "loss": 3.1274, + "step": 6050 + }, + { + "epoch": 8.86, + "grad_norm": 0.5008309483528137, + "learning_rate": 5.028577842786287e-06, + "loss": 3.1181, + "step": 6051 + }, + { + "epoch": 8.86, + "grad_norm": 0.726362943649292, + "learning_rate": 5.013901976249036e-06, + "loss": 3.1433, + "step": 6052 + }, + { + "epoch": 8.87, + "grad_norm": 0.2972591817378998, + "learning_rate": 4.999247006016783e-06, + "loss": 3.138, + "step": 6053 + }, + { + "epoch": 8.87, + "grad_norm": 0.3870035409927368, + "learning_rate": 4.984612935313515e-06, + "loss": 3.2039, + "step": 6054 + }, + { + "epoch": 8.87, + "grad_norm": 0.6251978278160095, + "learning_rate": 4.969999767358646e-06, + "loss": 3.1232, + "step": 6055 + }, + { + "epoch": 8.87, + "grad_norm": 0.4746638238430023, + "learning_rate": 4.9554075053669605e-06, + "loss": 3.1353, + "step": 6056 + }, + { + "epoch": 8.87, + "grad_norm": 0.26788651943206787, + "learning_rate": 4.9408361525486665e-06, + "loss": 3.1221, + "step": 6057 + } + ], + "logging_steps": 1, + "max_steps": 6730, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "save_steps": 673, + "total_flos": 1.4959528555776993e+20, + "train_batch_size": 2, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-6057/training_args.bin b/checkpoint-6057/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..c9318904b092ceb1505e2e8f2b5300dd991baa7f --- /dev/null +++ b/checkpoint-6057/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a4ab9da40c5b7630b88beccc23069550f831ccb9b69b82772516407dd51150c +size 5688 diff --git a/checkpoint-6730/README.md b/checkpoint-6730/README.md new file mode 100644 index 0000000000000000000000000000000000000000..dbeb1c7761226042e1d72de6378a4fa69fdb3f06 --- /dev/null +++ b/checkpoint-6730/README.md @@ -0,0 +1,202 @@ +--- +library_name: peft +base_model: unsloth/gemma-7b +--- + +# Model Card for Model ID + + + + + +## Model Details + +### Model Description + + + + + +- **Developed by:** [More Information Needed] +- **Funded by [optional]:** [More Information Needed] +- **Shared by [optional]:** [More Information Needed] +- **Model type:** [More Information Needed] +- **Language(s) (NLP):** [More Information Needed] +- **License:** [More Information Needed] +- **Finetuned from model [optional]:** [More Information Needed] + +### Model Sources [optional] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations. + +## How to Get Started with the Model + +Use the code below to get started with the model. + +[More Information Needed] + +## Training Details + +### Training Data + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). + +- **Hardware Type:** [More Information Needed] +- **Hours used:** [More Information Needed] +- **Cloud Provider:** [More Information Needed] +- **Compute Region:** [More Information Needed] +- **Carbon Emitted:** [More Information Needed] + +## Technical Specifications [optional] + +### Model Architecture and Objective + +[More Information Needed] + +### Compute Infrastructure + +[More Information Needed] + +#### Hardware + +[More Information Needed] + +#### Software + +[More Information Needed] + +## Citation [optional] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[More Information Needed] + +## More Information [optional] + +[More Information Needed] + +## Model Card Authors [optional] + +[More Information Needed] + +## Model Card Contact + +[More Information Needed] +### Framework versions + +- PEFT 0.9.0 \ No newline at end of file diff --git a/checkpoint-6730/adapter_config.json b/checkpoint-6730/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..35971161ab187cbdc8af6c44079edb2d2b677921 --- /dev/null +++ b/checkpoint-6730/adapter_config.json @@ -0,0 +1,33 @@ +{ + "alpha_pattern": {}, + "auto_mapping": null, + "base_model_name_or_path": "unsloth/gemma-7b", + "bias": "none", + "fan_in_fan_out": null, + "inference_mode": true, + "init_lora_weights": true, + "layers_pattern": null, + "layers_to_transform": null, + "loftq_config": {}, + "lora_alpha": 16, + "lora_dropout": 0.05, + "megatron_config": null, + "megatron_core": "megatron.core", + "modules_to_save": null, + "peft_type": "LORA", + "r": 32, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "v_proj", + "o_proj", + "down_proj", + "k_proj", + "up_proj", + "gate_proj", + "q_proj" + ], + "task_type": "CAUSAL_LM", + "use_dora": false, + "use_rslora": false +} \ No newline at end of file diff --git a/checkpoint-6730/adapter_model.safetensors b/checkpoint-6730/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1b9951526b63b0d03b0872d48fa5ba3f8fb9cecb --- /dev/null +++ b/checkpoint-6730/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58ce96e81ec854cab49155c3e49332b00651e7bde0a5073948f61d7cbab7ffa4 +size 400084608 diff --git a/checkpoint-6730/optimizer.pt b/checkpoint-6730/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..84a375b706583a64e28fd153522f224d131ebf24 --- /dev/null +++ b/checkpoint-6730/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:456e14c0759fe8e5290596ff6211ecbbfe1f752d5ed68035623f7fa5a9f5dad0 +size 200866804 diff --git a/checkpoint-6730/rng_state_0.pth b/checkpoint-6730/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..9b2a7a5f1e49fbf2df9f43c6447aed676fe86e52 --- /dev/null +++ b/checkpoint-6730/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c15ab8332c8225148d2ec325bc308f6c348c350eb46b7442325ffde9ad04dabe +size 15984 diff --git a/checkpoint-6730/rng_state_1.pth b/checkpoint-6730/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..132cad1c2688f6308aee9885819855a98086bc17 --- /dev/null +++ b/checkpoint-6730/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da5ade7d5a0f5e0764875871ff439488a0f0db8be0afb2026a3ef9018bc8d9be +size 15984 diff --git a/checkpoint-6730/rng_state_2.pth b/checkpoint-6730/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..a74290895a5a8a72caa6091fe27a5118b829dd18 --- /dev/null +++ b/checkpoint-6730/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:604b6d8d4dca91b6504d1cc4fdf6b927e9fa5871d5856b4649bca20a100b06b7 +size 15984 diff --git a/checkpoint-6730/rng_state_3.pth b/checkpoint-6730/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..3efdbbb4fbe4f31e419e8959d1538eb597a08056 --- /dev/null +++ b/checkpoint-6730/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bc4a85169005685d032c1e2623dff3b07e48d44e2d1647ff29e408d056609249 +size 15984 diff --git a/checkpoint-6730/rng_state_4.pth b/checkpoint-6730/rng_state_4.pth new file mode 100644 index 0000000000000000000000000000000000000000..884e7013981b5500098ea3da7682d1e54fce0b14 --- /dev/null +++ b/checkpoint-6730/rng_state_4.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44cefe476e62c35059784212f7936931b2deccfc242251a7d0290fb563d8467f +size 15984 diff --git a/checkpoint-6730/rng_state_5.pth b/checkpoint-6730/rng_state_5.pth new file mode 100644 index 0000000000000000000000000000000000000000..23306da8322bccb9f29862d7a352367fe01391fb --- /dev/null +++ b/checkpoint-6730/rng_state_5.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e6172cef09d7fe3c5abc28758069c87cffa7686b91cfffa19a33811295c36eb3 +size 15984 diff --git a/checkpoint-6730/rng_state_6.pth b/checkpoint-6730/rng_state_6.pth new file mode 100644 index 0000000000000000000000000000000000000000..69a8cc548b7a4940532c315a5a7ad99286891754 --- /dev/null +++ b/checkpoint-6730/rng_state_6.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85ee33e43bd6fb2a97557b17cc04482b2b0b4aaf745a4b1ac5af0b22456f1ee4 +size 15984 diff --git a/checkpoint-6730/rng_state_7.pth b/checkpoint-6730/rng_state_7.pth new file mode 100644 index 0000000000000000000000000000000000000000..dffd05c410a0a830a1e280a566d8fa61a36f6b76 --- /dev/null +++ b/checkpoint-6730/rng_state_7.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ee33d2a6c9821fe1c280876ac43e11dcaf1146c1701030910de3c0613cf58e8a +size 15984 diff --git a/checkpoint-6730/scheduler.pt b/checkpoint-6730/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..d60bfb512fb142f64b82cb5f88e101e100db6f84 --- /dev/null +++ b/checkpoint-6730/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad0fd447a24ffe9ac63918d74e99dede665085199d64565ccdaf18700e734429 +size 1064 diff --git a/checkpoint-6730/special_tokens_map.json b/checkpoint-6730/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80 --- /dev/null +++ b/checkpoint-6730/special_tokens_map.json @@ -0,0 +1,34 @@ +{ + "additional_special_tokens": [ + "", + "" + ], + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/checkpoint-6730/tokenizer.json b/checkpoint-6730/tokenizer.json new file mode 100644 index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044 --- /dev/null +++ b/checkpoint-6730/tokenizer.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee +size 17477929 diff --git a/checkpoint-6730/tokenizer.model b/checkpoint-6730/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8 --- /dev/null +++ b/checkpoint-6730/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2 +size 4241003 diff --git a/checkpoint-6730/tokenizer_config.json b/checkpoint-6730/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..1451ca400a880c5056ab40cb960930cf827de2de --- /dev/null +++ b/checkpoint-6730/tokenizer_config.json @@ -0,0 +1,70 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "3": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "106": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "107": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "additional_special_tokens": [ + "", + "" + ], + "bos_token": "", + "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '' + role + '\n' + message['content'] | trim + '\n' }}{% endfor %}{% if add_generation_prompt %}{{'model\n'}}{% endif %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": null, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "tokenizer_class": "GemmaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/checkpoint-6730/trainer_state.json b/checkpoint-6730/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..8e77968f01b02d39612543cbcd4534c6c6a195b0 --- /dev/null +++ b/checkpoint-6730/trainer_state.json @@ -0,0 +1,47451 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 9.854093187302766, + "eval_steps": 169, + "global_step": 6730, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "grad_norm": 0.10426733642816544, + "learning_rate": 6.25e-06, + "loss": 39.5782, + "step": 1 + }, + { + "epoch": 0.0, + "eval_loss": 39.257938385009766, + "eval_runtime": 1143.223, + "eval_samples_per_second": 14.994, + "eval_steps_per_second": 0.938, + "step": 1 + }, + { + "epoch": 0.0, + "grad_norm": 0.10457848757505417, + "learning_rate": 1.25e-05, + "loss": 39.5213, + "step": 2 + }, + { + "epoch": 0.0, + "grad_norm": 0.10474710911512375, + "learning_rate": 1.8750000000000002e-05, + "loss": 39.4605, + "step": 3 + }, + { + "epoch": 0.01, + "grad_norm": 0.11379425972700119, + "learning_rate": 2.5e-05, + "loss": 39.4968, + "step": 4 + }, + { + "epoch": 0.01, + "grad_norm": 0.12108173966407776, + "learning_rate": 3.125e-05, + "loss": 39.4831, + "step": 5 + }, + { + "epoch": 0.01, + "grad_norm": 0.13807319104671478, + "learning_rate": 3.7500000000000003e-05, + "loss": 39.5751, + "step": 6 + }, + { + "epoch": 0.01, + "grad_norm": 0.1619432121515274, + "learning_rate": 4.375e-05, + "loss": 39.4458, + "step": 7 + }, + { + "epoch": 0.01, + "grad_norm": 0.1936173141002655, + "learning_rate": 5e-05, + "loss": 39.4696, + "step": 8 + }, + { + "epoch": 0.01, + "grad_norm": 0.23368136584758759, + "learning_rate": 5.6250000000000005e-05, + "loss": 39.5398, + "step": 9 + }, + { + "epoch": 0.01, + "grad_norm": 0.28147873282432556, + "learning_rate": 6.25e-05, + "loss": 39.5004, + "step": 10 + }, + { + "epoch": 0.02, + "grad_norm": 0.3371683359146118, + "learning_rate": 6.875e-05, + "loss": 39.4975, + "step": 11 + }, + { + "epoch": 0.02, + "grad_norm": 0.4020785391330719, + "learning_rate": 7.500000000000001e-05, + "loss": 39.5237, + "step": 12 + }, + { + "epoch": 0.02, + "grad_norm": 0.4757504463195801, + "learning_rate": 8.125000000000001e-05, + "loss": 39.5889, + "step": 13 + }, + { + "epoch": 0.02, + "grad_norm": 0.5553041100502014, + "learning_rate": 8.75e-05, + "loss": 39.5117, + "step": 14 + }, + { + "epoch": 0.02, + "grad_norm": 0.6490126848220825, + "learning_rate": 9.375e-05, + "loss": 39.5925, + "step": 15 + }, + { + "epoch": 0.02, + "grad_norm": 0.7478038668632507, + "learning_rate": 0.0001, + "loss": 39.5638, + "step": 16 + }, + { + "epoch": 0.03, + "grad_norm": 0.8497499227523804, + "learning_rate": 0.00010625000000000001, + "loss": 39.513, + "step": 17 + }, + { + "epoch": 0.03, + "grad_norm": 0.9426423907279968, + "learning_rate": 0.00011250000000000001, + "loss": 39.1585, + "step": 18 + }, + { + "epoch": 0.03, + "grad_norm": 0.9930742979049683, + "learning_rate": 0.00011875, + "loss": 38.4983, + "step": 19 + }, + { + "epoch": 0.03, + "grad_norm": 0.9656581878662109, + "learning_rate": 0.000125, + "loss": 37.7409, + "step": 20 + }, + { + "epoch": 0.03, + "grad_norm": 1.0122301578521729, + "learning_rate": 0.00013125000000000002, + "loss": 37.248, + "step": 21 + }, + { + "epoch": 0.03, + "grad_norm": 1.0719304084777832, + "learning_rate": 0.0001375, + "loss": 37.0875, + "step": 22 + }, + { + "epoch": 0.03, + "grad_norm": 1.1606364250183105, + "learning_rate": 0.00014375, + "loss": 36.8509, + "step": 23 + }, + { + "epoch": 0.04, + "grad_norm": 1.2129034996032715, + "learning_rate": 0.00015000000000000001, + "loss": 36.5097, + "step": 24 + }, + { + "epoch": 0.04, + "grad_norm": 1.3106859922409058, + "learning_rate": 0.00015625, + "loss": 36.101, + "step": 25 + }, + { + "epoch": 0.04, + "grad_norm": 1.395408272743225, + "learning_rate": 0.00016250000000000002, + "loss": 35.5219, + "step": 26 + }, + { + "epoch": 0.04, + "grad_norm": 1.5144782066345215, + "learning_rate": 0.00016875, + "loss": 35.1062, + "step": 27 + }, + { + "epoch": 0.04, + "grad_norm": 1.6357661485671997, + "learning_rate": 0.000175, + "loss": 34.562, + "step": 28 + }, + { + "epoch": 0.04, + "grad_norm": 1.7252875566482544, + "learning_rate": 0.00018125000000000001, + "loss": 33.7745, + "step": 29 + }, + { + "epoch": 0.04, + "grad_norm": 1.8169721364974976, + "learning_rate": 0.0001875, + "loss": 33.1849, + "step": 30 + }, + { + "epoch": 0.05, + "grad_norm": 1.8421262502670288, + "learning_rate": 0.00019375000000000002, + "loss": 32.4654, + "step": 31 + }, + { + "epoch": 0.05, + "grad_norm": 1.9798814058303833, + "learning_rate": 0.0002, + "loss": 31.5573, + "step": 32 + }, + { + "epoch": 0.05, + "grad_norm": 2.1344172954559326, + "learning_rate": 0.0001999999890003351, + "loss": 30.9011, + "step": 33 + }, + { + "epoch": 0.05, + "grad_norm": 2.1522722244262695, + "learning_rate": 0.00019999995600134275, + "loss": 30.0969, + "step": 34 + }, + { + "epoch": 0.05, + "grad_norm": 1.9861739873886108, + "learning_rate": 0.0001999999010030303, + "loss": 29.2185, + "step": 35 + }, + { + "epoch": 0.05, + "grad_norm": 1.967444896697998, + "learning_rate": 0.00019999982400540977, + "loss": 28.5804, + "step": 36 + }, + { + "epoch": 0.05, + "grad_norm": 1.9654927253723145, + "learning_rate": 0.00019999972500849814, + "loss": 27.8538, + "step": 37 + }, + { + "epoch": 0.06, + "grad_norm": 1.9123767614364624, + "learning_rate": 0.00019999960401231715, + "loss": 27.1902, + "step": 38 + }, + { + "epoch": 0.06, + "grad_norm": 1.7705096006393433, + "learning_rate": 0.00019999946101689346, + "loss": 26.4983, + "step": 39 + }, + { + "epoch": 0.06, + "grad_norm": 1.6448909044265747, + "learning_rate": 0.00019999929602225854, + "loss": 25.8354, + "step": 40 + }, + { + "epoch": 0.06, + "grad_norm": 1.5614784955978394, + "learning_rate": 0.00019999910902844862, + "loss": 25.2961, + "step": 41 + }, + { + "epoch": 0.06, + "grad_norm": 1.5540868043899536, + "learning_rate": 0.0001999989000355049, + "loss": 24.8097, + "step": 42 + }, + { + "epoch": 0.06, + "grad_norm": 1.4986692667007446, + "learning_rate": 0.00019999866904347332, + "loss": 24.3735, + "step": 43 + }, + { + "epoch": 0.07, + "grad_norm": 1.4384242296218872, + "learning_rate": 0.00019999841605240475, + "loss": 23.8929, + "step": 44 + }, + { + "epoch": 0.07, + "grad_norm": 1.290089726448059, + "learning_rate": 0.00019999814106235478, + "loss": 23.4627, + "step": 45 + }, + { + "epoch": 0.07, + "grad_norm": 1.3345509767532349, + "learning_rate": 0.0001999978440733839, + "loss": 23.1114, + "step": 46 + }, + { + "epoch": 0.07, + "grad_norm": 1.5627790689468384, + "learning_rate": 0.00019999752508555752, + "loss": 22.8306, + "step": 47 + }, + { + "epoch": 0.07, + "grad_norm": 1.67068612575531, + "learning_rate": 0.00019999718409894577, + "loss": 22.5982, + "step": 48 + }, + { + "epoch": 0.07, + "grad_norm": 1.6429119110107422, + "learning_rate": 0.00019999682111362368, + "loss": 22.2669, + "step": 49 + }, + { + "epoch": 0.07, + "grad_norm": 1.5682146549224854, + "learning_rate": 0.00019999643612967106, + "loss": 21.9562, + "step": 50 + }, + { + "epoch": 0.08, + "grad_norm": 1.3983449935913086, + "learning_rate": 0.00019999602914717266, + "loss": 21.6858, + "step": 51 + }, + { + "epoch": 0.08, + "grad_norm": 1.3296594619750977, + "learning_rate": 0.000199995600166218, + "loss": 21.4453, + "step": 52 + }, + { + "epoch": 0.08, + "grad_norm": 1.2462741136550903, + "learning_rate": 0.00019999514918690142, + "loss": 21.1379, + "step": 53 + }, + { + "epoch": 0.08, + "grad_norm": 1.2661601305007935, + "learning_rate": 0.0001999946762093222, + "loss": 20.8178, + "step": 54 + }, + { + "epoch": 0.08, + "grad_norm": 1.5054320096969604, + "learning_rate": 0.00019999418123358432, + "loss": 20.5357, + "step": 55 + }, + { + "epoch": 0.08, + "grad_norm": 1.8500558137893677, + "learning_rate": 0.0001999936642597967, + "loss": 20.2271, + "step": 56 + }, + { + "epoch": 0.08, + "grad_norm": 2.102200508117676, + "learning_rate": 0.00019999312528807308, + "loss": 19.9845, + "step": 57 + }, + { + "epoch": 0.09, + "grad_norm": 2.3885788917541504, + "learning_rate": 0.000199992564318532, + "loss": 19.6821, + "step": 58 + }, + { + "epoch": 0.09, + "grad_norm": 1.3269203901290894, + "learning_rate": 0.00019999198135129689, + "loss": 19.4197, + "step": 59 + }, + { + "epoch": 0.09, + "grad_norm": 1.0758070945739746, + "learning_rate": 0.000199991376386496, + "loss": 19.0985, + "step": 60 + }, + { + "epoch": 0.09, + "grad_norm": 1.1309177875518799, + "learning_rate": 0.00019999074942426246, + "loss": 18.8138, + "step": 61 + }, + { + "epoch": 0.09, + "grad_norm": 1.230851173400879, + "learning_rate": 0.00019999010046473412, + "loss": 18.5005, + "step": 62 + }, + { + "epoch": 0.09, + "grad_norm": 1.2155430316925049, + "learning_rate": 0.00019998942950805377, + "loss": 18.1152, + "step": 63 + }, + { + "epoch": 0.1, + "grad_norm": 1.2195416688919067, + "learning_rate": 0.00019998873655436905, + "loss": 17.7053, + "step": 64 + }, + { + "epoch": 0.1, + "grad_norm": 1.2259167432785034, + "learning_rate": 0.0001999880216038324, + "loss": 17.3123, + "step": 65 + }, + { + "epoch": 0.1, + "grad_norm": 1.2353639602661133, + "learning_rate": 0.00019998728465660105, + "loss": 16.9048, + "step": 66 + }, + { + "epoch": 0.1, + "grad_norm": 1.3028600215911865, + "learning_rate": 0.00019998652571283715, + "loss": 16.477, + "step": 67 + }, + { + "epoch": 0.1, + "grad_norm": 1.3070034980773926, + "learning_rate": 0.00019998574477270772, + "loss": 16.0892, + "step": 68 + }, + { + "epoch": 0.1, + "grad_norm": 1.231437087059021, + "learning_rate": 0.00019998494183638446, + "loss": 15.6727, + "step": 69 + }, + { + "epoch": 0.1, + "grad_norm": 1.143466591835022, + "learning_rate": 0.00019998411690404412, + "loss": 15.3314, + "step": 70 + }, + { + "epoch": 0.11, + "grad_norm": 1.4420065879821777, + "learning_rate": 0.0001999832699758681, + "loss": 15.1217, + "step": 71 + }, + { + "epoch": 0.11, + "grad_norm": 1.9184247255325317, + "learning_rate": 0.00019998240105204276, + "loss": 15.0641, + "step": 72 + }, + { + "epoch": 0.11, + "grad_norm": 2.11586332321167, + "learning_rate": 0.0001999815101327592, + "loss": 14.9107, + "step": 73 + }, + { + "epoch": 0.11, + "grad_norm": 2.1635212898254395, + "learning_rate": 0.0001999805972182135, + "loss": 14.6949, + "step": 74 + }, + { + "epoch": 0.11, + "grad_norm": 1.9647817611694336, + "learning_rate": 0.00019997966230860643, + "loss": 14.3436, + "step": 75 + }, + { + "epoch": 0.11, + "grad_norm": 1.6679894924163818, + "learning_rate": 0.00019997870540414367, + "loss": 13.917, + "step": 76 + }, + { + "epoch": 0.11, + "grad_norm": 1.407972812652588, + "learning_rate": 0.0001999777265050358, + "loss": 13.6019, + "step": 77 + }, + { + "epoch": 0.12, + "grad_norm": 1.442781686782837, + "learning_rate": 0.00019997672561149806, + "loss": 13.3002, + "step": 78 + }, + { + "epoch": 0.12, + "grad_norm": 1.4343981742858887, + "learning_rate": 0.00019997570272375074, + "loss": 13.111, + "step": 79 + }, + { + "epoch": 0.12, + "grad_norm": 1.4623301029205322, + "learning_rate": 0.00019997465784201882, + "loss": 12.9296, + "step": 80 + }, + { + "epoch": 0.12, + "grad_norm": 1.2028108835220337, + "learning_rate": 0.00019997359096653216, + "loss": 12.7946, + "step": 81 + }, + { + "epoch": 0.12, + "grad_norm": 0.7906871438026428, + "learning_rate": 0.0001999725020975255, + "loss": 12.6467, + "step": 82 + }, + { + "epoch": 0.12, + "grad_norm": 0.8495603799819946, + "learning_rate": 0.0001999713912352384, + "loss": 12.5962, + "step": 83 + }, + { + "epoch": 0.12, + "grad_norm": 1.952924370765686, + "learning_rate": 0.00019997025837991514, + "loss": 12.4475, + "step": 84 + }, + { + "epoch": 0.13, + "grad_norm": 2.176668405532837, + "learning_rate": 0.00019996910353180502, + "loss": 12.3597, + "step": 85 + }, + { + "epoch": 0.13, + "grad_norm": 1.4191577434539795, + "learning_rate": 0.0001999679266911621, + "loss": 12.2764, + "step": 86 + }, + { + "epoch": 0.13, + "grad_norm": 1.0766239166259766, + "learning_rate": 0.00019996672785824526, + "loss": 12.2163, + "step": 87 + }, + { + "epoch": 0.13, + "grad_norm": 0.8922137022018433, + "learning_rate": 0.00019996550703331825, + "loss": 12.004, + "step": 88 + }, + { + "epoch": 0.13, + "grad_norm": 0.8033492565155029, + "learning_rate": 0.00019996426421664963, + "loss": 11.8298, + "step": 89 + }, + { + "epoch": 0.13, + "grad_norm": 0.8409807682037354, + "learning_rate": 0.00019996299940851277, + "loss": 11.7005, + "step": 90 + }, + { + "epoch": 0.14, + "grad_norm": 1.0452065467834473, + "learning_rate": 0.000199961712609186, + "loss": 11.5059, + "step": 91 + }, + { + "epoch": 0.14, + "grad_norm": 1.1938868761062622, + "learning_rate": 0.00019996040381895237, + "loss": 11.3725, + "step": 92 + }, + { + "epoch": 0.14, + "grad_norm": 1.1088273525238037, + "learning_rate": 0.0001999590730380998, + "loss": 11.1496, + "step": 93 + }, + { + "epoch": 0.14, + "grad_norm": 0.9627100229263306, + "learning_rate": 0.00019995772026692107, + "loss": 10.9487, + "step": 94 + }, + { + "epoch": 0.14, + "grad_norm": 0.9412399530410767, + "learning_rate": 0.00019995634550571377, + "loss": 10.7797, + "step": 95 + }, + { + "epoch": 0.14, + "grad_norm": 0.962394654750824, + "learning_rate": 0.00019995494875478032, + "loss": 10.5443, + "step": 96 + }, + { + "epoch": 0.14, + "grad_norm": 0.7746237516403198, + "learning_rate": 0.000199953530014428, + "loss": 10.3263, + "step": 97 + }, + { + "epoch": 0.15, + "grad_norm": 0.821438729763031, + "learning_rate": 0.00019995208928496897, + "loss": 10.1812, + "step": 98 + }, + { + "epoch": 0.15, + "grad_norm": 1.4739282131195068, + "learning_rate": 0.00019995062656672012, + "loss": 9.9212, + "step": 99 + }, + { + "epoch": 0.15, + "grad_norm": 0.9201889038085938, + "learning_rate": 0.00019994914186000328, + "loss": 9.8514, + "step": 100 + }, + { + "epoch": 0.15, + "grad_norm": 0.909538745880127, + "learning_rate": 0.00019994763516514505, + "loss": 9.7503, + "step": 101 + }, + { + "epoch": 0.15, + "grad_norm": 0.8758969902992249, + "learning_rate": 0.0001999461064824769, + "loss": 9.6219, + "step": 102 + }, + { + "epoch": 0.15, + "grad_norm": 0.9529918432235718, + "learning_rate": 0.00019994455581233514, + "loss": 9.5321, + "step": 103 + }, + { + "epoch": 0.15, + "grad_norm": 1.0058927536010742, + "learning_rate": 0.0001999429831550609, + "loss": 9.3729, + "step": 104 + }, + { + "epoch": 0.16, + "grad_norm": 0.9739552140235901, + "learning_rate": 0.00019994138851100015, + "loss": 9.3018, + "step": 105 + }, + { + "epoch": 0.16, + "grad_norm": 0.8659080266952515, + "learning_rate": 0.0001999397718805037, + "loss": 9.1481, + "step": 106 + }, + { + "epoch": 0.16, + "grad_norm": 0.9334455132484436, + "learning_rate": 0.00019993813326392722, + "loss": 9.0994, + "step": 107 + }, + { + "epoch": 0.16, + "grad_norm": 0.9508799910545349, + "learning_rate": 0.00019993647266163117, + "loss": 9.0184, + "step": 108 + }, + { + "epoch": 0.16, + "grad_norm": 0.8480530381202698, + "learning_rate": 0.00019993479007398088, + "loss": 8.8884, + "step": 109 + }, + { + "epoch": 0.16, + "grad_norm": 0.7846106290817261, + "learning_rate": 0.0001999330855013465, + "loss": 8.7288, + "step": 110 + }, + { + "epoch": 0.16, + "grad_norm": 0.7611198425292969, + "learning_rate": 0.00019993135894410304, + "loss": 8.7147, + "step": 111 + }, + { + "epoch": 0.17, + "grad_norm": 0.7758539915084839, + "learning_rate": 0.00019992961040263033, + "loss": 8.7001, + "step": 112 + }, + { + "epoch": 0.17, + "grad_norm": 0.6993403434753418, + "learning_rate": 0.000199927839877313, + "loss": 8.5369, + "step": 113 + }, + { + "epoch": 0.17, + "grad_norm": 0.6209515929222107, + "learning_rate": 0.0001999260473685406, + "loss": 8.4196, + "step": 114 + }, + { + "epoch": 0.17, + "grad_norm": 0.6134201884269714, + "learning_rate": 0.00019992423287670748, + "loss": 8.3379, + "step": 115 + }, + { + "epoch": 0.17, + "grad_norm": 0.6407685875892639, + "learning_rate": 0.00019992239640221276, + "loss": 8.2981, + "step": 116 + }, + { + "epoch": 0.17, + "grad_norm": 0.5949248671531677, + "learning_rate": 0.00019992053794546047, + "loss": 8.266, + "step": 117 + }, + { + "epoch": 0.18, + "grad_norm": 0.5390576720237732, + "learning_rate": 0.0001999186575068595, + "loss": 8.2384, + "step": 118 + }, + { + "epoch": 0.18, + "grad_norm": 0.5707695484161377, + "learning_rate": 0.00019991675508682349, + "loss": 8.1337, + "step": 119 + }, + { + "epoch": 0.18, + "grad_norm": 0.6072248816490173, + "learning_rate": 0.00019991483068577097, + "loss": 8.1228, + "step": 120 + }, + { + "epoch": 0.18, + "grad_norm": 0.5436020493507385, + "learning_rate": 0.00019991288430412532, + "loss": 8.0083, + "step": 121 + }, + { + "epoch": 0.18, + "grad_norm": 0.43990185856819153, + "learning_rate": 0.00019991091594231464, + "loss": 7.943, + "step": 122 + }, + { + "epoch": 0.18, + "grad_norm": 0.7439638376235962, + "learning_rate": 0.0001999089256007721, + "loss": 7.9075, + "step": 123 + }, + { + "epoch": 0.18, + "grad_norm": 0.5559173822402954, + "learning_rate": 0.00019990691327993548, + "loss": 7.8031, + "step": 124 + }, + { + "epoch": 0.19, + "grad_norm": 0.5491620302200317, + "learning_rate": 0.0001999048789802475, + "loss": 7.7261, + "step": 125 + }, + { + "epoch": 0.19, + "grad_norm": 0.7270621657371521, + "learning_rate": 0.00019990282270215563, + "loss": 7.8013, + "step": 126 + }, + { + "epoch": 0.19, + "grad_norm": 0.5974888205528259, + "learning_rate": 0.0001999007444461123, + "loss": 7.6982, + "step": 127 + }, + { + "epoch": 0.19, + "grad_norm": 0.4842529892921448, + "learning_rate": 0.00019989864421257467, + "loss": 7.6482, + "step": 128 + }, + { + "epoch": 0.19, + "grad_norm": 0.6567961573600769, + "learning_rate": 0.00019989652200200485, + "loss": 7.6721, + "step": 129 + }, + { + "epoch": 0.19, + "grad_norm": 0.5840482711791992, + "learning_rate": 0.00019989437781486963, + "loss": 7.6075, + "step": 130 + }, + { + "epoch": 0.19, + "grad_norm": 0.41609910130500793, + "learning_rate": 0.0001998922116516408, + "loss": 7.5145, + "step": 131 + }, + { + "epoch": 0.2, + "grad_norm": 0.4465920329093933, + "learning_rate": 0.00019989002351279483, + "loss": 7.5013, + "step": 132 + }, + { + "epoch": 0.2, + "grad_norm": 0.6557760238647461, + "learning_rate": 0.0001998878133988131, + "loss": 7.5361, + "step": 133 + }, + { + "epoch": 0.2, + "grad_norm": 0.4710114598274231, + "learning_rate": 0.00019988558131018186, + "loss": 7.4989, + "step": 134 + }, + { + "epoch": 0.2, + "grad_norm": 0.67153000831604, + "learning_rate": 0.00019988332724739215, + "loss": 7.6135, + "step": 135 + }, + { + "epoch": 0.2, + "grad_norm": 1.6115596294403076, + "learning_rate": 0.00019988105121093982, + "loss": 7.8275, + "step": 136 + }, + { + "epoch": 0.2, + "grad_norm": 0.7458434700965881, + "learning_rate": 0.0001998787532013256, + "loss": 7.9861, + "step": 137 + }, + { + "epoch": 0.2, + "grad_norm": 0.6269705891609192, + "learning_rate": 0.000199876433219055, + "loss": 8.0249, + "step": 138 + }, + { + "epoch": 0.21, + "grad_norm": 0.7430456876754761, + "learning_rate": 0.00019987409126463846, + "loss": 8.0029, + "step": 139 + }, + { + "epoch": 0.21, + "grad_norm": 1.2338321208953857, + "learning_rate": 0.0001998717273385912, + "loss": 7.9936, + "step": 140 + }, + { + "epoch": 0.21, + "grad_norm": 0.7371585965156555, + "learning_rate": 0.00019986934144143316, + "loss": 7.8833, + "step": 141 + }, + { + "epoch": 0.21, + "grad_norm": 0.6107997894287109, + "learning_rate": 0.00019986693357368934, + "loss": 7.7912, + "step": 142 + }, + { + "epoch": 0.21, + "grad_norm": 1.231048583984375, + "learning_rate": 0.00019986450373588946, + "loss": 7.7342, + "step": 143 + }, + { + "epoch": 0.21, + "grad_norm": 0.6497113704681396, + "learning_rate": 0.00019986205192856795, + "loss": 7.7583, + "step": 144 + }, + { + "epoch": 0.22, + "grad_norm": 1.1496895551681519, + "learning_rate": 0.0001998595781522643, + "loss": 7.7372, + "step": 145 + }, + { + "epoch": 0.22, + "grad_norm": 0.949022114276886, + "learning_rate": 0.00019985708240752268, + "loss": 7.7063, + "step": 146 + }, + { + "epoch": 0.22, + "grad_norm": 3.4438350200653076, + "learning_rate": 0.00019985456469489213, + "loss": 7.6564, + "step": 147 + }, + { + "epoch": 0.22, + "grad_norm": 1.3692785501480103, + "learning_rate": 0.00019985202501492656, + "loss": 7.7306, + "step": 148 + }, + { + "epoch": 0.22, + "grad_norm": 1.341970443725586, + "learning_rate": 0.00019984946336818463, + "loss": 7.6558, + "step": 149 + }, + { + "epoch": 0.22, + "grad_norm": 0.8111562728881836, + "learning_rate": 0.00019984687975522993, + "loss": 7.5931, + "step": 150 + }, + { + "epoch": 0.22, + "grad_norm": 1.028735876083374, + "learning_rate": 0.00019984427417663085, + "loss": 7.5077, + "step": 151 + }, + { + "epoch": 0.23, + "grad_norm": 4.875123023986816, + "learning_rate": 0.00019984164663296058, + "loss": 7.4798, + "step": 152 + }, + { + "epoch": 0.23, + "grad_norm": 1.2443698644638062, + "learning_rate": 0.00019983899712479715, + "loss": 7.5151, + "step": 153 + }, + { + "epoch": 0.23, + "grad_norm": 38.03964614868164, + "learning_rate": 0.00019983632565272346, + "loss": 7.3952, + "step": 154 + }, + { + "epoch": 0.23, + "grad_norm": 2.3863110542297363, + "learning_rate": 0.00019983363221732716, + "loss": 7.37, + "step": 155 + }, + { + "epoch": 0.23, + "grad_norm": 1.0871200561523438, + "learning_rate": 0.00019983091681920084, + "loss": 7.4005, + "step": 156 + }, + { + "epoch": 0.23, + "grad_norm": 1.156063437461853, + "learning_rate": 0.0001998281794589419, + "loss": 7.3377, + "step": 157 + }, + { + "epoch": 0.23, + "grad_norm": 3.350292682647705, + "learning_rate": 0.00019982542013715244, + "loss": 7.0855, + "step": 158 + }, + { + "epoch": 0.24, + "grad_norm": 4.875136375427246, + "learning_rate": 0.00019982263885443956, + "loss": 7.0308, + "step": 159 + }, + { + "epoch": 0.24, + "grad_norm": 4.6525068283081055, + "learning_rate": 0.00019981983561141516, + "loss": 7.0757, + "step": 160 + }, + { + "epoch": 0.24, + "grad_norm": 2.9209208488464355, + "learning_rate": 0.00019981701040869585, + "loss": 7.1687, + "step": 161 + }, + { + "epoch": 0.24, + "grad_norm": 2.691115379333496, + "learning_rate": 0.00019981416324690317, + "loss": 7.2762, + "step": 162 + }, + { + "epoch": 0.24, + "grad_norm": 240.15269470214844, + "learning_rate": 0.00019981129412666358, + "loss": 7.3205, + "step": 163 + }, + { + "epoch": 0.24, + "grad_norm": 5.614213466644287, + "learning_rate": 0.00019980840304860813, + "loss": 7.2591, + "step": 164 + }, + { + "epoch": 0.25, + "grad_norm": 6.848481178283691, + "learning_rate": 0.0001998054900133729, + "loss": 7.3047, + "step": 165 + }, + { + "epoch": 0.25, + "grad_norm": 1.3473995923995972, + "learning_rate": 0.00019980255502159873, + "loss": 7.314, + "step": 166 + }, + { + "epoch": 0.25, + "grad_norm": 1.2133899927139282, + "learning_rate": 0.00019979959807393132, + "loss": 7.2487, + "step": 167 + }, + { + "epoch": 0.25, + "grad_norm": 46.43593215942383, + "learning_rate": 0.00019979661917102115, + "loss": 7.2762, + "step": 168 + }, + { + "epoch": 0.25, + "grad_norm": 1.915299892425537, + "learning_rate": 0.00019979361831352357, + "loss": 7.2511, + "step": 169 + }, + { + "epoch": 0.25, + "eval_loss": 7.077121257781982, + "eval_runtime": 1142.7781, + "eval_samples_per_second": 15.0, + "eval_steps_per_second": 0.938, + "step": 169 + }, + { + "epoch": 0.25, + "grad_norm": 2.1094720363616943, + "learning_rate": 0.00019979059550209876, + "loss": 7.2082, + "step": 170 + }, + { + "epoch": 0.25, + "grad_norm": 1.6243547201156616, + "learning_rate": 0.0001997875507374117, + "loss": 7.2193, + "step": 171 + }, + { + "epoch": 0.26, + "grad_norm": 1.493200659751892, + "learning_rate": 0.0001997844840201322, + "loss": 7.2206, + "step": 172 + }, + { + "epoch": 0.26, + "grad_norm": 5.889941215515137, + "learning_rate": 0.00019978139535093496, + "loss": 7.2216, + "step": 173 + }, + { + "epoch": 0.26, + "grad_norm": 16.500181198120117, + "learning_rate": 0.00019977828473049947, + "loss": 7.2762, + "step": 174 + }, + { + "epoch": 0.26, + "grad_norm": 1.3627996444702148, + "learning_rate": 0.00019977515215951, + "loss": 7.2056, + "step": 175 + }, + { + "epoch": 0.26, + "grad_norm": 3.392564535140991, + "learning_rate": 0.0001997719976386557, + "loss": 7.2832, + "step": 176 + }, + { + "epoch": 0.26, + "grad_norm": 4.66141939163208, + "learning_rate": 0.00019976882116863061, + "loss": 7.2547, + "step": 177 + }, + { + "epoch": 0.26, + "grad_norm": 1.6790283918380737, + "learning_rate": 0.00019976562275013344, + "loss": 7.1852, + "step": 178 + }, + { + "epoch": 0.27, + "grad_norm": 3.375530242919922, + "learning_rate": 0.00019976240238386787, + "loss": 7.2132, + "step": 179 + }, + { + "epoch": 0.27, + "grad_norm": 2.9105639457702637, + "learning_rate": 0.0001997591600705424, + "loss": 7.2096, + "step": 180 + }, + { + "epoch": 0.27, + "grad_norm": 3.507265329360962, + "learning_rate": 0.00019975589581087023, + "loss": 7.1184, + "step": 181 + }, + { + "epoch": 0.27, + "grad_norm": 3.2614657878875732, + "learning_rate": 0.0001997526096055695, + "loss": 7.174, + "step": 182 + }, + { + "epoch": 0.27, + "grad_norm": 20.52471923828125, + "learning_rate": 0.0001997493014553632, + "loss": 7.1458, + "step": 183 + }, + { + "epoch": 0.27, + "grad_norm": 7.462634563446045, + "learning_rate": 0.00019974597136097904, + "loss": 7.2018, + "step": 184 + }, + { + "epoch": 0.27, + "grad_norm": 3.244502544403076, + "learning_rate": 0.00019974261932314972, + "loss": 7.2068, + "step": 185 + }, + { + "epoch": 0.28, + "grad_norm": 3.8728437423706055, + "learning_rate": 0.0001997392453426125, + "loss": 7.2269, + "step": 186 + }, + { + "epoch": 0.28, + "grad_norm": 1.4866065979003906, + "learning_rate": 0.00019973584942010978, + "loss": 7.3578, + "step": 187 + }, + { + "epoch": 0.28, + "grad_norm": 11.506189346313477, + "learning_rate": 0.0001997324315563886, + "loss": 7.3122, + "step": 188 + }, + { + "epoch": 0.28, + "grad_norm": 30.93811798095703, + "learning_rate": 0.0001997289917522008, + "loss": 7.3178, + "step": 189 + }, + { + "epoch": 0.28, + "grad_norm": 27.9945125579834, + "learning_rate": 0.00019972553000830324, + "loss": 7.3128, + "step": 190 + }, + { + "epoch": 0.28, + "grad_norm": 144.04150390625, + "learning_rate": 0.00019972204632545736, + "loss": 7.2427, + "step": 191 + }, + { + "epoch": 0.29, + "grad_norm": 77.6709213256836, + "learning_rate": 0.00019971854070442962, + "loss": 7.1204, + "step": 192 + }, + { + "epoch": 0.29, + "grad_norm": 56.658504486083984, + "learning_rate": 0.00019971501314599118, + "loss": 7.1243, + "step": 193 + }, + { + "epoch": 0.29, + "grad_norm": 12.49243450164795, + "learning_rate": 0.00019971146365091813, + "loss": 7.0943, + "step": 194 + }, + { + "epoch": 0.29, + "grad_norm": 12.344218254089355, + "learning_rate": 0.0001997078922199913, + "loss": 7.0644, + "step": 195 + }, + { + "epoch": 0.29, + "grad_norm": 542.6148071289062, + "learning_rate": 0.0001997042988539964, + "loss": 7.1231, + "step": 196 + }, + { + "epoch": 0.29, + "grad_norm": 13.391535758972168, + "learning_rate": 0.00019970068355372393, + "loss": 7.0059, + "step": 197 + }, + { + "epoch": 0.29, + "grad_norm": 438.4042663574219, + "learning_rate": 0.00019969704631996927, + "loss": 7.0568, + "step": 198 + }, + { + "epoch": 0.3, + "grad_norm": 918.09326171875, + "learning_rate": 0.00019969338715353257, + "loss": 7.0436, + "step": 199 + }, + { + "epoch": 0.3, + "grad_norm": 8.38797664642334, + "learning_rate": 0.00019968970605521876, + "loss": 7.0259, + "step": 200 + }, + { + "epoch": 0.3, + "grad_norm": 99.3614501953125, + "learning_rate": 0.00019968600302583774, + "loss": 7.016, + "step": 201 + }, + { + "epoch": 0.3, + "grad_norm": 9.193567276000977, + "learning_rate": 0.0001996822780662041, + "loss": 7.0366, + "step": 202 + }, + { + "epoch": 0.3, + "grad_norm": 20.37701988220215, + "learning_rate": 0.00019967853117713739, + "loss": 7.0319, + "step": 203 + }, + { + "epoch": 0.3, + "grad_norm": 13.956972122192383, + "learning_rate": 0.00019967476235946175, + "loss": 7.1477, + "step": 204 + }, + { + "epoch": 0.3, + "grad_norm": 18.673934936523438, + "learning_rate": 0.00019967097161400642, + "loss": 7.0995, + "step": 205 + }, + { + "epoch": 0.31, + "grad_norm": 52.66343688964844, + "learning_rate": 0.0001996671589416053, + "loss": 7.1156, + "step": 206 + }, + { + "epoch": 0.31, + "grad_norm": 54.673892974853516, + "learning_rate": 0.00019966332434309716, + "loss": 7.0896, + "step": 207 + }, + { + "epoch": 0.31, + "grad_norm": 120.72140502929688, + "learning_rate": 0.00019965946781932558, + "loss": 7.0867, + "step": 208 + }, + { + "epoch": 0.31, + "grad_norm": 22.450502395629883, + "learning_rate": 0.00019965558937113894, + "loss": 7.1196, + "step": 209 + }, + { + "epoch": 0.31, + "grad_norm": 94.62016296386719, + "learning_rate": 0.00019965168899939052, + "loss": 7.1317, + "step": 210 + }, + { + "epoch": 0.31, + "grad_norm": 47.635772705078125, + "learning_rate": 0.00019964776670493837, + "loss": 7.1411, + "step": 211 + }, + { + "epoch": 0.31, + "grad_norm": 5.324208736419678, + "learning_rate": 0.00019964382248864536, + "loss": 7.1979, + "step": 212 + }, + { + "epoch": 0.32, + "grad_norm": 11.261038780212402, + "learning_rate": 0.00019963985635137915, + "loss": 7.1543, + "step": 213 + }, + { + "epoch": 0.32, + "grad_norm": 463.3558044433594, + "learning_rate": 0.0001996358682940123, + "loss": 7.1743, + "step": 214 + }, + { + "epoch": 0.32, + "grad_norm": 3.5599043369293213, + "learning_rate": 0.00019963185831742218, + "loss": 7.1325, + "step": 215 + }, + { + "epoch": 0.32, + "grad_norm": 3.592991828918457, + "learning_rate": 0.00019962782642249092, + "loss": 7.1094, + "step": 216 + }, + { + "epoch": 0.32, + "grad_norm": 59.75080108642578, + "learning_rate": 0.00019962377261010551, + "loss": 7.1571, + "step": 217 + }, + { + "epoch": 0.32, + "grad_norm": 3.108335494995117, + "learning_rate": 0.00019961969688115782, + "loss": 7.061, + "step": 218 + }, + { + "epoch": 0.33, + "grad_norm": 87.59527587890625, + "learning_rate": 0.0001996155992365444, + "loss": 7.0367, + "step": 219 + }, + { + "epoch": 0.33, + "grad_norm": 234.99960327148438, + "learning_rate": 0.00019961147967716673, + "loss": 7.1225, + "step": 220 + }, + { + "epoch": 0.33, + "grad_norm": 2.509389877319336, + "learning_rate": 0.00019960733820393114, + "loss": 7.0294, + "step": 221 + }, + { + "epoch": 0.33, + "grad_norm": 4.749375343322754, + "learning_rate": 0.00019960317481774865, + "loss": 6.9989, + "step": 222 + }, + { + "epoch": 0.33, + "grad_norm": 6.5035319328308105, + "learning_rate": 0.00019959898951953524, + "loss": 6.9879, + "step": 223 + }, + { + "epoch": 0.33, + "grad_norm": 15.936602592468262, + "learning_rate": 0.0001995947823102116, + "loss": 6.8966, + "step": 224 + }, + { + "epoch": 0.33, + "grad_norm": 27.570510864257812, + "learning_rate": 0.00019959055319070332, + "loss": 6.9171, + "step": 225 + }, + { + "epoch": 0.34, + "grad_norm": 31.393901824951172, + "learning_rate": 0.00019958630216194073, + "loss": 6.993, + "step": 226 + }, + { + "epoch": 0.34, + "grad_norm": 2.980260133743286, + "learning_rate": 0.0001995820292248591, + "loss": 7.0092, + "step": 227 + }, + { + "epoch": 0.34, + "grad_norm": 2.1385061740875244, + "learning_rate": 0.0001995777343803984, + "loss": 6.9541, + "step": 228 + }, + { + "epoch": 0.34, + "grad_norm": 4.7213053703308105, + "learning_rate": 0.00019957341762950344, + "loss": 6.8875, + "step": 229 + }, + { + "epoch": 0.34, + "grad_norm": 2.533954381942749, + "learning_rate": 0.000199569078973124, + "loss": 6.934, + "step": 230 + }, + { + "epoch": 0.34, + "grad_norm": 3.432656764984131, + "learning_rate": 0.00019956471841221438, + "loss": 6.8725, + "step": 231 + }, + { + "epoch": 0.34, + "grad_norm": 9.07863998413086, + "learning_rate": 0.000199560335947734, + "loss": 6.8456, + "step": 232 + }, + { + "epoch": 0.35, + "grad_norm": 2.4700138568878174, + "learning_rate": 0.0001995559315806469, + "loss": 6.8803, + "step": 233 + }, + { + "epoch": 0.35, + "grad_norm": 15.215128898620605, + "learning_rate": 0.0001995515053119221, + "loss": 6.7947, + "step": 234 + }, + { + "epoch": 0.35, + "grad_norm": 10.798171997070312, + "learning_rate": 0.00019954705714253328, + "loss": 6.8543, + "step": 235 + }, + { + "epoch": 0.35, + "grad_norm": 2.1929962635040283, + "learning_rate": 0.000199542587073459, + "loss": 6.7983, + "step": 236 + }, + { + "epoch": 0.35, + "grad_norm": 18.292057037353516, + "learning_rate": 0.00019953809510568271, + "loss": 6.704, + "step": 237 + }, + { + "epoch": 0.35, + "grad_norm": 0.7752164602279663, + "learning_rate": 0.00019953358124019253, + "loss": 6.7268, + "step": 238 + }, + { + "epoch": 0.35, + "grad_norm": 1.0236197710037231, + "learning_rate": 0.00019952904547798154, + "loss": 6.6618, + "step": 239 + }, + { + "epoch": 0.36, + "grad_norm": 1.3812897205352783, + "learning_rate": 0.00019952448782004756, + "loss": 6.6367, + "step": 240 + }, + { + "epoch": 0.36, + "grad_norm": 1.219442367553711, + "learning_rate": 0.00019951990826739323, + "loss": 6.6027, + "step": 241 + }, + { + "epoch": 0.36, + "grad_norm": 1.485183596611023, + "learning_rate": 0.00019951530682102604, + "loss": 6.5468, + "step": 242 + }, + { + "epoch": 0.36, + "grad_norm": 20.273746490478516, + "learning_rate": 0.00019951068348195828, + "loss": 6.5281, + "step": 243 + }, + { + "epoch": 0.36, + "grad_norm": 3.194899082183838, + "learning_rate": 0.00019950603825120706, + "loss": 6.5781, + "step": 244 + }, + { + "epoch": 0.36, + "grad_norm": 0.9288484454154968, + "learning_rate": 0.00019950137112979425, + "loss": 6.5043, + "step": 245 + }, + { + "epoch": 0.37, + "grad_norm": 0.9781804084777832, + "learning_rate": 0.00019949668211874665, + "loss": 6.4531, + "step": 246 + }, + { + "epoch": 0.37, + "grad_norm": 5.633649826049805, + "learning_rate": 0.00019949197121909575, + "loss": 6.4262, + "step": 247 + }, + { + "epoch": 0.37, + "grad_norm": 4.805322170257568, + "learning_rate": 0.00019948723843187796, + "loss": 6.3919, + "step": 248 + }, + { + "epoch": 0.37, + "grad_norm": 1.2492502927780151, + "learning_rate": 0.00019948248375813445, + "loss": 6.3459, + "step": 249 + }, + { + "epoch": 0.37, + "grad_norm": 1.5589463710784912, + "learning_rate": 0.0001994777071989112, + "loss": 6.2473, + "step": 250 + }, + { + "epoch": 0.37, + "grad_norm": 1.4179363250732422, + "learning_rate": 0.00019947290875525908, + "loss": 6.2753, + "step": 251 + }, + { + "epoch": 0.37, + "grad_norm": 1.064170479774475, + "learning_rate": 0.00019946808842823366, + "loss": 6.2785, + "step": 252 + }, + { + "epoch": 0.38, + "grad_norm": 1.806328296661377, + "learning_rate": 0.00019946324621889535, + "loss": 6.2752, + "step": 253 + }, + { + "epoch": 0.38, + "grad_norm": 1.2437814474105835, + "learning_rate": 0.0001994583821283095, + "loss": 6.1444, + "step": 254 + }, + { + "epoch": 0.38, + "grad_norm": 17.912715911865234, + "learning_rate": 0.00019945349615754608, + "loss": 6.1001, + "step": 255 + }, + { + "epoch": 0.38, + "grad_norm": 0.8929802775382996, + "learning_rate": 0.00019944858830768002, + "loss": 6.0709, + "step": 256 + }, + { + "epoch": 0.38, + "grad_norm": 2.8683338165283203, + "learning_rate": 0.000199443658579791, + "loss": 5.9837, + "step": 257 + }, + { + "epoch": 0.38, + "grad_norm": 1.3912129402160645, + "learning_rate": 0.00019943870697496357, + "loss": 5.9728, + "step": 258 + }, + { + "epoch": 0.38, + "grad_norm": 1.9068537950515747, + "learning_rate": 0.00019943373349428698, + "loss": 5.9657, + "step": 259 + }, + { + "epoch": 0.39, + "grad_norm": 2.0884199142456055, + "learning_rate": 0.0001994287381388554, + "loss": 5.963, + "step": 260 + }, + { + "epoch": 0.39, + "grad_norm": 0.8210564851760864, + "learning_rate": 0.00019942372090976779, + "loss": 5.8253, + "step": 261 + }, + { + "epoch": 0.39, + "grad_norm": 3.3612587451934814, + "learning_rate": 0.0001994186818081279, + "loss": 5.7743, + "step": 262 + }, + { + "epoch": 0.39, + "grad_norm": 2.1916985511779785, + "learning_rate": 0.00019941362083504425, + "loss": 5.7697, + "step": 263 + }, + { + "epoch": 0.39, + "grad_norm": 1.3054827451705933, + "learning_rate": 0.00019940853799163026, + "loss": 5.7867, + "step": 264 + }, + { + "epoch": 0.39, + "grad_norm": 12.717972755432129, + "learning_rate": 0.00019940343327900416, + "loss": 5.6708, + "step": 265 + }, + { + "epoch": 0.4, + "grad_norm": 3.4190762042999268, + "learning_rate": 0.00019939830669828888, + "loss": 5.6869, + "step": 266 + }, + { + "epoch": 0.4, + "grad_norm": 2.753885507583618, + "learning_rate": 0.00019939315825061223, + "loss": 5.5825, + "step": 267 + }, + { + "epoch": 0.4, + "grad_norm": 1.1797378063201904, + "learning_rate": 0.0001993879879371069, + "loss": 5.6165, + "step": 268 + }, + { + "epoch": 0.4, + "grad_norm": 2.9885246753692627, + "learning_rate": 0.00019938279575891033, + "loss": 5.5817, + "step": 269 + }, + { + "epoch": 0.4, + "grad_norm": 1.8179508447647095, + "learning_rate": 0.00019937758171716468, + "loss": 5.4549, + "step": 270 + }, + { + "epoch": 0.4, + "grad_norm": 7.133963584899902, + "learning_rate": 0.00019937234581301707, + "loss": 5.5333, + "step": 271 + }, + { + "epoch": 0.4, + "grad_norm": 7.718517780303955, + "learning_rate": 0.00019936708804761932, + "loss": 5.4891, + "step": 272 + }, + { + "epoch": 0.41, + "grad_norm": 1.2075421810150146, + "learning_rate": 0.00019936180842212818, + "loss": 5.3542, + "step": 273 + }, + { + "epoch": 0.41, + "grad_norm": 6.181577205657959, + "learning_rate": 0.000199356506937705, + "loss": 5.4501, + "step": 274 + }, + { + "epoch": 0.41, + "grad_norm": 6.531129837036133, + "learning_rate": 0.00019935118359551624, + "loss": 5.5179, + "step": 275 + }, + { + "epoch": 0.41, + "grad_norm": 4.549463272094727, + "learning_rate": 0.00019934583839673287, + "loss": 5.4167, + "step": 276 + }, + { + "epoch": 0.41, + "grad_norm": 1.8859062194824219, + "learning_rate": 0.00019934047134253081, + "loss": 5.2628, + "step": 277 + }, + { + "epoch": 0.41, + "grad_norm": 7.954891681671143, + "learning_rate": 0.00019933508243409086, + "loss": 5.3723, + "step": 278 + }, + { + "epoch": 0.41, + "grad_norm": 6.0001444816589355, + "learning_rate": 0.00019932967167259847, + "loss": 5.2979, + "step": 279 + }, + { + "epoch": 0.42, + "grad_norm": 6.450177192687988, + "learning_rate": 0.000199324239059244, + "loss": 5.2947, + "step": 280 + }, + { + "epoch": 0.42, + "grad_norm": 5.8015642166137695, + "learning_rate": 0.00019931878459522254, + "loss": 5.2849, + "step": 281 + }, + { + "epoch": 0.42, + "grad_norm": 8.846211433410645, + "learning_rate": 0.00019931330828173408, + "loss": 5.1841, + "step": 282 + }, + { + "epoch": 0.42, + "grad_norm": 4.6721086502075195, + "learning_rate": 0.00019930781011998338, + "loss": 5.2056, + "step": 283 + }, + { + "epoch": 0.42, + "grad_norm": 3.9085495471954346, + "learning_rate": 0.00019930229011118, + "loss": 5.1857, + "step": 284 + }, + { + "epoch": 0.42, + "grad_norm": 1.6466888189315796, + "learning_rate": 0.00019929674825653826, + "loss": 5.0631, + "step": 285 + }, + { + "epoch": 0.42, + "grad_norm": 2.1051926612854004, + "learning_rate": 0.0001992911845572774, + "loss": 5.033, + "step": 286 + }, + { + "epoch": 0.43, + "grad_norm": 3.247199058532715, + "learning_rate": 0.00019928559901462133, + "loss": 5.098, + "step": 287 + }, + { + "epoch": 0.43, + "grad_norm": 4.04020881652832, + "learning_rate": 0.00019927999162979885, + "loss": 5.0759, + "step": 288 + }, + { + "epoch": 0.43, + "grad_norm": 1.5520354509353638, + "learning_rate": 0.00019927436240404359, + "loss": 5.0395, + "step": 289 + }, + { + "epoch": 0.43, + "grad_norm": 1.2532438039779663, + "learning_rate": 0.00019926871133859388, + "loss": 4.9942, + "step": 290 + }, + { + "epoch": 0.43, + "grad_norm": 3.3151402473449707, + "learning_rate": 0.00019926303843469296, + "loss": 4.9396, + "step": 291 + }, + { + "epoch": 0.43, + "grad_norm": 1.311003565788269, + "learning_rate": 0.00019925734369358883, + "loss": 4.9452, + "step": 292 + }, + { + "epoch": 0.44, + "grad_norm": 3.930495500564575, + "learning_rate": 0.00019925162711653425, + "loss": 4.9488, + "step": 293 + }, + { + "epoch": 0.44, + "grad_norm": 1.0052605867385864, + "learning_rate": 0.00019924588870478687, + "loss": 4.8627, + "step": 294 + }, + { + "epoch": 0.44, + "grad_norm": 7.544490814208984, + "learning_rate": 0.00019924012845960908, + "loss": 4.9602, + "step": 295 + }, + { + "epoch": 0.44, + "grad_norm": 6.155104637145996, + "learning_rate": 0.00019923434638226812, + "loss": 4.9091, + "step": 296 + }, + { + "epoch": 0.44, + "grad_norm": 3.3449931144714355, + "learning_rate": 0.00019922854247403599, + "loss": 5.002, + "step": 297 + }, + { + "epoch": 0.44, + "grad_norm": 3.9289393424987793, + "learning_rate": 0.0001992227167361895, + "loss": 4.8124, + "step": 298 + }, + { + "epoch": 0.44, + "grad_norm": 6.716512680053711, + "learning_rate": 0.00019921686917001032, + "loss": 4.8421, + "step": 299 + }, + { + "epoch": 0.45, + "grad_norm": 4.2615132331848145, + "learning_rate": 0.00019921099977678484, + "loss": 4.8303, + "step": 300 + }, + { + "epoch": 0.45, + "grad_norm": 4.952175617218018, + "learning_rate": 0.00019920510855780428, + "loss": 4.8718, + "step": 301 + }, + { + "epoch": 0.45, + "grad_norm": 1.830321192741394, + "learning_rate": 0.00019919919551436468, + "loss": 4.8395, + "step": 302 + }, + { + "epoch": 0.45, + "grad_norm": 2.7058656215667725, + "learning_rate": 0.0001991932606477669, + "loss": 4.721, + "step": 303 + }, + { + "epoch": 0.45, + "grad_norm": 2.4603841304779053, + "learning_rate": 0.00019918730395931649, + "loss": 4.7365, + "step": 304 + }, + { + "epoch": 0.45, + "grad_norm": 1.5384434461593628, + "learning_rate": 0.00019918132545032396, + "loss": 4.6979, + "step": 305 + }, + { + "epoch": 0.45, + "grad_norm": 5.274871349334717, + "learning_rate": 0.0001991753251221045, + "loss": 4.7287, + "step": 306 + }, + { + "epoch": 0.46, + "grad_norm": 3.8402764797210693, + "learning_rate": 0.00019916930297597817, + "loss": 4.744, + "step": 307 + }, + { + "epoch": 0.46, + "grad_norm": 3.417302131652832, + "learning_rate": 0.0001991632590132698, + "loss": 4.6439, + "step": 308 + }, + { + "epoch": 0.46, + "grad_norm": 4.113287925720215, + "learning_rate": 0.00019915719323530898, + "loss": 4.6228, + "step": 309 + }, + { + "epoch": 0.46, + "grad_norm": 3.57785964012146, + "learning_rate": 0.0001991511056434302, + "loss": 4.6077, + "step": 310 + }, + { + "epoch": 0.46, + "grad_norm": 1.568590521812439, + "learning_rate": 0.00019914499623897264, + "loss": 4.6056, + "step": 311 + }, + { + "epoch": 0.46, + "grad_norm": 3.293492317199707, + "learning_rate": 0.0001991388650232804, + "loss": 4.5294, + "step": 312 + }, + { + "epoch": 0.46, + "grad_norm": 3.416661262512207, + "learning_rate": 0.00019913271199770217, + "loss": 4.6239, + "step": 313 + }, + { + "epoch": 0.47, + "grad_norm": 2.6851394176483154, + "learning_rate": 0.0001991265371635917, + "loss": 4.6183, + "step": 314 + }, + { + "epoch": 0.47, + "grad_norm": 3.812194585800171, + "learning_rate": 0.0001991203405223074, + "loss": 4.5595, + "step": 315 + }, + { + "epoch": 0.47, + "grad_norm": 1.8244884014129639, + "learning_rate": 0.0001991141220752124, + "loss": 4.5329, + "step": 316 + }, + { + "epoch": 0.47, + "grad_norm": 2.763340950012207, + "learning_rate": 0.0001991078818236748, + "loss": 4.5336, + "step": 317 + }, + { + "epoch": 0.47, + "grad_norm": 0.6997628808021545, + "learning_rate": 0.0001991016197690674, + "loss": 4.5623, + "step": 318 + }, + { + "epoch": 0.47, + "grad_norm": 6.273808479309082, + "learning_rate": 0.0001990953359127678, + "loss": 4.549, + "step": 319 + }, + { + "epoch": 0.48, + "grad_norm": 4.5749359130859375, + "learning_rate": 0.00019908903025615842, + "loss": 4.4894, + "step": 320 + }, + { + "epoch": 0.48, + "grad_norm": 4.788200378417969, + "learning_rate": 0.00019908270280062643, + "loss": 4.514, + "step": 321 + }, + { + "epoch": 0.48, + "grad_norm": 5.6347551345825195, + "learning_rate": 0.00019907635354756385, + "loss": 4.477, + "step": 322 + }, + { + "epoch": 0.48, + "grad_norm": 5.96523904800415, + "learning_rate": 0.00019906998249836745, + "loss": 4.4759, + "step": 323 + }, + { + "epoch": 0.48, + "grad_norm": 2.5423848628997803, + "learning_rate": 0.00019906358965443888, + "loss": 4.4077, + "step": 324 + }, + { + "epoch": 0.48, + "grad_norm": 1.0939940214157104, + "learning_rate": 0.00019905717501718444, + "loss": 4.372, + "step": 325 + }, + { + "epoch": 0.48, + "grad_norm": 2.4741616249084473, + "learning_rate": 0.00019905073858801536, + "loss": 4.4112, + "step": 326 + }, + { + "epoch": 0.49, + "grad_norm": 2.774264335632324, + "learning_rate": 0.00019904428036834763, + "loss": 4.4412, + "step": 327 + }, + { + "epoch": 0.49, + "grad_norm": 1.9465824365615845, + "learning_rate": 0.00019903780035960194, + "loss": 4.4442, + "step": 328 + }, + { + "epoch": 0.49, + "grad_norm": 1.6578731536865234, + "learning_rate": 0.0001990312985632039, + "loss": 4.4368, + "step": 329 + }, + { + "epoch": 0.49, + "grad_norm": 1.1936014890670776, + "learning_rate": 0.00019902477498058387, + "loss": 4.3445, + "step": 330 + }, + { + "epoch": 0.49, + "grad_norm": 1.1214628219604492, + "learning_rate": 0.00019901822961317697, + "loss": 4.4032, + "step": 331 + }, + { + "epoch": 0.49, + "grad_norm": 1.4763222932815552, + "learning_rate": 0.00019901166246242316, + "loss": 4.3713, + "step": 332 + }, + { + "epoch": 0.49, + "grad_norm": 3.219590425491333, + "learning_rate": 0.00019900507352976714, + "loss": 4.3248, + "step": 333 + }, + { + "epoch": 0.5, + "grad_norm": 1.5973373651504517, + "learning_rate": 0.00019899846281665843, + "loss": 4.3571, + "step": 334 + }, + { + "epoch": 0.5, + "grad_norm": 2.3610401153564453, + "learning_rate": 0.0001989918303245514, + "loss": 4.2991, + "step": 335 + }, + { + "epoch": 0.5, + "grad_norm": 2.8560168743133545, + "learning_rate": 0.0001989851760549051, + "loss": 4.3024, + "step": 336 + }, + { + "epoch": 0.5, + "grad_norm": 0.9701719880104065, + "learning_rate": 0.0001989785000091834, + "loss": 4.2602, + "step": 337 + }, + { + "epoch": 0.5, + "grad_norm": 1.6666287183761597, + "learning_rate": 0.00019897180218885507, + "loss": 4.2519, + "step": 338 + }, + { + "epoch": 0.5, + "eval_loss": 4.065431594848633, + "eval_runtime": 1149.6322, + "eval_samples_per_second": 14.911, + "eval_steps_per_second": 0.932, + "step": 338 + }, + { + "epoch": 0.5, + "grad_norm": 2.7144992351531982, + "learning_rate": 0.00019896508259539353, + "loss": 4.2504, + "step": 339 + }, + { + "epoch": 0.5, + "grad_norm": 1.088392734527588, + "learning_rate": 0.00019895834123027703, + "loss": 4.2467, + "step": 340 + }, + { + "epoch": 0.51, + "grad_norm": 1.7031919956207275, + "learning_rate": 0.00019895157809498868, + "loss": 4.2498, + "step": 341 + }, + { + "epoch": 0.51, + "grad_norm": 3.883507013320923, + "learning_rate": 0.00019894479319101626, + "loss": 4.289, + "step": 342 + }, + { + "epoch": 0.51, + "grad_norm": 1.1266521215438843, + "learning_rate": 0.00019893798651985246, + "loss": 4.2371, + "step": 343 + }, + { + "epoch": 0.51, + "grad_norm": 2.436180591583252, + "learning_rate": 0.00019893115808299468, + "loss": 4.213, + "step": 344 + }, + { + "epoch": 0.51, + "grad_norm": 6.882200241088867, + "learning_rate": 0.00019892430788194513, + "loss": 4.2289, + "step": 345 + }, + { + "epoch": 0.51, + "grad_norm": 4.444315433502197, + "learning_rate": 0.00019891743591821077, + "loss": 4.1858, + "step": 346 + }, + { + "epoch": 0.52, + "grad_norm": 2.7933340072631836, + "learning_rate": 0.00019891054219330344, + "loss": 4.1988, + "step": 347 + }, + { + "epoch": 0.52, + "grad_norm": 1.5705324411392212, + "learning_rate": 0.00019890362670873973, + "loss": 4.1639, + "step": 348 + }, + { + "epoch": 0.52, + "grad_norm": 4.15967321395874, + "learning_rate": 0.00019889668946604093, + "loss": 4.1391, + "step": 349 + }, + { + "epoch": 0.52, + "grad_norm": 3.6706178188323975, + "learning_rate": 0.00019888973046673324, + "loss": 4.1706, + "step": 350 + }, + { + "epoch": 0.52, + "grad_norm": 4.234200477600098, + "learning_rate": 0.00019888274971234757, + "loss": 4.3108, + "step": 351 + }, + { + "epoch": 0.52, + "grad_norm": 2.6971170902252197, + "learning_rate": 0.00019887574720441967, + "loss": 4.1871, + "step": 352 + }, + { + "epoch": 0.52, + "grad_norm": 1.4427449703216553, + "learning_rate": 0.00019886872294448997, + "loss": 4.1327, + "step": 353 + }, + { + "epoch": 0.53, + "grad_norm": 4.426568984985352, + "learning_rate": 0.00019886167693410384, + "loss": 4.195, + "step": 354 + }, + { + "epoch": 0.53, + "grad_norm": 4.152656555175781, + "learning_rate": 0.00019885460917481136, + "loss": 4.1858, + "step": 355 + }, + { + "epoch": 0.53, + "grad_norm": 5.731381893157959, + "learning_rate": 0.0001988475196681673, + "loss": 4.179, + "step": 356 + }, + { + "epoch": 0.53, + "grad_norm": 17.160917282104492, + "learning_rate": 0.0001988404084157314, + "loss": 4.0972, + "step": 357 + }, + { + "epoch": 0.53, + "grad_norm": 1.9398918151855469, + "learning_rate": 0.00019883327541906802, + "loss": 4.1238, + "step": 358 + }, + { + "epoch": 0.53, + "grad_norm": 2.0568935871124268, + "learning_rate": 0.00019882612067974643, + "loss": 4.0953, + "step": 359 + }, + { + "epoch": 0.53, + "grad_norm": 3.45890212059021, + "learning_rate": 0.00019881894419934056, + "loss": 4.1171, + "step": 360 + }, + { + "epoch": 0.54, + "grad_norm": 1.6760789155960083, + "learning_rate": 0.00019881174597942923, + "loss": 4.1128, + "step": 361 + }, + { + "epoch": 0.54, + "grad_norm": 307.13031005859375, + "learning_rate": 0.000198804526021596, + "loss": 4.0658, + "step": 362 + }, + { + "epoch": 0.54, + "grad_norm": 15.57320785522461, + "learning_rate": 0.00019879728432742918, + "loss": 4.2352, + "step": 363 + }, + { + "epoch": 0.54, + "grad_norm": 8.536359786987305, + "learning_rate": 0.00019879002089852197, + "loss": 4.2513, + "step": 364 + }, + { + "epoch": 0.54, + "grad_norm": 9.536478996276855, + "learning_rate": 0.00019878273573647218, + "loss": 4.2273, + "step": 365 + }, + { + "epoch": 0.54, + "grad_norm": 11.019887924194336, + "learning_rate": 0.00019877542884288256, + "loss": 4.2037, + "step": 366 + }, + { + "epoch": 0.55, + "grad_norm": 7.40153169631958, + "learning_rate": 0.00019876810021936053, + "loss": 4.1092, + "step": 367 + }, + { + "epoch": 0.55, + "grad_norm": 5.204662322998047, + "learning_rate": 0.0001987607498675184, + "loss": 4.0828, + "step": 368 + }, + { + "epoch": 0.55, + "grad_norm": 3.2306785583496094, + "learning_rate": 0.00019875337778897315, + "loss": 4.0828, + "step": 369 + }, + { + "epoch": 0.55, + "grad_norm": 9.242484092712402, + "learning_rate": 0.00019874598398534658, + "loss": 4.1478, + "step": 370 + }, + { + "epoch": 0.55, + "grad_norm": 10.423043251037598, + "learning_rate": 0.00019873856845826536, + "loss": 4.1709, + "step": 371 + }, + { + "epoch": 0.55, + "grad_norm": 2.9181222915649414, + "learning_rate": 0.00019873113120936074, + "loss": 4.0819, + "step": 372 + }, + { + "epoch": 0.55, + "grad_norm": 5.6168718338012695, + "learning_rate": 0.00019872367224026895, + "loss": 4.09, + "step": 373 + }, + { + "epoch": 0.56, + "grad_norm": 2.738548994064331, + "learning_rate": 0.00019871619155263087, + "loss": 4.1129, + "step": 374 + }, + { + "epoch": 0.56, + "grad_norm": 8.364383697509766, + "learning_rate": 0.0001987086891480922, + "loss": 4.0793, + "step": 375 + }, + { + "epoch": 0.56, + "grad_norm": 8.516968727111816, + "learning_rate": 0.00019870116502830346, + "loss": 4.1287, + "step": 376 + }, + { + "epoch": 0.56, + "grad_norm": 2.941425323486328, + "learning_rate": 0.00019869361919491987, + "loss": 4.1145, + "step": 377 + }, + { + "epoch": 0.56, + "grad_norm": 4.925361633300781, + "learning_rate": 0.00019868605164960145, + "loss": 4.1046, + "step": 378 + }, + { + "epoch": 0.56, + "grad_norm": 1.9746365547180176, + "learning_rate": 0.0001986784623940131, + "loss": 3.9944, + "step": 379 + }, + { + "epoch": 0.56, + "grad_norm": 8.789060592651367, + "learning_rate": 0.00019867085142982426, + "loss": 4.0482, + "step": 380 + }, + { + "epoch": 0.57, + "grad_norm": 7.4179301261901855, + "learning_rate": 0.0001986632187587094, + "loss": 4.0702, + "step": 381 + }, + { + "epoch": 0.57, + "grad_norm": 3.4922187328338623, + "learning_rate": 0.00019865556438234763, + "loss": 4.0579, + "step": 382 + }, + { + "epoch": 0.57, + "grad_norm": 5.763650417327881, + "learning_rate": 0.00019864788830242284, + "loss": 4.0776, + "step": 383 + }, + { + "epoch": 0.57, + "grad_norm": 3.3361189365386963, + "learning_rate": 0.00019864019052062375, + "loss": 4.0426, + "step": 384 + }, + { + "epoch": 0.57, + "grad_norm": 8.98303508758545, + "learning_rate": 0.0001986324710386438, + "loss": 4.0599, + "step": 385 + }, + { + "epoch": 0.57, + "grad_norm": 8.836548805236816, + "learning_rate": 0.00019862472985818124, + "loss": 4.0534, + "step": 386 + }, + { + "epoch": 0.57, + "grad_norm": 2.0397250652313232, + "learning_rate": 0.00019861696698093902, + "loss": 4.0105, + "step": 387 + }, + { + "epoch": 0.58, + "grad_norm": 3.091047763824463, + "learning_rate": 0.000198609182408625, + "loss": 3.9637, + "step": 388 + }, + { + "epoch": 0.58, + "grad_norm": 3.192955255508423, + "learning_rate": 0.00019860137614295168, + "loss": 3.9963, + "step": 389 + }, + { + "epoch": 0.58, + "grad_norm": 1.5237922668457031, + "learning_rate": 0.00019859354818563643, + "loss": 3.964, + "step": 390 + }, + { + "epoch": 0.58, + "grad_norm": 4.192271709442139, + "learning_rate": 0.00019858569853840131, + "loss": 3.9638, + "step": 391 + }, + { + "epoch": 0.58, + "grad_norm": 2.3870139122009277, + "learning_rate": 0.00019857782720297319, + "loss": 4.0272, + "step": 392 + }, + { + "epoch": 0.58, + "grad_norm": 6.654936790466309, + "learning_rate": 0.00019856993418108376, + "loss": 3.9616, + "step": 393 + }, + { + "epoch": 0.59, + "grad_norm": 4.117982387542725, + "learning_rate": 0.00019856201947446937, + "loss": 3.9983, + "step": 394 + }, + { + "epoch": 0.59, + "grad_norm": 5.394196033477783, + "learning_rate": 0.00019855408308487124, + "loss": 4.0334, + "step": 395 + }, + { + "epoch": 0.59, + "grad_norm": 5.410153865814209, + "learning_rate": 0.0001985461250140353, + "loss": 4.0342, + "step": 396 + }, + { + "epoch": 0.59, + "grad_norm": 1.220145583152771, + "learning_rate": 0.00019853814526371228, + "loss": 3.9964, + "step": 397 + }, + { + "epoch": 0.59, + "grad_norm": 1.3831374645233154, + "learning_rate": 0.00019853014383565772, + "loss": 3.8909, + "step": 398 + }, + { + "epoch": 0.59, + "grad_norm": 3.4708709716796875, + "learning_rate": 0.0001985221207316318, + "loss": 3.9548, + "step": 399 + }, + { + "epoch": 0.59, + "grad_norm": 1.6609090566635132, + "learning_rate": 0.00019851407595339958, + "loss": 3.9481, + "step": 400 + }, + { + "epoch": 0.6, + "grad_norm": 1.5527194738388062, + "learning_rate": 0.00019850600950273086, + "loss": 3.9567, + "step": 401 + }, + { + "epoch": 0.6, + "grad_norm": 6.021573066711426, + "learning_rate": 0.00019849792138140022, + "loss": 4.0133, + "step": 402 + }, + { + "epoch": 0.6, + "grad_norm": 4.150272846221924, + "learning_rate": 0.00019848981159118699, + "loss": 3.9614, + "step": 403 + }, + { + "epoch": 0.6, + "grad_norm": 7.311338901519775, + "learning_rate": 0.00019848168013387522, + "loss": 3.9789, + "step": 404 + }, + { + "epoch": 0.6, + "grad_norm": 7.797035217285156, + "learning_rate": 0.00019847352701125385, + "loss": 4.0036, + "step": 405 + }, + { + "epoch": 0.6, + "grad_norm": 2.067911386489868, + "learning_rate": 0.0001984653522251165, + "loss": 3.9421, + "step": 406 + }, + { + "epoch": 0.6, + "grad_norm": 5.364695072174072, + "learning_rate": 0.00019845715577726148, + "loss": 3.9846, + "step": 407 + }, + { + "epoch": 0.61, + "grad_norm": 3.663616895675659, + "learning_rate": 0.00019844893766949207, + "loss": 3.9255, + "step": 408 + }, + { + "epoch": 0.61, + "grad_norm": 8.939522743225098, + "learning_rate": 0.00019844069790361611, + "loss": 3.9433, + "step": 409 + }, + { + "epoch": 0.61, + "grad_norm": 9.791443824768066, + "learning_rate": 0.00019843243648144635, + "loss": 3.9879, + "step": 410 + }, + { + "epoch": 0.61, + "grad_norm": 1.5382161140441895, + "learning_rate": 0.00019842415340480025, + "loss": 3.9755, + "step": 411 + }, + { + "epoch": 0.61, + "grad_norm": 7.2470879554748535, + "learning_rate": 0.00019841584867549998, + "loss": 3.9603, + "step": 412 + }, + { + "epoch": 0.61, + "grad_norm": 7.930785655975342, + "learning_rate": 0.0001984075222953726, + "loss": 3.9416, + "step": 413 + }, + { + "epoch": 0.61, + "grad_norm": 3.421830654144287, + "learning_rate": 0.00019839917426624977, + "loss": 3.9023, + "step": 414 + }, + { + "epoch": 0.62, + "grad_norm": 13.186482429504395, + "learning_rate": 0.00019839080458996806, + "loss": 4.0143, + "step": 415 + }, + { + "epoch": 0.62, + "grad_norm": 18.97665023803711, + "learning_rate": 0.0001983824132683687, + "loss": 4.0361, + "step": 416 + }, + { + "epoch": 0.62, + "grad_norm": 12.884719848632812, + "learning_rate": 0.0001983740003032978, + "loss": 4.0086, + "step": 417 + }, + { + "epoch": 0.62, + "grad_norm": 2.3282957077026367, + "learning_rate": 0.00019836556569660608, + "loss": 3.8853, + "step": 418 + }, + { + "epoch": 0.62, + "grad_norm": 5.4361186027526855, + "learning_rate": 0.0001983571094501491, + "loss": 3.8995, + "step": 419 + }, + { + "epoch": 0.62, + "grad_norm": 2.480142116546631, + "learning_rate": 0.00019834863156578724, + "loss": 3.9204, + "step": 420 + }, + { + "epoch": 0.63, + "grad_norm": 8.683335304260254, + "learning_rate": 0.0001983401320453855, + "loss": 4.04, + "step": 421 + }, + { + "epoch": 0.63, + "grad_norm": 10.6290864944458, + "learning_rate": 0.00019833161089081381, + "loss": 3.9174, + "step": 422 + }, + { + "epoch": 0.63, + "grad_norm": 1.7693558931350708, + "learning_rate": 0.00019832306810394664, + "loss": 3.937, + "step": 423 + }, + { + "epoch": 0.63, + "grad_norm": 6.862966537475586, + "learning_rate": 0.00019831450368666348, + "loss": 3.9314, + "step": 424 + }, + { + "epoch": 0.63, + "grad_norm": 6.495865345001221, + "learning_rate": 0.0001983059176408484, + "loss": 3.8863, + "step": 425 + }, + { + "epoch": 0.63, + "grad_norm": 2.8406107425689697, + "learning_rate": 0.0001982973099683902, + "loss": 3.9525, + "step": 426 + }, + { + "epoch": 0.63, + "grad_norm": 10.725767135620117, + "learning_rate": 0.00019828868067118255, + "loss": 3.894, + "step": 427 + }, + { + "epoch": 0.64, + "grad_norm": 12.51372241973877, + "learning_rate": 0.0001982800297511239, + "loss": 4.0417, + "step": 428 + }, + { + "epoch": 0.64, + "grad_norm": 5.210300445556641, + "learning_rate": 0.00019827135721011736, + "loss": 3.9486, + "step": 429 + }, + { + "epoch": 0.64, + "grad_norm": 8.358055114746094, + "learning_rate": 0.0001982626630500708, + "loss": 4.0152, + "step": 430 + }, + { + "epoch": 0.64, + "grad_norm": 6.729689598083496, + "learning_rate": 0.0001982539472728969, + "loss": 3.8904, + "step": 431 + }, + { + "epoch": 0.64, + "grad_norm": 4.138473033905029, + "learning_rate": 0.00019824520988051306, + "loss": 3.8004, + "step": 432 + }, + { + "epoch": 0.64, + "grad_norm": 6.071408271789551, + "learning_rate": 0.00019823645087484147, + "loss": 4.0332, + "step": 433 + }, + { + "epoch": 0.64, + "grad_norm": 4.820131301879883, + "learning_rate": 0.000198227670257809, + "loss": 4.0076, + "step": 434 + }, + { + "epoch": 0.65, + "grad_norm": 3.6242270469665527, + "learning_rate": 0.00019821886803134738, + "loss": 3.9052, + "step": 435 + }, + { + "epoch": 0.65, + "grad_norm": 4.531324863433838, + "learning_rate": 0.00019821004419739307, + "loss": 3.8389, + "step": 436 + }, + { + "epoch": 0.65, + "grad_norm": 3.6086678504943848, + "learning_rate": 0.00019820119875788716, + "loss": 3.8785, + "step": 437 + }, + { + "epoch": 0.65, + "grad_norm": 3.600398063659668, + "learning_rate": 0.00019819233171477566, + "loss": 3.8425, + "step": 438 + }, + { + "epoch": 0.65, + "grad_norm": 3.8471834659576416, + "learning_rate": 0.00019818344307000925, + "loss": 3.8667, + "step": 439 + }, + { + "epoch": 0.65, + "grad_norm": 3.2653310298919678, + "learning_rate": 0.00019817453282554333, + "loss": 3.8744, + "step": 440 + }, + { + "epoch": 0.65, + "grad_norm": 4.489375591278076, + "learning_rate": 0.00019816560098333817, + "loss": 3.8705, + "step": 441 + }, + { + "epoch": 0.66, + "grad_norm": 3.816256284713745, + "learning_rate": 0.00019815664754535862, + "loss": 3.7886, + "step": 442 + }, + { + "epoch": 0.66, + "grad_norm": 6.624678134918213, + "learning_rate": 0.00019814767251357448, + "loss": 3.865, + "step": 443 + }, + { + "epoch": 0.66, + "grad_norm": 6.498514652252197, + "learning_rate": 0.00019813867588996012, + "loss": 3.8771, + "step": 444 + }, + { + "epoch": 0.66, + "grad_norm": 3.340604305267334, + "learning_rate": 0.00019812965767649474, + "loss": 3.8211, + "step": 445 + }, + { + "epoch": 0.66, + "grad_norm": 5.6404805183410645, + "learning_rate": 0.0001981206178751623, + "loss": 3.8503, + "step": 446 + }, + { + "epoch": 0.66, + "grad_norm": 2.282567024230957, + "learning_rate": 0.00019811155648795153, + "loss": 3.865, + "step": 447 + }, + { + "epoch": 0.67, + "grad_norm": 4.082645893096924, + "learning_rate": 0.00019810247351685587, + "loss": 3.7833, + "step": 448 + }, + { + "epoch": 0.67, + "grad_norm": 2.7963573932647705, + "learning_rate": 0.00019809336896387344, + "loss": 3.7991, + "step": 449 + }, + { + "epoch": 0.67, + "grad_norm": 8.172618865966797, + "learning_rate": 0.00019808424283100726, + "loss": 3.8369, + "step": 450 + }, + { + "epoch": 0.67, + "grad_norm": 8.013967514038086, + "learning_rate": 0.000198075095120265, + "loss": 3.8526, + "step": 451 + }, + { + "epoch": 0.67, + "grad_norm": 0.9244366884231567, + "learning_rate": 0.00019806592583365907, + "loss": 3.7882, + "step": 452 + }, + { + "epoch": 0.67, + "grad_norm": 4.72100305557251, + "learning_rate": 0.00019805673497320662, + "loss": 3.7314, + "step": 453 + }, + { + "epoch": 0.67, + "grad_norm": 2.556580066680908, + "learning_rate": 0.00019804752254092968, + "loss": 3.8743, + "step": 454 + }, + { + "epoch": 0.68, + "grad_norm": 10.014892578125, + "learning_rate": 0.00019803828853885485, + "loss": 3.8263, + "step": 455 + }, + { + "epoch": 0.68, + "grad_norm": 12.621180534362793, + "learning_rate": 0.00019802903296901357, + "loss": 3.8423, + "step": 456 + }, + { + "epoch": 0.68, + "grad_norm": 6.725605010986328, + "learning_rate": 0.000198019755833442, + "loss": 3.7592, + "step": 457 + }, + { + "epoch": 0.68, + "grad_norm": 5.9739603996276855, + "learning_rate": 0.00019801045713418102, + "loss": 3.8252, + "step": 458 + }, + { + "epoch": 0.68, + "grad_norm": 6.546073913574219, + "learning_rate": 0.00019800113687327633, + "loss": 3.8135, + "step": 459 + }, + { + "epoch": 0.68, + "grad_norm": 1.279313325881958, + "learning_rate": 0.0001979917950527783, + "loss": 3.8242, + "step": 460 + }, + { + "epoch": 0.68, + "grad_norm": 14.773170471191406, + "learning_rate": 0.00019798243167474207, + "loss": 3.9442, + "step": 461 + }, + { + "epoch": 0.69, + "grad_norm": 20.262290954589844, + "learning_rate": 0.00019797304674122753, + "loss": 3.9303, + "step": 462 + }, + { + "epoch": 0.69, + "grad_norm": 13.257652282714844, + "learning_rate": 0.00019796364025429926, + "loss": 3.8547, + "step": 463 + }, + { + "epoch": 0.69, + "grad_norm": 1.5865988731384277, + "learning_rate": 0.00019795421221602668, + "loss": 3.781, + "step": 464 + }, + { + "epoch": 0.69, + "grad_norm": 4.687791347503662, + "learning_rate": 0.00019794476262848388, + "loss": 3.7924, + "step": 465 + }, + { + "epoch": 0.69, + "grad_norm": 1.8210766315460205, + "learning_rate": 0.0001979352914937497, + "loss": 3.8706, + "step": 466 + }, + { + "epoch": 0.69, + "grad_norm": 10.269219398498535, + "learning_rate": 0.00019792579881390768, + "loss": 3.8314, + "step": 467 + }, + { + "epoch": 0.7, + "grad_norm": 11.758221626281738, + "learning_rate": 0.00019791628459104625, + "loss": 3.8405, + "step": 468 + }, + { + "epoch": 0.7, + "grad_norm": 3.8874833583831787, + "learning_rate": 0.0001979067488272584, + "loss": 3.7889, + "step": 469 + }, + { + "epoch": 0.7, + "grad_norm": 8.348686218261719, + "learning_rate": 0.00019789719152464192, + "loss": 3.815, + "step": 470 + }, + { + "epoch": 0.7, + "grad_norm": 11.831920623779297, + "learning_rate": 0.0001978876126852994, + "loss": 3.8381, + "step": 471 + }, + { + "epoch": 0.7, + "grad_norm": 7.59721040725708, + "learning_rate": 0.00019787801231133814, + "loss": 3.8394, + "step": 472 + }, + { + "epoch": 0.7, + "grad_norm": 4.788586616516113, + "learning_rate": 0.00019786839040487005, + "loss": 3.8047, + "step": 473 + }, + { + "epoch": 0.7, + "grad_norm": 9.041956901550293, + "learning_rate": 0.00019785874696801202, + "loss": 3.7394, + "step": 474 + }, + { + "epoch": 0.71, + "grad_norm": 3.841315746307373, + "learning_rate": 0.00019784908200288546, + "loss": 3.7839, + "step": 475 + }, + { + "epoch": 0.71, + "grad_norm": 6.382961750030518, + "learning_rate": 0.0001978393955116166, + "loss": 3.7502, + "step": 476 + }, + { + "epoch": 0.71, + "grad_norm": 8.265791893005371, + "learning_rate": 0.00019782968749633642, + "loss": 3.7888, + "step": 477 + }, + { + "epoch": 0.71, + "grad_norm": 4.210597038269043, + "learning_rate": 0.00019781995795918063, + "loss": 3.7096, + "step": 478 + }, + { + "epoch": 0.71, + "grad_norm": 10.053391456604004, + "learning_rate": 0.00019781020690228963, + "loss": 3.7276, + "step": 479 + }, + { + "epoch": 0.71, + "grad_norm": 14.174814224243164, + "learning_rate": 0.0001978004343278086, + "loss": 3.891, + "step": 480 + }, + { + "epoch": 0.71, + "grad_norm": 7.2850661277771, + "learning_rate": 0.00019779064023788746, + "loss": 3.8149, + "step": 481 + }, + { + "epoch": 0.72, + "grad_norm": 5.421473503112793, + "learning_rate": 0.00019778082463468085, + "loss": 3.8328, + "step": 482 + }, + { + "epoch": 0.72, + "grad_norm": 8.712848663330078, + "learning_rate": 0.00019777098752034807, + "loss": 3.826, + "step": 483 + }, + { + "epoch": 0.72, + "grad_norm": 4.683780670166016, + "learning_rate": 0.00019776112889705334, + "loss": 3.7366, + "step": 484 + }, + { + "epoch": 0.72, + "grad_norm": 5.190871715545654, + "learning_rate": 0.00019775124876696536, + "loss": 3.7527, + "step": 485 + }, + { + "epoch": 0.72, + "grad_norm": 8.225262641906738, + "learning_rate": 0.00019774134713225778, + "loss": 3.8112, + "step": 486 + }, + { + "epoch": 0.72, + "grad_norm": 2.200317859649658, + "learning_rate": 0.00019773142399510883, + "loss": 3.7121, + "step": 487 + }, + { + "epoch": 0.72, + "grad_norm": 9.15087604522705, + "learning_rate": 0.0001977214793577016, + "loss": 3.7827, + "step": 488 + }, + { + "epoch": 0.73, + "grad_norm": 11.307198524475098, + "learning_rate": 0.0001977115132222238, + "loss": 3.7558, + "step": 489 + }, + { + "epoch": 0.73, + "grad_norm": 8.0049409866333, + "learning_rate": 0.00019770152559086794, + "loss": 3.7133, + "step": 490 + }, + { + "epoch": 0.73, + "grad_norm": 4.207780361175537, + "learning_rate": 0.0001976915164658312, + "loss": 3.7517, + "step": 491 + }, + { + "epoch": 0.73, + "grad_norm": 9.044132232666016, + "learning_rate": 0.00019768148584931554, + "loss": 3.7111, + "step": 492 + }, + { + "epoch": 0.73, + "grad_norm": 6.856382846832275, + "learning_rate": 0.00019767143374352762, + "loss": 3.7271, + "step": 493 + }, + { + "epoch": 0.73, + "grad_norm": 3.427558183670044, + "learning_rate": 0.00019766136015067886, + "loss": 3.7357, + "step": 494 + }, + { + "epoch": 0.74, + "grad_norm": 4.354124069213867, + "learning_rate": 0.00019765126507298534, + "loss": 3.7107, + "step": 495 + }, + { + "epoch": 0.74, + "grad_norm": 3.53426194190979, + "learning_rate": 0.00019764114851266796, + "loss": 3.7585, + "step": 496 + }, + { + "epoch": 0.74, + "grad_norm": 2.1305716037750244, + "learning_rate": 0.00019763101047195225, + "loss": 3.688, + "step": 497 + }, + { + "epoch": 0.74, + "grad_norm": 108.82621002197266, + "learning_rate": 0.00019762085095306856, + "loss": 3.6435, + "step": 498 + }, + { + "epoch": 0.74, + "grad_norm": 8.900164604187012, + "learning_rate": 0.00019761066995825183, + "loss": 3.7853, + "step": 499 + }, + { + "epoch": 0.74, + "grad_norm": 8.347526550292969, + "learning_rate": 0.0001976004674897419, + "loss": 3.7845, + "step": 500 + }, + { + "epoch": 0.74, + "grad_norm": 3.418429374694824, + "learning_rate": 0.00019759024354978322, + "loss": 3.7529, + "step": 501 + }, + { + "epoch": 0.75, + "grad_norm": 14.40571403503418, + "learning_rate": 0.00019757999814062495, + "loss": 3.795, + "step": 502 + }, + { + "epoch": 0.75, + "grad_norm": 21.624300003051758, + "learning_rate": 0.00019756973126452108, + "loss": 4.0035, + "step": 503 + }, + { + "epoch": 0.75, + "grad_norm": 14.849161148071289, + "learning_rate": 0.00019755944292373017, + "loss": 3.9246, + "step": 504 + }, + { + "epoch": 0.75, + "grad_norm": 7.422677993774414, + "learning_rate": 0.00019754913312051568, + "loss": 3.9148, + "step": 505 + }, + { + "epoch": 0.75, + "grad_norm": 11.331315994262695, + "learning_rate": 0.00019753880185714563, + "loss": 3.7997, + "step": 506 + }, + { + "epoch": 0.75, + "grad_norm": 34.16909408569336, + "learning_rate": 0.00019752844913589285, + "loss": 3.7348, + "step": 507 + }, + { + "epoch": 0.75, + "eval_loss": 3.593658447265625, + "eval_runtime": 1156.954, + "eval_samples_per_second": 14.816, + "eval_steps_per_second": 0.927, + "step": 507 + }, + { + "epoch": 0.75, + "grad_norm": 1562.1485595703125, + "learning_rate": 0.00019751807495903484, + "loss": 3.8149, + "step": 508 + }, + { + "epoch": 0.76, + "grad_norm": 12.771462440490723, + "learning_rate": 0.0001975076793288539, + "loss": 3.9449, + "step": 509 + }, + { + "epoch": 0.76, + "grad_norm": 7.921811580657959, + "learning_rate": 0.00019749726224763697, + "loss": 3.9061, + "step": 510 + }, + { + "epoch": 0.76, + "grad_norm": 4.624162197113037, + "learning_rate": 0.00019748682371767575, + "loss": 3.8572, + "step": 511 + }, + { + "epoch": 0.76, + "grad_norm": 33.93694305419922, + "learning_rate": 0.00019747636374126663, + "loss": 3.8013, + "step": 512 + }, + { + "epoch": 0.76, + "grad_norm": 67.59894561767578, + "learning_rate": 0.00019746588232071076, + "loss": 3.9281, + "step": 513 + }, + { + "epoch": 0.76, + "grad_norm": 18.626140594482422, + "learning_rate": 0.00019745537945831397, + "loss": 4.0322, + "step": 514 + }, + { + "epoch": 0.76, + "grad_norm": 9.813076972961426, + "learning_rate": 0.00019744485515638677, + "loss": 4.0781, + "step": 515 + }, + { + "epoch": 0.77, + "grad_norm": 12.949821472167969, + "learning_rate": 0.00019743430941724453, + "loss": 4.0615, + "step": 516 + }, + { + "epoch": 0.77, + "grad_norm": 9.89267635345459, + "learning_rate": 0.00019742374224320717, + "loss": 3.9752, + "step": 517 + }, + { + "epoch": 0.77, + "grad_norm": 4.396897792816162, + "learning_rate": 0.00019741315363659942, + "loss": 3.768, + "step": 518 + }, + { + "epoch": 0.77, + "grad_norm": 3.6423070430755615, + "learning_rate": 0.00019740254359975072, + "loss": 3.7554, + "step": 519 + }, + { + "epoch": 0.77, + "grad_norm": 5.352052688598633, + "learning_rate": 0.0001973919121349952, + "loss": 3.8254, + "step": 520 + }, + { + "epoch": 0.77, + "grad_norm": 4.971579074859619, + "learning_rate": 0.00019738125924467168, + "loss": 3.9223, + "step": 521 + }, + { + "epoch": 0.78, + "grad_norm": 3.2794764041900635, + "learning_rate": 0.00019737058493112376, + "loss": 3.7504, + "step": 522 + }, + { + "epoch": 0.78, + "grad_norm": 3.714524269104004, + "learning_rate": 0.00019735988919669968, + "loss": 3.7884, + "step": 523 + }, + { + "epoch": 0.78, + "grad_norm": 5.291292190551758, + "learning_rate": 0.0001973491720437525, + "loss": 3.8398, + "step": 524 + }, + { + "epoch": 0.78, + "grad_norm": 4.59107780456543, + "learning_rate": 0.00019733843347463984, + "loss": 3.9075, + "step": 525 + }, + { + "epoch": 0.78, + "grad_norm": 6.6806793212890625, + "learning_rate": 0.00019732767349172413, + "loss": 3.7432, + "step": 526 + }, + { + "epoch": 0.78, + "grad_norm": 48.79610824584961, + "learning_rate": 0.00019731689209737257, + "loss": 3.7873, + "step": 527 + }, + { + "epoch": 0.78, + "grad_norm": 7.798208236694336, + "learning_rate": 0.0001973060892939569, + "loss": 3.8176, + "step": 528 + }, + { + "epoch": 0.79, + "grad_norm": 7.387636661529541, + "learning_rate": 0.00019729526508385369, + "loss": 3.892, + "step": 529 + }, + { + "epoch": 0.79, + "grad_norm": 3.599566698074341, + "learning_rate": 0.00019728441946944422, + "loss": 3.754, + "step": 530 + }, + { + "epoch": 0.79, + "grad_norm": 6.45220947265625, + "learning_rate": 0.00019727355245311446, + "loss": 3.7462, + "step": 531 + }, + { + "epoch": 0.79, + "grad_norm": 6.4421563148498535, + "learning_rate": 0.00019726266403725503, + "loss": 3.8014, + "step": 532 + }, + { + "epoch": 0.79, + "grad_norm": 9.638814926147461, + "learning_rate": 0.00019725175422426135, + "loss": 3.9211, + "step": 533 + }, + { + "epoch": 0.79, + "grad_norm": 4.907318592071533, + "learning_rate": 0.00019724082301653345, + "loss": 3.8431, + "step": 534 + }, + { + "epoch": 0.79, + "grad_norm": 3.3494908809661865, + "learning_rate": 0.00019722987041647622, + "loss": 3.8407, + "step": 535 + }, + { + "epoch": 0.8, + "grad_norm": 2.774167060852051, + "learning_rate": 0.00019721889642649912, + "loss": 3.7459, + "step": 536 + }, + { + "epoch": 0.8, + "grad_norm": 4.917426586151123, + "learning_rate": 0.0001972079010490163, + "loss": 3.8871, + "step": 537 + }, + { + "epoch": 0.8, + "grad_norm": 2.627586841583252, + "learning_rate": 0.0001971968842864467, + "loss": 3.7812, + "step": 538 + }, + { + "epoch": 0.8, + "grad_norm": 6.549022197723389, + "learning_rate": 0.00019718584614121396, + "loss": 3.7288, + "step": 539 + }, + { + "epoch": 0.8, + "grad_norm": 13.424997329711914, + "learning_rate": 0.00019717478661574638, + "loss": 3.7931, + "step": 540 + }, + { + "epoch": 0.8, + "grad_norm": 10.896014213562012, + "learning_rate": 0.00019716370571247696, + "loss": 3.8668, + "step": 541 + }, + { + "epoch": 0.8, + "grad_norm": 7.297813415527344, + "learning_rate": 0.00019715260343384347, + "loss": 3.8884, + "step": 542 + }, + { + "epoch": 0.81, + "grad_norm": 3.8043060302734375, + "learning_rate": 0.00019714147978228832, + "loss": 3.8604, + "step": 543 + }, + { + "epoch": 0.81, + "grad_norm": 25.301651000976562, + "learning_rate": 0.0001971303347602586, + "loss": 3.8794, + "step": 544 + }, + { + "epoch": 0.81, + "grad_norm": 215.01597595214844, + "learning_rate": 0.0001971191683702062, + "loss": 3.9304, + "step": 545 + }, + { + "epoch": 0.81, + "grad_norm": 253.02139282226562, + "learning_rate": 0.0001971079806145876, + "loss": 4.3177, + "step": 546 + }, + { + "epoch": 0.81, + "grad_norm": 41.9672966003418, + "learning_rate": 0.00019709677149586409, + "loss": 4.7431, + "step": 547 + }, + { + "epoch": 0.81, + "grad_norm": 34.384029388427734, + "learning_rate": 0.0001970855410165015, + "loss": 5.1589, + "step": 548 + }, + { + "epoch": 0.82, + "grad_norm": 18.074525833129883, + "learning_rate": 0.00019707428917897058, + "loss": 5.4518, + "step": 549 + }, + { + "epoch": 0.82, + "grad_norm": 10.624735832214355, + "learning_rate": 0.0001970630159857466, + "loss": 5.0911, + "step": 550 + }, + { + "epoch": 0.82, + "grad_norm": 15.860513687133789, + "learning_rate": 0.00019705172143930953, + "loss": 4.8221, + "step": 551 + }, + { + "epoch": 0.82, + "grad_norm": 9.580286979675293, + "learning_rate": 0.00019704040554214422, + "loss": 4.7721, + "step": 552 + }, + { + "epoch": 0.82, + "grad_norm": 16.33072280883789, + "learning_rate": 0.00019702906829674002, + "loss": 4.7985, + "step": 553 + }, + { + "epoch": 0.82, + "grad_norm": 15.135660171508789, + "learning_rate": 0.00019701770970559105, + "loss": 4.7531, + "step": 554 + }, + { + "epoch": 0.82, + "grad_norm": 15.646889686584473, + "learning_rate": 0.0001970063297711961, + "loss": 4.6453, + "step": 555 + }, + { + "epoch": 0.83, + "grad_norm": 13.652432441711426, + "learning_rate": 0.00019699492849605874, + "loss": 4.671, + "step": 556 + }, + { + "epoch": 0.83, + "grad_norm": 18.923864364624023, + "learning_rate": 0.00019698350588268713, + "loss": 4.7306, + "step": 557 + }, + { + "epoch": 0.83, + "grad_norm": 15.4611177444458, + "learning_rate": 0.0001969720619335942, + "loss": 4.6082, + "step": 558 + }, + { + "epoch": 0.83, + "grad_norm": 6.054503440856934, + "learning_rate": 0.0001969605966512975, + "loss": 4.4091, + "step": 559 + }, + { + "epoch": 0.83, + "grad_norm": 11.314775466918945, + "learning_rate": 0.00019694911003831933, + "loss": 4.1236, + "step": 560 + }, + { + "epoch": 0.83, + "grad_norm": 12.968482971191406, + "learning_rate": 0.0001969376020971867, + "loss": 4.023, + "step": 561 + }, + { + "epoch": 0.83, + "grad_norm": 8.118325233459473, + "learning_rate": 0.00019692607283043124, + "loss": 3.9331, + "step": 562 + }, + { + "epoch": 0.84, + "grad_norm": 4.455306529998779, + "learning_rate": 0.00019691452224058933, + "loss": 3.9195, + "step": 563 + }, + { + "epoch": 0.84, + "grad_norm": 5.133811950683594, + "learning_rate": 0.000196902950330202, + "loss": 3.8438, + "step": 564 + }, + { + "epoch": 0.84, + "grad_norm": 4.062127590179443, + "learning_rate": 0.00019689135710181503, + "loss": 3.7435, + "step": 565 + }, + { + "epoch": 0.84, + "grad_norm": 4.880388259887695, + "learning_rate": 0.00019687974255797885, + "loss": 3.7828, + "step": 566 + }, + { + "epoch": 0.84, + "grad_norm": 5.3774027824401855, + "learning_rate": 0.00019686810670124856, + "loss": 3.7977, + "step": 567 + }, + { + "epoch": 0.84, + "grad_norm": 5.431920528411865, + "learning_rate": 0.00019685644953418397, + "loss": 3.8113, + "step": 568 + }, + { + "epoch": 0.84, + "grad_norm": 3.2028257846832275, + "learning_rate": 0.00019684477105934956, + "loss": 3.7359, + "step": 569 + }, + { + "epoch": 0.85, + "grad_norm": 3.04168701171875, + "learning_rate": 0.00019683307127931458, + "loss": 3.6924, + "step": 570 + }, + { + "epoch": 0.85, + "grad_norm": 3.9302849769592285, + "learning_rate": 0.00019682135019665286, + "loss": 3.7543, + "step": 571 + }, + { + "epoch": 0.85, + "grad_norm": 3.5611188411712646, + "learning_rate": 0.00019680960781394297, + "loss": 3.751, + "step": 572 + }, + { + "epoch": 0.85, + "grad_norm": 2.1138036251068115, + "learning_rate": 0.00019679784413376811, + "loss": 3.7196, + "step": 573 + }, + { + "epoch": 0.85, + "grad_norm": 3.7343742847442627, + "learning_rate": 0.0001967860591587163, + "loss": 3.6888, + "step": 574 + }, + { + "epoch": 0.85, + "grad_norm": 3.8373682498931885, + "learning_rate": 0.00019677425289138007, + "loss": 3.6819, + "step": 575 + }, + { + "epoch": 0.86, + "grad_norm": 3.580359935760498, + "learning_rate": 0.00019676242533435678, + "loss": 3.662, + "step": 576 + }, + { + "epoch": 0.86, + "grad_norm": 3.4737274646759033, + "learning_rate": 0.00019675057649024837, + "loss": 3.6293, + "step": 577 + }, + { + "epoch": 0.86, + "grad_norm": 2.3330490589141846, + "learning_rate": 0.00019673870636166156, + "loss": 3.6554, + "step": 578 + }, + { + "epoch": 0.86, + "grad_norm": 4.6455512046813965, + "learning_rate": 0.00019672681495120763, + "loss": 3.6991, + "step": 579 + }, + { + "epoch": 0.86, + "grad_norm": 2.803642988204956, + "learning_rate": 0.00019671490226150264, + "loss": 3.6473, + "step": 580 + }, + { + "epoch": 0.86, + "grad_norm": 5.52310848236084, + "learning_rate": 0.00019670296829516732, + "loss": 3.6435, + "step": 581 + }, + { + "epoch": 0.86, + "grad_norm": 5.066474914550781, + "learning_rate": 0.00019669101305482705, + "loss": 3.6769, + "step": 582 + }, + { + "epoch": 0.87, + "grad_norm": 2.860041379928589, + "learning_rate": 0.00019667903654311194, + "loss": 3.6205, + "step": 583 + }, + { + "epoch": 0.87, + "grad_norm": 2.413363456726074, + "learning_rate": 0.00019666703876265666, + "loss": 3.6287, + "step": 584 + }, + { + "epoch": 0.87, + "grad_norm": 4.433047771453857, + "learning_rate": 0.0001966550197161007, + "loss": 3.6912, + "step": 585 + }, + { + "epoch": 0.87, + "grad_norm": 117.41354370117188, + "learning_rate": 0.00019664297940608812, + "loss": 3.6996, + "step": 586 + }, + { + "epoch": 0.87, + "grad_norm": 2.949551582336426, + "learning_rate": 0.0001966309178352678, + "loss": 3.6726, + "step": 587 + }, + { + "epoch": 0.87, + "grad_norm": 6.152908802032471, + "learning_rate": 0.0001966188350062931, + "loss": 3.6525, + "step": 588 + }, + { + "epoch": 0.87, + "grad_norm": 6.42512321472168, + "learning_rate": 0.00019660673092182225, + "loss": 3.6739, + "step": 589 + }, + { + "epoch": 0.88, + "grad_norm": 3.4023945331573486, + "learning_rate": 0.000196594605584518, + "loss": 3.6757, + "step": 590 + }, + { + "epoch": 0.88, + "grad_norm": 4.428458213806152, + "learning_rate": 0.00019658245899704789, + "loss": 3.6438, + "step": 591 + }, + { + "epoch": 0.88, + "grad_norm": 4.402431488037109, + "learning_rate": 0.00019657029116208405, + "loss": 3.6902, + "step": 592 + }, + { + "epoch": 0.88, + "grad_norm": 4.742648124694824, + "learning_rate": 0.00019655810208230333, + "loss": 3.5998, + "step": 593 + }, + { + "epoch": 0.88, + "grad_norm": 1.4105639457702637, + "learning_rate": 0.00019654589176038728, + "loss": 3.5836, + "step": 594 + }, + { + "epoch": 0.88, + "grad_norm": 5.09644079208374, + "learning_rate": 0.00019653366019902206, + "loss": 3.616, + "step": 595 + }, + { + "epoch": 0.89, + "grad_norm": 5.025322914123535, + "learning_rate": 0.00019652140740089854, + "loss": 3.6783, + "step": 596 + }, + { + "epoch": 0.89, + "grad_norm": 4.536464691162109, + "learning_rate": 0.00019650913336871222, + "loss": 3.6274, + "step": 597 + }, + { + "epoch": 0.89, + "grad_norm": 3.1868488788604736, + "learning_rate": 0.00019649683810516336, + "loss": 3.6113, + "step": 598 + }, + { + "epoch": 0.89, + "grad_norm": 4.735551834106445, + "learning_rate": 0.00019648452161295682, + "loss": 3.5986, + "step": 599 + }, + { + "epoch": 0.89, + "grad_norm": 4.411576747894287, + "learning_rate": 0.0001964721838948021, + "loss": 3.5865, + "step": 600 + }, + { + "epoch": 0.89, + "grad_norm": 2.665224075317383, + "learning_rate": 0.00019645982495341348, + "loss": 3.5546, + "step": 601 + }, + { + "epoch": 0.89, + "grad_norm": 4.546963691711426, + "learning_rate": 0.00019644744479150981, + "loss": 3.6332, + "step": 602 + }, + { + "epoch": 0.9, + "grad_norm": 1.664005160331726, + "learning_rate": 0.00019643504341181466, + "loss": 3.6299, + "step": 603 + }, + { + "epoch": 0.9, + "grad_norm": 2.6972901821136475, + "learning_rate": 0.0001964226208170562, + "loss": 3.6158, + "step": 604 + }, + { + "epoch": 0.9, + "grad_norm": 3.175882339477539, + "learning_rate": 0.00019641017700996736, + "loss": 3.5895, + "step": 605 + }, + { + "epoch": 0.9, + "grad_norm": 3.3170359134674072, + "learning_rate": 0.00019639771199328574, + "loss": 3.5989, + "step": 606 + }, + { + "epoch": 0.9, + "grad_norm": 3.245979070663452, + "learning_rate": 0.00019638522576975346, + "loss": 3.5763, + "step": 607 + }, + { + "epoch": 0.9, + "grad_norm": 1.0223109722137451, + "learning_rate": 0.00019637271834211748, + "loss": 3.5668, + "step": 608 + }, + { + "epoch": 0.9, + "grad_norm": 1.9072734117507935, + "learning_rate": 0.00019636018971312932, + "loss": 3.5738, + "step": 609 + }, + { + "epoch": 0.91, + "grad_norm": 2.925701141357422, + "learning_rate": 0.00019634763988554522, + "loss": 3.616, + "step": 610 + }, + { + "epoch": 0.91, + "grad_norm": 2.6939620971679688, + "learning_rate": 0.00019633506886212603, + "loss": 3.5764, + "step": 611 + }, + { + "epoch": 0.91, + "grad_norm": 2.020845413208008, + "learning_rate": 0.0001963224766456373, + "loss": 3.5868, + "step": 612 + }, + { + "epoch": 0.91, + "grad_norm": 3.805741548538208, + "learning_rate": 0.0001963098632388492, + "loss": 3.5979, + "step": 613 + }, + { + "epoch": 0.91, + "grad_norm": 1.6603795289993286, + "learning_rate": 0.00019629722864453663, + "loss": 3.5792, + "step": 614 + }, + { + "epoch": 0.91, + "grad_norm": 8.938972473144531, + "learning_rate": 0.00019628457286547914, + "loss": 3.5978, + "step": 615 + }, + { + "epoch": 0.91, + "grad_norm": 9.437355041503906, + "learning_rate": 0.0001962718959044609, + "loss": 3.6115, + "step": 616 + }, + { + "epoch": 0.92, + "grad_norm": 1.7519173622131348, + "learning_rate": 0.0001962591977642707, + "loss": 3.5745, + "step": 617 + }, + { + "epoch": 0.92, + "grad_norm": 8.598402976989746, + "learning_rate": 0.00019624647844770213, + "loss": 3.5939, + "step": 618 + }, + { + "epoch": 0.92, + "grad_norm": 11.709813117980957, + "learning_rate": 0.0001962337379575533, + "loss": 3.7499, + "step": 619 + }, + { + "epoch": 0.92, + "grad_norm": 8.366117477416992, + "learning_rate": 0.00019622097629662708, + "loss": 3.5727, + "step": 620 + }, + { + "epoch": 0.92, + "grad_norm": 2.4304378032684326, + "learning_rate": 0.0001962081934677309, + "loss": 3.6257, + "step": 621 + }, + { + "epoch": 0.92, + "grad_norm": 6.7245025634765625, + "learning_rate": 0.00019619538947367692, + "loss": 3.5817, + "step": 622 + }, + { + "epoch": 0.93, + "grad_norm": 2.964221239089966, + "learning_rate": 0.00019618256431728194, + "loss": 3.5903, + "step": 623 + }, + { + "epoch": 0.93, + "grad_norm": 5.846991539001465, + "learning_rate": 0.00019616971800136734, + "loss": 3.5198, + "step": 624 + }, + { + "epoch": 0.93, + "grad_norm": 6.497872829437256, + "learning_rate": 0.00019615685052875933, + "loss": 3.5556, + "step": 625 + }, + { + "epoch": 0.93, + "grad_norm": 1.711037039756775, + "learning_rate": 0.00019614396190228864, + "loss": 3.5108, + "step": 626 + }, + { + "epoch": 0.93, + "grad_norm": 9.816991806030273, + "learning_rate": 0.0001961310521247906, + "loss": 3.5854, + "step": 627 + }, + { + "epoch": 0.93, + "grad_norm": 12.794452667236328, + "learning_rate": 0.00019611812119910537, + "loss": 3.6549, + "step": 628 + }, + { + "epoch": 0.93, + "grad_norm": 6.7332916259765625, + "learning_rate": 0.0001961051691280776, + "loss": 3.6605, + "step": 629 + }, + { + "epoch": 0.94, + "grad_norm": 6.012147426605225, + "learning_rate": 0.00019609219591455675, + "loss": 3.6014, + "step": 630 + }, + { + "epoch": 0.94, + "grad_norm": 6.8346381187438965, + "learning_rate": 0.00019607920156139671, + "loss": 3.5713, + "step": 631 + }, + { + "epoch": 0.94, + "grad_norm": 3.088141679763794, + "learning_rate": 0.0001960661860714563, + "loss": 3.5767, + "step": 632 + }, + { + "epoch": 0.94, + "grad_norm": 8.472504615783691, + "learning_rate": 0.0001960531494475987, + "loss": 3.6236, + "step": 633 + }, + { + "epoch": 0.94, + "grad_norm": 9.718921661376953, + "learning_rate": 0.00019604009169269193, + "loss": 3.5673, + "step": 634 + }, + { + "epoch": 0.94, + "grad_norm": 2.083310127258301, + "learning_rate": 0.00019602701280960866, + "loss": 3.5786, + "step": 635 + }, + { + "epoch": 0.94, + "grad_norm": 8.658992767333984, + "learning_rate": 0.0001960139128012261, + "loss": 3.6109, + "step": 636 + }, + { + "epoch": 0.95, + "grad_norm": 10.503134727478027, + "learning_rate": 0.0001960007916704262, + "loss": 3.6035, + "step": 637 + }, + { + "epoch": 0.95, + "grad_norm": 7.250131130218506, + "learning_rate": 0.00019598764942009548, + "loss": 3.6211, + "step": 638 + }, + { + "epoch": 0.95, + "grad_norm": 4.671655178070068, + "learning_rate": 0.00019597448605312515, + "loss": 3.5175, + "step": 639 + }, + { + "epoch": 0.95, + "grad_norm": 9.631921768188477, + "learning_rate": 0.00019596130157241112, + "loss": 3.6144, + "step": 640 + }, + { + "epoch": 0.95, + "grad_norm": 5.835899353027344, + "learning_rate": 0.00019594809598085381, + "loss": 3.5662, + "step": 641 + }, + { + "epoch": 0.95, + "grad_norm": 5.0995893478393555, + "learning_rate": 0.00019593486928135841, + "loss": 3.5438, + "step": 642 + }, + { + "epoch": 0.95, + "grad_norm": 6.497475624084473, + "learning_rate": 0.00019592162147683473, + "loss": 3.536, + "step": 643 + }, + { + "epoch": 0.96, + "grad_norm": 2.704535722732544, + "learning_rate": 0.00019590835257019714, + "loss": 3.5268, + "step": 644 + }, + { + "epoch": 0.96, + "grad_norm": 8.847089767456055, + "learning_rate": 0.00019589506256436473, + "loss": 3.6095, + "step": 645 + }, + { + "epoch": 0.96, + "grad_norm": 10.879183769226074, + "learning_rate": 0.00019588175146226122, + "loss": 3.657, + "step": 646 + }, + { + "epoch": 0.96, + "grad_norm": 4.468491077423096, + "learning_rate": 0.00019586841926681495, + "loss": 3.5589, + "step": 647 + }, + { + "epoch": 0.96, + "grad_norm": 6.55201530456543, + "learning_rate": 0.00019585506598095897, + "loss": 3.5587, + "step": 648 + }, + { + "epoch": 0.96, + "grad_norm": 8.522590637207031, + "learning_rate": 0.0001958416916076308, + "loss": 3.5354, + "step": 649 + }, + { + "epoch": 0.97, + "grad_norm": 4.8020806312561035, + "learning_rate": 0.00019582829614977283, + "loss": 3.506, + "step": 650 + }, + { + "epoch": 0.97, + "grad_norm": 7.987406253814697, + "learning_rate": 0.00019581487961033188, + "loss": 3.5615, + "step": 651 + }, + { + "epoch": 0.97, + "grad_norm": 11.70871353149414, + "learning_rate": 0.00019580144199225953, + "loss": 3.5993, + "step": 652 + }, + { + "epoch": 0.97, + "grad_norm": 4.69012975692749, + "learning_rate": 0.00019578798329851203, + "loss": 3.542, + "step": 653 + }, + { + "epoch": 0.97, + "grad_norm": 6.301438331604004, + "learning_rate": 0.0001957745035320501, + "loss": 3.5313, + "step": 654 + }, + { + "epoch": 0.97, + "grad_norm": 8.368910789489746, + "learning_rate": 0.0001957610026958393, + "loss": 3.5855, + "step": 655 + }, + { + "epoch": 0.97, + "grad_norm": 5.793453693389893, + "learning_rate": 0.00019574748079284962, + "loss": 3.5403, + "step": 656 + }, + { + "epoch": 0.98, + "grad_norm": 4.990311622619629, + "learning_rate": 0.00019573393782605582, + "loss": 3.5196, + "step": 657 + }, + { + "epoch": 0.98, + "grad_norm": 9.033957481384277, + "learning_rate": 0.0001957203737984373, + "loss": 3.528, + "step": 658 + }, + { + "epoch": 0.98, + "grad_norm": 3.766148805618286, + "learning_rate": 0.00019570678871297804, + "loss": 3.5015, + "step": 659 + }, + { + "epoch": 0.98, + "grad_norm": 7.3363518714904785, + "learning_rate": 0.00019569318257266666, + "loss": 3.5311, + "step": 660 + }, + { + "epoch": 0.98, + "grad_norm": 9.119929313659668, + "learning_rate": 0.00019567955538049643, + "loss": 3.5767, + "step": 661 + }, + { + "epoch": 0.98, + "grad_norm": 5.856534957885742, + "learning_rate": 0.00019566590713946522, + "loss": 3.5836, + "step": 662 + }, + { + "epoch": 0.98, + "grad_norm": 4.941824436187744, + "learning_rate": 0.00019565223785257557, + "loss": 3.5366, + "step": 663 + }, + { + "epoch": 0.99, + "grad_norm": 8.7649507522583, + "learning_rate": 0.00019563854752283465, + "loss": 3.5562, + "step": 664 + }, + { + "epoch": 0.99, + "grad_norm": 3.0613012313842773, + "learning_rate": 0.00019562483615325418, + "loss": 3.5715, + "step": 665 + }, + { + "epoch": 0.99, + "grad_norm": 6.858389377593994, + "learning_rate": 0.00019561110374685062, + "loss": 3.4948, + "step": 666 + }, + { + "epoch": 0.99, + "grad_norm": 8.862404823303223, + "learning_rate": 0.00019559735030664496, + "loss": 3.5523, + "step": 667 + }, + { + "epoch": 0.99, + "grad_norm": 5.061213970184326, + "learning_rate": 0.00019558357583566297, + "loss": 3.4475, + "step": 668 + }, + { + "epoch": 0.99, + "grad_norm": 7.312166690826416, + "learning_rate": 0.0001955697803369348, + "loss": 3.5274, + "step": 669 + }, + { + "epoch": 0.99, + "grad_norm": 11.606597900390625, + "learning_rate": 0.00019555596381349547, + "loss": 3.5426, + "step": 670 + }, + { + "epoch": 1.0, + "grad_norm": 5.651201248168945, + "learning_rate": 0.0001955421262683845, + "loss": 3.533, + "step": 671 + }, + { + "epoch": 1.0, + "grad_norm": 5.392771244049072, + "learning_rate": 0.00019552826770464598, + "loss": 3.4873, + "step": 672 + }, + { + "epoch": 1.0, + "grad_norm": 7.194543361663818, + "learning_rate": 0.00019551438812532882, + "loss": 3.5313, + "step": 673 + }, + { + "epoch": 1.0, + "grad_norm": 3.8677308559417725, + "learning_rate": 0.00019550048753348638, + "loss": 3.5539, + "step": 674 + }, + { + "epoch": 1.0, + "grad_norm": 8.319040298461914, + "learning_rate": 0.0001954865659321767, + "loss": 3.5204, + "step": 675 + }, + { + "epoch": 1.0, + "grad_norm": 12.673453330993652, + "learning_rate": 0.00019547262332446241, + "loss": 3.4573, + "step": 676 + }, + { + "epoch": 1.0, + "eval_loss": 3.3125858306884766, + "eval_runtime": 1160.3441, + "eval_samples_per_second": 14.773, + "eval_steps_per_second": 0.924, + "step": 676 + }, + { + "epoch": 1.01, + "grad_norm": 6.508894920349121, + "learning_rate": 0.00019545865971341085, + "loss": 3.5108, + "step": 677 + }, + { + "epoch": 1.01, + "grad_norm": 5.700242519378662, + "learning_rate": 0.00019544467510209388, + "loss": 3.4928, + "step": 678 + }, + { + "epoch": 1.01, + "grad_norm": 7.889042854309082, + "learning_rate": 0.00019543066949358803, + "loss": 3.5309, + "step": 679 + }, + { + "epoch": 1.01, + "grad_norm": 4.5509724617004395, + "learning_rate": 0.0001954166428909744, + "loss": 3.5496, + "step": 680 + }, + { + "epoch": 1.01, + "grad_norm": 7.286065578460693, + "learning_rate": 0.00019540259529733883, + "loss": 3.5685, + "step": 681 + }, + { + "epoch": 1.01, + "grad_norm": 10.92095947265625, + "learning_rate": 0.00019538852671577165, + "loss": 3.5438, + "step": 682 + }, + { + "epoch": 1.01, + "grad_norm": 4.9438557624816895, + "learning_rate": 0.00019537443714936787, + "loss": 3.4908, + "step": 683 + }, + { + "epoch": 1.0, + "grad_norm": 6.148499011993408, + "learning_rate": 0.00019536032660122707, + "loss": 3.5438, + "step": 684 + }, + { + "epoch": 1.0, + "grad_norm": 8.24433708190918, + "learning_rate": 0.00019534619507445352, + "loss": 3.6016, + "step": 685 + }, + { + "epoch": 1.0, + "grad_norm": 5.07331657409668, + "learning_rate": 0.000195332042572156, + "loss": 3.6093, + "step": 686 + }, + { + "epoch": 1.01, + "grad_norm": 7.187658309936523, + "learning_rate": 0.00019531786909744803, + "loss": 3.5025, + "step": 687 + }, + { + "epoch": 1.01, + "grad_norm": 10.835110664367676, + "learning_rate": 0.0001953036746534476, + "loss": 3.5561, + "step": 688 + }, + { + "epoch": 1.01, + "grad_norm": 4.704497814178467, + "learning_rate": 0.00019528945924327751, + "loss": 3.5617, + "step": 689 + }, + { + "epoch": 1.01, + "grad_norm": 6.398228168487549, + "learning_rate": 0.00019527522287006495, + "loss": 3.5302, + "step": 690 + }, + { + "epoch": 1.01, + "grad_norm": 8.401050567626953, + "learning_rate": 0.00019526096553694187, + "loss": 3.5882, + "step": 691 + }, + { + "epoch": 1.01, + "grad_norm": 4.914839267730713, + "learning_rate": 0.00019524668724704478, + "loss": 3.4931, + "step": 692 + }, + { + "epoch": 1.01, + "grad_norm": 6.374270915985107, + "learning_rate": 0.00019523238800351483, + "loss": 3.4807, + "step": 693 + }, + { + "epoch": 1.02, + "grad_norm": 9.898938179016113, + "learning_rate": 0.0001952180678094977, + "loss": 3.5138, + "step": 694 + }, + { + "epoch": 1.02, + "grad_norm": 4.583561897277832, + "learning_rate": 0.00019520372666814378, + "loss": 3.4739, + "step": 695 + }, + { + "epoch": 1.02, + "grad_norm": 6.307168483734131, + "learning_rate": 0.00019518936458260807, + "loss": 3.5473, + "step": 696 + }, + { + "epoch": 1.02, + "grad_norm": 8.191679000854492, + "learning_rate": 0.00019517498155605003, + "loss": 3.5002, + "step": 697 + }, + { + "epoch": 1.02, + "grad_norm": 5.078609466552734, + "learning_rate": 0.0001951605775916339, + "loss": 3.4731, + "step": 698 + }, + { + "epoch": 1.02, + "grad_norm": 6.6793060302734375, + "learning_rate": 0.00019514615269252845, + "loss": 3.5543, + "step": 699 + }, + { + "epoch": 1.02, + "grad_norm": 10.510246276855469, + "learning_rate": 0.00019513170686190706, + "loss": 3.5589, + "step": 700 + }, + { + "epoch": 1.03, + "grad_norm": 4.769564151763916, + "learning_rate": 0.00019511724010294766, + "loss": 3.4587, + "step": 701 + }, + { + "epoch": 1.03, + "grad_norm": 6.157220363616943, + "learning_rate": 0.00019510275241883292, + "loss": 3.4658, + "step": 702 + }, + { + "epoch": 1.03, + "grad_norm": 8.317898750305176, + "learning_rate": 0.00019508824381274996, + "loss": 3.5147, + "step": 703 + }, + { + "epoch": 1.03, + "grad_norm": 4.906617164611816, + "learning_rate": 0.00019507371428789064, + "loss": 3.5319, + "step": 704 + }, + { + "epoch": 1.03, + "grad_norm": 6.44058084487915, + "learning_rate": 0.00019505916384745134, + "loss": 3.5268, + "step": 705 + }, + { + "epoch": 1.03, + "grad_norm": 10.473220825195312, + "learning_rate": 0.00019504459249463304, + "loss": 3.535, + "step": 706 + }, + { + "epoch": 1.04, + "grad_norm": 4.5666656494140625, + "learning_rate": 0.00019503000023264136, + "loss": 3.4976, + "step": 707 + }, + { + "epoch": 1.04, + "grad_norm": 6.2709221839904785, + "learning_rate": 0.0001950153870646865, + "loss": 3.5246, + "step": 708 + }, + { + "epoch": 1.04, + "grad_norm": 8.397104263305664, + "learning_rate": 0.00019500075299398324, + "loss": 3.5369, + "step": 709 + }, + { + "epoch": 1.04, + "grad_norm": 4.958198547363281, + "learning_rate": 0.00019498609802375097, + "loss": 3.4644, + "step": 710 + }, + { + "epoch": 1.04, + "grad_norm": 6.521581172943115, + "learning_rate": 0.00019497142215721373, + "loss": 3.4931, + "step": 711 + }, + { + "epoch": 1.04, + "grad_norm": 10.69038200378418, + "learning_rate": 0.00019495672539760007, + "loss": 3.4746, + "step": 712 + }, + { + "epoch": 1.04, + "grad_norm": 5.08820915222168, + "learning_rate": 0.0001949420077481432, + "loss": 3.4963, + "step": 713 + }, + { + "epoch": 1.05, + "grad_norm": 5.867018222808838, + "learning_rate": 0.00019492726921208092, + "loss": 3.4735, + "step": 714 + }, + { + "epoch": 1.05, + "grad_norm": 7.802560329437256, + "learning_rate": 0.00019491250979265554, + "loss": 3.514, + "step": 715 + }, + { + "epoch": 1.05, + "grad_norm": 4.548806190490723, + "learning_rate": 0.0001948977294931141, + "loss": 3.4695, + "step": 716 + }, + { + "epoch": 1.05, + "grad_norm": 7.293185710906982, + "learning_rate": 0.00019488292831670818, + "loss": 3.5301, + "step": 717 + }, + { + "epoch": 1.05, + "grad_norm": 10.711014747619629, + "learning_rate": 0.00019486810626669388, + "loss": 3.5166, + "step": 718 + }, + { + "epoch": 1.05, + "grad_norm": 5.207925796508789, + "learning_rate": 0.00019485326334633198, + "loss": 3.4741, + "step": 719 + }, + { + "epoch": 1.05, + "grad_norm": 5.915352821350098, + "learning_rate": 0.00019483839955888784, + "loss": 3.4837, + "step": 720 + }, + { + "epoch": 1.06, + "grad_norm": 8.018194198608398, + "learning_rate": 0.00019482351490763135, + "loss": 3.4991, + "step": 721 + }, + { + "epoch": 1.06, + "grad_norm": 4.433293342590332, + "learning_rate": 0.00019480860939583705, + "loss": 3.417, + "step": 722 + }, + { + "epoch": 1.06, + "grad_norm": 6.9495038986206055, + "learning_rate": 0.0001947936830267841, + "loss": 3.5061, + "step": 723 + }, + { + "epoch": 1.06, + "grad_norm": 10.787668228149414, + "learning_rate": 0.00019477873580375614, + "loss": 3.5036, + "step": 724 + }, + { + "epoch": 1.06, + "grad_norm": 5.106097221374512, + "learning_rate": 0.0001947637677300415, + "loss": 3.4732, + "step": 725 + }, + { + "epoch": 1.06, + "grad_norm": 5.890613555908203, + "learning_rate": 0.00019474877880893302, + "loss": 3.5222, + "step": 726 + }, + { + "epoch": 1.06, + "grad_norm": 7.86448335647583, + "learning_rate": 0.0001947337690437282, + "loss": 3.4886, + "step": 727 + }, + { + "epoch": 1.07, + "grad_norm": 4.345864772796631, + "learning_rate": 0.00019471873843772904, + "loss": 3.4902, + "step": 728 + }, + { + "epoch": 1.07, + "grad_norm": 7.271783351898193, + "learning_rate": 0.00019470368699424218, + "loss": 3.4832, + "step": 729 + }, + { + "epoch": 1.07, + "grad_norm": 11.227890968322754, + "learning_rate": 0.0001946886147165789, + "loss": 3.4815, + "step": 730 + }, + { + "epoch": 1.07, + "grad_norm": 5.0216264724731445, + "learning_rate": 0.00019467352160805498, + "loss": 3.4912, + "step": 731 + }, + { + "epoch": 1.07, + "grad_norm": 5.998202800750732, + "learning_rate": 0.00019465840767199073, + "loss": 3.4653, + "step": 732 + }, + { + "epoch": 1.07, + "grad_norm": 7.844666004180908, + "learning_rate": 0.00019464327291171117, + "loss": 3.5708, + "step": 733 + }, + { + "epoch": 1.08, + "grad_norm": 4.441153526306152, + "learning_rate": 0.00019462811733054588, + "loss": 3.479, + "step": 734 + }, + { + "epoch": 1.08, + "grad_norm": 7.58824348449707, + "learning_rate": 0.0001946129409318289, + "loss": 3.4787, + "step": 735 + }, + { + "epoch": 1.08, + "grad_norm": 11.293692588806152, + "learning_rate": 0.00019459774371889901, + "loss": 3.5108, + "step": 736 + }, + { + "epoch": 1.08, + "grad_norm": 6.113391399383545, + "learning_rate": 0.00019458252569509946, + "loss": 3.4456, + "step": 737 + }, + { + "epoch": 1.08, + "grad_norm": 4.897597312927246, + "learning_rate": 0.00019456728686377813, + "loss": 3.5001, + "step": 738 + }, + { + "epoch": 1.08, + "grad_norm": 6.808078765869141, + "learning_rate": 0.00019455202722828743, + "loss": 3.5318, + "step": 739 + }, + { + "epoch": 1.08, + "grad_norm": 2.7436025142669678, + "learning_rate": 0.0001945367467919844, + "loss": 3.452, + "step": 740 + }, + { + "epoch": 1.09, + "grad_norm": 10.119528770446777, + "learning_rate": 0.00019452144555823065, + "loss": 3.5012, + "step": 741 + }, + { + "epoch": 1.09, + "grad_norm": 15.063286781311035, + "learning_rate": 0.00019450612353039235, + "loss": 3.5615, + "step": 742 + }, + { + "epoch": 1.09, + "grad_norm": 9.542817115783691, + "learning_rate": 0.0001944907807118402, + "loss": 3.5215, + "step": 743 + }, + { + "epoch": 1.09, + "grad_norm": 3.414539098739624, + "learning_rate": 0.00019447541710594952, + "loss": 3.4854, + "step": 744 + }, + { + "epoch": 1.09, + "grad_norm": 5.853447437286377, + "learning_rate": 0.00019446003271610026, + "loss": 3.4727, + "step": 745 + }, + { + "epoch": 1.09, + "grad_norm": 2.5720930099487305, + "learning_rate": 0.00019444462754567682, + "loss": 3.4853, + "step": 746 + }, + { + "epoch": 1.09, + "grad_norm": 8.554189682006836, + "learning_rate": 0.0001944292015980683, + "loss": 3.4712, + "step": 747 + }, + { + "epoch": 1.1, + "grad_norm": 11.265769958496094, + "learning_rate": 0.00019441375487666822, + "loss": 3.5236, + "step": 748 + }, + { + "epoch": 1.1, + "grad_norm": 4.981095314025879, + "learning_rate": 0.00019439828738487483, + "loss": 3.5258, + "step": 749 + }, + { + "epoch": 1.1, + "grad_norm": 5.837092876434326, + "learning_rate": 0.00019438279912609082, + "loss": 3.5503, + "step": 750 + }, + { + "epoch": 1.1, + "grad_norm": 8.067007064819336, + "learning_rate": 0.00019436729010372355, + "loss": 3.5129, + "step": 751 + }, + { + "epoch": 1.1, + "grad_norm": 4.977865219116211, + "learning_rate": 0.00019435176032118486, + "loss": 3.4686, + "step": 752 + }, + { + "epoch": 1.1, + "grad_norm": 6.810112953186035, + "learning_rate": 0.00019433620978189122, + "loss": 3.4424, + "step": 753 + }, + { + "epoch": 1.1, + "grad_norm": 10.288273811340332, + "learning_rate": 0.00019432063848926365, + "loss": 3.4829, + "step": 754 + }, + { + "epoch": 1.11, + "grad_norm": 4.5541839599609375, + "learning_rate": 0.0001943050464467277, + "loss": 3.4592, + "step": 755 + }, + { + "epoch": 1.11, + "grad_norm": 5.883044719696045, + "learning_rate": 0.0001942894336577136, + "loss": 3.5044, + "step": 756 + }, + { + "epoch": 1.11, + "grad_norm": 7.786341667175293, + "learning_rate": 0.00019427380012565593, + "loss": 3.5129, + "step": 757 + }, + { + "epoch": 1.11, + "grad_norm": 4.628783226013184, + "learning_rate": 0.00019425814585399405, + "loss": 3.43, + "step": 758 + }, + { + "epoch": 1.11, + "grad_norm": 6.876189708709717, + "learning_rate": 0.00019424247084617177, + "loss": 3.4606, + "step": 759 + }, + { + "epoch": 1.11, + "grad_norm": 10.343745231628418, + "learning_rate": 0.00019422677510563748, + "loss": 3.5079, + "step": 760 + }, + { + "epoch": 1.12, + "grad_norm": 4.678600311279297, + "learning_rate": 0.00019421105863584413, + "loss": 3.5135, + "step": 761 + }, + { + "epoch": 1.12, + "grad_norm": 6.044957160949707, + "learning_rate": 0.0001941953214402493, + "loss": 3.5067, + "step": 762 + }, + { + "epoch": 1.12, + "grad_norm": 7.979968070983887, + "learning_rate": 0.00019417956352231502, + "loss": 3.4955, + "step": 763 + }, + { + "epoch": 1.12, + "grad_norm": 4.172381401062012, + "learning_rate": 0.00019416378488550787, + "loss": 3.4781, + "step": 764 + }, + { + "epoch": 1.12, + "grad_norm": 7.783565998077393, + "learning_rate": 0.00019414798553329916, + "loss": 3.4649, + "step": 765 + }, + { + "epoch": 1.12, + "grad_norm": 11.02284049987793, + "learning_rate": 0.00019413216546916457, + "loss": 3.5242, + "step": 766 + }, + { + "epoch": 1.12, + "grad_norm": 5.28484582901001, + "learning_rate": 0.00019411632469658443, + "loss": 3.4645, + "step": 767 + }, + { + "epoch": 1.13, + "grad_norm": 6.049994945526123, + "learning_rate": 0.00019410046321904356, + "loss": 3.4601, + "step": 768 + }, + { + "epoch": 1.13, + "grad_norm": 7.867702960968018, + "learning_rate": 0.00019408458104003143, + "loss": 3.4853, + "step": 769 + }, + { + "epoch": 1.13, + "grad_norm": 5.088557720184326, + "learning_rate": 0.00019406867816304204, + "loss": 3.5329, + "step": 770 + }, + { + "epoch": 1.13, + "grad_norm": 6.444199085235596, + "learning_rate": 0.00019405275459157383, + "loss": 3.4624, + "step": 771 + }, + { + "epoch": 1.13, + "grad_norm": 9.823208808898926, + "learning_rate": 0.00019403681032912994, + "loss": 3.4496, + "step": 772 + }, + { + "epoch": 1.13, + "grad_norm": 4.309892654418945, + "learning_rate": 0.00019402084537921798, + "loss": 3.4607, + "step": 773 + }, + { + "epoch": 1.13, + "grad_norm": 6.139223575592041, + "learning_rate": 0.00019400485974535014, + "loss": 3.4246, + "step": 774 + }, + { + "epoch": 1.14, + "grad_norm": 8.018882751464844, + "learning_rate": 0.00019398885343104315, + "loss": 3.4731, + "step": 775 + }, + { + "epoch": 1.14, + "grad_norm": 4.7042555809021, + "learning_rate": 0.00019397282643981826, + "loss": 3.4863, + "step": 776 + }, + { + "epoch": 1.14, + "grad_norm": 6.583994388580322, + "learning_rate": 0.00019395677877520137, + "loss": 3.4354, + "step": 777 + }, + { + "epoch": 1.14, + "grad_norm": 9.998130798339844, + "learning_rate": 0.00019394071044072283, + "loss": 3.4882, + "step": 778 + }, + { + "epoch": 1.14, + "grad_norm": 4.592165946960449, + "learning_rate": 0.00019392462143991753, + "loss": 3.4696, + "step": 779 + }, + { + "epoch": 1.14, + "grad_norm": 6.2591681480407715, + "learning_rate": 0.00019390851177632497, + "loss": 3.4393, + "step": 780 + }, + { + "epoch": 1.14, + "grad_norm": 8.170223236083984, + "learning_rate": 0.00019389238145348916, + "loss": 3.421, + "step": 781 + }, + { + "epoch": 1.15, + "grad_norm": 5.1918559074401855, + "learning_rate": 0.00019387623047495867, + "loss": 3.452, + "step": 782 + }, + { + "epoch": 1.15, + "grad_norm": 6.183322429656982, + "learning_rate": 0.0001938600588442866, + "loss": 3.5035, + "step": 783 + }, + { + "epoch": 1.15, + "grad_norm": 9.819025039672852, + "learning_rate": 0.00019384386656503064, + "loss": 3.4897, + "step": 784 + }, + { + "epoch": 1.15, + "grad_norm": 4.417552471160889, + "learning_rate": 0.00019382765364075292, + "loss": 3.5126, + "step": 785 + }, + { + "epoch": 1.15, + "grad_norm": 6.277004241943359, + "learning_rate": 0.0001938114200750202, + "loss": 3.4848, + "step": 786 + }, + { + "epoch": 1.15, + "grad_norm": 8.565668106079102, + "learning_rate": 0.0001937951658714038, + "loss": 3.5328, + "step": 787 + }, + { + "epoch": 1.16, + "grad_norm": 5.04703426361084, + "learning_rate": 0.00019377889103347945, + "loss": 3.4752, + "step": 788 + }, + { + "epoch": 1.16, + "grad_norm": 5.934195518493652, + "learning_rate": 0.0001937625955648276, + "loss": 3.5186, + "step": 789 + }, + { + "epoch": 1.16, + "grad_norm": 9.426629066467285, + "learning_rate": 0.00019374627946903305, + "loss": 3.4537, + "step": 790 + }, + { + "epoch": 1.16, + "grad_norm": 3.8387088775634766, + "learning_rate": 0.00019372994274968527, + "loss": 3.4172, + "step": 791 + }, + { + "epoch": 1.16, + "grad_norm": 6.488461971282959, + "learning_rate": 0.00019371358541037825, + "loss": 3.5177, + "step": 792 + }, + { + "epoch": 1.16, + "grad_norm": 8.451127052307129, + "learning_rate": 0.0001936972074547105, + "loss": 3.4701, + "step": 793 + }, + { + "epoch": 1.16, + "grad_norm": 5.298417568206787, + "learning_rate": 0.00019368080888628502, + "loss": 3.4579, + "step": 794 + }, + { + "epoch": 1.17, + "grad_norm": 6.457042694091797, + "learning_rate": 0.00019366438970870942, + "loss": 3.4666, + "step": 795 + }, + { + "epoch": 1.17, + "grad_norm": 9.55510139465332, + "learning_rate": 0.00019364794992559578, + "loss": 3.4449, + "step": 796 + }, + { + "epoch": 1.17, + "grad_norm": 3.862884998321533, + "learning_rate": 0.00019363148954056077, + "loss": 3.4185, + "step": 797 + }, + { + "epoch": 1.17, + "grad_norm": 6.591372966766357, + "learning_rate": 0.00019361500855722555, + "loss": 3.537, + "step": 798 + }, + { + "epoch": 1.17, + "grad_norm": 8.397138595581055, + "learning_rate": 0.0001935985069792158, + "loss": 3.4205, + "step": 799 + }, + { + "epoch": 1.17, + "grad_norm": 5.131923675537109, + "learning_rate": 0.00019358198481016182, + "loss": 3.4916, + "step": 800 + }, + { + "epoch": 1.17, + "grad_norm": 5.620177745819092, + "learning_rate": 0.00019356544205369834, + "loss": 3.4604, + "step": 801 + }, + { + "epoch": 1.18, + "grad_norm": 9.219866752624512, + "learning_rate": 0.00019354887871346463, + "loss": 3.467, + "step": 802 + }, + { + "epoch": 1.18, + "grad_norm": 4.237349033355713, + "learning_rate": 0.00019353229479310456, + "loss": 3.4847, + "step": 803 + }, + { + "epoch": 1.18, + "grad_norm": 6.673110485076904, + "learning_rate": 0.00019351569029626644, + "loss": 3.4412, + "step": 804 + }, + { + "epoch": 1.18, + "grad_norm": 8.31098747253418, + "learning_rate": 0.0001934990652266032, + "loss": 3.4838, + "step": 805 + }, + { + "epoch": 1.18, + "grad_norm": 5.009383201599121, + "learning_rate": 0.0001934824195877722, + "loss": 3.4408, + "step": 806 + }, + { + "epoch": 1.18, + "grad_norm": 5.483675956726074, + "learning_rate": 0.00019346575338343537, + "loss": 3.4663, + "step": 807 + }, + { + "epoch": 1.19, + "grad_norm": 9.371169090270996, + "learning_rate": 0.00019344906661725918, + "loss": 3.5116, + "step": 808 + }, + { + "epoch": 1.19, + "grad_norm": 3.6695902347564697, + "learning_rate": 0.00019343235929291462, + "loss": 3.4776, + "step": 809 + }, + { + "epoch": 1.19, + "grad_norm": 6.492842197418213, + "learning_rate": 0.00019341563141407714, + "loss": 3.4725, + "step": 810 + }, + { + "epoch": 1.19, + "grad_norm": 8.520849227905273, + "learning_rate": 0.00019339888298442682, + "loss": 3.4845, + "step": 811 + }, + { + "epoch": 1.19, + "grad_norm": 4.986138820648193, + "learning_rate": 0.00019338211400764816, + "loss": 3.4313, + "step": 812 + }, + { + "epoch": 1.19, + "grad_norm": 6.272872447967529, + "learning_rate": 0.00019336532448743023, + "loss": 3.4337, + "step": 813 + }, + { + "epoch": 1.19, + "grad_norm": 10.374552726745605, + "learning_rate": 0.00019334851442746664, + "loss": 3.4586, + "step": 814 + }, + { + "epoch": 1.2, + "grad_norm": 4.865425109863281, + "learning_rate": 0.00019333168383145547, + "loss": 3.5045, + "step": 815 + }, + { + "epoch": 1.2, + "grad_norm": 6.018640995025635, + "learning_rate": 0.0001933148327030993, + "loss": 3.4509, + "step": 816 + }, + { + "epoch": 1.2, + "grad_norm": 8.015512466430664, + "learning_rate": 0.00019329796104610537, + "loss": 3.4657, + "step": 817 + }, + { + "epoch": 1.2, + "grad_norm": 4.071451187133789, + "learning_rate": 0.00019328106886418522, + "loss": 3.4079, + "step": 818 + }, + { + "epoch": 1.2, + "grad_norm": 7.4237823486328125, + "learning_rate": 0.00019326415616105507, + "loss": 3.4906, + "step": 819 + }, + { + "epoch": 1.2, + "grad_norm": 10.776564598083496, + "learning_rate": 0.00019324722294043558, + "loss": 3.4639, + "step": 820 + }, + { + "epoch": 1.2, + "grad_norm": 5.143712997436523, + "learning_rate": 0.00019323026920605197, + "loss": 3.4356, + "step": 821 + }, + { + "epoch": 1.21, + "grad_norm": 5.7633891105651855, + "learning_rate": 0.00019321329496163397, + "loss": 3.428, + "step": 822 + }, + { + "epoch": 1.21, + "grad_norm": 7.827281475067139, + "learning_rate": 0.0001931963002109157, + "loss": 3.4913, + "step": 823 + }, + { + "epoch": 1.21, + "grad_norm": 4.470641136169434, + "learning_rate": 0.00019317928495763602, + "loss": 3.4441, + "step": 824 + }, + { + "epoch": 1.21, + "grad_norm": 7.5875091552734375, + "learning_rate": 0.00019316224920553807, + "loss": 3.468, + "step": 825 + }, + { + "epoch": 1.21, + "grad_norm": 10.94041919708252, + "learning_rate": 0.00019314519295836964, + "loss": 3.4932, + "step": 826 + }, + { + "epoch": 1.21, + "grad_norm": 5.624682426452637, + "learning_rate": 0.000193128116219883, + "loss": 3.4411, + "step": 827 + }, + { + "epoch": 1.21, + "grad_norm": 5.68918514251709, + "learning_rate": 0.00019311101899383492, + "loss": 3.4896, + "step": 828 + }, + { + "epoch": 1.22, + "grad_norm": 7.500881671905518, + "learning_rate": 0.00019309390128398665, + "loss": 3.4429, + "step": 829 + }, + { + "epoch": 1.22, + "grad_norm": 4.407763481140137, + "learning_rate": 0.000193076763094104, + "loss": 3.4694, + "step": 830 + }, + { + "epoch": 1.22, + "grad_norm": 7.296525955200195, + "learning_rate": 0.0001930596044279572, + "loss": 3.4065, + "step": 831 + }, + { + "epoch": 1.22, + "grad_norm": 11.3180513381958, + "learning_rate": 0.00019304242528932113, + "loss": 3.5025, + "step": 832 + }, + { + "epoch": 1.22, + "grad_norm": 5.404650688171387, + "learning_rate": 0.00019302522568197502, + "loss": 3.4129, + "step": 833 + }, + { + "epoch": 1.22, + "grad_norm": 5.595582485198975, + "learning_rate": 0.0001930080056097027, + "loss": 3.5212, + "step": 834 + }, + { + "epoch": 1.23, + "grad_norm": 7.934787273406982, + "learning_rate": 0.00019299076507629245, + "loss": 3.5551, + "step": 835 + }, + { + "epoch": 1.23, + "grad_norm": 4.486216068267822, + "learning_rate": 0.00019297350408553708, + "loss": 3.441, + "step": 836 + }, + { + "epoch": 1.23, + "grad_norm": 7.029922962188721, + "learning_rate": 0.0001929562226412339, + "loss": 3.4982, + "step": 837 + }, + { + "epoch": 1.23, + "grad_norm": 10.693680763244629, + "learning_rate": 0.0001929389207471847, + "loss": 3.4494, + "step": 838 + }, + { + "epoch": 1.23, + "grad_norm": 4.879317760467529, + "learning_rate": 0.00019292159840719576, + "loss": 3.4357, + "step": 839 + }, + { + "epoch": 1.23, + "grad_norm": 6.057565212249756, + "learning_rate": 0.00019290425562507792, + "loss": 3.4738, + "step": 840 + }, + { + "epoch": 1.23, + "grad_norm": 7.925189018249512, + "learning_rate": 0.0001928868924046465, + "loss": 3.4606, + "step": 841 + }, + { + "epoch": 1.24, + "grad_norm": 4.621508598327637, + "learning_rate": 0.0001928695087497212, + "loss": 3.4088, + "step": 842 + }, + { + "epoch": 1.24, + "grad_norm": 6.142604351043701, + "learning_rate": 0.00019285210466412635, + "loss": 3.4484, + "step": 843 + }, + { + "epoch": 1.24, + "grad_norm": 10.20028305053711, + "learning_rate": 0.00019283468015169076, + "loss": 3.5059, + "step": 844 + }, + { + "epoch": 1.24, + "grad_norm": 4.734007358551025, + "learning_rate": 0.0001928172352162477, + "loss": 3.4299, + "step": 845 + }, + { + "epoch": 1.24, + "eval_loss": 3.2429304122924805, + "eval_runtime": 1163.5436, + "eval_samples_per_second": 14.733, + "eval_steps_per_second": 0.921, + "step": 845 + }, + { + "epoch": 1.24, + "grad_norm": 6.036166191101074, + "learning_rate": 0.00019279976986163493, + "loss": 3.3913, + "step": 846 + }, + { + "epoch": 1.24, + "grad_norm": 8.014019012451172, + "learning_rate": 0.00019278228409169468, + "loss": 3.4981, + "step": 847 + }, + { + "epoch": 1.24, + "grad_norm": 4.856557369232178, + "learning_rate": 0.00019276477791027374, + "loss": 3.4562, + "step": 848 + }, + { + "epoch": 1.25, + "grad_norm": 6.650398254394531, + "learning_rate": 0.00019274725132122336, + "loss": 3.4407, + "step": 849 + }, + { + "epoch": 1.25, + "grad_norm": 10.177898406982422, + "learning_rate": 0.00019272970432839922, + "loss": 3.4437, + "step": 850 + }, + { + "epoch": 1.25, + "grad_norm": 4.559935569763184, + "learning_rate": 0.0001927121369356616, + "loss": 3.418, + "step": 851 + }, + { + "epoch": 1.25, + "grad_norm": 5.8679423332214355, + "learning_rate": 0.00019269454914687517, + "loss": 3.4338, + "step": 852 + }, + { + "epoch": 1.25, + "grad_norm": 8.044879913330078, + "learning_rate": 0.00019267694096590914, + "loss": 3.4146, + "step": 853 + }, + { + "epoch": 1.25, + "grad_norm": 4.654249668121338, + "learning_rate": 0.0001926593123966372, + "loss": 3.4353, + "step": 854 + }, + { + "epoch": 1.25, + "grad_norm": 6.326594352722168, + "learning_rate": 0.0001926416634429375, + "loss": 3.447, + "step": 855 + }, + { + "epoch": 1.26, + "grad_norm": 10.194587707519531, + "learning_rate": 0.0001926239941086927, + "loss": 3.4594, + "step": 856 + }, + { + "epoch": 1.26, + "grad_norm": 4.672416687011719, + "learning_rate": 0.00019260630439778994, + "loss": 3.4292, + "step": 857 + }, + { + "epoch": 1.26, + "grad_norm": 5.656294822692871, + "learning_rate": 0.00019258859431412083, + "loss": 3.4478, + "step": 858 + }, + { + "epoch": 1.26, + "grad_norm": 8.051904678344727, + "learning_rate": 0.00019257086386158148, + "loss": 3.4809, + "step": 859 + }, + { + "epoch": 1.26, + "grad_norm": 4.461113452911377, + "learning_rate": 0.00019255311304407243, + "loss": 3.4361, + "step": 860 + }, + { + "epoch": 1.26, + "grad_norm": 6.34873104095459, + "learning_rate": 0.00019253534186549883, + "loss": 3.424, + "step": 861 + }, + { + "epoch": 1.27, + "grad_norm": 10.255434036254883, + "learning_rate": 0.00019251755032977013, + "loss": 3.4578, + "step": 862 + }, + { + "epoch": 1.27, + "grad_norm": 4.895047187805176, + "learning_rate": 0.00019249973844080037, + "loss": 3.4606, + "step": 863 + }, + { + "epoch": 1.27, + "grad_norm": 6.042967319488525, + "learning_rate": 0.00019248190620250807, + "loss": 3.4398, + "step": 864 + }, + { + "epoch": 1.27, + "grad_norm": 7.916470050811768, + "learning_rate": 0.00019246405361881622, + "loss": 3.4921, + "step": 865 + }, + { + "epoch": 1.27, + "grad_norm": 4.985673904418945, + "learning_rate": 0.0001924461806936522, + "loss": 3.4425, + "step": 866 + }, + { + "epoch": 1.27, + "grad_norm": 6.128829002380371, + "learning_rate": 0.000192428287430948, + "loss": 3.4719, + "step": 867 + }, + { + "epoch": 1.27, + "grad_norm": 9.511234283447266, + "learning_rate": 0.00019241037383463998, + "loss": 3.4018, + "step": 868 + }, + { + "epoch": 1.28, + "grad_norm": 3.816689968109131, + "learning_rate": 0.000192392439908669, + "loss": 3.375, + "step": 869 + }, + { + "epoch": 1.28, + "grad_norm": 6.579503059387207, + "learning_rate": 0.00019237448565698046, + "loss": 3.4039, + "step": 870 + }, + { + "epoch": 1.28, + "grad_norm": 8.701712608337402, + "learning_rate": 0.00019235651108352416, + "loss": 3.3992, + "step": 871 + }, + { + "epoch": 1.28, + "grad_norm": 5.65900993347168, + "learning_rate": 0.00019233851619225432, + "loss": 3.4225, + "step": 872 + }, + { + "epoch": 1.28, + "grad_norm": 4.935523509979248, + "learning_rate": 0.00019232050098712974, + "loss": 3.4198, + "step": 873 + }, + { + "epoch": 1.28, + "grad_norm": 8.355915069580078, + "learning_rate": 0.00019230246547211367, + "loss": 3.4283, + "step": 874 + }, + { + "epoch": 1.28, + "grad_norm": 3.1533217430114746, + "learning_rate": 0.0001922844096511738, + "loss": 3.4058, + "step": 875 + }, + { + "epoch": 1.29, + "grad_norm": 7.010662078857422, + "learning_rate": 0.00019226633352828224, + "loss": 3.3914, + "step": 876 + }, + { + "epoch": 1.29, + "grad_norm": 8.548685073852539, + "learning_rate": 0.00019224823710741569, + "loss": 3.5008, + "step": 877 + }, + { + "epoch": 1.29, + "grad_norm": 5.439549446105957, + "learning_rate": 0.00019223012039255516, + "loss": 3.4421, + "step": 878 + }, + { + "epoch": 1.29, + "grad_norm": 5.446386814117432, + "learning_rate": 0.00019221198338768622, + "loss": 3.4071, + "step": 879 + }, + { + "epoch": 1.29, + "grad_norm": 9.378525733947754, + "learning_rate": 0.00019219382609679894, + "loss": 3.409, + "step": 880 + }, + { + "epoch": 1.29, + "grad_norm": 4.573349475860596, + "learning_rate": 0.0001921756485238878, + "loss": 3.4156, + "step": 881 + }, + { + "epoch": 1.29, + "grad_norm": 5.857920169830322, + "learning_rate": 0.00019215745067295169, + "loss": 3.433, + "step": 882 + }, + { + "epoch": 1.3, + "grad_norm": 7.654170036315918, + "learning_rate": 0.00019213923254799402, + "loss": 3.4011, + "step": 883 + }, + { + "epoch": 1.3, + "grad_norm": 4.262071132659912, + "learning_rate": 0.0001921209941530227, + "loss": 3.401, + "step": 884 + }, + { + "epoch": 1.3, + "grad_norm": 7.949652194976807, + "learning_rate": 0.00019210273549205003, + "loss": 3.5334, + "step": 885 + }, + { + "epoch": 1.3, + "grad_norm": 11.388256072998047, + "learning_rate": 0.0001920844565690928, + "loss": 3.4325, + "step": 886 + }, + { + "epoch": 1.3, + "grad_norm": 6.005974292755127, + "learning_rate": 0.00019206615738817225, + "loss": 3.4264, + "step": 887 + }, + { + "epoch": 1.3, + "grad_norm": 5.020950794219971, + "learning_rate": 0.0001920478379533141, + "loss": 3.426, + "step": 888 + }, + { + "epoch": 1.31, + "grad_norm": 6.8441386222839355, + "learning_rate": 0.00019202949826854844, + "loss": 3.4234, + "step": 889 + }, + { + "epoch": 1.31, + "grad_norm": 3.4975130558013916, + "learning_rate": 0.0001920111383379099, + "loss": 3.4667, + "step": 890 + }, + { + "epoch": 1.31, + "grad_norm": 7.94019889831543, + "learning_rate": 0.00019199275816543762, + "loss": 3.4208, + "step": 891 + }, + { + "epoch": 1.31, + "grad_norm": 11.041260719299316, + "learning_rate": 0.00019197435775517501, + "loss": 3.4392, + "step": 892 + }, + { + "epoch": 1.31, + "grad_norm": 5.074487209320068, + "learning_rate": 0.0001919559371111701, + "loss": 3.3906, + "step": 893 + }, + { + "epoch": 1.31, + "grad_norm": 5.909255027770996, + "learning_rate": 0.00019193749623747525, + "loss": 3.4508, + "step": 894 + }, + { + "epoch": 1.31, + "grad_norm": 7.794407367706299, + "learning_rate": 0.0001919190351381474, + "loss": 3.3957, + "step": 895 + }, + { + "epoch": 1.32, + "grad_norm": 4.833076000213623, + "learning_rate": 0.00019190055381724781, + "loss": 3.403, + "step": 896 + }, + { + "epoch": 1.32, + "grad_norm": 6.273806095123291, + "learning_rate": 0.0001918820522788423, + "loss": 3.4114, + "step": 897 + }, + { + "epoch": 1.32, + "grad_norm": 9.626479148864746, + "learning_rate": 0.00019186353052700104, + "loss": 3.4555, + "step": 898 + }, + { + "epoch": 1.32, + "grad_norm": 4.462099075317383, + "learning_rate": 0.00019184498856579868, + "loss": 3.4001, + "step": 899 + }, + { + "epoch": 1.32, + "grad_norm": 5.839742660522461, + "learning_rate": 0.00019182642639931438, + "loss": 3.4041, + "step": 900 + }, + { + "epoch": 1.32, + "grad_norm": 7.5265583992004395, + "learning_rate": 0.00019180784403163167, + "loss": 3.3898, + "step": 901 + }, + { + "epoch": 1.32, + "grad_norm": 4.089066028594971, + "learning_rate": 0.0001917892414668385, + "loss": 3.4162, + "step": 902 + }, + { + "epoch": 1.33, + "grad_norm": 7.329177379608154, + "learning_rate": 0.0001917706187090274, + "loss": 3.5439, + "step": 903 + }, + { + "epoch": 1.33, + "grad_norm": 10.022590637207031, + "learning_rate": 0.0001917519757622952, + "loss": 3.4009, + "step": 904 + }, + { + "epoch": 1.33, + "grad_norm": 5.030527591705322, + "learning_rate": 0.00019173331263074324, + "loss": 3.4314, + "step": 905 + }, + { + "epoch": 1.33, + "grad_norm": 5.504607677459717, + "learning_rate": 0.00019171462931847724, + "loss": 3.4463, + "step": 906 + }, + { + "epoch": 1.33, + "grad_norm": 7.219310283660889, + "learning_rate": 0.00019169592582960742, + "loss": 3.4878, + "step": 907 + }, + { + "epoch": 1.33, + "grad_norm": 4.046157360076904, + "learning_rate": 0.00019167720216824847, + "loss": 3.3643, + "step": 908 + }, + { + "epoch": 1.34, + "grad_norm": 7.463158130645752, + "learning_rate": 0.00019165845833851941, + "loss": 3.472, + "step": 909 + }, + { + "epoch": 1.34, + "grad_norm": 9.688533782958984, + "learning_rate": 0.00019163969434454382, + "loss": 3.4541, + "step": 910 + }, + { + "epoch": 1.34, + "grad_norm": 4.718085765838623, + "learning_rate": 0.00019162091019044958, + "loss": 3.4252, + "step": 911 + }, + { + "epoch": 1.34, + "grad_norm": 6.263822555541992, + "learning_rate": 0.00019160210588036911, + "loss": 3.4105, + "step": 912 + }, + { + "epoch": 1.34, + "grad_norm": 7.928195953369141, + "learning_rate": 0.00019158328141843928, + "loss": 3.4249, + "step": 913 + }, + { + "epoch": 1.34, + "grad_norm": 4.594407081604004, + "learning_rate": 0.00019156443680880126, + "loss": 3.453, + "step": 914 + }, + { + "epoch": 1.34, + "grad_norm": 6.126087188720703, + "learning_rate": 0.0001915455720556008, + "loss": 3.3944, + "step": 915 + }, + { + "epoch": 1.35, + "grad_norm": 9.558304786682129, + "learning_rate": 0.000191526687162988, + "loss": 3.3926, + "step": 916 + }, + { + "epoch": 1.35, + "grad_norm": 4.792506217956543, + "learning_rate": 0.0001915077821351174, + "loss": 3.4081, + "step": 917 + }, + { + "epoch": 1.35, + "grad_norm": 5.998870372772217, + "learning_rate": 0.00019148885697614797, + "loss": 3.4326, + "step": 918 + }, + { + "epoch": 1.35, + "grad_norm": 7.541780948638916, + "learning_rate": 0.00019146991169024315, + "loss": 3.4647, + "step": 919 + }, + { + "epoch": 1.35, + "grad_norm": 4.514527320861816, + "learning_rate": 0.00019145094628157078, + "loss": 3.3601, + "step": 920 + }, + { + "epoch": 1.35, + "grad_norm": 6.531849384307861, + "learning_rate": 0.0001914319607543031, + "loss": 3.4428, + "step": 921 + }, + { + "epoch": 1.35, + "grad_norm": 9.34991455078125, + "learning_rate": 0.00019141295511261678, + "loss": 3.393, + "step": 922 + }, + { + "epoch": 1.36, + "grad_norm": 4.035532474517822, + "learning_rate": 0.000191393929360693, + "loss": 3.3938, + "step": 923 + }, + { + "epoch": 1.36, + "grad_norm": 6.159792900085449, + "learning_rate": 0.0001913748835027172, + "loss": 3.4207, + "step": 924 + }, + { + "epoch": 1.36, + "grad_norm": 8.069183349609375, + "learning_rate": 0.00019135581754287944, + "loss": 3.4143, + "step": 925 + }, + { + "epoch": 1.36, + "grad_norm": 4.528735160827637, + "learning_rate": 0.00019133673148537405, + "loss": 3.4282, + "step": 926 + }, + { + "epoch": 1.36, + "grad_norm": 6.5554399490356445, + "learning_rate": 0.00019131762533439983, + "loss": 3.4149, + "step": 927 + }, + { + "epoch": 1.36, + "grad_norm": 9.77618408203125, + "learning_rate": 0.00019129849909416005, + "loss": 3.4509, + "step": 928 + }, + { + "epoch": 1.36, + "grad_norm": 4.406791687011719, + "learning_rate": 0.0001912793527688623, + "loss": 3.3955, + "step": 929 + }, + { + "epoch": 1.37, + "grad_norm": 6.1382622718811035, + "learning_rate": 0.00019126018636271867, + "loss": 3.4066, + "step": 930 + }, + { + "epoch": 1.37, + "grad_norm": 8.133618354797363, + "learning_rate": 0.00019124099987994565, + "loss": 3.4536, + "step": 931 + }, + { + "epoch": 1.37, + "grad_norm": 4.887189865112305, + "learning_rate": 0.0001912217933247641, + "loss": 3.4516, + "step": 932 + }, + { + "epoch": 1.37, + "grad_norm": 5.946772575378418, + "learning_rate": 0.0001912025667013994, + "loss": 3.3842, + "step": 933 + }, + { + "epoch": 1.37, + "grad_norm": 9.542755126953125, + "learning_rate": 0.00019118332001408122, + "loss": 3.3833, + "step": 934 + }, + { + "epoch": 1.37, + "grad_norm": 4.512613296508789, + "learning_rate": 0.00019116405326704373, + "loss": 3.3754, + "step": 935 + }, + { + "epoch": 1.38, + "grad_norm": 6.039595603942871, + "learning_rate": 0.00019114476646452547, + "loss": 3.4341, + "step": 936 + }, + { + "epoch": 1.38, + "grad_norm": 7.820281982421875, + "learning_rate": 0.0001911254596107694, + "loss": 3.3616, + "step": 937 + }, + { + "epoch": 1.38, + "grad_norm": 4.0570759773254395, + "learning_rate": 0.0001911061327100229, + "loss": 3.3999, + "step": 938 + }, + { + "epoch": 1.38, + "grad_norm": 6.659790992736816, + "learning_rate": 0.0001910867857665378, + "loss": 3.361, + "step": 939 + }, + { + "epoch": 1.38, + "grad_norm": 10.188064575195312, + "learning_rate": 0.00019106741878457026, + "loss": 3.3805, + "step": 940 + }, + { + "epoch": 1.38, + "grad_norm": 4.66610860824585, + "learning_rate": 0.0001910480317683809, + "loss": 3.3881, + "step": 941 + }, + { + "epoch": 1.38, + "grad_norm": 5.840310573577881, + "learning_rate": 0.0001910286247222347, + "loss": 3.4162, + "step": 942 + }, + { + "epoch": 1.39, + "grad_norm": 7.710467338562012, + "learning_rate": 0.00019100919765040113, + "loss": 3.396, + "step": 943 + }, + { + "epoch": 1.39, + "grad_norm": 4.53704833984375, + "learning_rate": 0.00019098975055715397, + "loss": 3.3605, + "step": 944 + }, + { + "epoch": 1.39, + "grad_norm": 6.515756607055664, + "learning_rate": 0.00019097028344677153, + "loss": 3.4119, + "step": 945 + }, + { + "epoch": 1.39, + "grad_norm": 10.256503105163574, + "learning_rate": 0.00019095079632353633, + "loss": 3.4173, + "step": 946 + }, + { + "epoch": 1.39, + "grad_norm": 4.8957672119140625, + "learning_rate": 0.0001909312891917355, + "loss": 3.3812, + "step": 947 + }, + { + "epoch": 1.39, + "grad_norm": 5.8118205070495605, + "learning_rate": 0.00019091176205566043, + "loss": 3.3734, + "step": 948 + }, + { + "epoch": 1.39, + "grad_norm": 7.7631025314331055, + "learning_rate": 0.00019089221491960695, + "loss": 3.4171, + "step": 949 + }, + { + "epoch": 1.4, + "grad_norm": 4.2992377281188965, + "learning_rate": 0.00019087264778787534, + "loss": 3.3724, + "step": 950 + }, + { + "epoch": 1.4, + "grad_norm": 6.633007049560547, + "learning_rate": 0.00019085306066477021, + "loss": 3.4198, + "step": 951 + }, + { + "epoch": 1.4, + "grad_norm": 9.867729187011719, + "learning_rate": 0.00019083345355460062, + "loss": 3.4247, + "step": 952 + }, + { + "epoch": 1.4, + "grad_norm": 4.476417541503906, + "learning_rate": 0.00019081382646167998, + "loss": 3.3617, + "step": 953 + }, + { + "epoch": 1.4, + "grad_norm": 5.963509559631348, + "learning_rate": 0.00019079417939032612, + "loss": 3.3945, + "step": 954 + }, + { + "epoch": 1.4, + "grad_norm": 7.921098232269287, + "learning_rate": 0.00019077451234486128, + "loss": 3.4305, + "step": 955 + }, + { + "epoch": 1.4, + "grad_norm": 4.772178649902344, + "learning_rate": 0.00019075482532961204, + "loss": 3.3836, + "step": 956 + }, + { + "epoch": 1.41, + "grad_norm": 6.6379265785217285, + "learning_rate": 0.00019073511834890949, + "loss": 3.3646, + "step": 957 + }, + { + "epoch": 1.41, + "grad_norm": 9.883600234985352, + "learning_rate": 0.00019071539140708894, + "loss": 3.398, + "step": 958 + }, + { + "epoch": 1.41, + "grad_norm": 4.0690178871154785, + "learning_rate": 0.00019069564450849025, + "loss": 3.3394, + "step": 959 + }, + { + "epoch": 1.41, + "grad_norm": 6.045716762542725, + "learning_rate": 0.00019067587765745757, + "loss": 3.4695, + "step": 960 + }, + { + "epoch": 1.41, + "grad_norm": 7.893038272857666, + "learning_rate": 0.00019065609085833947, + "loss": 3.4336, + "step": 961 + }, + { + "epoch": 1.41, + "grad_norm": 4.4083075523376465, + "learning_rate": 0.00019063628411548897, + "loss": 3.3471, + "step": 962 + }, + { + "epoch": 1.42, + "grad_norm": 6.27316951751709, + "learning_rate": 0.00019061645743326337, + "loss": 3.4311, + "step": 963 + }, + { + "epoch": 1.42, + "grad_norm": 9.92757797241211, + "learning_rate": 0.00019059661081602443, + "loss": 3.4114, + "step": 964 + }, + { + "epoch": 1.42, + "grad_norm": 5.099198341369629, + "learning_rate": 0.00019057674426813822, + "loss": 3.4386, + "step": 965 + }, + { + "epoch": 1.42, + "grad_norm": 5.424035549163818, + "learning_rate": 0.00019055685779397532, + "loss": 3.4275, + "step": 966 + }, + { + "epoch": 1.42, + "grad_norm": 7.513932704925537, + "learning_rate": 0.0001905369513979106, + "loss": 3.4135, + "step": 967 + }, + { + "epoch": 1.42, + "grad_norm": 4.388814449310303, + "learning_rate": 0.0001905170250843233, + "loss": 3.3889, + "step": 968 + }, + { + "epoch": 1.42, + "grad_norm": 6.554171562194824, + "learning_rate": 0.00019049707885759713, + "loss": 3.4184, + "step": 969 + }, + { + "epoch": 1.43, + "grad_norm": 9.21511173248291, + "learning_rate": 0.00019047711272212006, + "loss": 3.393, + "step": 970 + }, + { + "epoch": 1.43, + "grad_norm": 4.280778408050537, + "learning_rate": 0.00019045712668228456, + "loss": 3.3808, + "step": 971 + }, + { + "epoch": 1.43, + "grad_norm": 5.782205581665039, + "learning_rate": 0.0001904371207424874, + "loss": 3.3168, + "step": 972 + }, + { + "epoch": 1.43, + "grad_norm": 7.35162878036499, + "learning_rate": 0.00019041709490712978, + "loss": 3.4373, + "step": 973 + }, + { + "epoch": 1.43, + "grad_norm": 3.4769937992095947, + "learning_rate": 0.0001903970491806172, + "loss": 3.4039, + "step": 974 + }, + { + "epoch": 1.43, + "grad_norm": 8.196279525756836, + "learning_rate": 0.00019037698356735963, + "loss": 3.38, + "step": 975 + }, + { + "epoch": 1.43, + "grad_norm": 11.59374713897705, + "learning_rate": 0.00019035689807177136, + "loss": 3.4377, + "step": 976 + }, + { + "epoch": 1.44, + "grad_norm": 6.463580131530762, + "learning_rate": 0.00019033679269827105, + "loss": 3.3706, + "step": 977 + }, + { + "epoch": 1.44, + "grad_norm": 4.297844409942627, + "learning_rate": 0.00019031666745128177, + "loss": 3.4047, + "step": 978 + }, + { + "epoch": 1.44, + "grad_norm": 6.652123928070068, + "learning_rate": 0.00019029652233523093, + "loss": 3.3968, + "step": 979 + }, + { + "epoch": 1.44, + "grad_norm": 3.399991750717163, + "learning_rate": 0.00019027635735455027, + "loss": 3.3552, + "step": 980 + }, + { + "epoch": 1.44, + "grad_norm": 7.520792007446289, + "learning_rate": 0.00019025617251367604, + "loss": 3.3979, + "step": 981 + }, + { + "epoch": 1.44, + "grad_norm": 10.621402740478516, + "learning_rate": 0.00019023596781704872, + "loss": 3.4131, + "step": 982 + }, + { + "epoch": 1.44, + "grad_norm": 5.002892971038818, + "learning_rate": 0.0001902157432691132, + "loss": 3.3583, + "step": 983 + }, + { + "epoch": 1.45, + "grad_norm": 5.647467136383057, + "learning_rate": 0.00019019549887431877, + "loss": 3.384, + "step": 984 + }, + { + "epoch": 1.45, + "grad_norm": 7.81645393371582, + "learning_rate": 0.00019017523463711906, + "loss": 3.4198, + "step": 985 + }, + { + "epoch": 1.45, + "grad_norm": 4.508808135986328, + "learning_rate": 0.00019015495056197207, + "loss": 3.4205, + "step": 986 + }, + { + "epoch": 1.45, + "grad_norm": 6.004973411560059, + "learning_rate": 0.00019013464665334012, + "loss": 3.396, + "step": 987 + }, + { + "epoch": 1.45, + "grad_norm": 9.286575317382812, + "learning_rate": 0.00019011432291568995, + "loss": 3.4053, + "step": 988 + }, + { + "epoch": 1.45, + "grad_norm": 4.02089786529541, + "learning_rate": 0.00019009397935349268, + "loss": 3.3216, + "step": 989 + }, + { + "epoch": 1.46, + "grad_norm": 6.502218246459961, + "learning_rate": 0.00019007361597122373, + "loss": 3.4367, + "step": 990 + }, + { + "epoch": 1.46, + "grad_norm": 8.069449424743652, + "learning_rate": 0.0001900532327733629, + "loss": 3.3666, + "step": 991 + }, + { + "epoch": 1.46, + "grad_norm": 4.8657917976379395, + "learning_rate": 0.0001900328297643944, + "loss": 3.3401, + "step": 992 + }, + { + "epoch": 1.46, + "grad_norm": 5.57826042175293, + "learning_rate": 0.00019001240694880672, + "loss": 3.393, + "step": 993 + }, + { + "epoch": 1.46, + "grad_norm": 8.596120834350586, + "learning_rate": 0.00018999196433109274, + "loss": 3.416, + "step": 994 + }, + { + "epoch": 1.46, + "grad_norm": 4.3653130531311035, + "learning_rate": 0.0001899715019157497, + "loss": 3.3789, + "step": 995 + }, + { + "epoch": 1.46, + "grad_norm": 6.067572593688965, + "learning_rate": 0.0001899510197072792, + "loss": 3.3577, + "step": 996 + }, + { + "epoch": 1.47, + "grad_norm": 7.723760604858398, + "learning_rate": 0.00018993051771018719, + "loss": 3.4, + "step": 997 + }, + { + "epoch": 1.47, + "grad_norm": 4.619333744049072, + "learning_rate": 0.00018990999592898396, + "loss": 3.3979, + "step": 998 + }, + { + "epoch": 1.47, + "grad_norm": 6.598495006561279, + "learning_rate": 0.00018988945436818422, + "loss": 3.4331, + "step": 999 + }, + { + "epoch": 1.47, + "grad_norm": 9.450194358825684, + "learning_rate": 0.0001898688930323069, + "loss": 3.4156, + "step": 1000 + }, + { + "epoch": 1.47, + "grad_norm": 4.10836935043335, + "learning_rate": 0.00018984831192587538, + "loss": 3.3787, + "step": 1001 + }, + { + "epoch": 1.47, + "grad_norm": 6.087796211242676, + "learning_rate": 0.00018982771105341737, + "loss": 3.3969, + "step": 1002 + }, + { + "epoch": 1.47, + "grad_norm": 7.915310859680176, + "learning_rate": 0.00018980709041946497, + "loss": 3.3948, + "step": 1003 + }, + { + "epoch": 1.48, + "grad_norm": 4.227004528045654, + "learning_rate": 0.0001897864500285545, + "loss": 3.3238, + "step": 1004 + }, + { + "epoch": 1.48, + "grad_norm": 6.717720031738281, + "learning_rate": 0.00018976578988522675, + "loss": 3.3781, + "step": 1005 + }, + { + "epoch": 1.48, + "grad_norm": 10.050957679748535, + "learning_rate": 0.0001897451099940268, + "loss": 3.4204, + "step": 1006 + }, + { + "epoch": 1.48, + "grad_norm": 4.475175857543945, + "learning_rate": 0.0001897244103595041, + "loss": 3.4221, + "step": 1007 + }, + { + "epoch": 1.48, + "grad_norm": 5.719995975494385, + "learning_rate": 0.00018970369098621244, + "loss": 3.3555, + "step": 1008 + }, + { + "epoch": 1.48, + "grad_norm": 7.594791889190674, + "learning_rate": 0.00018968295187870996, + "loss": 3.4073, + "step": 1009 + }, + { + "epoch": 1.49, + "grad_norm": 4.333056926727295, + "learning_rate": 0.00018966219304155906, + "loss": 3.3556, + "step": 1010 + }, + { + "epoch": 1.49, + "grad_norm": 6.078528881072998, + "learning_rate": 0.00018964141447932656, + "loss": 3.3484, + "step": 1011 + }, + { + "epoch": 1.49, + "grad_norm": 8.950298309326172, + "learning_rate": 0.00018962061619658366, + "loss": 3.4055, + "step": 1012 + }, + { + "epoch": 1.49, + "grad_norm": 4.340825080871582, + "learning_rate": 0.0001895997981979058, + "loss": 3.334, + "step": 1013 + }, + { + "epoch": 1.49, + "grad_norm": 6.103879451751709, + "learning_rate": 0.00018957896048787278, + "loss": 3.4908, + "step": 1014 + }, + { + "epoch": 1.49, + "eval_loss": 3.206282138824463, + "eval_runtime": 1164.0205, + "eval_samples_per_second": 14.727, + "eval_steps_per_second": 0.921, + "step": 1014 + }, + { + "epoch": 1.49, + "grad_norm": 7.772497653961182, + "learning_rate": 0.00018955810307106881, + "loss": 3.3821, + "step": 1015 + }, + { + "epoch": 1.49, + "grad_norm": 4.639206886291504, + "learning_rate": 0.00018953722595208236, + "loss": 3.3776, + "step": 1016 + }, + { + "epoch": 1.5, + "grad_norm": 5.658569812774658, + "learning_rate": 0.00018951632913550626, + "loss": 3.339, + "step": 1017 + }, + { + "epoch": 1.5, + "grad_norm": 8.45776081085205, + "learning_rate": 0.00018949541262593762, + "loss": 3.4214, + "step": 1018 + }, + { + "epoch": 1.5, + "grad_norm": 3.2400307655334473, + "learning_rate": 0.000189474476427978, + "loss": 3.3666, + "step": 1019 + }, + { + "epoch": 1.5, + "grad_norm": 6.397870063781738, + "learning_rate": 0.0001894535205462332, + "loss": 3.4133, + "step": 1020 + }, + { + "epoch": 1.5, + "grad_norm": 8.270928382873535, + "learning_rate": 0.00018943254498531337, + "loss": 3.4133, + "step": 1021 + }, + { + "epoch": 1.5, + "grad_norm": 5.132183074951172, + "learning_rate": 0.00018941154974983298, + "loss": 3.3984, + "step": 1022 + }, + { + "epoch": 1.5, + "grad_norm": 5.842728137969971, + "learning_rate": 0.0001893905348444109, + "loss": 3.4019, + "step": 1023 + }, + { + "epoch": 1.51, + "grad_norm": 8.618907928466797, + "learning_rate": 0.00018936950027367017, + "loss": 3.4036, + "step": 1024 + }, + { + "epoch": 1.51, + "grad_norm": 3.155228853225708, + "learning_rate": 0.00018934844604223835, + "loss": 3.3512, + "step": 1025 + }, + { + "epoch": 1.51, + "grad_norm": 6.209192752838135, + "learning_rate": 0.00018932737215474714, + "loss": 3.388, + "step": 1026 + }, + { + "epoch": 1.51, + "grad_norm": 7.953395843505859, + "learning_rate": 0.00018930627861583275, + "loss": 3.402, + "step": 1027 + }, + { + "epoch": 1.51, + "grad_norm": 4.540157794952393, + "learning_rate": 0.00018928516543013552, + "loss": 3.3162, + "step": 1028 + }, + { + "epoch": 1.51, + "grad_norm": 6.285725116729736, + "learning_rate": 0.0001892640326023003, + "loss": 3.4329, + "step": 1029 + }, + { + "epoch": 1.51, + "grad_norm": 9.266477584838867, + "learning_rate": 0.0001892428801369761, + "loss": 3.381, + "step": 1030 + }, + { + "epoch": 1.52, + "grad_norm": 4.977963924407959, + "learning_rate": 0.00018922170803881635, + "loss": 3.3758, + "step": 1031 + }, + { + "epoch": 1.52, + "grad_norm": 5.400416374206543, + "learning_rate": 0.0001892005163124788, + "loss": 3.3279, + "step": 1032 + }, + { + "epoch": 1.52, + "grad_norm": 6.933363914489746, + "learning_rate": 0.00018917930496262542, + "loss": 3.3655, + "step": 1033 + }, + { + "epoch": 1.52, + "grad_norm": 3.6254875659942627, + "learning_rate": 0.00018915807399392262, + "loss": 3.3451, + "step": 1034 + }, + { + "epoch": 1.52, + "grad_norm": 7.3330841064453125, + "learning_rate": 0.00018913682341104105, + "loss": 3.4291, + "step": 1035 + }, + { + "epoch": 1.52, + "grad_norm": 10.801877975463867, + "learning_rate": 0.0001891155532186557, + "loss": 3.4253, + "step": 1036 + }, + { + "epoch": 1.53, + "grad_norm": 5.197689533233643, + "learning_rate": 0.00018909426342144587, + "loss": 3.3764, + "step": 1037 + }, + { + "epoch": 1.53, + "grad_norm": 5.4762797355651855, + "learning_rate": 0.00018907295402409514, + "loss": 3.3669, + "step": 1038 + }, + { + "epoch": 1.53, + "grad_norm": 9.155226707458496, + "learning_rate": 0.0001890516250312915, + "loss": 3.3748, + "step": 1039 + }, + { + "epoch": 1.53, + "grad_norm": 4.974143028259277, + "learning_rate": 0.00018903027644772715, + "loss": 3.3812, + "step": 1040 + }, + { + "epoch": 1.53, + "grad_norm": 3.935563802719116, + "learning_rate": 0.00018900890827809862, + "loss": 3.3748, + "step": 1041 + }, + { + "epoch": 1.53, + "grad_norm": 5.06904935836792, + "learning_rate": 0.0001889875205271068, + "loss": 3.4328, + "step": 1042 + }, + { + "epoch": 1.53, + "grad_norm": 1.7735739946365356, + "learning_rate": 0.0001889661131994568, + "loss": 3.3331, + "step": 1043 + }, + { + "epoch": 1.54, + "grad_norm": 3.266242504119873, + "learning_rate": 0.00018894468629985814, + "loss": 3.3797, + "step": 1044 + }, + { + "epoch": 1.54, + "grad_norm": 1.9417365789413452, + "learning_rate": 0.0001889232398330246, + "loss": 3.3556, + "step": 1045 + }, + { + "epoch": 1.54, + "grad_norm": 2.2055392265319824, + "learning_rate": 0.00018890177380367419, + "loss": 3.3695, + "step": 1046 + }, + { + "epoch": 1.54, + "grad_norm": 3.9984004497528076, + "learning_rate": 0.00018888028821652936, + "loss": 3.4158, + "step": 1047 + }, + { + "epoch": 1.54, + "grad_norm": 2.413911819458008, + "learning_rate": 0.0001888587830763168, + "loss": 3.4492, + "step": 1048 + }, + { + "epoch": 1.54, + "grad_norm": 4.597828388214111, + "learning_rate": 0.00018883725838776742, + "loss": 3.3518, + "step": 1049 + }, + { + "epoch": 1.54, + "grad_norm": 4.952722549438477, + "learning_rate": 0.00018881571415561656, + "loss": 3.371, + "step": 1050 + }, + { + "epoch": 1.55, + "grad_norm": 2.318856954574585, + "learning_rate": 0.00018879415038460383, + "loss": 3.3974, + "step": 1051 + }, + { + "epoch": 1.55, + "grad_norm": 2.227694034576416, + "learning_rate": 0.00018877256707947306, + "loss": 3.3784, + "step": 1052 + }, + { + "epoch": 1.55, + "grad_norm": 4.430450916290283, + "learning_rate": 0.00018875096424497248, + "loss": 3.3792, + "step": 1053 + }, + { + "epoch": 1.55, + "grad_norm": 2.9697399139404297, + "learning_rate": 0.00018872934188585454, + "loss": 3.4059, + "step": 1054 + }, + { + "epoch": 1.55, + "grad_norm": 4.346686840057373, + "learning_rate": 0.000188707700006876, + "loss": 3.4155, + "step": 1055 + }, + { + "epoch": 1.55, + "grad_norm": 4.2472381591796875, + "learning_rate": 0.000188686038612798, + "loss": 3.4242, + "step": 1056 + }, + { + "epoch": 1.55, + "grad_norm": 3.157696008682251, + "learning_rate": 0.0001886643577083858, + "loss": 3.4913, + "step": 1057 + }, + { + "epoch": 1.56, + "grad_norm": 2.940396308898926, + "learning_rate": 0.00018864265729840913, + "loss": 3.4734, + "step": 1058 + }, + { + "epoch": 1.56, + "grad_norm": 4.0289154052734375, + "learning_rate": 0.00018862093738764192, + "loss": 3.443, + "step": 1059 + }, + { + "epoch": 1.56, + "grad_norm": 3.4877893924713135, + "learning_rate": 0.00018859919798086238, + "loss": 3.5322, + "step": 1060 + }, + { + "epoch": 1.56, + "grad_norm": 4.22461462020874, + "learning_rate": 0.00018857743908285305, + "loss": 3.4562, + "step": 1061 + }, + { + "epoch": 1.56, + "grad_norm": 4.031467437744141, + "learning_rate": 0.00018855566069840077, + "loss": 3.4653, + "step": 1062 + }, + { + "epoch": 1.56, + "grad_norm": 3.602692127227783, + "learning_rate": 0.00018853386283229658, + "loss": 3.4555, + "step": 1063 + }, + { + "epoch": 1.57, + "grad_norm": 2.7948570251464844, + "learning_rate": 0.00018851204548933593, + "loss": 3.4255, + "step": 1064 + }, + { + "epoch": 1.57, + "grad_norm": 4.970945358276367, + "learning_rate": 0.00018849020867431842, + "loss": 3.4463, + "step": 1065 + }, + { + "epoch": 1.57, + "grad_norm": 4.409501552581787, + "learning_rate": 0.00018846835239204804, + "loss": 3.3992, + "step": 1066 + }, + { + "epoch": 1.57, + "grad_norm": 2.8853704929351807, + "learning_rate": 0.000188446476647333, + "loss": 3.4709, + "step": 1067 + }, + { + "epoch": 1.57, + "grad_norm": 2.3841402530670166, + "learning_rate": 0.00018842458144498588, + "loss": 3.4342, + "step": 1068 + }, + { + "epoch": 1.57, + "grad_norm": 4.642658710479736, + "learning_rate": 0.00018840266678982342, + "loss": 3.4102, + "step": 1069 + }, + { + "epoch": 1.57, + "grad_norm": 4.265832424163818, + "learning_rate": 0.00018838073268666672, + "loss": 3.4367, + "step": 1070 + }, + { + "epoch": 1.58, + "grad_norm": 2.900568723678589, + "learning_rate": 0.00018835877914034113, + "loss": 3.4052, + "step": 1071 + }, + { + "epoch": 1.58, + "grad_norm": 2.4887197017669678, + "learning_rate": 0.00018833680615567628, + "loss": 3.4175, + "step": 1072 + }, + { + "epoch": 1.58, + "grad_norm": 5.449963092803955, + "learning_rate": 0.0001883148137375061, + "loss": 3.4521, + "step": 1073 + }, + { + "epoch": 1.58, + "grad_norm": 5.6257643699646, + "learning_rate": 0.00018829280189066873, + "loss": 3.4741, + "step": 1074 + }, + { + "epoch": 1.58, + "grad_norm": 2.799276828765869, + "learning_rate": 0.00018827077062000668, + "loss": 3.4431, + "step": 1075 + }, + { + "epoch": 1.58, + "grad_norm": 2.863265037536621, + "learning_rate": 0.00018824871993036664, + "loss": 3.3978, + "step": 1076 + }, + { + "epoch": 1.58, + "grad_norm": 4.204624176025391, + "learning_rate": 0.00018822664982659964, + "loss": 3.4559, + "step": 1077 + }, + { + "epoch": 1.59, + "grad_norm": 3.0599143505096436, + "learning_rate": 0.00018820456031356093, + "loss": 3.3883, + "step": 1078 + }, + { + "epoch": 1.59, + "grad_norm": 4.109722137451172, + "learning_rate": 0.00018818245139611011, + "loss": 3.4014, + "step": 1079 + }, + { + "epoch": 1.59, + "grad_norm": 3.4017956256866455, + "learning_rate": 0.00018816032307911092, + "loss": 3.4348, + "step": 1080 + }, + { + "epoch": 1.59, + "grad_norm": 4.216831207275391, + "learning_rate": 0.00018813817536743148, + "loss": 3.452, + "step": 1081 + }, + { + "epoch": 1.59, + "grad_norm": 3.819659471511841, + "learning_rate": 0.00018811600826594416, + "loss": 3.3937, + "step": 1082 + }, + { + "epoch": 1.59, + "grad_norm": 3.5902295112609863, + "learning_rate": 0.00018809382177952552, + "loss": 3.3992, + "step": 1083 + }, + { + "epoch": 1.59, + "grad_norm": 2.6973037719726562, + "learning_rate": 0.0001880716159130565, + "loss": 3.4278, + "step": 1084 + }, + { + "epoch": 1.6, + "grad_norm": 5.055513858795166, + "learning_rate": 0.0001880493906714222, + "loss": 3.3913, + "step": 1085 + }, + { + "epoch": 1.6, + "grad_norm": 4.533970832824707, + "learning_rate": 0.00018802714605951199, + "loss": 3.4047, + "step": 1086 + }, + { + "epoch": 1.6, + "grad_norm": 2.5329983234405518, + "learning_rate": 0.00018800488208221963, + "loss": 3.4291, + "step": 1087 + }, + { + "epoch": 1.6, + "grad_norm": 1.993481993675232, + "learning_rate": 0.00018798259874444297, + "loss": 3.4334, + "step": 1088 + }, + { + "epoch": 1.6, + "grad_norm": 6.799105167388916, + "learning_rate": 0.00018796029605108424, + "loss": 3.4107, + "step": 1089 + }, + { + "epoch": 1.6, + "grad_norm": 6.37121057510376, + "learning_rate": 0.00018793797400704985, + "loss": 3.4237, + "step": 1090 + }, + { + "epoch": 1.61, + "grad_norm": 1.2064040899276733, + "learning_rate": 0.0001879156326172505, + "loss": 3.4487, + "step": 1091 + }, + { + "epoch": 1.61, + "grad_norm": 0.6396639943122864, + "learning_rate": 0.00018789327188660116, + "loss": 3.3509, + "step": 1092 + }, + { + "epoch": 1.61, + "grad_norm": 2.529014825820923, + "learning_rate": 0.00018787089182002105, + "loss": 3.4014, + "step": 1093 + }, + { + "epoch": 1.61, + "grad_norm": 2.6367127895355225, + "learning_rate": 0.00018784849242243362, + "loss": 3.3676, + "step": 1094 + }, + { + "epoch": 1.61, + "grad_norm": 0.968987226486206, + "learning_rate": 0.00018782607369876659, + "loss": 3.4156, + "step": 1095 + }, + { + "epoch": 1.61, + "grad_norm": 5.669402599334717, + "learning_rate": 0.00018780363565395192, + "loss": 3.4195, + "step": 1096 + }, + { + "epoch": 1.61, + "grad_norm": 4.468107223510742, + "learning_rate": 0.00018778117829292587, + "loss": 3.4453, + "step": 1097 + }, + { + "epoch": 1.62, + "grad_norm": 3.340261220932007, + "learning_rate": 0.00018775870162062886, + "loss": 3.4024, + "step": 1098 + }, + { + "epoch": 1.62, + "grad_norm": 2.798548698425293, + "learning_rate": 0.00018773620564200565, + "loss": 3.503, + "step": 1099 + }, + { + "epoch": 1.62, + "grad_norm": 5.848763465881348, + "learning_rate": 0.00018771369036200513, + "loss": 3.3962, + "step": 1100 + }, + { + "epoch": 1.62, + "grad_norm": 6.020981788635254, + "learning_rate": 0.00018769115578558062, + "loss": 3.4708, + "step": 1101 + }, + { + "epoch": 1.62, + "grad_norm": 2.3334991931915283, + "learning_rate": 0.00018766860191768946, + "loss": 3.3861, + "step": 1102 + }, + { + "epoch": 1.62, + "grad_norm": 3.1461942195892334, + "learning_rate": 0.00018764602876329345, + "loss": 3.3828, + "step": 1103 + }, + { + "epoch": 1.62, + "grad_norm": 2.833292007446289, + "learning_rate": 0.00018762343632735848, + "loss": 3.4628, + "step": 1104 + }, + { + "epoch": 1.63, + "grad_norm": 1.1478309631347656, + "learning_rate": 0.00018760082461485476, + "loss": 3.3512, + "step": 1105 + }, + { + "epoch": 1.63, + "grad_norm": 4.6011199951171875, + "learning_rate": 0.00018757819363075665, + "loss": 3.4301, + "step": 1106 + }, + { + "epoch": 1.63, + "grad_norm": 3.997016191482544, + "learning_rate": 0.0001875555433800429, + "loss": 3.4099, + "step": 1107 + }, + { + "epoch": 1.63, + "grad_norm": 4.506150722503662, + "learning_rate": 0.00018753287386769638, + "loss": 3.4169, + "step": 1108 + }, + { + "epoch": 1.63, + "grad_norm": 4.197047710418701, + "learning_rate": 0.0001875101850987042, + "loss": 3.3518, + "step": 1109 + }, + { + "epoch": 1.63, + "grad_norm": 3.504765033721924, + "learning_rate": 0.0001874874770780578, + "loss": 3.3838, + "step": 1110 + }, + { + "epoch": 1.64, + "grad_norm": 2.7405476570129395, + "learning_rate": 0.00018746474981075272, + "loss": 3.376, + "step": 1111 + }, + { + "epoch": 1.64, + "grad_norm": 4.3409037590026855, + "learning_rate": 0.00018744200330178887, + "loss": 3.4184, + "step": 1112 + }, + { + "epoch": 1.64, + "grad_norm": 4.318120002746582, + "learning_rate": 0.0001874192375561703, + "loss": 3.4744, + "step": 1113 + }, + { + "epoch": 1.64, + "grad_norm": 3.5253307819366455, + "learning_rate": 0.00018739645257890533, + "loss": 3.4058, + "step": 1114 + }, + { + "epoch": 1.64, + "grad_norm": 3.368002414703369, + "learning_rate": 0.0001873736483750065, + "loss": 3.4013, + "step": 1115 + }, + { + "epoch": 1.64, + "grad_norm": 3.804229736328125, + "learning_rate": 0.00018735082494949058, + "loss": 3.3625, + "step": 1116 + }, + { + "epoch": 1.64, + "grad_norm": 3.6312084197998047, + "learning_rate": 0.00018732798230737855, + "loss": 3.3568, + "step": 1117 + }, + { + "epoch": 1.65, + "grad_norm": 3.4278273582458496, + "learning_rate": 0.00018730512045369568, + "loss": 3.4077, + "step": 1118 + }, + { + "epoch": 1.65, + "grad_norm": 3.177262306213379, + "learning_rate": 0.0001872822393934714, + "loss": 3.41, + "step": 1119 + }, + { + "epoch": 1.65, + "grad_norm": 3.8335347175598145, + "learning_rate": 0.00018725933913173938, + "loss": 3.3968, + "step": 1120 + }, + { + "epoch": 1.65, + "grad_norm": 3.43869948387146, + "learning_rate": 0.00018723641967353756, + "loss": 3.4295, + "step": 1121 + }, + { + "epoch": 1.65, + "grad_norm": 4.323538780212402, + "learning_rate": 0.00018721348102390803, + "loss": 3.3734, + "step": 1122 + }, + { + "epoch": 1.65, + "grad_norm": 3.6614975929260254, + "learning_rate": 0.00018719052318789717, + "loss": 3.3522, + "step": 1123 + }, + { + "epoch": 1.65, + "grad_norm": 3.442986249923706, + "learning_rate": 0.0001871675461705555, + "loss": 3.3743, + "step": 1124 + }, + { + "epoch": 1.66, + "grad_norm": 3.7068862915039062, + "learning_rate": 0.00018714454997693783, + "loss": 3.3611, + "step": 1125 + }, + { + "epoch": 1.66, + "grad_norm": 3.6128504276275635, + "learning_rate": 0.00018712153461210326, + "loss": 3.3678, + "step": 1126 + }, + { + "epoch": 1.66, + "grad_norm": 3.2057278156280518, + "learning_rate": 0.00018709850008111488, + "loss": 3.4118, + "step": 1127 + }, + { + "epoch": 1.66, + "grad_norm": 4.726739406585693, + "learning_rate": 0.0001870754463890402, + "loss": 3.3669, + "step": 1128 + }, + { + "epoch": 1.66, + "grad_norm": 4.606038570404053, + "learning_rate": 0.00018705237354095085, + "loss": 3.3316, + "step": 1129 + }, + { + "epoch": 1.66, + "grad_norm": 2.9650778770446777, + "learning_rate": 0.00018702928154192274, + "loss": 3.3614, + "step": 1130 + }, + { + "epoch": 1.66, + "grad_norm": 2.7509312629699707, + "learning_rate": 0.00018700617039703595, + "loss": 3.3797, + "step": 1131 + }, + { + "epoch": 1.67, + "grad_norm": 5.575433731079102, + "learning_rate": 0.00018698304011137475, + "loss": 3.384, + "step": 1132 + }, + { + "epoch": 1.67, + "grad_norm": 4.589602947235107, + "learning_rate": 0.00018695989069002764, + "loss": 3.3729, + "step": 1133 + }, + { + "epoch": 1.67, + "grad_norm": 2.780677318572998, + "learning_rate": 0.00018693672213808738, + "loss": 3.3704, + "step": 1134 + }, + { + "epoch": 1.67, + "grad_norm": 1.9864367246627808, + "learning_rate": 0.00018691353446065087, + "loss": 3.3718, + "step": 1135 + }, + { + "epoch": 1.67, + "grad_norm": 5.260591983795166, + "learning_rate": 0.00018689032766281924, + "loss": 3.3856, + "step": 1136 + }, + { + "epoch": 1.67, + "grad_norm": 4.4715447425842285, + "learning_rate": 0.00018686710174969787, + "loss": 3.3433, + "step": 1137 + }, + { + "epoch": 1.68, + "grad_norm": 3.0349228382110596, + "learning_rate": 0.00018684385672639622, + "loss": 3.3361, + "step": 1138 + }, + { + "epoch": 1.68, + "grad_norm": 2.5944900512695312, + "learning_rate": 0.0001868205925980281, + "loss": 3.3379, + "step": 1139 + }, + { + "epoch": 1.68, + "grad_norm": 4.727088928222656, + "learning_rate": 0.00018679730936971148, + "loss": 3.3889, + "step": 1140 + }, + { + "epoch": 1.68, + "grad_norm": 5.072381019592285, + "learning_rate": 0.0001867740070465685, + "loss": 3.4185, + "step": 1141 + }, + { + "epoch": 1.68, + "grad_norm": 2.782468318939209, + "learning_rate": 0.00018675068563372545, + "loss": 3.3635, + "step": 1142 + }, + { + "epoch": 1.68, + "grad_norm": 2.7385001182556152, + "learning_rate": 0.000186727345136313, + "loss": 3.4115, + "step": 1143 + }, + { + "epoch": 1.68, + "grad_norm": 4.811634540557861, + "learning_rate": 0.0001867039855594658, + "loss": 3.4155, + "step": 1144 + }, + { + "epoch": 1.69, + "grad_norm": 4.225504398345947, + "learning_rate": 0.00018668060690832285, + "loss": 3.3729, + "step": 1145 + }, + { + "epoch": 1.69, + "grad_norm": 3.26381254196167, + "learning_rate": 0.0001866572091880273, + "loss": 3.3982, + "step": 1146 + }, + { + "epoch": 1.69, + "grad_norm": 2.842423915863037, + "learning_rate": 0.00018663379240372647, + "loss": 3.387, + "step": 1147 + }, + { + "epoch": 1.69, + "grad_norm": 4.7502217292785645, + "learning_rate": 0.0001866103565605719, + "loss": 3.377, + "step": 1148 + }, + { + "epoch": 1.69, + "grad_norm": 4.280509948730469, + "learning_rate": 0.00018658690166371933, + "loss": 3.3774, + "step": 1149 + }, + { + "epoch": 1.69, + "grad_norm": 3.5164103507995605, + "learning_rate": 0.00018656342771832866, + "loss": 3.357, + "step": 1150 + }, + { + "epoch": 1.69, + "grad_norm": 2.891474485397339, + "learning_rate": 0.00018653993472956404, + "loss": 3.3755, + "step": 1151 + }, + { + "epoch": 1.7, + "grad_norm": 4.281400203704834, + "learning_rate": 0.00018651642270259373, + "loss": 3.3511, + "step": 1152 + }, + { + "epoch": 1.7, + "grad_norm": 4.510589122772217, + "learning_rate": 0.00018649289164259024, + "loss": 3.4113, + "step": 1153 + }, + { + "epoch": 1.7, + "grad_norm": 2.6343142986297607, + "learning_rate": 0.00018646934155473022, + "loss": 3.3291, + "step": 1154 + }, + { + "epoch": 1.7, + "grad_norm": 2.3589959144592285, + "learning_rate": 0.00018644577244419455, + "loss": 3.3638, + "step": 1155 + }, + { + "epoch": 1.7, + "grad_norm": 5.584714889526367, + "learning_rate": 0.00018642218431616828, + "loss": 3.32, + "step": 1156 + }, + { + "epoch": 1.7, + "grad_norm": 4.909949779510498, + "learning_rate": 0.00018639857717584065, + "loss": 3.3947, + "step": 1157 + }, + { + "epoch": 1.7, + "grad_norm": 2.9206135272979736, + "learning_rate": 0.00018637495102840506, + "loss": 3.3826, + "step": 1158 + }, + { + "epoch": 1.71, + "grad_norm": 2.58848237991333, + "learning_rate": 0.0001863513058790591, + "loss": 3.3237, + "step": 1159 + }, + { + "epoch": 1.71, + "grad_norm": 5.167590141296387, + "learning_rate": 0.00018632764173300457, + "loss": 3.3972, + "step": 1160 + }, + { + "epoch": 1.71, + "grad_norm": 5.174654483795166, + "learning_rate": 0.00018630395859544735, + "loss": 3.357, + "step": 1161 + }, + { + "epoch": 1.71, + "grad_norm": 2.867471218109131, + "learning_rate": 0.00018628025647159766, + "loss": 3.3406, + "step": 1162 + }, + { + "epoch": 1.71, + "grad_norm": 2.780794620513916, + "learning_rate": 0.00018625653536666974, + "loss": 3.3527, + "step": 1163 + }, + { + "epoch": 1.71, + "grad_norm": 4.502468585968018, + "learning_rate": 0.00018623279528588214, + "loss": 3.3216, + "step": 1164 + }, + { + "epoch": 1.72, + "grad_norm": 3.549867868423462, + "learning_rate": 0.00018620903623445747, + "loss": 3.3949, + "step": 1165 + }, + { + "epoch": 1.72, + "grad_norm": 3.7050774097442627, + "learning_rate": 0.00018618525821762256, + "loss": 3.395, + "step": 1166 + }, + { + "epoch": 1.72, + "grad_norm": 3.520019769668579, + "learning_rate": 0.00018616146124060846, + "loss": 3.3757, + "step": 1167 + }, + { + "epoch": 1.72, + "grad_norm": 3.870093822479248, + "learning_rate": 0.00018613764530865028, + "loss": 3.3712, + "step": 1168 + }, + { + "epoch": 1.72, + "grad_norm": 3.196948766708374, + "learning_rate": 0.0001861138104269874, + "loss": 3.3337, + "step": 1169 + }, + { + "epoch": 1.72, + "grad_norm": 4.416697025299072, + "learning_rate": 0.00018608995660086337, + "loss": 3.4137, + "step": 1170 + }, + { + "epoch": 1.72, + "grad_norm": 3.9429569244384766, + "learning_rate": 0.00018606608383552582, + "loss": 3.389, + "step": 1171 + }, + { + "epoch": 1.73, + "grad_norm": 3.530494213104248, + "learning_rate": 0.0001860421921362266, + "loss": 3.3752, + "step": 1172 + }, + { + "epoch": 1.73, + "grad_norm": 3.766986608505249, + "learning_rate": 0.00018601828150822177, + "loss": 3.3386, + "step": 1173 + }, + { + "epoch": 1.73, + "grad_norm": 3.8910138607025146, + "learning_rate": 0.00018599435195677146, + "loss": 3.348, + "step": 1174 + }, + { + "epoch": 1.73, + "grad_norm": 3.5032382011413574, + "learning_rate": 0.00018597040348714003, + "loss": 3.408, + "step": 1175 + }, + { + "epoch": 1.73, + "grad_norm": 3.5396668910980225, + "learning_rate": 0.000185946436104596, + "loss": 3.317, + "step": 1176 + }, + { + "epoch": 1.73, + "grad_norm": 3.7018423080444336, + "learning_rate": 0.00018592244981441197, + "loss": 3.3456, + "step": 1177 + }, + { + "epoch": 1.73, + "grad_norm": 3.41996169090271, + "learning_rate": 0.00018589844462186484, + "loss": 3.4147, + "step": 1178 + }, + { + "epoch": 1.74, + "grad_norm": 2.907388210296631, + "learning_rate": 0.00018587442053223557, + "loss": 3.3377, + "step": 1179 + }, + { + "epoch": 1.74, + "grad_norm": 4.651727676391602, + "learning_rate": 0.00018585037755080926, + "loss": 3.3401, + "step": 1180 + }, + { + "epoch": 1.74, + "grad_norm": 4.285563945770264, + "learning_rate": 0.00018582631568287523, + "loss": 3.3671, + "step": 1181 + }, + { + "epoch": 1.74, + "grad_norm": 3.3015687465667725, + "learning_rate": 0.00018580223493372695, + "loss": 3.3049, + "step": 1182 + }, + { + "epoch": 1.74, + "grad_norm": 2.8395988941192627, + "learning_rate": 0.000185778135308662, + "loss": 3.3588, + "step": 1183 + }, + { + "epoch": 1.74, + "eval_loss": 3.1614086627960205, + "eval_runtime": 1168.1043, + "eval_samples_per_second": 14.675, + "eval_steps_per_second": 0.918, + "step": 1183 + }, + { + "epoch": 1.74, + "grad_norm": 4.551041603088379, + "learning_rate": 0.00018575401681298216, + "loss": 3.3579, + "step": 1184 + }, + { + "epoch": 1.74, + "grad_norm": 4.050691604614258, + "learning_rate": 0.00018572987945199328, + "loss": 3.3174, + "step": 1185 + }, + { + "epoch": 1.75, + "grad_norm": 3.483586072921753, + "learning_rate": 0.0001857057232310055, + "loss": 3.3689, + "step": 1186 + }, + { + "epoch": 1.75, + "grad_norm": 3.430351734161377, + "learning_rate": 0.00018568154815533293, + "loss": 3.3624, + "step": 1187 + }, + { + "epoch": 1.75, + "grad_norm": 4.206818103790283, + "learning_rate": 0.00018565735423029404, + "loss": 3.3365, + "step": 1188 + }, + { + "epoch": 1.75, + "grad_norm": 3.6076526641845703, + "learning_rate": 0.00018563314146121122, + "loss": 3.3728, + "step": 1189 + }, + { + "epoch": 1.75, + "grad_norm": 3.844933271408081, + "learning_rate": 0.0001856089098534112, + "loss": 3.3577, + "step": 1190 + }, + { + "epoch": 1.75, + "grad_norm": 3.1116933822631836, + "learning_rate": 0.00018558465941222472, + "loss": 3.3242, + "step": 1191 + }, + { + "epoch": 1.76, + "grad_norm": 4.367002010345459, + "learning_rate": 0.00018556039014298672, + "loss": 3.3382, + "step": 1192 + }, + { + "epoch": 1.76, + "grad_norm": 4.229283332824707, + "learning_rate": 0.0001855361020510363, + "loss": 3.3318, + "step": 1193 + }, + { + "epoch": 1.76, + "grad_norm": 3.6189987659454346, + "learning_rate": 0.00018551179514171668, + "loss": 3.3879, + "step": 1194 + }, + { + "epoch": 1.76, + "grad_norm": 3.045738935470581, + "learning_rate": 0.0001854874694203752, + "loss": 3.3407, + "step": 1195 + }, + { + "epoch": 1.76, + "grad_norm": 4.931478977203369, + "learning_rate": 0.00018546312489236335, + "loss": 3.3428, + "step": 1196 + }, + { + "epoch": 1.76, + "grad_norm": 4.312923908233643, + "learning_rate": 0.00018543876156303676, + "loss": 3.3768, + "step": 1197 + }, + { + "epoch": 1.76, + "grad_norm": 3.236755132675171, + "learning_rate": 0.00018541437943775523, + "loss": 3.3637, + "step": 1198 + }, + { + "epoch": 1.77, + "grad_norm": 2.9794223308563232, + "learning_rate": 0.00018538997852188266, + "loss": 3.2961, + "step": 1199 + }, + { + "epoch": 1.77, + "grad_norm": 4.777314186096191, + "learning_rate": 0.00018536555882078706, + "loss": 3.3838, + "step": 1200 + }, + { + "epoch": 1.77, + "grad_norm": 4.084963321685791, + "learning_rate": 0.0001853411203398406, + "loss": 3.3525, + "step": 1201 + }, + { + "epoch": 1.77, + "grad_norm": 3.598912477493286, + "learning_rate": 0.0001853166630844196, + "loss": 3.3362, + "step": 1202 + }, + { + "epoch": 1.77, + "grad_norm": 3.44163179397583, + "learning_rate": 0.0001852921870599045, + "loss": 3.4129, + "step": 1203 + }, + { + "epoch": 1.77, + "grad_norm": 3.8749027252197266, + "learning_rate": 0.00018526769227167984, + "loss": 3.3069, + "step": 1204 + }, + { + "epoch": 1.77, + "grad_norm": 3.8452203273773193, + "learning_rate": 0.00018524317872513433, + "loss": 3.3084, + "step": 1205 + }, + { + "epoch": 1.78, + "grad_norm": 3.8901963233947754, + "learning_rate": 0.00018521864642566076, + "loss": 3.322, + "step": 1206 + }, + { + "epoch": 1.78, + "grad_norm": 3.340991973876953, + "learning_rate": 0.00018519409537865608, + "loss": 3.3373, + "step": 1207 + }, + { + "epoch": 1.78, + "grad_norm": 4.287864685058594, + "learning_rate": 0.0001851695255895214, + "loss": 3.2985, + "step": 1208 + }, + { + "epoch": 1.78, + "grad_norm": 3.954977512359619, + "learning_rate": 0.00018514493706366184, + "loss": 3.3286, + "step": 1209 + }, + { + "epoch": 1.78, + "grad_norm": 3.5455312728881836, + "learning_rate": 0.00018512032980648674, + "loss": 3.3191, + "step": 1210 + }, + { + "epoch": 1.78, + "grad_norm": 3.2507898807525635, + "learning_rate": 0.00018509570382340955, + "loss": 3.3356, + "step": 1211 + }, + { + "epoch": 1.79, + "grad_norm": 4.234015464782715, + "learning_rate": 0.0001850710591198478, + "loss": 3.365, + "step": 1212 + }, + { + "epoch": 1.79, + "grad_norm": 3.4031906127929688, + "learning_rate": 0.00018504639570122316, + "loss": 3.3405, + "step": 1213 + }, + { + "epoch": 1.79, + "grad_norm": 3.9651551246643066, + "learning_rate": 0.00018502171357296144, + "loss": 3.3953, + "step": 1214 + }, + { + "epoch": 1.79, + "grad_norm": 3.6755764484405518, + "learning_rate": 0.00018499701274049248, + "loss": 3.3102, + "step": 1215 + }, + { + "epoch": 1.79, + "grad_norm": 3.6327264308929443, + "learning_rate": 0.00018497229320925038, + "loss": 3.318, + "step": 1216 + }, + { + "epoch": 1.79, + "grad_norm": 3.4605491161346436, + "learning_rate": 0.0001849475549846732, + "loss": 3.3059, + "step": 1217 + }, + { + "epoch": 1.79, + "grad_norm": 3.7450873851776123, + "learning_rate": 0.00018492279807220326, + "loss": 3.3303, + "step": 1218 + }, + { + "epoch": 1.8, + "grad_norm": 3.5053107738494873, + "learning_rate": 0.00018489802247728684, + "loss": 3.3211, + "step": 1219 + }, + { + "epoch": 1.8, + "grad_norm": 3.7314960956573486, + "learning_rate": 0.00018487322820537444, + "loss": 3.3359, + "step": 1220 + }, + { + "epoch": 1.8, + "grad_norm": 3.397026777267456, + "learning_rate": 0.00018484841526192065, + "loss": 3.3199, + "step": 1221 + }, + { + "epoch": 1.8, + "grad_norm": 3.9689042568206787, + "learning_rate": 0.00018482358365238413, + "loss": 3.3249, + "step": 1222 + }, + { + "epoch": 1.8, + "grad_norm": 3.4028615951538086, + "learning_rate": 0.00018479873338222763, + "loss": 3.3234, + "step": 1223 + }, + { + "epoch": 1.8, + "grad_norm": 4.0719780921936035, + "learning_rate": 0.00018477386445691813, + "loss": 3.3755, + "step": 1224 + }, + { + "epoch": 1.8, + "grad_norm": 3.8761250972747803, + "learning_rate": 0.00018474897688192655, + "loss": 3.3313, + "step": 1225 + }, + { + "epoch": 1.81, + "grad_norm": 3.763307809829712, + "learning_rate": 0.00018472407066272804, + "loss": 3.3112, + "step": 1226 + }, + { + "epoch": 1.81, + "grad_norm": 3.3675754070281982, + "learning_rate": 0.00018469914580480177, + "loss": 3.373, + "step": 1227 + }, + { + "epoch": 1.81, + "grad_norm": 3.7994701862335205, + "learning_rate": 0.00018467420231363103, + "loss": 3.3034, + "step": 1228 + }, + { + "epoch": 1.81, + "grad_norm": 3.433627128601074, + "learning_rate": 0.00018464924019470327, + "loss": 3.2695, + "step": 1229 + }, + { + "epoch": 1.81, + "grad_norm": 3.7266974449157715, + "learning_rate": 0.00018462425945350996, + "loss": 3.3498, + "step": 1230 + }, + { + "epoch": 1.81, + "grad_norm": 3.44728422164917, + "learning_rate": 0.00018459926009554666, + "loss": 3.3682, + "step": 1231 + }, + { + "epoch": 1.81, + "grad_norm": 4.175528049468994, + "learning_rate": 0.00018457424212631312, + "loss": 3.3374, + "step": 1232 + }, + { + "epoch": 1.82, + "grad_norm": 4.267003059387207, + "learning_rate": 0.0001845492055513131, + "loss": 3.2985, + "step": 1233 + }, + { + "epoch": 1.82, + "grad_norm": 3.53112530708313, + "learning_rate": 0.0001845241503760545, + "loss": 3.3654, + "step": 1234 + }, + { + "epoch": 1.82, + "grad_norm": 3.1011736392974854, + "learning_rate": 0.00018449907660604927, + "loss": 3.3169, + "step": 1235 + }, + { + "epoch": 1.82, + "grad_norm": 4.862052917480469, + "learning_rate": 0.0001844739842468135, + "loss": 3.329, + "step": 1236 + }, + { + "epoch": 1.82, + "grad_norm": 4.767108917236328, + "learning_rate": 0.00018444887330386727, + "loss": 3.3546, + "step": 1237 + }, + { + "epoch": 1.82, + "grad_norm": 3.3005921840667725, + "learning_rate": 0.00018442374378273488, + "loss": 3.338, + "step": 1238 + }, + { + "epoch": 1.83, + "grad_norm": 3.2703089714050293, + "learning_rate": 0.00018439859568894463, + "loss": 3.281, + "step": 1239 + }, + { + "epoch": 1.83, + "grad_norm": 4.4580888748168945, + "learning_rate": 0.00018437342902802897, + "loss": 3.3748, + "step": 1240 + }, + { + "epoch": 1.83, + "grad_norm": 3.4954566955566406, + "learning_rate": 0.00018434824380552433, + "loss": 3.3291, + "step": 1241 + }, + { + "epoch": 1.83, + "grad_norm": 3.5984458923339844, + "learning_rate": 0.00018432304002697137, + "loss": 3.3834, + "step": 1242 + }, + { + "epoch": 1.83, + "grad_norm": 3.217467784881592, + "learning_rate": 0.00018429781769791468, + "loss": 3.3814, + "step": 1243 + }, + { + "epoch": 1.83, + "grad_norm": 4.149362087249756, + "learning_rate": 0.00018427257682390304, + "loss": 3.3808, + "step": 1244 + }, + { + "epoch": 1.83, + "grad_norm": 3.7322371006011963, + "learning_rate": 0.0001842473174104893, + "loss": 3.3546, + "step": 1245 + }, + { + "epoch": 1.84, + "grad_norm": 3.667970895767212, + "learning_rate": 0.0001842220394632303, + "loss": 3.3144, + "step": 1246 + }, + { + "epoch": 1.84, + "grad_norm": 3.353289842605591, + "learning_rate": 0.00018419674298768705, + "loss": 3.367, + "step": 1247 + }, + { + "epoch": 1.84, + "grad_norm": 4.078193664550781, + "learning_rate": 0.00018417142798942462, + "loss": 3.3284, + "step": 1248 + }, + { + "epoch": 1.84, + "grad_norm": 3.791130304336548, + "learning_rate": 0.0001841460944740121, + "loss": 3.3223, + "step": 1249 + }, + { + "epoch": 1.84, + "grad_norm": 3.9178853034973145, + "learning_rate": 0.00018412074244702274, + "loss": 3.4204, + "step": 1250 + }, + { + "epoch": 1.84, + "grad_norm": 3.1923162937164307, + "learning_rate": 0.00018409537191403378, + "loss": 3.2976, + "step": 1251 + }, + { + "epoch": 1.84, + "grad_norm": 4.827831745147705, + "learning_rate": 0.00018406998288062658, + "loss": 3.3697, + "step": 1252 + }, + { + "epoch": 1.85, + "grad_norm": 5.54097843170166, + "learning_rate": 0.00018404457535238657, + "loss": 3.3319, + "step": 1253 + }, + { + "epoch": 1.85, + "grad_norm": 2.249396562576294, + "learning_rate": 0.00018401914933490322, + "loss": 3.3106, + "step": 1254 + }, + { + "epoch": 1.85, + "grad_norm": 1.4117510318756104, + "learning_rate": 0.0001839937048337701, + "loss": 3.3664, + "step": 1255 + }, + { + "epoch": 1.85, + "grad_norm": 5.818370819091797, + "learning_rate": 0.0001839682418545848, + "loss": 3.3226, + "step": 1256 + }, + { + "epoch": 1.85, + "grad_norm": 4.127283573150635, + "learning_rate": 0.00018394276040294903, + "loss": 3.2937, + "step": 1257 + }, + { + "epoch": 1.85, + "grad_norm": 3.759086847305298, + "learning_rate": 0.00018391726048446852, + "loss": 3.3379, + "step": 1258 + }, + { + "epoch": 1.85, + "grad_norm": 4.152595520019531, + "learning_rate": 0.0001838917421047531, + "loss": 3.3433, + "step": 1259 + }, + { + "epoch": 1.86, + "grad_norm": 2.377365827560425, + "learning_rate": 0.00018386620526941662, + "loss": 3.3103, + "step": 1260 + }, + { + "epoch": 1.86, + "grad_norm": 1.4664210081100464, + "learning_rate": 0.00018384064998407706, + "loss": 3.3038, + "step": 1261 + }, + { + "epoch": 1.86, + "grad_norm": 5.424790859222412, + "learning_rate": 0.00018381507625435633, + "loss": 3.3518, + "step": 1262 + }, + { + "epoch": 1.86, + "grad_norm": 4.665937900543213, + "learning_rate": 0.00018378948408588058, + "loss": 3.3437, + "step": 1263 + }, + { + "epoch": 1.86, + "grad_norm": 2.893569231033325, + "learning_rate": 0.00018376387348427984, + "loss": 3.2693, + "step": 1264 + }, + { + "epoch": 1.86, + "grad_norm": 3.674450159072876, + "learning_rate": 0.0001837382444551883, + "loss": 3.3324, + "step": 1265 + }, + { + "epoch": 1.87, + "grad_norm": 2.98838210105896, + "learning_rate": 0.00018371259700424416, + "loss": 3.3242, + "step": 1266 + }, + { + "epoch": 1.87, + "grad_norm": 2.417462110519409, + "learning_rate": 0.00018368693113708968, + "loss": 3.3771, + "step": 1267 + }, + { + "epoch": 1.87, + "grad_norm": 4.9014458656311035, + "learning_rate": 0.00018366124685937122, + "loss": 3.3235, + "step": 1268 + }, + { + "epoch": 1.87, + "grad_norm": 4.8541693687438965, + "learning_rate": 0.00018363554417673915, + "loss": 3.3149, + "step": 1269 + }, + { + "epoch": 1.87, + "grad_norm": 3.2948014736175537, + "learning_rate": 0.0001836098230948478, + "loss": 3.3211, + "step": 1270 + }, + { + "epoch": 1.87, + "grad_norm": 3.0249054431915283, + "learning_rate": 0.00018358408361935573, + "loss": 3.3073, + "step": 1271 + }, + { + "epoch": 1.87, + "grad_norm": 4.5272722244262695, + "learning_rate": 0.00018355832575592541, + "loss": 3.2809, + "step": 1272 + }, + { + "epoch": 1.88, + "grad_norm": 3.790753126144409, + "learning_rate": 0.0001835325495102234, + "loss": 3.3502, + "step": 1273 + }, + { + "epoch": 1.88, + "grad_norm": 3.7797255516052246, + "learning_rate": 0.00018350675488792035, + "loss": 3.2818, + "step": 1274 + }, + { + "epoch": 1.88, + "grad_norm": 3.5615875720977783, + "learning_rate": 0.00018348094189469082, + "loss": 3.3617, + "step": 1275 + }, + { + "epoch": 1.88, + "grad_norm": 3.750861167907715, + "learning_rate": 0.00018345511053621355, + "loss": 3.3649, + "step": 1276 + }, + { + "epoch": 1.88, + "grad_norm": 3.202655076980591, + "learning_rate": 0.00018342926081817127, + "loss": 3.3013, + "step": 1277 + }, + { + "epoch": 1.88, + "grad_norm": 3.6577436923980713, + "learning_rate": 0.0001834033927462507, + "loss": 3.3074, + "step": 1278 + }, + { + "epoch": 1.88, + "grad_norm": 3.164114236831665, + "learning_rate": 0.0001833775063261427, + "loss": 3.3211, + "step": 1279 + }, + { + "epoch": 1.89, + "grad_norm": 3.7693588733673096, + "learning_rate": 0.00018335160156354206, + "loss": 3.3229, + "step": 1280 + }, + { + "epoch": 1.89, + "grad_norm": 3.5220518112182617, + "learning_rate": 0.00018332567846414768, + "loss": 3.304, + "step": 1281 + }, + { + "epoch": 1.89, + "grad_norm": 3.884568691253662, + "learning_rate": 0.00018329973703366242, + "loss": 3.2905, + "step": 1282 + }, + { + "epoch": 1.89, + "grad_norm": 3.471947431564331, + "learning_rate": 0.0001832737772777933, + "loss": 3.3356, + "step": 1283 + }, + { + "epoch": 1.89, + "grad_norm": 3.8152236938476562, + "learning_rate": 0.00018324779920225127, + "loss": 3.3036, + "step": 1284 + }, + { + "epoch": 1.89, + "grad_norm": 3.813997507095337, + "learning_rate": 0.0001832218028127513, + "loss": 3.3465, + "step": 1285 + }, + { + "epoch": 1.89, + "grad_norm": 3.5465102195739746, + "learning_rate": 0.00018319578811501244, + "loss": 3.3241, + "step": 1286 + }, + { + "epoch": 1.9, + "grad_norm": 3.25008487701416, + "learning_rate": 0.00018316975511475776, + "loss": 3.3111, + "step": 1287 + }, + { + "epoch": 1.9, + "grad_norm": 4.218858242034912, + "learning_rate": 0.00018314370381771432, + "loss": 3.2914, + "step": 1288 + }, + { + "epoch": 1.9, + "grad_norm": 3.7324070930480957, + "learning_rate": 0.00018311763422961325, + "loss": 3.3188, + "step": 1289 + }, + { + "epoch": 1.9, + "grad_norm": 3.4570472240448, + "learning_rate": 0.00018309154635618965, + "loss": 3.3034, + "step": 1290 + }, + { + "epoch": 1.9, + "grad_norm": 2.9619879722595215, + "learning_rate": 0.00018306544020318274, + "loss": 3.3231, + "step": 1291 + }, + { + "epoch": 1.9, + "grad_norm": 4.235932350158691, + "learning_rate": 0.00018303931577633566, + "loss": 3.3226, + "step": 1292 + }, + { + "epoch": 1.91, + "grad_norm": 3.5511438846588135, + "learning_rate": 0.00018301317308139561, + "loss": 3.3305, + "step": 1293 + }, + { + "epoch": 1.91, + "grad_norm": 3.656703233718872, + "learning_rate": 0.00018298701212411382, + "loss": 3.3261, + "step": 1294 + }, + { + "epoch": 1.91, + "grad_norm": 3.338941812515259, + "learning_rate": 0.00018296083291024552, + "loss": 3.3564, + "step": 1295 + }, + { + "epoch": 1.91, + "grad_norm": 3.9150142669677734, + "learning_rate": 0.00018293463544554993, + "loss": 3.3292, + "step": 1296 + }, + { + "epoch": 1.91, + "grad_norm": 10.699624061584473, + "learning_rate": 0.00018290841973579039, + "loss": 3.3042, + "step": 1297 + }, + { + "epoch": 1.91, + "grad_norm": 1.4264583587646484, + "learning_rate": 0.0001828821857867341, + "loss": 3.3268, + "step": 1298 + }, + { + "epoch": 1.91, + "grad_norm": 6.050600528717041, + "learning_rate": 0.0001828559336041524, + "loss": 3.3629, + "step": 1299 + }, + { + "epoch": 1.92, + "grad_norm": 6.324507236480713, + "learning_rate": 0.00018282966319382057, + "loss": 3.3402, + "step": 1300 + }, + { + "epoch": 1.92, + "grad_norm": 1.820309042930603, + "learning_rate": 0.00018280337456151797, + "loss": 3.2955, + "step": 1301 + }, + { + "epoch": 1.92, + "grad_norm": 7.433422565460205, + "learning_rate": 0.00018277706771302788, + "loss": 3.3309, + "step": 1302 + }, + { + "epoch": 1.92, + "grad_norm": 9.409677505493164, + "learning_rate": 0.00018275074265413762, + "loss": 3.4307, + "step": 1303 + }, + { + "epoch": 1.92, + "grad_norm": 5.268000602722168, + "learning_rate": 0.00018272439939063856, + "loss": 3.4344, + "step": 1304 + }, + { + "epoch": 1.92, + "grad_norm": 5.51171875, + "learning_rate": 0.00018269803792832603, + "loss": 3.3385, + "step": 1305 + }, + { + "epoch": 1.92, + "grad_norm": 5.903563976287842, + "learning_rate": 0.00018267165827299934, + "loss": 3.3243, + "step": 1306 + }, + { + "epoch": 1.93, + "grad_norm": 3.7264065742492676, + "learning_rate": 0.00018264526043046192, + "loss": 3.3759, + "step": 1307 + }, + { + "epoch": 1.93, + "grad_norm": 5.053271293640137, + "learning_rate": 0.00018261884440652106, + "loss": 3.3202, + "step": 1308 + }, + { + "epoch": 1.93, + "grad_norm": 2.33587646484375, + "learning_rate": 0.0001825924102069881, + "loss": 3.2899, + "step": 1309 + }, + { + "epoch": 1.93, + "grad_norm": 4.696453094482422, + "learning_rate": 0.00018256595783767843, + "loss": 3.359, + "step": 1310 + }, + { + "epoch": 1.93, + "grad_norm": 4.715218544006348, + "learning_rate": 0.00018253948730441132, + "loss": 3.3466, + "step": 1311 + }, + { + "epoch": 1.93, + "grad_norm": 2.377490997314453, + "learning_rate": 0.0001825129986130102, + "loss": 3.3496, + "step": 1312 + }, + { + "epoch": 1.94, + "grad_norm": 2.269477367401123, + "learning_rate": 0.00018248649176930235, + "loss": 3.3404, + "step": 1313 + }, + { + "epoch": 1.94, + "grad_norm": 4.577181816101074, + "learning_rate": 0.0001824599667791191, + "loss": 3.3864, + "step": 1314 + }, + { + "epoch": 1.94, + "grad_norm": 4.408148288726807, + "learning_rate": 0.0001824334236482958, + "loss": 3.3405, + "step": 1315 + }, + { + "epoch": 1.94, + "grad_norm": 2.9572770595550537, + "learning_rate": 0.00018240686238267173, + "loss": 3.323, + "step": 1316 + }, + { + "epoch": 1.94, + "grad_norm": 4.4048380851745605, + "learning_rate": 0.0001823802829880902, + "loss": 3.2676, + "step": 1317 + }, + { + "epoch": 1.94, + "grad_norm": 1.4889674186706543, + "learning_rate": 0.0001823536854703985, + "loss": 3.2902, + "step": 1318 + }, + { + "epoch": 1.94, + "grad_norm": 1.6742371320724487, + "learning_rate": 0.00018232706983544792, + "loss": 3.2945, + "step": 1319 + }, + { + "epoch": 1.95, + "grad_norm": 3.9357426166534424, + "learning_rate": 0.0001823004360890937, + "loss": 3.3486, + "step": 1320 + }, + { + "epoch": 1.95, + "grad_norm": 2.596331834793091, + "learning_rate": 0.0001822737842371951, + "loss": 3.3539, + "step": 1321 + }, + { + "epoch": 1.95, + "grad_norm": 4.866861820220947, + "learning_rate": 0.00018224711428561533, + "loss": 3.3449, + "step": 1322 + }, + { + "epoch": 1.95, + "grad_norm": 5.0658955574035645, + "learning_rate": 0.0001822204262402216, + "loss": 3.3107, + "step": 1323 + }, + { + "epoch": 1.95, + "grad_norm": 1.9593414068222046, + "learning_rate": 0.00018219372010688515, + "loss": 3.348, + "step": 1324 + }, + { + "epoch": 1.95, + "grad_norm": 2.256303310394287, + "learning_rate": 0.0001821669958914811, + "loss": 3.3012, + "step": 1325 + }, + { + "epoch": 1.95, + "grad_norm": 3.2346420288085938, + "learning_rate": 0.0001821402535998886, + "loss": 3.3372, + "step": 1326 + }, + { + "epoch": 1.96, + "grad_norm": 1.7795019149780273, + "learning_rate": 0.00018211349323799078, + "loss": 3.3459, + "step": 1327 + }, + { + "epoch": 1.96, + "grad_norm": 7.262651443481445, + "learning_rate": 0.00018208671481167478, + "loss": 3.3454, + "step": 1328 + }, + { + "epoch": 1.96, + "grad_norm": 8.501165390014648, + "learning_rate": 0.00018205991832683164, + "loss": 3.3952, + "step": 1329 + }, + { + "epoch": 1.96, + "grad_norm": 2.0047447681427, + "learning_rate": 0.00018203310378935638, + "loss": 3.3304, + "step": 1330 + }, + { + "epoch": 1.96, + "grad_norm": 6.586772918701172, + "learning_rate": 0.00018200627120514807, + "loss": 3.3235, + "step": 1331 + }, + { + "epoch": 1.96, + "grad_norm": 6.3678083419799805, + "learning_rate": 0.00018197942058010967, + "loss": 3.4105, + "step": 1332 + }, + { + "epoch": 1.96, + "grad_norm": 3.969449520111084, + "learning_rate": 0.00018195255192014816, + "loss": 3.3689, + "step": 1333 + }, + { + "epoch": 1.97, + "grad_norm": 5.26508092880249, + "learning_rate": 0.0001819256652311744, + "loss": 3.3417, + "step": 1334 + }, + { + "epoch": 1.97, + "grad_norm": 5.3484673500061035, + "learning_rate": 0.00018189876051910337, + "loss": 3.32, + "step": 1335 + }, + { + "epoch": 1.97, + "grad_norm": 3.664024829864502, + "learning_rate": 0.00018187183778985388, + "loss": 3.4004, + "step": 1336 + }, + { + "epoch": 1.97, + "grad_norm": 5.414148807525635, + "learning_rate": 0.00018184489704934877, + "loss": 3.2964, + "step": 1337 + }, + { + "epoch": 1.97, + "grad_norm": 1.5016615390777588, + "learning_rate": 0.0001818179383035148, + "loss": 3.3444, + "step": 1338 + }, + { + "epoch": 1.97, + "grad_norm": 7.925120830535889, + "learning_rate": 0.00018179096155828274, + "loss": 3.3881, + "step": 1339 + }, + { + "epoch": 1.98, + "grad_norm": 9.725903511047363, + "learning_rate": 0.00018176396681958724, + "loss": 3.3698, + "step": 1340 + }, + { + "epoch": 1.98, + "grad_norm": 5.562051773071289, + "learning_rate": 0.00018173695409336702, + "loss": 3.3285, + "step": 1341 + }, + { + "epoch": 1.98, + "grad_norm": 4.413424491882324, + "learning_rate": 0.00018170992338556468, + "loss": 3.3209, + "step": 1342 + }, + { + "epoch": 1.98, + "grad_norm": 5.979745864868164, + "learning_rate": 0.00018168287470212677, + "loss": 3.3173, + "step": 1343 + }, + { + "epoch": 1.98, + "grad_norm": 2.968111753463745, + "learning_rate": 0.00018165580804900384, + "loss": 3.3304, + "step": 1344 + }, + { + "epoch": 1.98, + "grad_norm": 6.121416091918945, + "learning_rate": 0.00018162872343215037, + "loss": 3.325, + "step": 1345 + }, + { + "epoch": 1.98, + "grad_norm": 5.006643772125244, + "learning_rate": 0.0001816016208575248, + "loss": 3.3311, + "step": 1346 + }, + { + "epoch": 1.99, + "grad_norm": 2.833677291870117, + "learning_rate": 0.00018157450033108954, + "loss": 3.3484, + "step": 1347 + }, + { + "epoch": 1.99, + "grad_norm": 3.661172389984131, + "learning_rate": 0.00018154736185881088, + "loss": 3.3119, + "step": 1348 + }, + { + "epoch": 1.99, + "grad_norm": 1.4361642599105835, + "learning_rate": 0.0001815202054466591, + "loss": 3.3744, + "step": 1349 + }, + { + "epoch": 1.99, + "grad_norm": 2.236370801925659, + "learning_rate": 0.00018149303110060846, + "loss": 3.285, + "step": 1350 + }, + { + "epoch": 1.99, + "grad_norm": 3.9948387145996094, + "learning_rate": 0.00018146583882663712, + "loss": 3.3107, + "step": 1351 + }, + { + "epoch": 1.99, + "grad_norm": 2.2754180431365967, + "learning_rate": 0.0001814386286307272, + "loss": 3.3646, + "step": 1352 + }, + { + "epoch": 1.99, + "eval_loss": 3.131267786026001, + "eval_runtime": 1163.1, + "eval_samples_per_second": 14.738, + "eval_steps_per_second": 0.922, + "step": 1352 + }, + { + "epoch": 1.99, + "grad_norm": 4.385812282562256, + "learning_rate": 0.00018141140051886474, + "loss": 3.3364, + "step": 1353 + }, + { + "epoch": 2.0, + "grad_norm": 2.7103238105773926, + "learning_rate": 0.0001813841544970398, + "loss": 3.3096, + "step": 1354 + }, + { + "epoch": 2.0, + "grad_norm": 4.108714580535889, + "learning_rate": 0.00018135689057124625, + "loss": 3.3506, + "step": 1355 + }, + { + "epoch": 2.0, + "grad_norm": 3.6100189685821533, + "learning_rate": 0.000181329608747482, + "loss": 3.2999, + "step": 1356 + }, + { + "epoch": 2.0, + "grad_norm": 3.728175640106201, + "learning_rate": 0.0001813023090317489, + "loss": 3.2945, + "step": 1357 + }, + { + "epoch": 2.0, + "grad_norm": 3.231055974960327, + "learning_rate": 0.00018127499143005268, + "loss": 3.3229, + "step": 1358 + }, + { + "epoch": 2.0, + "grad_norm": 3.6905510425567627, + "learning_rate": 0.00018124765594840298, + "loss": 3.3186, + "step": 1359 + }, + { + "epoch": 2.0, + "grad_norm": 3.330751657485962, + "learning_rate": 0.0001812203025928135, + "loss": 3.3526, + "step": 1360 + }, + { + "epoch": 2.01, + "grad_norm": 3.546685218811035, + "learning_rate": 0.00018119293136930175, + "loss": 3.2914, + "step": 1361 + }, + { + "epoch": 2.01, + "grad_norm": 3.202545404434204, + "learning_rate": 0.00018116554228388924, + "loss": 3.3155, + "step": 1362 + }, + { + "epoch": 2.01, + "grad_norm": 3.762091875076294, + "learning_rate": 0.00018113813534260137, + "loss": 3.2653, + "step": 1363 + }, + { + "epoch": 2.01, + "grad_norm": 3.2883477210998535, + "learning_rate": 0.00018111071055146752, + "loss": 3.2802, + "step": 1364 + }, + { + "epoch": 2.01, + "grad_norm": 4.180018424987793, + "learning_rate": 0.0001810832679165209, + "loss": 3.2743, + "step": 1365 + }, + { + "epoch": 2.01, + "grad_norm": 3.009850025177002, + "learning_rate": 0.00018105580744379873, + "loss": 3.3396, + "step": 1366 + }, + { + "epoch": 2.02, + "grad_norm": 3.3239614963531494, + "learning_rate": 0.00018102832913934215, + "loss": 3.302, + "step": 1367 + }, + { + "epoch": 2.0, + "grad_norm": 2.7158830165863037, + "learning_rate": 0.00018100083300919616, + "loss": 3.2927, + "step": 1368 + }, + { + "epoch": 2.0, + "grad_norm": 4.113391399383545, + "learning_rate": 0.00018097331905940977, + "loss": 3.258, + "step": 1369 + }, + { + "epoch": 2.0, + "grad_norm": 4.452334880828857, + "learning_rate": 0.00018094578729603583, + "loss": 3.2988, + "step": 1370 + }, + { + "epoch": 2.0, + "grad_norm": 2.8618204593658447, + "learning_rate": 0.00018091823772513118, + "loss": 3.2938, + "step": 1371 + }, + { + "epoch": 2.01, + "grad_norm": 2.671171188354492, + "learning_rate": 0.00018089067035275652, + "loss": 3.2756, + "step": 1372 + }, + { + "epoch": 2.01, + "grad_norm": 4.292674541473389, + "learning_rate": 0.0001808630851849765, + "loss": 3.2655, + "step": 1373 + }, + { + "epoch": 2.01, + "grad_norm": 4.163233757019043, + "learning_rate": 0.00018083548222785965, + "loss": 3.2825, + "step": 1374 + }, + { + "epoch": 2.01, + "grad_norm": 2.8590521812438965, + "learning_rate": 0.00018080786148747842, + "loss": 3.3127, + "step": 1375 + }, + { + "epoch": 2.01, + "grad_norm": 2.4040653705596924, + "learning_rate": 0.00018078022296990923, + "loss": 3.348, + "step": 1376 + }, + { + "epoch": 2.01, + "grad_norm": 5.037558555603027, + "learning_rate": 0.00018075256668123238, + "loss": 3.275, + "step": 1377 + }, + { + "epoch": 2.02, + "grad_norm": 4.536971569061279, + "learning_rate": 0.000180724892627532, + "loss": 3.2533, + "step": 1378 + }, + { + "epoch": 2.02, + "grad_norm": 2.9320995807647705, + "learning_rate": 0.00018069720081489624, + "loss": 3.3223, + "step": 1379 + }, + { + "epoch": 2.02, + "grad_norm": 2.6839616298675537, + "learning_rate": 0.00018066949124941712, + "loss": 3.3313, + "step": 1380 + }, + { + "epoch": 2.02, + "grad_norm": 3.9813764095306396, + "learning_rate": 0.00018064176393719055, + "loss": 3.2821, + "step": 1381 + }, + { + "epoch": 2.02, + "grad_norm": 3.4426040649414062, + "learning_rate": 0.00018061401888431634, + "loss": 3.3005, + "step": 1382 + }, + { + "epoch": 2.02, + "grad_norm": 3.2250571250915527, + "learning_rate": 0.00018058625609689824, + "loss": 3.2879, + "step": 1383 + }, + { + "epoch": 2.02, + "grad_norm": 2.6666173934936523, + "learning_rate": 0.00018055847558104383, + "loss": 3.2522, + "step": 1384 + }, + { + "epoch": 2.03, + "grad_norm": 4.8121843338012695, + "learning_rate": 0.00018053067734286472, + "loss": 3.3785, + "step": 1385 + }, + { + "epoch": 2.03, + "grad_norm": 4.593588352203369, + "learning_rate": 0.00018050286138847628, + "loss": 3.2831, + "step": 1386 + }, + { + "epoch": 2.03, + "grad_norm": 3.0059878826141357, + "learning_rate": 0.0001804750277239978, + "loss": 3.2606, + "step": 1387 + }, + { + "epoch": 2.03, + "grad_norm": 2.9211039543151855, + "learning_rate": 0.00018044717635555259, + "loss": 3.2735, + "step": 1388 + }, + { + "epoch": 2.03, + "grad_norm": 3.4243338108062744, + "learning_rate": 0.00018041930728926768, + "loss": 3.2794, + "step": 1389 + }, + { + "epoch": 2.03, + "grad_norm": 3.398374557495117, + "learning_rate": 0.00018039142053127412, + "loss": 3.3066, + "step": 1390 + }, + { + "epoch": 2.03, + "grad_norm": 3.260343074798584, + "learning_rate": 0.0001803635160877068, + "loss": 3.3, + "step": 1391 + }, + { + "epoch": 2.04, + "grad_norm": 2.542720079421997, + "learning_rate": 0.00018033559396470454, + "loss": 3.2654, + "step": 1392 + }, + { + "epoch": 2.04, + "grad_norm": 4.721008777618408, + "learning_rate": 0.00018030765416840997, + "loss": 3.3286, + "step": 1393 + }, + { + "epoch": 2.04, + "grad_norm": 4.219196319580078, + "learning_rate": 0.00018027969670496966, + "loss": 3.2769, + "step": 1394 + }, + { + "epoch": 2.04, + "grad_norm": 2.8509461879730225, + "learning_rate": 0.00018025172158053413, + "loss": 3.2763, + "step": 1395 + }, + { + "epoch": 2.04, + "grad_norm": 3.1871495246887207, + "learning_rate": 0.00018022372880125765, + "loss": 3.3038, + "step": 1396 + }, + { + "epoch": 2.04, + "grad_norm": 4.717142105102539, + "learning_rate": 0.00018019571837329847, + "loss": 3.2649, + "step": 1397 + }, + { + "epoch": 2.04, + "grad_norm": 3.9490749835968018, + "learning_rate": 0.00018016769030281872, + "loss": 3.2709, + "step": 1398 + }, + { + "epoch": 2.05, + "grad_norm": 3.04848051071167, + "learning_rate": 0.00018013964459598434, + "loss": 3.2638, + "step": 1399 + }, + { + "epoch": 2.05, + "grad_norm": 2.971536636352539, + "learning_rate": 0.00018011158125896522, + "loss": 3.3796, + "step": 1400 + }, + { + "epoch": 2.05, + "grad_norm": 3.643239736557007, + "learning_rate": 0.0001800835002979351, + "loss": 3.2766, + "step": 1401 + }, + { + "epoch": 2.05, + "grad_norm": 3.0714406967163086, + "learning_rate": 0.00018005540171907162, + "loss": 3.2691, + "step": 1402 + }, + { + "epoch": 2.05, + "grad_norm": 3.928359031677246, + "learning_rate": 0.00018002728552855629, + "loss": 3.2842, + "step": 1403 + }, + { + "epoch": 2.05, + "grad_norm": 3.7628214359283447, + "learning_rate": 0.00017999915173257444, + "loss": 3.3371, + "step": 1404 + }, + { + "epoch": 2.06, + "grad_norm": 3.1093904972076416, + "learning_rate": 0.00017997100033731534, + "loss": 3.2866, + "step": 1405 + }, + { + "epoch": 2.06, + "grad_norm": 2.735902786254883, + "learning_rate": 0.0001799428313489721, + "loss": 3.2598, + "step": 1406 + }, + { + "epoch": 2.06, + "grad_norm": 3.890085458755493, + "learning_rate": 0.00017991464477374173, + "loss": 3.2565, + "step": 1407 + }, + { + "epoch": 2.06, + "grad_norm": 3.5979950428009033, + "learning_rate": 0.00017988644061782508, + "loss": 3.3153, + "step": 1408 + }, + { + "epoch": 2.06, + "grad_norm": 3.224681854248047, + "learning_rate": 0.00017985821888742685, + "loss": 3.2969, + "step": 1409 + }, + { + "epoch": 2.06, + "grad_norm": 3.645235776901245, + "learning_rate": 0.00017982997958875565, + "loss": 3.2823, + "step": 1410 + }, + { + "epoch": 2.06, + "grad_norm": 3.3547754287719727, + "learning_rate": 0.000179801722728024, + "loss": 3.3546, + "step": 1411 + }, + { + "epoch": 2.07, + "grad_norm": 2.8428351879119873, + "learning_rate": 0.00017977344831144807, + "loss": 3.3096, + "step": 1412 + }, + { + "epoch": 2.07, + "grad_norm": 4.488086223602295, + "learning_rate": 0.00017974515634524817, + "loss": 3.2783, + "step": 1413 + }, + { + "epoch": 2.07, + "grad_norm": 4.082129955291748, + "learning_rate": 0.0001797168468356483, + "loss": 3.3131, + "step": 1414 + }, + { + "epoch": 2.07, + "grad_norm": 3.037198305130005, + "learning_rate": 0.00017968851978887636, + "loss": 3.3435, + "step": 1415 + }, + { + "epoch": 2.07, + "grad_norm": 2.8374056816101074, + "learning_rate": 0.0001796601752111641, + "loss": 3.3026, + "step": 1416 + }, + { + "epoch": 2.07, + "grad_norm": 4.270615577697754, + "learning_rate": 0.00017963181310874716, + "loss": 3.3384, + "step": 1417 + }, + { + "epoch": 2.07, + "grad_norm": 4.1699323654174805, + "learning_rate": 0.00017960343348786504, + "loss": 3.2869, + "step": 1418 + }, + { + "epoch": 2.08, + "grad_norm": 2.7383670806884766, + "learning_rate": 0.00017957503635476099, + "loss": 3.2882, + "step": 1419 + }, + { + "epoch": 2.08, + "grad_norm": 2.707303762435913, + "learning_rate": 0.00017954662171568226, + "loss": 3.3616, + "step": 1420 + }, + { + "epoch": 2.08, + "grad_norm": 5.056338787078857, + "learning_rate": 0.00017951818957687981, + "loss": 3.381, + "step": 1421 + }, + { + "epoch": 2.08, + "grad_norm": 4.49586820602417, + "learning_rate": 0.0001794897399446086, + "loss": 3.2837, + "step": 1422 + }, + { + "epoch": 2.08, + "grad_norm": 2.5468156337738037, + "learning_rate": 0.0001794612728251273, + "loss": 3.2547, + "step": 1423 + }, + { + "epoch": 2.08, + "grad_norm": 2.2053098678588867, + "learning_rate": 0.00017943278822469852, + "loss": 3.3216, + "step": 1424 + }, + { + "epoch": 2.09, + "grad_norm": 4.875980854034424, + "learning_rate": 0.00017940428614958865, + "loss": 3.3319, + "step": 1425 + }, + { + "epoch": 2.09, + "grad_norm": 4.466098308563232, + "learning_rate": 0.000179375766606068, + "loss": 3.276, + "step": 1426 + }, + { + "epoch": 2.09, + "grad_norm": 2.8682076930999756, + "learning_rate": 0.0001793472296004106, + "loss": 3.2971, + "step": 1427 + }, + { + "epoch": 2.09, + "grad_norm": 2.2296552658081055, + "learning_rate": 0.00017931867513889448, + "loss": 3.2946, + "step": 1428 + }, + { + "epoch": 2.09, + "grad_norm": 4.954530239105225, + "learning_rate": 0.0001792901032278014, + "loss": 3.269, + "step": 1429 + }, + { + "epoch": 2.09, + "grad_norm": 4.960486888885498, + "learning_rate": 0.00017926151387341697, + "loss": 3.2616, + "step": 1430 + }, + { + "epoch": 2.09, + "grad_norm": 2.2952184677124023, + "learning_rate": 0.0001792329070820307, + "loss": 3.3348, + "step": 1431 + }, + { + "epoch": 2.1, + "grad_norm": 1.9920496940612793, + "learning_rate": 0.00017920428285993583, + "loss": 3.3137, + "step": 1432 + }, + { + "epoch": 2.1, + "grad_norm": 4.211342811584473, + "learning_rate": 0.00017917564121342958, + "loss": 3.3246, + "step": 1433 + }, + { + "epoch": 2.1, + "grad_norm": 3.4940345287323, + "learning_rate": 0.00017914698214881284, + "loss": 3.3421, + "step": 1434 + }, + { + "epoch": 2.1, + "grad_norm": 3.7058966159820557, + "learning_rate": 0.00017911830567239048, + "loss": 3.3377, + "step": 1435 + }, + { + "epoch": 2.1, + "grad_norm": 3.2200000286102295, + "learning_rate": 0.00017908961179047105, + "loss": 3.3125, + "step": 1436 + }, + { + "epoch": 2.1, + "grad_norm": 3.529256582260132, + "learning_rate": 0.00017906090050936712, + "loss": 3.2833, + "step": 1437 + }, + { + "epoch": 2.1, + "grad_norm": 3.189732313156128, + "learning_rate": 0.0001790321718353949, + "loss": 3.2662, + "step": 1438 + }, + { + "epoch": 2.11, + "grad_norm": 3.6067655086517334, + "learning_rate": 0.0001790034257748745, + "loss": 3.2795, + "step": 1439 + }, + { + "epoch": 2.11, + "grad_norm": 3.4113893508911133, + "learning_rate": 0.0001789746623341299, + "loss": 3.2666, + "step": 1440 + }, + { + "epoch": 2.11, + "grad_norm": 3.466991662979126, + "learning_rate": 0.00017894588151948885, + "loss": 3.325, + "step": 1441 + }, + { + "epoch": 2.11, + "grad_norm": 3.057206153869629, + "learning_rate": 0.00017891708333728296, + "loss": 3.2636, + "step": 1442 + }, + { + "epoch": 2.11, + "grad_norm": 3.8873162269592285, + "learning_rate": 0.0001788882677938476, + "loss": 3.3139, + "step": 1443 + }, + { + "epoch": 2.11, + "grad_norm": 3.4515161514282227, + "learning_rate": 0.000178859434895522, + "loss": 3.2884, + "step": 1444 + }, + { + "epoch": 2.11, + "grad_norm": 14.146556854248047, + "learning_rate": 0.00017883058464864923, + "loss": 3.2571, + "step": 1445 + }, + { + "epoch": 2.12, + "grad_norm": 9.11373519897461, + "learning_rate": 0.00017880171705957613, + "loss": 3.295, + "step": 1446 + }, + { + "epoch": 2.12, + "grad_norm": 5.7905049324035645, + "learning_rate": 0.0001787728321346534, + "loss": 3.2532, + "step": 1447 + }, + { + "epoch": 2.12, + "grad_norm": 3.7080447673797607, + "learning_rate": 0.0001787439298802355, + "loss": 3.3201, + "step": 1448 + }, + { + "epoch": 2.12, + "grad_norm": 4.587584018707275, + "learning_rate": 0.00017871501030268072, + "loss": 3.2702, + "step": 1449 + }, + { + "epoch": 2.12, + "grad_norm": 0.9951494932174683, + "learning_rate": 0.0001786860734083512, + "loss": 3.2862, + "step": 1450 + }, + { + "epoch": 2.12, + "grad_norm": 0.8612810373306274, + "learning_rate": 0.00017865711920361292, + "loss": 3.2968, + "step": 1451 + }, + { + "epoch": 2.13, + "grad_norm": 2.446154832839966, + "learning_rate": 0.00017862814769483549, + "loss": 3.2937, + "step": 1452 + }, + { + "epoch": 2.13, + "grad_norm": 1.2420624494552612, + "learning_rate": 0.00017859915888839253, + "loss": 3.2922, + "step": 1453 + }, + { + "epoch": 2.13, + "grad_norm": 1.9885892868041992, + "learning_rate": 0.00017857015279066135, + "loss": 3.3043, + "step": 1454 + }, + { + "epoch": 2.13, + "grad_norm": 2.127197504043579, + "learning_rate": 0.00017854112940802314, + "loss": 3.3503, + "step": 1455 + }, + { + "epoch": 2.13, + "grad_norm": 1.1510310173034668, + "learning_rate": 0.0001785120887468628, + "loss": 3.2314, + "step": 1456 + }, + { + "epoch": 2.13, + "grad_norm": 2.6665754318237305, + "learning_rate": 0.00017848303081356912, + "loss": 3.2788, + "step": 1457 + }, + { + "epoch": 2.13, + "grad_norm": 1.0793408155441284, + "learning_rate": 0.0001784539556145346, + "loss": 3.2623, + "step": 1458 + }, + { + "epoch": 2.14, + "grad_norm": 2.212766647338867, + "learning_rate": 0.00017842486315615563, + "loss": 3.3542, + "step": 1459 + }, + { + "epoch": 2.14, + "grad_norm": 1.8812615871429443, + "learning_rate": 0.00017839575344483238, + "loss": 3.2481, + "step": 1460 + }, + { + "epoch": 2.14, + "grad_norm": 1.537009358406067, + "learning_rate": 0.0001783666264869687, + "loss": 3.2492, + "step": 1461 + }, + { + "epoch": 2.14, + "grad_norm": 1.0844135284423828, + "learning_rate": 0.00017833748228897241, + "loss": 3.3093, + "step": 1462 + }, + { + "epoch": 2.14, + "grad_norm": 1.4660224914550781, + "learning_rate": 0.000178308320857255, + "loss": 3.2598, + "step": 1463 + }, + { + "epoch": 2.14, + "grad_norm": 0.9054508209228516, + "learning_rate": 0.00017827914219823183, + "loss": 3.3132, + "step": 1464 + }, + { + "epoch": 2.14, + "grad_norm": 5.584422588348389, + "learning_rate": 0.00017824994631832197, + "loss": 3.2961, + "step": 1465 + }, + { + "epoch": 2.15, + "grad_norm": 1.9765104055404663, + "learning_rate": 0.0001782207332239483, + "loss": 3.2963, + "step": 1466 + }, + { + "epoch": 2.15, + "grad_norm": 1.4504390954971313, + "learning_rate": 0.00017819150292153757, + "loss": 3.3335, + "step": 1467 + }, + { + "epoch": 2.15, + "grad_norm": 1.2761657238006592, + "learning_rate": 0.0001781622554175202, + "loss": 3.2779, + "step": 1468 + }, + { + "epoch": 2.15, + "grad_norm": 2.1837310791015625, + "learning_rate": 0.00017813299071833046, + "loss": 3.2791, + "step": 1469 + }, + { + "epoch": 2.15, + "grad_norm": 2.7105600833892822, + "learning_rate": 0.00017810370883040636, + "loss": 3.2983, + "step": 1470 + }, + { + "epoch": 2.15, + "grad_norm": 1.1672013998031616, + "learning_rate": 0.00017807440976018977, + "loss": 3.2989, + "step": 1471 + }, + { + "epoch": 2.15, + "grad_norm": 4.486381530761719, + "learning_rate": 0.00017804509351412629, + "loss": 3.3132, + "step": 1472 + }, + { + "epoch": 2.16, + "grad_norm": 3.2585389614105225, + "learning_rate": 0.0001780157600986652, + "loss": 3.3249, + "step": 1473 + }, + { + "epoch": 2.16, + "grad_norm": 3.1671297550201416, + "learning_rate": 0.00017798640952025978, + "loss": 3.31, + "step": 1474 + }, + { + "epoch": 2.16, + "grad_norm": 2.345313310623169, + "learning_rate": 0.0001779570417853669, + "loss": 3.2382, + "step": 1475 + }, + { + "epoch": 2.16, + "grad_norm": 5.773190498352051, + "learning_rate": 0.0001779276569004473, + "loss": 3.2903, + "step": 1476 + }, + { + "epoch": 2.16, + "grad_norm": 4.879696846008301, + "learning_rate": 0.00017789825487196537, + "loss": 3.3107, + "step": 1477 + }, + { + "epoch": 2.16, + "grad_norm": 4.470544815063477, + "learning_rate": 0.00017786883570638942, + "loss": 3.3, + "step": 1478 + }, + { + "epoch": 2.17, + "grad_norm": 5.051728248596191, + "learning_rate": 0.00017783939941019153, + "loss": 3.2964, + "step": 1479 + }, + { + "epoch": 2.17, + "grad_norm": 1.2672648429870605, + "learning_rate": 0.00017780994598984737, + "loss": 3.297, + "step": 1480 + }, + { + "epoch": 2.17, + "grad_norm": 5.323600769042969, + "learning_rate": 0.0001777804754518366, + "loss": 3.2878, + "step": 1481 + }, + { + "epoch": 2.17, + "grad_norm": 4.052299976348877, + "learning_rate": 0.00017775098780264243, + "loss": 3.3135, + "step": 1482 + }, + { + "epoch": 2.17, + "grad_norm": 3.538057565689087, + "learning_rate": 0.00017772148304875205, + "loss": 3.3209, + "step": 1483 + }, + { + "epoch": 2.17, + "grad_norm": 2.5743610858917236, + "learning_rate": 0.0001776919611966563, + "loss": 3.2468, + "step": 1484 + }, + { + "epoch": 2.17, + "grad_norm": 6.518936634063721, + "learning_rate": 0.00017766242225284966, + "loss": 3.2914, + "step": 1485 + }, + { + "epoch": 2.18, + "grad_norm": 5.448997497558594, + "learning_rate": 0.00017763286622383067, + "loss": 3.3127, + "step": 1486 + }, + { + "epoch": 2.18, + "grad_norm": 2.5418717861175537, + "learning_rate": 0.00017760329311610135, + "loss": 3.2913, + "step": 1487 + }, + { + "epoch": 2.18, + "grad_norm": 4.122663974761963, + "learning_rate": 0.00017757370293616762, + "loss": 3.3257, + "step": 1488 + }, + { + "epoch": 2.18, + "grad_norm": 3.3550655841827393, + "learning_rate": 0.00017754409569053917, + "loss": 3.271, + "step": 1489 + }, + { + "epoch": 2.18, + "grad_norm": 2.8279335498809814, + "learning_rate": 0.0001775144713857293, + "loss": 3.2989, + "step": 1490 + }, + { + "epoch": 2.18, + "grad_norm": 2.4097659587860107, + "learning_rate": 0.0001774848300282552, + "loss": 3.2557, + "step": 1491 + }, + { + "epoch": 2.18, + "grad_norm": 2.702836275100708, + "learning_rate": 0.00017745517162463776, + "loss": 3.2719, + "step": 1492 + }, + { + "epoch": 2.19, + "grad_norm": 1.937203288078308, + "learning_rate": 0.00017742549618140167, + "loss": 3.2801, + "step": 1493 + }, + { + "epoch": 2.19, + "grad_norm": 2.2853002548217773, + "learning_rate": 0.00017739580370507532, + "loss": 3.4001, + "step": 1494 + }, + { + "epoch": 2.19, + "grad_norm": 3.39026141166687, + "learning_rate": 0.00017736609420219082, + "loss": 3.2562, + "step": 1495 + }, + { + "epoch": 2.19, + "grad_norm": 2.192948341369629, + "learning_rate": 0.00017733636767928407, + "loss": 3.2548, + "step": 1496 + }, + { + "epoch": 2.19, + "grad_norm": 3.966799259185791, + "learning_rate": 0.00017730662414289472, + "loss": 3.2744, + "step": 1497 + }, + { + "epoch": 2.19, + "grad_norm": 2.6247634887695312, + "learning_rate": 0.00017727686359956615, + "loss": 3.287, + "step": 1498 + }, + { + "epoch": 2.19, + "grad_norm": 2.959963083267212, + "learning_rate": 0.00017724708605584545, + "loss": 3.2864, + "step": 1499 + }, + { + "epoch": 2.2, + "grad_norm": 1.7757418155670166, + "learning_rate": 0.00017721729151828355, + "loss": 3.3003, + "step": 1500 + }, + { + "epoch": 2.2, + "grad_norm": 3.2801480293273926, + "learning_rate": 0.00017718747999343498, + "loss": 3.2662, + "step": 1501 + }, + { + "epoch": 2.2, + "grad_norm": 2.6395089626312256, + "learning_rate": 0.0001771576514878581, + "loss": 3.2991, + "step": 1502 + }, + { + "epoch": 2.2, + "grad_norm": 2.687575340270996, + "learning_rate": 0.00017712780600811497, + "loss": 3.2676, + "step": 1503 + }, + { + "epoch": 2.2, + "grad_norm": 2.702707052230835, + "learning_rate": 0.00017709794356077143, + "loss": 3.2862, + "step": 1504 + }, + { + "epoch": 2.2, + "grad_norm": 2.10098934173584, + "learning_rate": 0.000177068064152397, + "loss": 3.2425, + "step": 1505 + }, + { + "epoch": 2.21, + "grad_norm": 1.3301888704299927, + "learning_rate": 0.0001770381677895649, + "loss": 3.2325, + "step": 1506 + }, + { + "epoch": 2.21, + "grad_norm": 3.560911178588867, + "learning_rate": 0.0001770082544788522, + "loss": 3.2914, + "step": 1507 + }, + { + "epoch": 2.21, + "grad_norm": 2.01891827583313, + "learning_rate": 0.0001769783242268396, + "loss": 3.2213, + "step": 1508 + }, + { + "epoch": 2.21, + "grad_norm": 3.6540088653564453, + "learning_rate": 0.00017694837704011155, + "loss": 3.3066, + "step": 1509 + }, + { + "epoch": 2.21, + "grad_norm": 3.1314940452575684, + "learning_rate": 0.00017691841292525625, + "loss": 3.2576, + "step": 1510 + }, + { + "epoch": 2.21, + "grad_norm": 3.816842555999756, + "learning_rate": 0.0001768884318888656, + "loss": 3.2784, + "step": 1511 + }, + { + "epoch": 2.21, + "grad_norm": 3.5961854457855225, + "learning_rate": 0.0001768584339375352, + "loss": 3.2562, + "step": 1512 + }, + { + "epoch": 2.22, + "grad_norm": 3.9628171920776367, + "learning_rate": 0.0001768284190778644, + "loss": 3.2561, + "step": 1513 + }, + { + "epoch": 2.22, + "grad_norm": 4.071855068206787, + "learning_rate": 0.00017679838731645632, + "loss": 3.2924, + "step": 1514 + }, + { + "epoch": 2.22, + "grad_norm": 2.176236152648926, + "learning_rate": 0.00017676833865991765, + "loss": 3.309, + "step": 1515 + }, + { + "epoch": 2.22, + "grad_norm": 1.6997188329696655, + "learning_rate": 0.000176738273114859, + "loss": 3.2835, + "step": 1516 + }, + { + "epoch": 2.22, + "grad_norm": 3.275249719619751, + "learning_rate": 0.00017670819068789454, + "loss": 3.2506, + "step": 1517 + }, + { + "epoch": 2.22, + "grad_norm": 1.3623771667480469, + "learning_rate": 0.0001766780913856422, + "loss": 3.2054, + "step": 1518 + }, + { + "epoch": 2.22, + "grad_norm": 5.944447994232178, + "learning_rate": 0.00017664797521472364, + "loss": 3.2794, + "step": 1519 + }, + { + "epoch": 2.23, + "grad_norm": 5.1751484870910645, + "learning_rate": 0.0001766178421817642, + "loss": 3.2498, + "step": 1520 + }, + { + "epoch": 2.23, + "grad_norm": 2.410102128982544, + "learning_rate": 0.00017658769229339294, + "loss": 3.2672, + "step": 1521 + }, + { + "epoch": 2.23, + "eval_loss": 3.0885374546051025, + "eval_runtime": 1164.3656, + "eval_samples_per_second": 14.722, + "eval_steps_per_second": 0.921, + "step": 1521 + }, + { + "epoch": 2.23, + "grad_norm": 2.7669081687927246, + "learning_rate": 0.00017655752555624263, + "loss": 3.2525, + "step": 1522 + }, + { + "epoch": 2.23, + "grad_norm": 4.010193347930908, + "learning_rate": 0.00017652734197694977, + "loss": 3.2827, + "step": 1523 + }, + { + "epoch": 2.23, + "grad_norm": 3.233734130859375, + "learning_rate": 0.00017649714156215456, + "loss": 3.2372, + "step": 1524 + }, + { + "epoch": 2.23, + "grad_norm": 3.2297024726867676, + "learning_rate": 0.00017646692431850085, + "loss": 3.2763, + "step": 1525 + }, + { + "epoch": 2.24, + "grad_norm": 2.7911462783813477, + "learning_rate": 0.00017643669025263625, + "loss": 3.2702, + "step": 1526 + }, + { + "epoch": 2.24, + "grad_norm": 2.8759915828704834, + "learning_rate": 0.00017640643937121207, + "loss": 3.2338, + "step": 1527 + }, + { + "epoch": 2.24, + "grad_norm": 2.3804805278778076, + "learning_rate": 0.00017637617168088325, + "loss": 3.2646, + "step": 1528 + }, + { + "epoch": 2.24, + "grad_norm": 4.322963237762451, + "learning_rate": 0.00017634588718830853, + "loss": 3.2743, + "step": 1529 + }, + { + "epoch": 2.24, + "grad_norm": 3.6001713275909424, + "learning_rate": 0.00017631558590015028, + "loss": 3.3172, + "step": 1530 + }, + { + "epoch": 2.24, + "grad_norm": 3.2861557006835938, + "learning_rate": 0.00017628526782307455, + "loss": 3.2723, + "step": 1531 + }, + { + "epoch": 2.24, + "grad_norm": 3.6503138542175293, + "learning_rate": 0.0001762549329637512, + "loss": 3.234, + "step": 1532 + }, + { + "epoch": 2.25, + "grad_norm": 3.1032047271728516, + "learning_rate": 0.00017622458132885357, + "loss": 3.2845, + "step": 1533 + }, + { + "epoch": 2.25, + "grad_norm": 2.5854177474975586, + "learning_rate": 0.00017619421292505892, + "loss": 3.2251, + "step": 1534 + }, + { + "epoch": 2.25, + "grad_norm": 4.162723064422607, + "learning_rate": 0.00017616382775904806, + "loss": 3.2519, + "step": 1535 + }, + { + "epoch": 2.25, + "grad_norm": 3.7609925270080566, + "learning_rate": 0.0001761334258375055, + "loss": 3.334, + "step": 1536 + }, + { + "epoch": 2.25, + "grad_norm": 3.426297664642334, + "learning_rate": 0.00017610300716711948, + "loss": 3.2644, + "step": 1537 + }, + { + "epoch": 2.25, + "grad_norm": 3.2877280712127686, + "learning_rate": 0.0001760725717545819, + "loss": 3.2466, + "step": 1538 + }, + { + "epoch": 2.25, + "grad_norm": 3.2899694442749023, + "learning_rate": 0.00017604211960658837, + "loss": 3.2245, + "step": 1539 + }, + { + "epoch": 2.26, + "grad_norm": 3.202381134033203, + "learning_rate": 0.0001760116507298381, + "loss": 3.2512, + "step": 1540 + }, + { + "epoch": 2.26, + "grad_norm": 9.299320220947266, + "learning_rate": 0.0001759811651310341, + "loss": 3.2295, + "step": 1541 + }, + { + "epoch": 2.26, + "grad_norm": 6.585042476654053, + "learning_rate": 0.000175950662816883, + "loss": 3.2845, + "step": 1542 + }, + { + "epoch": 2.26, + "grad_norm": 3.786369562149048, + "learning_rate": 0.00017592014379409502, + "loss": 3.2796, + "step": 1543 + }, + { + "epoch": 2.26, + "grad_norm": 5.25980281829834, + "learning_rate": 0.00017588960806938424, + "loss": 3.2427, + "step": 1544 + }, + { + "epoch": 2.26, + "grad_norm": 7.450433254241943, + "learning_rate": 0.00017585905564946827, + "loss": 3.3147, + "step": 1545 + }, + { + "epoch": 2.26, + "grad_norm": 3.1463863849639893, + "learning_rate": 0.00017582848654106845, + "loss": 3.2311, + "step": 1546 + }, + { + "epoch": 2.27, + "grad_norm": 5.310190200805664, + "learning_rate": 0.00017579790075090976, + "loss": 3.3092, + "step": 1547 + }, + { + "epoch": 2.27, + "grad_norm": 5.930779933929443, + "learning_rate": 0.0001757672982857209, + "loss": 3.2788, + "step": 1548 + }, + { + "epoch": 2.27, + "grad_norm": 2.841097831726074, + "learning_rate": 0.00017573667915223414, + "loss": 3.1998, + "step": 1549 + }, + { + "epoch": 2.27, + "grad_norm": 7.447371959686279, + "learning_rate": 0.0001757060433571856, + "loss": 3.2633, + "step": 1550 + }, + { + "epoch": 2.27, + "grad_norm": 7.628240585327148, + "learning_rate": 0.00017567539090731482, + "loss": 3.2626, + "step": 1551 + }, + { + "epoch": 2.27, + "grad_norm": 1.3023099899291992, + "learning_rate": 0.0001756447218093652, + "loss": 3.2525, + "step": 1552 + }, + { + "epoch": 2.28, + "grad_norm": 8.924667358398438, + "learning_rate": 0.00017561403607008376, + "loss": 3.2892, + "step": 1553 + }, + { + "epoch": 2.28, + "grad_norm": 11.807168006896973, + "learning_rate": 0.00017558333369622112, + "loss": 3.3097, + "step": 1554 + }, + { + "epoch": 2.28, + "grad_norm": 9.348532676696777, + "learning_rate": 0.0001755526146945316, + "loss": 3.3024, + "step": 1555 + }, + { + "epoch": 2.28, + "grad_norm": 1.3859421014785767, + "learning_rate": 0.00017552187907177315, + "loss": 3.2629, + "step": 1556 + }, + { + "epoch": 2.28, + "grad_norm": 12.93152141571045, + "learning_rate": 0.00017549112683470748, + "loss": 3.3065, + "step": 1557 + }, + { + "epoch": 2.28, + "grad_norm": 18.00539779663086, + "learning_rate": 0.00017546035799009978, + "loss": 3.3714, + "step": 1558 + }, + { + "epoch": 2.28, + "grad_norm": 13.751264572143555, + "learning_rate": 0.00017542957254471908, + "loss": 3.3417, + "step": 1559 + }, + { + "epoch": 2.29, + "grad_norm": 4.594606876373291, + "learning_rate": 0.0001753987705053379, + "loss": 3.343, + "step": 1560 + }, + { + "epoch": 2.29, + "grad_norm": 7.840448379516602, + "learning_rate": 0.0001753679518787325, + "loss": 3.3199, + "step": 1561 + }, + { + "epoch": 2.29, + "grad_norm": 11.789816856384277, + "learning_rate": 0.0001753371166716828, + "loss": 3.3301, + "step": 1562 + }, + { + "epoch": 2.29, + "grad_norm": 11.082632064819336, + "learning_rate": 0.0001753062648909723, + "loss": 3.3078, + "step": 1563 + }, + { + "epoch": 2.29, + "grad_norm": 5.516722679138184, + "learning_rate": 0.0001752753965433882, + "loss": 3.2917, + "step": 1564 + }, + { + "epoch": 2.29, + "grad_norm": 8.504383087158203, + "learning_rate": 0.00017524451163572136, + "loss": 3.3195, + "step": 1565 + }, + { + "epoch": 2.29, + "grad_norm": 14.364538192749023, + "learning_rate": 0.0001752136101747662, + "loss": 3.3471, + "step": 1566 + }, + { + "epoch": 2.3, + "grad_norm": 12.66904067993164, + "learning_rate": 0.00017518269216732087, + "loss": 3.3574, + "step": 1567 + }, + { + "epoch": 2.3, + "grad_norm": 4.23660945892334, + "learning_rate": 0.00017515175762018714, + "loss": 3.3283, + "step": 1568 + }, + { + "epoch": 2.3, + "grad_norm": 6.301276206970215, + "learning_rate": 0.00017512080654017036, + "loss": 3.3064, + "step": 1569 + }, + { + "epoch": 2.3, + "grad_norm": 8.942106246948242, + "learning_rate": 0.00017508983893407958, + "loss": 3.2761, + "step": 1570 + }, + { + "epoch": 2.3, + "grad_norm": 6.445720672607422, + "learning_rate": 0.00017505885480872745, + "loss": 3.3033, + "step": 1571 + }, + { + "epoch": 2.3, + "grad_norm": 3.0577292442321777, + "learning_rate": 0.0001750278541709303, + "loss": 3.2916, + "step": 1572 + }, + { + "epoch": 2.3, + "grad_norm": 5.473748683929443, + "learning_rate": 0.00017499683702750803, + "loss": 3.2609, + "step": 1573 + }, + { + "epoch": 2.31, + "grad_norm": 2.3021624088287354, + "learning_rate": 0.00017496580338528426, + "loss": 3.2478, + "step": 1574 + }, + { + "epoch": 2.31, + "grad_norm": 6.796987056732178, + "learning_rate": 0.00017493475325108611, + "loss": 3.3177, + "step": 1575 + }, + { + "epoch": 2.31, + "grad_norm": 8.425848007202148, + "learning_rate": 0.00017490368663174443, + "loss": 3.2999, + "step": 1576 + }, + { + "epoch": 2.31, + "grad_norm": 4.728596210479736, + "learning_rate": 0.0001748726035340937, + "loss": 3.2693, + "step": 1577 + }, + { + "epoch": 2.31, + "grad_norm": 6.588244915008545, + "learning_rate": 0.00017484150396497193, + "loss": 3.2474, + "step": 1578 + }, + { + "epoch": 2.31, + "grad_norm": 10.9650239944458, + "learning_rate": 0.00017481038793122088, + "loss": 3.3038, + "step": 1579 + }, + { + "epoch": 2.32, + "grad_norm": 5.750149726867676, + "learning_rate": 0.00017477925543968586, + "loss": 3.315, + "step": 1580 + }, + { + "epoch": 2.32, + "grad_norm": 4.661263465881348, + "learning_rate": 0.00017474810649721576, + "loss": 3.2819, + "step": 1581 + }, + { + "epoch": 2.32, + "grad_norm": 6.578607082366943, + "learning_rate": 0.0001747169411106632, + "loss": 3.268, + "step": 1582 + }, + { + "epoch": 2.32, + "grad_norm": 3.776803970336914, + "learning_rate": 0.0001746857592868843, + "loss": 3.2414, + "step": 1583 + }, + { + "epoch": 2.32, + "grad_norm": 5.648257732391357, + "learning_rate": 0.00017465456103273885, + "loss": 3.2601, + "step": 1584 + }, + { + "epoch": 2.32, + "grad_norm": 7.4345011711120605, + "learning_rate": 0.00017462334635509033, + "loss": 3.3107, + "step": 1585 + }, + { + "epoch": 2.32, + "grad_norm": 2.6171629428863525, + "learning_rate": 0.0001745921152608057, + "loss": 3.226, + "step": 1586 + }, + { + "epoch": 2.33, + "grad_norm": 6.692351341247559, + "learning_rate": 0.00017456086775675562, + "loss": 3.283, + "step": 1587 + }, + { + "epoch": 2.33, + "grad_norm": 7.625064373016357, + "learning_rate": 0.00017452960384981432, + "loss": 3.2796, + "step": 1588 + }, + { + "epoch": 2.33, + "grad_norm": 4.043941974639893, + "learning_rate": 0.00017449832354685961, + "loss": 3.2948, + "step": 1589 + }, + { + "epoch": 2.33, + "grad_norm": 6.564191818237305, + "learning_rate": 0.00017446702685477303, + "loss": 3.2824, + "step": 1590 + }, + { + "epoch": 2.33, + "grad_norm": 10.199933052062988, + "learning_rate": 0.00017443571378043959, + "loss": 3.3447, + "step": 1591 + }, + { + "epoch": 2.33, + "grad_norm": 6.0920729637146, + "learning_rate": 0.00017440438433074794, + "loss": 3.2682, + "step": 1592 + }, + { + "epoch": 2.33, + "grad_norm": 3.95531964302063, + "learning_rate": 0.00017437303851259038, + "loss": 3.2439, + "step": 1593 + }, + { + "epoch": 2.34, + "grad_norm": 6.185577869415283, + "learning_rate": 0.0001743416763328628, + "loss": 3.3002, + "step": 1594 + }, + { + "epoch": 2.34, + "grad_norm": 2.757133960723877, + "learning_rate": 0.00017431029779846463, + "loss": 3.2716, + "step": 1595 + }, + { + "epoch": 2.34, + "grad_norm": 8.002574920654297, + "learning_rate": 0.00017427890291629893, + "loss": 3.2844, + "step": 1596 + }, + { + "epoch": 2.34, + "grad_norm": 10.051774024963379, + "learning_rate": 0.0001742474916932724, + "loss": 3.2388, + "step": 1597 + }, + { + "epoch": 2.34, + "grad_norm": 5.608651638031006, + "learning_rate": 0.00017421606413629527, + "loss": 3.226, + "step": 1598 + }, + { + "epoch": 2.34, + "grad_norm": 4.5539231300354, + "learning_rate": 0.00017418462025228145, + "loss": 3.2424, + "step": 1599 + }, + { + "epoch": 2.34, + "grad_norm": 6.649083614349365, + "learning_rate": 0.0001741531600481483, + "loss": 3.2537, + "step": 1600 + }, + { + "epoch": 2.35, + "grad_norm": 3.6664934158325195, + "learning_rate": 0.00017412168353081686, + "loss": 3.243, + "step": 1601 + }, + { + "epoch": 2.35, + "grad_norm": 6.799222469329834, + "learning_rate": 0.00017409019070721183, + "loss": 3.2853, + "step": 1602 + }, + { + "epoch": 2.35, + "grad_norm": 9.875252723693848, + "learning_rate": 0.00017405868158426135, + "loss": 3.2505, + "step": 1603 + }, + { + "epoch": 2.35, + "grad_norm": 4.99692440032959, + "learning_rate": 0.00017402715616889728, + "loss": 3.3176, + "step": 1604 + }, + { + "epoch": 2.35, + "grad_norm": 6.484687328338623, + "learning_rate": 0.00017399561446805492, + "loss": 3.2547, + "step": 1605 + }, + { + "epoch": 2.35, + "grad_norm": 8.336297988891602, + "learning_rate": 0.00017396405648867328, + "loss": 3.2543, + "step": 1606 + }, + { + "epoch": 2.36, + "grad_norm": 5.791290283203125, + "learning_rate": 0.00017393248223769491, + "loss": 3.2538, + "step": 1607 + }, + { + "epoch": 2.36, + "grad_norm": 2.4937000274658203, + "learning_rate": 0.00017390089172206592, + "loss": 3.2665, + "step": 1608 + }, + { + "epoch": 2.36, + "grad_norm": 5.263038635253906, + "learning_rate": 0.00017386928494873602, + "loss": 3.2899, + "step": 1609 + }, + { + "epoch": 2.36, + "grad_norm": 0.9264150857925415, + "learning_rate": 0.00017383766192465846, + "loss": 3.2284, + "step": 1610 + }, + { + "epoch": 2.36, + "grad_norm": 7.343024730682373, + "learning_rate": 0.00017380602265679014, + "loss": 3.3191, + "step": 1611 + }, + { + "epoch": 2.36, + "grad_norm": 8.518827438354492, + "learning_rate": 0.00017377436715209144, + "loss": 3.3296, + "step": 1612 + }, + { + "epoch": 2.36, + "grad_norm": 5.042328357696533, + "learning_rate": 0.00017374269541752638, + "loss": 3.2711, + "step": 1613 + }, + { + "epoch": 2.37, + "grad_norm": 5.329464435577393, + "learning_rate": 0.00017371100746006252, + "loss": 3.2886, + "step": 1614 + }, + { + "epoch": 2.37, + "grad_norm": 8.302837371826172, + "learning_rate": 0.00017367930328667105, + "loss": 3.2565, + "step": 1615 + }, + { + "epoch": 2.37, + "grad_norm": 4.5046539306640625, + "learning_rate": 0.0001736475829043266, + "loss": 3.2309, + "step": 1616 + }, + { + "epoch": 2.37, + "grad_norm": 5.1940789222717285, + "learning_rate": 0.0001736158463200075, + "loss": 3.2687, + "step": 1617 + }, + { + "epoch": 2.37, + "grad_norm": 7.008042335510254, + "learning_rate": 0.00017358409354069554, + "loss": 3.2845, + "step": 1618 + }, + { + "epoch": 2.37, + "grad_norm": 3.637481689453125, + "learning_rate": 0.00017355232457337615, + "loss": 3.2221, + "step": 1619 + }, + { + "epoch": 2.37, + "grad_norm": 6.158731937408447, + "learning_rate": 0.00017352053942503828, + "loss": 3.2727, + "step": 1620 + }, + { + "epoch": 2.38, + "grad_norm": 9.95582103729248, + "learning_rate": 0.00017348873810267443, + "loss": 3.2826, + "step": 1621 + }, + { + "epoch": 2.38, + "grad_norm": 5.530641555786133, + "learning_rate": 0.0001734569206132807, + "loss": 3.291, + "step": 1622 + }, + { + "epoch": 2.38, + "grad_norm": 3.868178367614746, + "learning_rate": 0.00017342508696385674, + "loss": 3.2794, + "step": 1623 + }, + { + "epoch": 2.38, + "grad_norm": 6.0412516593933105, + "learning_rate": 0.00017339323716140573, + "loss": 3.2546, + "step": 1624 + }, + { + "epoch": 2.38, + "grad_norm": 2.217486619949341, + "learning_rate": 0.00017336137121293436, + "loss": 3.2526, + "step": 1625 + }, + { + "epoch": 2.38, + "grad_norm": 8.725723266601562, + "learning_rate": 0.000173329489125453, + "loss": 3.253, + "step": 1626 + }, + { + "epoch": 2.39, + "grad_norm": 12.251547813415527, + "learning_rate": 0.00017329759090597548, + "loss": 3.2593, + "step": 1627 + }, + { + "epoch": 2.39, + "grad_norm": 7.2429070472717285, + "learning_rate": 0.00017326567656151916, + "loss": 3.2555, + "step": 1628 + }, + { + "epoch": 2.39, + "grad_norm": 2.993010997772217, + "learning_rate": 0.000173233746099105, + "loss": 3.2165, + "step": 1629 + }, + { + "epoch": 2.39, + "grad_norm": 5.157336711883545, + "learning_rate": 0.0001732017995257575, + "loss": 3.3013, + "step": 1630 + }, + { + "epoch": 2.39, + "grad_norm": 2.0820674896240234, + "learning_rate": 0.00017316983684850467, + "loss": 3.2088, + "step": 1631 + }, + { + "epoch": 2.39, + "grad_norm": 8.305849075317383, + "learning_rate": 0.0001731378580743781, + "loss": 3.2902, + "step": 1632 + }, + { + "epoch": 2.39, + "grad_norm": 10.2841157913208, + "learning_rate": 0.00017310586321041288, + "loss": 3.2437, + "step": 1633 + }, + { + "epoch": 2.4, + "grad_norm": 4.831589698791504, + "learning_rate": 0.00017307385226364772, + "loss": 3.2177, + "step": 1634 + }, + { + "epoch": 2.4, + "grad_norm": 4.824088096618652, + "learning_rate": 0.00017304182524112478, + "loss": 3.2779, + "step": 1635 + }, + { + "epoch": 2.4, + "grad_norm": 7.442866802215576, + "learning_rate": 0.00017300978214988976, + "loss": 3.2716, + "step": 1636 + }, + { + "epoch": 2.4, + "grad_norm": 4.1812968254089355, + "learning_rate": 0.000172977722996992, + "loss": 3.2746, + "step": 1637 + }, + { + "epoch": 2.4, + "grad_norm": 5.722605228424072, + "learning_rate": 0.00017294564778948422, + "loss": 3.2238, + "step": 1638 + }, + { + "epoch": 2.4, + "grad_norm": 8.271087646484375, + "learning_rate": 0.00017291355653442278, + "loss": 3.3323, + "step": 1639 + }, + { + "epoch": 2.4, + "grad_norm": 3.2383551597595215, + "learning_rate": 0.00017288144923886757, + "loss": 3.2318, + "step": 1640 + }, + { + "epoch": 2.41, + "grad_norm": 6.346744537353516, + "learning_rate": 0.00017284932590988197, + "loss": 3.2569, + "step": 1641 + }, + { + "epoch": 2.41, + "grad_norm": 8.613646507263184, + "learning_rate": 0.00017281718655453286, + "loss": 3.2684, + "step": 1642 + }, + { + "epoch": 2.41, + "grad_norm": 5.7413129806518555, + "learning_rate": 0.00017278503117989073, + "loss": 3.2374, + "step": 1643 + }, + { + "epoch": 2.41, + "grad_norm": 3.2419638633728027, + "learning_rate": 0.0001727528597930295, + "loss": 3.2127, + "step": 1644 + }, + { + "epoch": 2.41, + "grad_norm": 6.177962779998779, + "learning_rate": 0.0001727206724010267, + "loss": 3.2859, + "step": 1645 + }, + { + "epoch": 2.41, + "grad_norm": 1.907907247543335, + "learning_rate": 0.0001726884690109633, + "loss": 3.2221, + "step": 1646 + }, + { + "epoch": 2.41, + "grad_norm": 6.6156415939331055, + "learning_rate": 0.0001726562496299239, + "loss": 3.2339, + "step": 1647 + }, + { + "epoch": 2.42, + "grad_norm": 8.058192253112793, + "learning_rate": 0.00017262401426499646, + "loss": 3.2718, + "step": 1648 + }, + { + "epoch": 2.42, + "grad_norm": 4.197020530700684, + "learning_rate": 0.00017259176292327259, + "loss": 3.3056, + "step": 1649 + }, + { + "epoch": 2.42, + "grad_norm": 5.8328166007995605, + "learning_rate": 0.00017255949561184738, + "loss": 3.2645, + "step": 1650 + }, + { + "epoch": 2.42, + "grad_norm": 9.331326484680176, + "learning_rate": 0.00017252721233781938, + "loss": 3.3117, + "step": 1651 + }, + { + "epoch": 2.42, + "grad_norm": 5.499820232391357, + "learning_rate": 0.00017249491310829078, + "loss": 3.278, + "step": 1652 + }, + { + "epoch": 2.42, + "grad_norm": 4.50565242767334, + "learning_rate": 0.00017246259793036707, + "loss": 3.2662, + "step": 1653 + }, + { + "epoch": 2.43, + "grad_norm": 6.363594055175781, + "learning_rate": 0.00017243026681115748, + "loss": 3.2784, + "step": 1654 + }, + { + "epoch": 2.43, + "grad_norm": 2.7347965240478516, + "learning_rate": 0.00017239791975777458, + "loss": 3.2907, + "step": 1655 + }, + { + "epoch": 2.43, + "grad_norm": 7.49398946762085, + "learning_rate": 0.00017236555677733453, + "loss": 3.3239, + "step": 1656 + }, + { + "epoch": 2.43, + "grad_norm": 10.45259952545166, + "learning_rate": 0.00017233317787695697, + "loss": 3.316, + "step": 1657 + }, + { + "epoch": 2.43, + "grad_norm": 5.672200679779053, + "learning_rate": 0.00017230078306376502, + "loss": 3.2043, + "step": 1658 + }, + { + "epoch": 2.43, + "grad_norm": 4.071951866149902, + "learning_rate": 0.00017226837234488534, + "loss": 3.2218, + "step": 1659 + }, + { + "epoch": 2.43, + "grad_norm": 5.800225257873535, + "learning_rate": 0.00017223594572744806, + "loss": 3.2586, + "step": 1660 + }, + { + "epoch": 2.44, + "grad_norm": 2.592320442199707, + "learning_rate": 0.0001722035032185868, + "loss": 3.244, + "step": 1661 + }, + { + "epoch": 2.44, + "grad_norm": 7.940516471862793, + "learning_rate": 0.00017217104482543876, + "loss": 3.2501, + "step": 1662 + }, + { + "epoch": 2.44, + "grad_norm": 10.571852684020996, + "learning_rate": 0.0001721385705551445, + "loss": 3.2631, + "step": 1663 + }, + { + "epoch": 2.44, + "grad_norm": 5.3643574714660645, + "learning_rate": 0.0001721060804148482, + "loss": 3.2116, + "step": 1664 + }, + { + "epoch": 2.44, + "grad_norm": 4.0979905128479, + "learning_rate": 0.00017207357441169742, + "loss": 3.1886, + "step": 1665 + }, + { + "epoch": 2.44, + "grad_norm": 6.099160194396973, + "learning_rate": 0.0001720410525528433, + "loss": 3.2502, + "step": 1666 + }, + { + "epoch": 2.44, + "grad_norm": 3.1805484294891357, + "learning_rate": 0.00017200851484544037, + "loss": 3.213, + "step": 1667 + }, + { + "epoch": 2.45, + "grad_norm": 7.588128089904785, + "learning_rate": 0.00017197596129664682, + "loss": 3.2935, + "step": 1668 + }, + { + "epoch": 2.45, + "grad_norm": 10.550485610961914, + "learning_rate": 0.0001719433919136241, + "loss": 3.2823, + "step": 1669 + }, + { + "epoch": 2.45, + "grad_norm": 5.52534818649292, + "learning_rate": 0.00017191080670353733, + "loss": 3.2452, + "step": 1670 + }, + { + "epoch": 2.45, + "grad_norm": 4.016864776611328, + "learning_rate": 0.00017187820567355497, + "loss": 3.2412, + "step": 1671 + }, + { + "epoch": 2.45, + "grad_norm": 6.307382583618164, + "learning_rate": 0.0001718455888308491, + "loss": 3.2699, + "step": 1672 + }, + { + "epoch": 2.45, + "grad_norm": 2.715496301651001, + "learning_rate": 0.00017181295618259515, + "loss": 3.2695, + "step": 1673 + }, + { + "epoch": 2.45, + "grad_norm": 7.303908348083496, + "learning_rate": 0.00017178030773597213, + "loss": 3.2392, + "step": 1674 + }, + { + "epoch": 2.46, + "grad_norm": 10.16519546508789, + "learning_rate": 0.00017174764349816245, + "loss": 3.2635, + "step": 1675 + }, + { + "epoch": 2.46, + "grad_norm": 5.080345630645752, + "learning_rate": 0.00017171496347635206, + "loss": 3.2302, + "step": 1676 + }, + { + "epoch": 2.46, + "grad_norm": 5.1449408531188965, + "learning_rate": 0.00017168226767773028, + "loss": 3.2135, + "step": 1677 + }, + { + "epoch": 2.46, + "grad_norm": 7.340461254119873, + "learning_rate": 0.00017164955610949004, + "loss": 3.2561, + "step": 1678 + }, + { + "epoch": 2.46, + "grad_norm": 3.821507453918457, + "learning_rate": 0.00017161682877882758, + "loss": 3.2111, + "step": 1679 + }, + { + "epoch": 2.46, + "grad_norm": 6.017604351043701, + "learning_rate": 0.0001715840856929428, + "loss": 3.3209, + "step": 1680 + }, + { + "epoch": 2.47, + "grad_norm": 8.912531852722168, + "learning_rate": 0.00017155132685903887, + "loss": 3.2441, + "step": 1681 + }, + { + "epoch": 2.47, + "grad_norm": 4.574288368225098, + "learning_rate": 0.00017151855228432253, + "loss": 3.2489, + "step": 1682 + }, + { + "epoch": 2.47, + "grad_norm": 4.964592456817627, + "learning_rate": 0.00017148576197600402, + "loss": 3.2607, + "step": 1683 + }, + { + "epoch": 2.47, + "grad_norm": 7.388319969177246, + "learning_rate": 0.00017145295594129693, + "loss": 3.2537, + "step": 1684 + }, + { + "epoch": 2.47, + "grad_norm": 4.345710277557373, + "learning_rate": 0.0001714201341874184, + "loss": 3.2435, + "step": 1685 + }, + { + "epoch": 2.47, + "grad_norm": 5.151455402374268, + "learning_rate": 0.000171387296721589, + "loss": 3.2198, + "step": 1686 + }, + { + "epoch": 2.47, + "grad_norm": 9.679162979125977, + "learning_rate": 0.0001713544435510327, + "loss": 3.3015, + "step": 1687 + }, + { + "epoch": 2.48, + "grad_norm": 4.5911383628845215, + "learning_rate": 0.00017132157468297704, + "loss": 3.2534, + "step": 1688 + }, + { + "epoch": 2.48, + "grad_norm": 3.5955042839050293, + "learning_rate": 0.00017128869012465292, + "loss": 3.2479, + "step": 1689 + }, + { + "epoch": 2.48, + "grad_norm": 4.769717693328857, + "learning_rate": 0.00017125578988329472, + "loss": 3.2706, + "step": 1690 + }, + { + "epoch": 2.48, + "eval_loss": 3.0677621364593506, + "eval_runtime": 1164.46, + "eval_samples_per_second": 14.721, + "eval_steps_per_second": 0.921, + "step": 1690 + }, + { + "epoch": 2.48, + "grad_norm": 0.9628227949142456, + "learning_rate": 0.0001712228739661403, + "loss": 3.2333, + "step": 1691 + }, + { + "epoch": 2.48, + "grad_norm": 9.899592399597168, + "learning_rate": 0.00017118994238043088, + "loss": 3.287, + "step": 1692 + }, + { + "epoch": 2.48, + "grad_norm": 13.610316276550293, + "learning_rate": 0.00017115699513341127, + "loss": 3.3172, + "step": 1693 + }, + { + "epoch": 2.48, + "grad_norm": 8.333012580871582, + "learning_rate": 0.00017112403223232964, + "loss": 3.2552, + "step": 1694 + }, + { + "epoch": 2.49, + "grad_norm": 1.985851764678955, + "learning_rate": 0.0001710910536844375, + "loss": 3.2494, + "step": 1695 + }, + { + "epoch": 2.49, + "grad_norm": 4.960620403289795, + "learning_rate": 0.00017105805949699, + "loss": 3.2411, + "step": 1696 + }, + { + "epoch": 2.49, + "grad_norm": 2.77993106842041, + "learning_rate": 0.00017102504967724563, + "loss": 3.264, + "step": 1697 + }, + { + "epoch": 2.49, + "grad_norm": 6.050425052642822, + "learning_rate": 0.0001709920242324663, + "loss": 3.2725, + "step": 1698 + }, + { + "epoch": 2.49, + "grad_norm": 6.0771307945251465, + "learning_rate": 0.00017095898316991744, + "loss": 3.199, + "step": 1699 + }, + { + "epoch": 2.49, + "grad_norm": 1.5405139923095703, + "learning_rate": 0.0001709259264968678, + "loss": 3.2523, + "step": 1700 + }, + { + "epoch": 2.49, + "grad_norm": 5.121715068817139, + "learning_rate": 0.00017089285422058968, + "loss": 3.2944, + "step": 1701 + }, + { + "epoch": 2.5, + "grad_norm": 4.806593894958496, + "learning_rate": 0.00017085976634835873, + "loss": 3.2461, + "step": 1702 + }, + { + "epoch": 2.5, + "grad_norm": 3.7719340324401855, + "learning_rate": 0.00017082666288745403, + "loss": 3.2067, + "step": 1703 + }, + { + "epoch": 2.5, + "grad_norm": 3.7723586559295654, + "learning_rate": 0.0001707935438451582, + "loss": 3.2715, + "step": 1704 + }, + { + "epoch": 2.5, + "grad_norm": 2.5075759887695312, + "learning_rate": 0.00017076040922875713, + "loss": 3.2053, + "step": 1705 + }, + { + "epoch": 2.5, + "grad_norm": 2.315262794494629, + "learning_rate": 0.00017072725904554025, + "loss": 3.2529, + "step": 1706 + }, + { + "epoch": 2.5, + "grad_norm": 3.849294900894165, + "learning_rate": 0.0001706940933028004, + "loss": 3.2501, + "step": 1707 + }, + { + "epoch": 2.51, + "grad_norm": 3.2091238498687744, + "learning_rate": 0.00017066091200783375, + "loss": 3.1893, + "step": 1708 + }, + { + "epoch": 2.51, + "grad_norm": 3.587019443511963, + "learning_rate": 0.00017062771516794002, + "loss": 3.2435, + "step": 1709 + }, + { + "epoch": 2.51, + "grad_norm": 2.367382526397705, + "learning_rate": 0.00017059450279042232, + "loss": 3.2453, + "step": 1710 + }, + { + "epoch": 2.51, + "grad_norm": 4.0962700843811035, + "learning_rate": 0.00017056127488258705, + "loss": 3.2043, + "step": 1711 + }, + { + "epoch": 2.51, + "grad_norm": 4.637650966644287, + "learning_rate": 0.0001705280314517442, + "loss": 3.2506, + "step": 1712 + }, + { + "epoch": 2.51, + "grad_norm": 1.244682788848877, + "learning_rate": 0.00017049477250520714, + "loss": 3.2169, + "step": 1713 + }, + { + "epoch": 2.51, + "grad_norm": 0.7056719660758972, + "learning_rate": 0.00017046149805029253, + "loss": 3.212, + "step": 1714 + }, + { + "epoch": 2.52, + "grad_norm": 2.6713907718658447, + "learning_rate": 0.00017042820809432056, + "loss": 3.2665, + "step": 1715 + }, + { + "epoch": 2.52, + "grad_norm": 1.15023934841156, + "learning_rate": 0.00017039490264461477, + "loss": 3.2735, + "step": 1716 + }, + { + "epoch": 2.52, + "grad_norm": 3.00900936126709, + "learning_rate": 0.0001703615817085022, + "loss": 3.2198, + "step": 1717 + }, + { + "epoch": 2.52, + "grad_norm": 3.382124423980713, + "learning_rate": 0.0001703282452933132, + "loss": 3.2279, + "step": 1718 + }, + { + "epoch": 2.52, + "grad_norm": 1.6238938570022583, + "learning_rate": 0.00017029489340638154, + "loss": 3.2521, + "step": 1719 + }, + { + "epoch": 2.52, + "grad_norm": 4.034934997558594, + "learning_rate": 0.00017026152605504442, + "loss": 3.2477, + "step": 1720 + }, + { + "epoch": 2.52, + "grad_norm": 4.463752269744873, + "learning_rate": 0.00017022814324664244, + "loss": 3.2478, + "step": 1721 + }, + { + "epoch": 2.53, + "grad_norm": 2.6497182846069336, + "learning_rate": 0.00017019474498851958, + "loss": 3.2774, + "step": 1722 + }, + { + "epoch": 2.53, + "grad_norm": 2.8634886741638184, + "learning_rate": 0.0001701613312880233, + "loss": 3.2704, + "step": 1723 + }, + { + "epoch": 2.53, + "grad_norm": 3.318505048751831, + "learning_rate": 0.00017012790215250432, + "loss": 3.2681, + "step": 1724 + }, + { + "epoch": 2.53, + "grad_norm": 2.9322783946990967, + "learning_rate": 0.0001700944575893168, + "loss": 3.2489, + "step": 1725 + }, + { + "epoch": 2.53, + "grad_norm": 1.9499863386154175, + "learning_rate": 0.00017006099760581838, + "loss": 3.2197, + "step": 1726 + }, + { + "epoch": 2.53, + "grad_norm": 2.5505218505859375, + "learning_rate": 0.00017002752220937004, + "loss": 3.2399, + "step": 1727 + }, + { + "epoch": 2.53, + "grad_norm": 2.46716570854187, + "learning_rate": 0.0001699940314073361, + "loss": 3.2078, + "step": 1728 + }, + { + "epoch": 2.54, + "grad_norm": 1.8553292751312256, + "learning_rate": 0.00016996052520708434, + "loss": 3.292, + "step": 1729 + }, + { + "epoch": 2.54, + "grad_norm": 2.041597366333008, + "learning_rate": 0.00016992700361598585, + "loss": 3.201, + "step": 1730 + }, + { + "epoch": 2.54, + "grad_norm": 2.173717737197876, + "learning_rate": 0.00016989346664141524, + "loss": 3.2707, + "step": 1731 + }, + { + "epoch": 2.54, + "grad_norm": 1.5214920043945312, + "learning_rate": 0.00016985991429075036, + "loss": 3.2218, + "step": 1732 + }, + { + "epoch": 2.54, + "grad_norm": 6.029373645782471, + "learning_rate": 0.00016982634657137252, + "loss": 3.3117, + "step": 1733 + }, + { + "epoch": 2.54, + "grad_norm": 5.91231107711792, + "learning_rate": 0.00016979276349066638, + "loss": 3.2202, + "step": 1734 + }, + { + "epoch": 2.55, + "grad_norm": 2.8163087368011475, + "learning_rate": 0.00016975916505602, + "loss": 3.2803, + "step": 1735 + }, + { + "epoch": 2.55, + "grad_norm": 3.7836880683898926, + "learning_rate": 0.00016972555127482483, + "loss": 3.2399, + "step": 1736 + }, + { + "epoch": 2.55, + "grad_norm": 3.339012622833252, + "learning_rate": 0.00016969192215447566, + "loss": 3.2619, + "step": 1737 + }, + { + "epoch": 2.55, + "grad_norm": 3.779021739959717, + "learning_rate": 0.00016965827770237067, + "loss": 3.2205, + "step": 1738 + }, + { + "epoch": 2.55, + "grad_norm": 0.7899141311645508, + "learning_rate": 0.00016962461792591137, + "loss": 3.2228, + "step": 1739 + }, + { + "epoch": 2.55, + "grad_norm": 2.307349443435669, + "learning_rate": 0.00016959094283250278, + "loss": 3.2412, + "step": 1740 + }, + { + "epoch": 2.55, + "grad_norm": 1.5662769079208374, + "learning_rate": 0.00016955725242955315, + "loss": 3.257, + "step": 1741 + }, + { + "epoch": 2.56, + "grad_norm": 1.4268426895141602, + "learning_rate": 0.0001695235467244741, + "loss": 3.2718, + "step": 1742 + }, + { + "epoch": 2.56, + "grad_norm": 1.7694865465164185, + "learning_rate": 0.0001694898257246807, + "loss": 3.2414, + "step": 1743 + }, + { + "epoch": 2.56, + "grad_norm": 2.555623769760132, + "learning_rate": 0.00016945608943759138, + "loss": 3.2384, + "step": 1744 + }, + { + "epoch": 2.56, + "grad_norm": 1.4633605480194092, + "learning_rate": 0.0001694223378706278, + "loss": 3.2164, + "step": 1745 + }, + { + "epoch": 2.56, + "grad_norm": 2.0434815883636475, + "learning_rate": 0.0001693885710312152, + "loss": 3.2547, + "step": 1746 + }, + { + "epoch": 2.56, + "grad_norm": 2.4210758209228516, + "learning_rate": 0.00016935478892678195, + "loss": 3.2367, + "step": 1747 + }, + { + "epoch": 2.56, + "grad_norm": 0.8179725408554077, + "learning_rate": 0.00016932099156475995, + "loss": 3.2619, + "step": 1748 + }, + { + "epoch": 2.57, + "grad_norm": 3.3028626441955566, + "learning_rate": 0.00016928717895258437, + "loss": 3.2256, + "step": 1749 + }, + { + "epoch": 2.57, + "grad_norm": 1.852142572402954, + "learning_rate": 0.00016925335109769374, + "loss": 3.2402, + "step": 1750 + }, + { + "epoch": 2.57, + "grad_norm": 2.416368007659912, + "learning_rate": 0.00016921950800753002, + "loss": 3.234, + "step": 1751 + }, + { + "epoch": 2.57, + "grad_norm": 1.5477232933044434, + "learning_rate": 0.00016918564968953842, + "loss": 3.2142, + "step": 1752 + }, + { + "epoch": 2.57, + "grad_norm": 1.5452466011047363, + "learning_rate": 0.00016915177615116754, + "loss": 3.1941, + "step": 1753 + }, + { + "epoch": 2.57, + "grad_norm": 3.213561773300171, + "learning_rate": 0.00016911788739986934, + "loss": 3.2154, + "step": 1754 + }, + { + "epoch": 2.58, + "grad_norm": 1.2881815433502197, + "learning_rate": 0.0001690839834430991, + "loss": 3.2909, + "step": 1755 + }, + { + "epoch": 2.58, + "grad_norm": 3.375774383544922, + "learning_rate": 0.00016905006428831552, + "loss": 3.2096, + "step": 1756 + }, + { + "epoch": 2.58, + "grad_norm": 2.6476569175720215, + "learning_rate": 0.00016901612994298053, + "loss": 3.2082, + "step": 1757 + }, + { + "epoch": 2.58, + "grad_norm": 3.5997047424316406, + "learning_rate": 0.0001689821804145595, + "loss": 3.1989, + "step": 1758 + }, + { + "epoch": 2.58, + "grad_norm": 2.435394763946533, + "learning_rate": 0.00016894821571052104, + "loss": 3.2109, + "step": 1759 + }, + { + "epoch": 2.58, + "grad_norm": 5.273232936859131, + "learning_rate": 0.00016891423583833724, + "loss": 3.2358, + "step": 1760 + }, + { + "epoch": 2.58, + "grad_norm": 4.958769798278809, + "learning_rate": 0.00016888024080548335, + "loss": 3.2315, + "step": 1761 + }, + { + "epoch": 2.59, + "grad_norm": 2.171224594116211, + "learning_rate": 0.00016884623061943813, + "loss": 3.2587, + "step": 1762 + }, + { + "epoch": 2.59, + "grad_norm": 1.9900020360946655, + "learning_rate": 0.00016881220528768356, + "loss": 3.2082, + "step": 1763 + }, + { + "epoch": 2.59, + "grad_norm": 2.3555827140808105, + "learning_rate": 0.00016877816481770496, + "loss": 3.2356, + "step": 1764 + }, + { + "epoch": 2.59, + "grad_norm": 1.4140695333480835, + "learning_rate": 0.00016874410921699105, + "loss": 3.1722, + "step": 1765 + }, + { + "epoch": 2.59, + "grad_norm": 4.323090076446533, + "learning_rate": 0.00016871003849303382, + "loss": 3.2533, + "step": 1766 + }, + { + "epoch": 2.59, + "grad_norm": 2.4270732402801514, + "learning_rate": 0.00016867595265332859, + "loss": 3.2092, + "step": 1767 + }, + { + "epoch": 2.59, + "grad_norm": 3.9084079265594482, + "learning_rate": 0.000168641851705374, + "loss": 3.2595, + "step": 1768 + }, + { + "epoch": 2.6, + "grad_norm": 3.986670970916748, + "learning_rate": 0.00016860773565667206, + "loss": 3.2302, + "step": 1769 + }, + { + "epoch": 2.6, + "grad_norm": 3.4050045013427734, + "learning_rate": 0.0001685736045147281, + "loss": 3.2589, + "step": 1770 + }, + { + "epoch": 2.6, + "grad_norm": 4.226673126220703, + "learning_rate": 0.0001685394582870507, + "loss": 3.235, + "step": 1771 + }, + { + "epoch": 2.6, + "grad_norm": 1.5907903909683228, + "learning_rate": 0.00016850529698115175, + "loss": 3.3035, + "step": 1772 + }, + { + "epoch": 2.6, + "grad_norm": 2.343108654022217, + "learning_rate": 0.00016847112060454665, + "loss": 3.1848, + "step": 1773 + }, + { + "epoch": 2.6, + "grad_norm": 3.3092055320739746, + "learning_rate": 0.00016843692916475384, + "loss": 3.2648, + "step": 1774 + }, + { + "epoch": 2.6, + "grad_norm": 1.7477455139160156, + "learning_rate": 0.00016840272266929523, + "loss": 3.206, + "step": 1775 + }, + { + "epoch": 2.61, + "grad_norm": 4.240764617919922, + "learning_rate": 0.00016836850112569613, + "loss": 3.2177, + "step": 1776 + }, + { + "epoch": 2.61, + "grad_norm": 4.2325758934021, + "learning_rate": 0.0001683342645414849, + "loss": 3.2154, + "step": 1777 + }, + { + "epoch": 2.61, + "grad_norm": 2.993154525756836, + "learning_rate": 0.00016830001292419345, + "loss": 3.1972, + "step": 1778 + }, + { + "epoch": 2.61, + "grad_norm": 3.2484753131866455, + "learning_rate": 0.00016826574628135688, + "loss": 3.232, + "step": 1779 + }, + { + "epoch": 2.61, + "grad_norm": 2.438793420791626, + "learning_rate": 0.00016823146462051364, + "loss": 3.2292, + "step": 1780 + }, + { + "epoch": 2.61, + "grad_norm": 2.0203206539154053, + "learning_rate": 0.0001681971679492054, + "loss": 3.1791, + "step": 1781 + }, + { + "epoch": 2.62, + "grad_norm": 4.8010125160217285, + "learning_rate": 0.0001681628562749773, + "loss": 3.2012, + "step": 1782 + }, + { + "epoch": 2.62, + "grad_norm": 4.596678256988525, + "learning_rate": 0.00016812852960537762, + "loss": 3.2204, + "step": 1783 + }, + { + "epoch": 2.62, + "grad_norm": 2.4350953102111816, + "learning_rate": 0.00016809418794795796, + "loss": 3.2175, + "step": 1784 + }, + { + "epoch": 2.62, + "grad_norm": 15.600502014160156, + "learning_rate": 0.00016805983131027333, + "loss": 3.256, + "step": 1785 + }, + { + "epoch": 2.62, + "grad_norm": 3.7827413082122803, + "learning_rate": 0.00016802545969988195, + "loss": 3.2165, + "step": 1786 + }, + { + "epoch": 2.62, + "grad_norm": 3.0212562084198, + "learning_rate": 0.0001679910731243453, + "loss": 3.2537, + "step": 1787 + }, + { + "epoch": 2.62, + "grad_norm": 2.496250629425049, + "learning_rate": 0.00016795667159122824, + "loss": 3.212, + "step": 1788 + }, + { + "epoch": 2.63, + "grad_norm": 3.222059488296509, + "learning_rate": 0.00016792225510809883, + "loss": 3.2665, + "step": 1789 + }, + { + "epoch": 2.63, + "grad_norm": 3.4902381896972656, + "learning_rate": 0.0001678878236825285, + "loss": 3.2326, + "step": 1790 + }, + { + "epoch": 2.63, + "grad_norm": 0.7913928627967834, + "learning_rate": 0.00016785337732209193, + "loss": 3.1662, + "step": 1791 + }, + { + "epoch": 2.63, + "grad_norm": 3.9567112922668457, + "learning_rate": 0.00016781891603436706, + "loss": 3.2217, + "step": 1792 + }, + { + "epoch": 2.63, + "grad_norm": 1.712324619293213, + "learning_rate": 0.00016778443982693519, + "loss": 3.1855, + "step": 1793 + }, + { + "epoch": 2.63, + "grad_norm": 1.2269927263259888, + "learning_rate": 0.0001677499487073808, + "loss": 3.2306, + "step": 1794 + }, + { + "epoch": 2.63, + "grad_norm": 3.1048622131347656, + "learning_rate": 0.00016771544268329177, + "loss": 3.1975, + "step": 1795 + }, + { + "epoch": 2.64, + "grad_norm": 1.18289315700531, + "learning_rate": 0.00016768092176225915, + "loss": 3.1904, + "step": 1796 + }, + { + "epoch": 2.64, + "grad_norm": 1.7925835847854614, + "learning_rate": 0.0001676463859518773, + "loss": 3.2143, + "step": 1797 + }, + { + "epoch": 2.64, + "grad_norm": 2.179051637649536, + "learning_rate": 0.0001676118352597439, + "loss": 3.2297, + "step": 1798 + }, + { + "epoch": 2.64, + "grad_norm": 1.2693228721618652, + "learning_rate": 0.00016757726969345984, + "loss": 3.2181, + "step": 1799 + }, + { + "epoch": 2.64, + "grad_norm": 5.075618743896484, + "learning_rate": 0.00016754268926062938, + "loss": 3.237, + "step": 1800 + }, + { + "epoch": 2.64, + "grad_norm": 5.278939723968506, + "learning_rate": 0.0001675080939688599, + "loss": 3.2266, + "step": 1801 + }, + { + "epoch": 2.64, + "grad_norm": 2.5977909564971924, + "learning_rate": 0.00016747348382576213, + "loss": 3.2409, + "step": 1802 + }, + { + "epoch": 2.65, + "grad_norm": 3.318488597869873, + "learning_rate": 0.00016743885883895013, + "loss": 3.231, + "step": 1803 + }, + { + "epoch": 2.65, + "grad_norm": 2.22575306892395, + "learning_rate": 0.00016740421901604118, + "loss": 3.2313, + "step": 1804 + }, + { + "epoch": 2.65, + "grad_norm": 2.436943769454956, + "learning_rate": 0.00016736956436465573, + "loss": 3.2268, + "step": 1805 + }, + { + "epoch": 2.65, + "grad_norm": 2.966697931289673, + "learning_rate": 0.0001673348948924176, + "loss": 3.2139, + "step": 1806 + }, + { + "epoch": 2.65, + "grad_norm": 2.458636999130249, + "learning_rate": 0.0001673002106069539, + "loss": 3.2228, + "step": 1807 + }, + { + "epoch": 2.65, + "grad_norm": 3.647672414779663, + "learning_rate": 0.00016726551151589485, + "loss": 3.256, + "step": 1808 + }, + { + "epoch": 2.66, + "grad_norm": 2.713827133178711, + "learning_rate": 0.0001672307976268741, + "loss": 3.2256, + "step": 1809 + }, + { + "epoch": 2.66, + "grad_norm": 3.452815294265747, + "learning_rate": 0.0001671960689475284, + "loss": 3.2875, + "step": 1810 + }, + { + "epoch": 2.66, + "grad_norm": 2.377556085586548, + "learning_rate": 0.00016716132548549786, + "loss": 3.2188, + "step": 1811 + }, + { + "epoch": 2.66, + "grad_norm": 3.791208028793335, + "learning_rate": 0.0001671265672484258, + "loss": 3.234, + "step": 1812 + }, + { + "epoch": 2.66, + "grad_norm": 3.7197184562683105, + "learning_rate": 0.00016709179424395884, + "loss": 3.1895, + "step": 1813 + }, + { + "epoch": 2.66, + "grad_norm": 2.4916012287139893, + "learning_rate": 0.00016705700647974675, + "loss": 3.2053, + "step": 1814 + }, + { + "epoch": 2.66, + "grad_norm": 2.569413185119629, + "learning_rate": 0.00016702220396344267, + "loss": 3.2358, + "step": 1815 + }, + { + "epoch": 2.67, + "grad_norm": 3.520432949066162, + "learning_rate": 0.00016698738670270285, + "loss": 3.2369, + "step": 1816 + }, + { + "epoch": 2.67, + "grad_norm": 2.880115032196045, + "learning_rate": 0.00016695255470518686, + "loss": 3.2232, + "step": 1817 + }, + { + "epoch": 2.67, + "grad_norm": 3.42354416847229, + "learning_rate": 0.00016691770797855758, + "loss": 3.1885, + "step": 1818 + }, + { + "epoch": 2.67, + "grad_norm": 3.31255841255188, + "learning_rate": 0.00016688284653048097, + "loss": 3.21, + "step": 1819 + }, + { + "epoch": 2.67, + "grad_norm": 3.157924175262451, + "learning_rate": 0.00016684797036862635, + "loss": 3.2055, + "step": 1820 + }, + { + "epoch": 2.67, + "grad_norm": 2.820952892303467, + "learning_rate": 0.00016681307950066628, + "loss": 3.1809, + "step": 1821 + }, + { + "epoch": 2.67, + "grad_norm": 3.921417236328125, + "learning_rate": 0.00016677817393427645, + "loss": 3.1902, + "step": 1822 + }, + { + "epoch": 2.68, + "grad_norm": 3.5899341106414795, + "learning_rate": 0.00016674325367713592, + "loss": 3.2222, + "step": 1823 + }, + { + "epoch": 2.68, + "grad_norm": 2.645939350128174, + "learning_rate": 0.00016670831873692683, + "loss": 3.1834, + "step": 1824 + }, + { + "epoch": 2.68, + "grad_norm": 2.4159257411956787, + "learning_rate": 0.00016667336912133468, + "loss": 3.2018, + "step": 1825 + }, + { + "epoch": 2.68, + "grad_norm": 3.56545090675354, + "learning_rate": 0.0001666384048380482, + "loss": 3.2096, + "step": 1826 + }, + { + "epoch": 2.68, + "grad_norm": 2.857130765914917, + "learning_rate": 0.0001666034258947592, + "loss": 3.2102, + "step": 1827 + }, + { + "epoch": 2.68, + "grad_norm": 3.4373629093170166, + "learning_rate": 0.00016656843229916285, + "loss": 3.2017, + "step": 1828 + }, + { + "epoch": 2.68, + "grad_norm": 2.5930848121643066, + "learning_rate": 0.00016653342405895756, + "loss": 3.2265, + "step": 1829 + }, + { + "epoch": 2.69, + "grad_norm": 4.415797710418701, + "learning_rate": 0.0001664984011818448, + "loss": 3.1589, + "step": 1830 + }, + { + "epoch": 2.69, + "grad_norm": 4.656808853149414, + "learning_rate": 0.0001664633636755295, + "loss": 3.2189, + "step": 1831 + }, + { + "epoch": 2.69, + "grad_norm": 1.3080215454101562, + "learning_rate": 0.00016642831154771957, + "loss": 3.2178, + "step": 1832 + }, + { + "epoch": 2.69, + "grad_norm": 1.491684913635254, + "learning_rate": 0.0001663932448061263, + "loss": 3.1922, + "step": 1833 + }, + { + "epoch": 2.69, + "grad_norm": 5.020033359527588, + "learning_rate": 0.0001663581634584641, + "loss": 3.2153, + "step": 1834 + }, + { + "epoch": 2.69, + "grad_norm": 4.078526973724365, + "learning_rate": 0.0001663230675124507, + "loss": 3.2165, + "step": 1835 + }, + { + "epoch": 2.7, + "grad_norm": 2.787774085998535, + "learning_rate": 0.00016628795697580686, + "loss": 3.1903, + "step": 1836 + }, + { + "epoch": 2.7, + "grad_norm": 2.4265925884246826, + "learning_rate": 0.0001662528318562568, + "loss": 3.1732, + "step": 1837 + }, + { + "epoch": 2.7, + "grad_norm": 3.579639434814453, + "learning_rate": 0.00016621769216152766, + "loss": 3.1891, + "step": 1838 + }, + { + "epoch": 2.7, + "grad_norm": 2.984905958175659, + "learning_rate": 0.00016618253789935005, + "loss": 3.2348, + "step": 1839 + }, + { + "epoch": 2.7, + "grad_norm": 4.036933898925781, + "learning_rate": 0.00016614736907745768, + "loss": 3.1677, + "step": 1840 + }, + { + "epoch": 2.7, + "grad_norm": 3.9433419704437256, + "learning_rate": 0.00016611218570358735, + "loss": 3.18, + "step": 1841 + }, + { + "epoch": 2.7, + "grad_norm": 2.285982370376587, + "learning_rate": 0.00016607698778547925, + "loss": 3.2298, + "step": 1842 + }, + { + "epoch": 2.71, + "grad_norm": 2.241262197494507, + "learning_rate": 0.00016604177533087667, + "loss": 3.1903, + "step": 1843 + }, + { + "epoch": 2.71, + "grad_norm": 3.7759506702423096, + "learning_rate": 0.00016600654834752611, + "loss": 3.2093, + "step": 1844 + }, + { + "epoch": 2.71, + "grad_norm": 3.2228505611419678, + "learning_rate": 0.00016597130684317725, + "loss": 3.1934, + "step": 1845 + }, + { + "epoch": 2.71, + "grad_norm": 3.3804237842559814, + "learning_rate": 0.00016593605082558305, + "loss": 3.2345, + "step": 1846 + }, + { + "epoch": 2.71, + "grad_norm": 2.6020681858062744, + "learning_rate": 0.0001659007803024995, + "loss": 3.1868, + "step": 1847 + }, + { + "epoch": 2.71, + "grad_norm": 3.3784382343292236, + "learning_rate": 0.00016586549528168596, + "loss": 3.1865, + "step": 1848 + }, + { + "epoch": 2.71, + "grad_norm": 2.881960391998291, + "learning_rate": 0.0001658301957709049, + "loss": 3.1906, + "step": 1849 + }, + { + "epoch": 2.72, + "grad_norm": 3.543879270553589, + "learning_rate": 0.00016579488177792187, + "loss": 3.2124, + "step": 1850 + }, + { + "epoch": 2.72, + "grad_norm": 3.8492488861083984, + "learning_rate": 0.00016575955331050584, + "loss": 3.3212, + "step": 1851 + }, + { + "epoch": 2.72, + "grad_norm": 2.2801854610443115, + "learning_rate": 0.00016572421037642875, + "loss": 3.2374, + "step": 1852 + }, + { + "epoch": 2.72, + "grad_norm": 2.0372071266174316, + "learning_rate": 0.00016568885298346586, + "loss": 3.2105, + "step": 1853 + }, + { + "epoch": 2.72, + "grad_norm": 4.137518405914307, + "learning_rate": 0.00016565348113939556, + "loss": 3.1718, + "step": 1854 + }, + { + "epoch": 2.72, + "grad_norm": 3.7536251544952393, + "learning_rate": 0.00016561809485199938, + "loss": 3.2118, + "step": 1855 + }, + { + "epoch": 2.73, + "grad_norm": 3.2339396476745605, + "learning_rate": 0.00016558269412906206, + "loss": 3.1746, + "step": 1856 + }, + { + "epoch": 2.73, + "grad_norm": 2.7833776473999023, + "learning_rate": 0.0001655472789783716, + "loss": 3.1876, + "step": 1857 + }, + { + "epoch": 2.73, + "grad_norm": 3.6448328495025635, + "learning_rate": 0.000165511849407719, + "loss": 3.2143, + "step": 1858 + }, + { + "epoch": 2.73, + "grad_norm": 3.4505040645599365, + "learning_rate": 0.0001654764054248986, + "loss": 3.173, + "step": 1859 + }, + { + "epoch": 2.73, + "eval_loss": 3.040992498397827, + "eval_runtime": 1164.9923, + "eval_samples_per_second": 14.714, + "eval_steps_per_second": 0.92, + "step": 1859 + }, + { + "epoch": 2.73, + "grad_norm": 2.8391804695129395, + "learning_rate": 0.0001654409470377078, + "loss": 3.2099, + "step": 1860 + }, + { + "epoch": 2.73, + "grad_norm": 2.7543411254882812, + "learning_rate": 0.0001654054742539472, + "loss": 3.1918, + "step": 1861 + }, + { + "epoch": 2.73, + "grad_norm": 3.5177597999572754, + "learning_rate": 0.00016536998708142062, + "loss": 3.245, + "step": 1862 + }, + { + "epoch": 2.74, + "grad_norm": 3.0664377212524414, + "learning_rate": 0.00016533448552793494, + "loss": 3.1616, + "step": 1863 + }, + { + "epoch": 2.74, + "grad_norm": 2.924795389175415, + "learning_rate": 0.00016529896960130032, + "loss": 3.229, + "step": 1864 + }, + { + "epoch": 2.74, + "grad_norm": 3.104602575302124, + "learning_rate": 0.00016526343930933, + "loss": 3.2294, + "step": 1865 + }, + { + "epoch": 2.74, + "grad_norm": 3.728653907775879, + "learning_rate": 0.0001652278946598404, + "loss": 3.2088, + "step": 1866 + }, + { + "epoch": 2.74, + "grad_norm": 3.854654312133789, + "learning_rate": 0.00016519233566065112, + "loss": 3.1829, + "step": 1867 + }, + { + "epoch": 2.74, + "grad_norm": 2.6493520736694336, + "learning_rate": 0.0001651567623195849, + "loss": 3.1807, + "step": 1868 + }, + { + "epoch": 2.74, + "grad_norm": 2.057978391647339, + "learning_rate": 0.00016512117464446762, + "loss": 3.141, + "step": 1869 + }, + { + "epoch": 2.75, + "grad_norm": 4.454994201660156, + "learning_rate": 0.00016508557264312835, + "loss": 3.1908, + "step": 1870 + }, + { + "epoch": 2.75, + "grad_norm": 4.243129253387451, + "learning_rate": 0.00016504995632339928, + "loss": 3.1819, + "step": 1871 + }, + { + "epoch": 2.75, + "grad_norm": 2.3739356994628906, + "learning_rate": 0.00016501432569311574, + "loss": 3.2002, + "step": 1872 + }, + { + "epoch": 2.75, + "grad_norm": 1.8129799365997314, + "learning_rate": 0.00016497868076011628, + "loss": 3.2084, + "step": 1873 + }, + { + "epoch": 2.75, + "grad_norm": 4.162952899932861, + "learning_rate": 0.0001649430215322425, + "loss": 3.1766, + "step": 1874 + }, + { + "epoch": 2.75, + "grad_norm": 3.205064535140991, + "learning_rate": 0.00016490734801733923, + "loss": 3.1601, + "step": 1875 + }, + { + "epoch": 2.75, + "grad_norm": 3.1646533012390137, + "learning_rate": 0.0001648716602232544, + "loss": 3.2347, + "step": 1876 + }, + { + "epoch": 2.76, + "grad_norm": 3.2153899669647217, + "learning_rate": 0.00016483595815783904, + "loss": 3.2166, + "step": 1877 + }, + { + "epoch": 2.76, + "grad_norm": 2.5029168128967285, + "learning_rate": 0.00016480024182894743, + "loss": 3.2366, + "step": 1878 + }, + { + "epoch": 2.76, + "grad_norm": 1.944775938987732, + "learning_rate": 0.0001647645112444369, + "loss": 3.1608, + "step": 1879 + }, + { + "epoch": 2.76, + "grad_norm": 4.4062700271606445, + "learning_rate": 0.00016472876641216786, + "loss": 3.226, + "step": 1880 + }, + { + "epoch": 2.76, + "grad_norm": 3.8181326389312744, + "learning_rate": 0.00016469300734000403, + "loss": 3.1934, + "step": 1881 + }, + { + "epoch": 2.76, + "grad_norm": 2.7566487789154053, + "learning_rate": 0.00016465723403581218, + "loss": 3.2031, + "step": 1882 + }, + { + "epoch": 2.77, + "grad_norm": 2.438339948654175, + "learning_rate": 0.00016462144650746213, + "loss": 3.2028, + "step": 1883 + }, + { + "epoch": 2.77, + "grad_norm": 3.6031992435455322, + "learning_rate": 0.0001645856447628269, + "loss": 3.2054, + "step": 1884 + }, + { + "epoch": 2.77, + "grad_norm": 2.86798095703125, + "learning_rate": 0.00016454982880978266, + "loss": 3.2189, + "step": 1885 + }, + { + "epoch": 2.77, + "grad_norm": 3.7704923152923584, + "learning_rate": 0.0001645139986562087, + "loss": 3.1854, + "step": 1886 + }, + { + "epoch": 2.77, + "grad_norm": 3.522306442260742, + "learning_rate": 0.0001644781543099874, + "loss": 3.2357, + "step": 1887 + }, + { + "epoch": 2.77, + "grad_norm": 2.8147850036621094, + "learning_rate": 0.00016444229577900423, + "loss": 3.2041, + "step": 1888 + }, + { + "epoch": 2.77, + "grad_norm": 4.882560729980469, + "learning_rate": 0.00016440642307114786, + "loss": 3.1773, + "step": 1889 + }, + { + "epoch": 2.78, + "grad_norm": 1.0316693782806396, + "learning_rate": 0.00016437053619431006, + "loss": 3.195, + "step": 1890 + }, + { + "epoch": 2.78, + "grad_norm": 1.5295097827911377, + "learning_rate": 0.0001643346351563857, + "loss": 3.2313, + "step": 1891 + }, + { + "epoch": 2.78, + "grad_norm": 1.8384897708892822, + "learning_rate": 0.00016429871996527276, + "loss": 3.1694, + "step": 1892 + }, + { + "epoch": 2.78, + "grad_norm": 1.1532273292541504, + "learning_rate": 0.00016426279062887233, + "loss": 3.1872, + "step": 1893 + }, + { + "epoch": 2.78, + "grad_norm": 0.97287917137146, + "learning_rate": 0.0001642268471550886, + "loss": 3.2283, + "step": 1894 + }, + { + "epoch": 2.78, + "grad_norm": 3.258812665939331, + "learning_rate": 0.00016419088955182897, + "loss": 3.2109, + "step": 1895 + }, + { + "epoch": 2.78, + "grad_norm": 3.9376001358032227, + "learning_rate": 0.00016415491782700384, + "loss": 3.213, + "step": 1896 + }, + { + "epoch": 2.79, + "grad_norm": 3.313781261444092, + "learning_rate": 0.00016411893198852668, + "loss": 3.2419, + "step": 1897 + }, + { + "epoch": 2.79, + "grad_norm": 1.2715352773666382, + "learning_rate": 0.00016408293204431425, + "loss": 3.1853, + "step": 1898 + }, + { + "epoch": 2.79, + "grad_norm": 4.407071113586426, + "learning_rate": 0.00016404691800228616, + "loss": 3.1633, + "step": 1899 + }, + { + "epoch": 2.79, + "grad_norm": 2.881059408187866, + "learning_rate": 0.00016401088987036538, + "loss": 3.1711, + "step": 1900 + }, + { + "epoch": 2.79, + "grad_norm": 213.681884765625, + "learning_rate": 0.0001639748476564778, + "loss": 3.219, + "step": 1901 + }, + { + "epoch": 2.79, + "grad_norm": 15.1211576461792, + "learning_rate": 0.00016393879136855248, + "loss": 3.3071, + "step": 1902 + }, + { + "epoch": 2.79, + "grad_norm": 227.103759765625, + "learning_rate": 0.00016390272101452156, + "loss": 3.3632, + "step": 1903 + }, + { + "epoch": 2.8, + "grad_norm": 94.93277740478516, + "learning_rate": 0.00016386663660232026, + "loss": 3.3738, + "step": 1904 + }, + { + "epoch": 2.8, + "grad_norm": 21.8365421295166, + "learning_rate": 0.00016383053813988694, + "loss": 3.4633, + "step": 1905 + }, + { + "epoch": 2.8, + "grad_norm": 461.8182373046875, + "learning_rate": 0.00016379442563516303, + "loss": 3.4446, + "step": 1906 + }, + { + "epoch": 2.8, + "grad_norm": 73.93539428710938, + "learning_rate": 0.00016375829909609296, + "loss": 3.511, + "step": 1907 + }, + { + "epoch": 2.8, + "grad_norm": 26.853099822998047, + "learning_rate": 0.00016372215853062438, + "loss": 3.4775, + "step": 1908 + }, + { + "epoch": 2.8, + "grad_norm": 449.6775207519531, + "learning_rate": 0.000163686003946708, + "loss": 3.5932, + "step": 1909 + }, + { + "epoch": 2.81, + "grad_norm": 1817.4749755859375, + "learning_rate": 0.00016364983535229754, + "loss": 3.6702, + "step": 1910 + }, + { + "epoch": 2.81, + "grad_norm": 431.15777587890625, + "learning_rate": 0.00016361365275534986, + "loss": 4.165, + "step": 1911 + }, + { + "epoch": 2.81, + "grad_norm": 1153.154296875, + "learning_rate": 0.0001635774561638249, + "loss": 4.4609, + "step": 1912 + }, + { + "epoch": 2.81, + "grad_norm": 378.61279296875, + "learning_rate": 0.00016354124558568567, + "loss": 4.9769, + "step": 1913 + }, + { + "epoch": 2.81, + "grad_norm": 228.8455352783203, + "learning_rate": 0.00016350502102889826, + "loss": 5.4494, + "step": 1914 + }, + { + "epoch": 2.81, + "grad_norm": 2911.531005859375, + "learning_rate": 0.0001634687825014318, + "loss": 5.2172, + "step": 1915 + }, + { + "epoch": 2.81, + "grad_norm": 8.003608703613281, + "learning_rate": 0.00016343253001125852, + "loss": 4.3254, + "step": 1916 + }, + { + "epoch": 2.82, + "grad_norm": 122.48676300048828, + "learning_rate": 0.00016339626356635372, + "loss": 3.9329, + "step": 1917 + }, + { + "epoch": 2.82, + "grad_norm": 165.21929931640625, + "learning_rate": 0.00016335998317469586, + "loss": 4.2273, + "step": 1918 + }, + { + "epoch": 2.82, + "grad_norm": 335.0215148925781, + "learning_rate": 0.00016332368884426626, + "loss": 4.668, + "step": 1919 + }, + { + "epoch": 2.82, + "grad_norm": 3772.19091796875, + "learning_rate": 0.00016328738058304956, + "loss": 5.2298, + "step": 1920 + }, + { + "epoch": 2.82, + "grad_norm": 34.06552505493164, + "learning_rate": 0.0001632510583990332, + "loss": 7.4421, + "step": 1921 + }, + { + "epoch": 2.82, + "grad_norm": 3216.72998046875, + "learning_rate": 0.00016321472230020794, + "loss": 7.2781, + "step": 1922 + }, + { + "epoch": 2.82, + "grad_norm": 9.185710906982422, + "learning_rate": 0.0001631783722945674, + "loss": 6.4889, + "step": 1923 + }, + { + "epoch": 2.83, + "grad_norm": 25.194772720336914, + "learning_rate": 0.00016314200839010834, + "loss": 6.3019, + "step": 1924 + }, + { + "epoch": 2.83, + "grad_norm": 35.877498626708984, + "learning_rate": 0.0001631056305948306, + "loss": 6.1576, + "step": 1925 + }, + { + "epoch": 2.83, + "grad_norm": 35.11639404296875, + "learning_rate": 0.00016306923891673707, + "loss": 5.7687, + "step": 1926 + }, + { + "epoch": 2.83, + "grad_norm": 7.56588077545166, + "learning_rate": 0.0001630328333638336, + "loss": 5.4341, + "step": 1927 + }, + { + "epoch": 2.83, + "grad_norm": 11.163580894470215, + "learning_rate": 0.00016299641394412927, + "loss": 5.2751, + "step": 1928 + }, + { + "epoch": 2.83, + "grad_norm": 9.635797500610352, + "learning_rate": 0.00016295998066563603, + "loss": 5.1351, + "step": 1929 + }, + { + "epoch": 2.83, + "grad_norm": 7.929530143737793, + "learning_rate": 0.000162923533536369, + "loss": 4.9651, + "step": 1930 + }, + { + "epoch": 2.84, + "grad_norm": 12.908093452453613, + "learning_rate": 0.00016288707256434623, + "loss": 4.9926, + "step": 1931 + }, + { + "epoch": 2.84, + "grad_norm": 17.120487213134766, + "learning_rate": 0.000162850597757589, + "loss": 4.9808, + "step": 1932 + }, + { + "epoch": 2.84, + "grad_norm": 12.659987449645996, + "learning_rate": 0.00016281410912412144, + "loss": 4.8239, + "step": 1933 + }, + { + "epoch": 2.84, + "grad_norm": 6.449459552764893, + "learning_rate": 0.00016277760667197087, + "loss": 4.7229, + "step": 1934 + }, + { + "epoch": 2.84, + "grad_norm": 11.50796127319336, + "learning_rate": 0.00016274109040916753, + "loss": 4.5966, + "step": 1935 + }, + { + "epoch": 2.84, + "grad_norm": 10.945295333862305, + "learning_rate": 0.00016270456034374474, + "loss": 4.5114, + "step": 1936 + }, + { + "epoch": 2.85, + "grad_norm": 6.565453052520752, + "learning_rate": 0.00016266801648373894, + "loss": 4.4361, + "step": 1937 + }, + { + "epoch": 2.85, + "grad_norm": 6.896590709686279, + "learning_rate": 0.0001626314588371895, + "loss": 4.3267, + "step": 1938 + }, + { + "epoch": 2.85, + "grad_norm": 6.761734485626221, + "learning_rate": 0.00016259488741213884, + "loss": 4.336, + "step": 1939 + }, + { + "epoch": 2.85, + "grad_norm": 7.6577301025390625, + "learning_rate": 0.00016255830221663243, + "loss": 4.2871, + "step": 1940 + }, + { + "epoch": 2.85, + "grad_norm": 5.5717620849609375, + "learning_rate": 0.00016252170325871878, + "loss": 4.2744, + "step": 1941 + }, + { + "epoch": 2.85, + "grad_norm": 6.425532341003418, + "learning_rate": 0.00016248509054644943, + "loss": 4.2368, + "step": 1942 + }, + { + "epoch": 2.85, + "grad_norm": 6.052181720733643, + "learning_rate": 0.00016244846408787895, + "loss": 4.1986, + "step": 1943 + }, + { + "epoch": 2.86, + "grad_norm": 4.308633804321289, + "learning_rate": 0.00016241182389106483, + "loss": 4.2131, + "step": 1944 + }, + { + "epoch": 2.86, + "grad_norm": 3.014502763748169, + "learning_rate": 0.00016237516996406775, + "loss": 4.1368, + "step": 1945 + }, + { + "epoch": 2.86, + "grad_norm": 3.824622869491577, + "learning_rate": 0.00016233850231495126, + "loss": 4.1542, + "step": 1946 + }, + { + "epoch": 2.86, + "grad_norm": 4.700507640838623, + "learning_rate": 0.00016230182095178206, + "loss": 4.032, + "step": 1947 + }, + { + "epoch": 2.86, + "grad_norm": 2.4007976055145264, + "learning_rate": 0.0001622651258826298, + "loss": 4.043, + "step": 1948 + }, + { + "epoch": 2.86, + "grad_norm": 2.202371835708618, + "learning_rate": 0.0001622284171155671, + "loss": 4.0561, + "step": 1949 + }, + { + "epoch": 2.86, + "grad_norm": 3.510922431945801, + "learning_rate": 0.00016219169465866965, + "loss": 4.0186, + "step": 1950 + }, + { + "epoch": 2.87, + "grad_norm": 3.223702907562256, + "learning_rate": 0.0001621549585200162, + "loss": 4.0432, + "step": 1951 + }, + { + "epoch": 2.87, + "grad_norm": 2.4226481914520264, + "learning_rate": 0.00016211820870768842, + "loss": 4.0018, + "step": 1952 + }, + { + "epoch": 2.87, + "grad_norm": 4.14686393737793, + "learning_rate": 0.000162081445229771, + "loss": 3.9202, + "step": 1953 + }, + { + "epoch": 2.87, + "grad_norm": 4.985760688781738, + "learning_rate": 0.0001620446680943517, + "loss": 4.0172, + "step": 1954 + }, + { + "epoch": 2.87, + "grad_norm": 4.1533427238464355, + "learning_rate": 0.00016200787730952124, + "loss": 3.9214, + "step": 1955 + }, + { + "epoch": 2.87, + "grad_norm": 2.108228921890259, + "learning_rate": 0.00016197107288337328, + "loss": 3.9857, + "step": 1956 + }, + { + "epoch": 2.88, + "grad_norm": 2.6513140201568604, + "learning_rate": 0.00016193425482400464, + "loss": 3.9326, + "step": 1957 + }, + { + "epoch": 2.88, + "grad_norm": 3.7441959381103516, + "learning_rate": 0.00016189742313951498, + "loss": 3.8733, + "step": 1958 + }, + { + "epoch": 2.88, + "grad_norm": 2.4456493854522705, + "learning_rate": 0.00016186057783800704, + "loss": 3.9193, + "step": 1959 + }, + { + "epoch": 2.88, + "grad_norm": 2.1966073513031006, + "learning_rate": 0.00016182371892758658, + "loss": 3.9049, + "step": 1960 + }, + { + "epoch": 2.88, + "grad_norm": 3.958984613418579, + "learning_rate": 0.00016178684641636226, + "loss": 3.8627, + "step": 1961 + }, + { + "epoch": 2.88, + "grad_norm": 2.8413755893707275, + "learning_rate": 0.00016174996031244581, + "loss": 3.9005, + "step": 1962 + }, + { + "epoch": 2.88, + "grad_norm": 2.8592469692230225, + "learning_rate": 0.00016171306062395192, + "loss": 3.8878, + "step": 1963 + }, + { + "epoch": 2.89, + "grad_norm": 5.321715831756592, + "learning_rate": 0.00016167614735899827, + "loss": 3.8257, + "step": 1964 + }, + { + "epoch": 2.89, + "grad_norm": 2.1376495361328125, + "learning_rate": 0.00016163922052570554, + "loss": 3.8189, + "step": 1965 + }, + { + "epoch": 2.89, + "grad_norm": 6.535236835479736, + "learning_rate": 0.0001616022801321974, + "loss": 3.8569, + "step": 1966 + }, + { + "epoch": 2.89, + "grad_norm": 4.882155895233154, + "learning_rate": 0.00016156532618660045, + "loss": 3.8699, + "step": 1967 + }, + { + "epoch": 2.89, + "grad_norm": 5.889296531677246, + "learning_rate": 0.00016152835869704432, + "loss": 3.8201, + "step": 1968 + }, + { + "epoch": 2.89, + "grad_norm": 5.7923102378845215, + "learning_rate": 0.00016149137767166165, + "loss": 3.8467, + "step": 1969 + }, + { + "epoch": 2.89, + "grad_norm": 4.793107509613037, + "learning_rate": 0.000161454383118588, + "loss": 3.789, + "step": 1970 + }, + { + "epoch": 2.9, + "grad_norm": 5.585588455200195, + "learning_rate": 0.00016141737504596186, + "loss": 3.8359, + "step": 1971 + }, + { + "epoch": 2.9, + "grad_norm": 1.328306794166565, + "learning_rate": 0.00016138035346192483, + "loss": 3.781, + "step": 1972 + }, + { + "epoch": 2.9, + "grad_norm": 1.4749982357025146, + "learning_rate": 0.00016134331837462142, + "loss": 3.7788, + "step": 1973 + }, + { + "epoch": 2.9, + "grad_norm": 4.906754016876221, + "learning_rate": 0.00016130626979219906, + "loss": 3.8094, + "step": 1974 + }, + { + "epoch": 2.9, + "grad_norm": 2.958904266357422, + "learning_rate": 0.00016126920772280818, + "loss": 3.8173, + "step": 1975 + }, + { + "epoch": 2.9, + "grad_norm": 8.751588821411133, + "learning_rate": 0.00016123213217460224, + "loss": 3.7913, + "step": 1976 + }, + { + "epoch": 2.9, + "grad_norm": 10.918651580810547, + "learning_rate": 0.00016119504315573755, + "loss": 3.7906, + "step": 1977 + }, + { + "epoch": 2.91, + "grad_norm": 2.1029868125915527, + "learning_rate": 0.0001611579406743735, + "loss": 3.8032, + "step": 1978 + }, + { + "epoch": 2.91, + "grad_norm": 9.132828712463379, + "learning_rate": 0.00016112082473867235, + "loss": 3.8242, + "step": 1979 + }, + { + "epoch": 2.91, + "grad_norm": 10.694255828857422, + "learning_rate": 0.0001610836953567994, + "loss": 3.8359, + "step": 1980 + }, + { + "epoch": 2.91, + "grad_norm": 6.331207752227783, + "learning_rate": 0.0001610465525369228, + "loss": 3.7975, + "step": 1981 + }, + { + "epoch": 2.91, + "grad_norm": 7.4574055671691895, + "learning_rate": 0.00016100939628721375, + "loss": 3.7936, + "step": 1982 + }, + { + "epoch": 2.91, + "grad_norm": 12.698988914489746, + "learning_rate": 0.00016097222661584643, + "loss": 3.8415, + "step": 1983 + }, + { + "epoch": 2.92, + "grad_norm": 7.2185750007629395, + "learning_rate": 0.00016093504353099783, + "loss": 3.8274, + "step": 1984 + }, + { + "epoch": 2.92, + "grad_norm": 5.854779243469238, + "learning_rate": 0.00016089784704084804, + "loss": 3.8396, + "step": 1985 + }, + { + "epoch": 2.92, + "grad_norm": 8.324865341186523, + "learning_rate": 0.00016086063715358003, + "loss": 3.8204, + "step": 1986 + }, + { + "epoch": 2.92, + "grad_norm": 3.311805009841919, + "learning_rate": 0.0001608234138773797, + "loss": 3.7652, + "step": 1987 + }, + { + "epoch": 2.92, + "grad_norm": 11.10246467590332, + "learning_rate": 0.00016078617722043595, + "loss": 3.8391, + "step": 1988 + }, + { + "epoch": 2.92, + "grad_norm": 15.776145935058594, + "learning_rate": 0.00016074892719094056, + "loss": 3.799, + "step": 1989 + }, + { + "epoch": 2.92, + "grad_norm": 7.907688617706299, + "learning_rate": 0.00016071166379708833, + "loss": 3.7895, + "step": 1990 + }, + { + "epoch": 2.93, + "grad_norm": 5.6883649826049805, + "learning_rate": 0.00016067438704707695, + "loss": 3.794, + "step": 1991 + }, + { + "epoch": 2.93, + "grad_norm": 8.36189079284668, + "learning_rate": 0.00016063709694910701, + "loss": 3.8036, + "step": 1992 + }, + { + "epoch": 2.93, + "grad_norm": 4.989264488220215, + "learning_rate": 0.00016059979351138213, + "loss": 3.7555, + "step": 1993 + }, + { + "epoch": 2.93, + "grad_norm": 7.9405670166015625, + "learning_rate": 0.0001605624767421088, + "loss": 3.7995, + "step": 1994 + }, + { + "epoch": 2.93, + "grad_norm": 11.919158935546875, + "learning_rate": 0.00016052514664949646, + "loss": 3.8064, + "step": 1995 + }, + { + "epoch": 2.93, + "grad_norm": 15.537467002868652, + "learning_rate": 0.00016048780324175746, + "loss": 3.8284, + "step": 1996 + }, + { + "epoch": 2.93, + "grad_norm": 3.631410837173462, + "learning_rate": 0.00016045044652710712, + "loss": 3.838, + "step": 1997 + }, + { + "epoch": 2.94, + "grad_norm": 2.047104597091675, + "learning_rate": 0.0001604130765137637, + "loss": 3.7354, + "step": 1998 + }, + { + "epoch": 2.94, + "grad_norm": 8.390944480895996, + "learning_rate": 0.00016037569320994827, + "loss": 3.8143, + "step": 1999 + }, + { + "epoch": 2.94, + "grad_norm": 8.165762901306152, + "learning_rate": 0.00016033829662388498, + "loss": 3.7676, + "step": 2000 + }, + { + "epoch": 2.94, + "grad_norm": 2.4152541160583496, + "learning_rate": 0.0001603008867638008, + "loss": 3.7495, + "step": 2001 + }, + { + "epoch": 2.94, + "grad_norm": 5.799648761749268, + "learning_rate": 0.00016026346363792567, + "loss": 3.7996, + "step": 2002 + }, + { + "epoch": 2.94, + "grad_norm": 3.751105785369873, + "learning_rate": 0.00016022602725449238, + "loss": 3.7672, + "step": 2003 + }, + { + "epoch": 2.94, + "grad_norm": 7.3117780685424805, + "learning_rate": 0.0001601885776217367, + "loss": 3.7477, + "step": 2004 + }, + { + "epoch": 2.95, + "grad_norm": 7.469902992248535, + "learning_rate": 0.00016015111474789733, + "loss": 3.7555, + "step": 2005 + }, + { + "epoch": 2.95, + "grad_norm": 2.65592622756958, + "learning_rate": 0.00016011363864121582, + "loss": 3.7347, + "step": 2006 + }, + { + "epoch": 2.95, + "grad_norm": 3.7348546981811523, + "learning_rate": 0.0001600761493099367, + "loss": 3.7731, + "step": 2007 + }, + { + "epoch": 2.95, + "grad_norm": 2.6096608638763428, + "learning_rate": 0.0001600386467623073, + "loss": 3.7667, + "step": 2008 + }, + { + "epoch": 2.95, + "grad_norm": 1.8336734771728516, + "learning_rate": 0.000160001131006578, + "loss": 3.7475, + "step": 2009 + }, + { + "epoch": 2.95, + "grad_norm": 9.613265991210938, + "learning_rate": 0.000159963602051002, + "loss": 3.7442, + "step": 2010 + }, + { + "epoch": 2.96, + "grad_norm": 4.934391021728516, + "learning_rate": 0.00015992605990383538, + "loss": 3.7025, + "step": 2011 + }, + { + "epoch": 2.96, + "grad_norm": 2.538320779800415, + "learning_rate": 0.0001598885045733372, + "loss": 3.7681, + "step": 2012 + }, + { + "epoch": 2.96, + "grad_norm": 2.606849193572998, + "learning_rate": 0.00015985093606776937, + "loss": 3.7237, + "step": 2013 + }, + { + "epoch": 2.96, + "grad_norm": 3.6826469898223877, + "learning_rate": 0.0001598133543953967, + "loss": 3.7382, + "step": 2014 + }, + { + "epoch": 2.96, + "grad_norm": 1.016801118850708, + "learning_rate": 0.0001597757595644869, + "loss": 3.7059, + "step": 2015 + }, + { + "epoch": 2.96, + "grad_norm": 6.855797290802002, + "learning_rate": 0.00015973815158331064, + "loss": 3.7315, + "step": 2016 + }, + { + "epoch": 2.96, + "grad_norm": 4.787756443023682, + "learning_rate": 0.00015970053046014137, + "loss": 3.7273, + "step": 2017 + }, + { + "epoch": 2.97, + "grad_norm": 5.884622573852539, + "learning_rate": 0.00015966289620325546, + "loss": 3.7079, + "step": 2018 + }, + { + "epoch": 2.97, + "grad_norm": 7.18374490737915, + "learning_rate": 0.00015962524882093227, + "loss": 3.7161, + "step": 2019 + }, + { + "epoch": 2.97, + "grad_norm": 1.6866484880447388, + "learning_rate": 0.00015958758832145388, + "loss": 3.6481, + "step": 2020 + }, + { + "epoch": 2.97, + "grad_norm": 10.191508293151855, + "learning_rate": 0.00015954991471310543, + "loss": 3.7034, + "step": 2021 + }, + { + "epoch": 2.97, + "grad_norm": 14.30430793762207, + "learning_rate": 0.0001595122280041748, + "loss": 3.7654, + "step": 2022 + }, + { + "epoch": 2.97, + "grad_norm": 5.003686904907227, + "learning_rate": 0.00015947452820295292, + "loss": 3.7148, + "step": 2023 + }, + { + "epoch": 2.97, + "grad_norm": 6.288287162780762, + "learning_rate": 0.00015943681531773333, + "loss": 3.7528, + "step": 2024 + }, + { + "epoch": 2.98, + "grad_norm": 8.045740127563477, + "learning_rate": 0.00015939908935681273, + "loss": 3.6763, + "step": 2025 + }, + { + "epoch": 2.98, + "grad_norm": 3.2493703365325928, + "learning_rate": 0.0001593613503284906, + "loss": 3.6939, + "step": 2026 + }, + { + "epoch": 2.98, + "grad_norm": 10.84262752532959, + "learning_rate": 0.0001593235982410691, + "loss": 3.7516, + "step": 2027 + }, + { + "epoch": 2.98, + "grad_norm": 15.299811363220215, + "learning_rate": 0.00015928583310285366, + "loss": 3.7319, + "step": 2028 + }, + { + "epoch": 2.98, + "eval_loss": 3.5392327308654785, + "eval_runtime": 1153.2794, + "eval_samples_per_second": 14.864, + "eval_steps_per_second": 0.93, + "step": 2028 + }, + { + "epoch": 2.98, + "grad_norm": 8.231132507324219, + "learning_rate": 0.0001592480549221522, + "loss": 3.7177, + "step": 2029 + }, + { + "epoch": 2.98, + "grad_norm": 5.159106254577637, + "learning_rate": 0.00015921026370727573, + "loss": 3.7017, + "step": 2030 + }, + { + "epoch": 2.98, + "grad_norm": 22.314449310302734, + "learning_rate": 0.00015917245946653807, + "loss": 3.6842, + "step": 2031 + }, + { + "epoch": 2.99, + "grad_norm": 8.433747291564941, + "learning_rate": 0.00015913464220825588, + "loss": 3.6786, + "step": 2032 + }, + { + "epoch": 2.99, + "grad_norm": 3.3623647689819336, + "learning_rate": 0.00015909681194074867, + "loss": 3.678, + "step": 2033 + }, + { + "epoch": 2.99, + "grad_norm": 10.62174129486084, + "learning_rate": 0.00015905896867233894, + "loss": 3.7098, + "step": 2034 + }, + { + "epoch": 2.99, + "grad_norm": 15.299424171447754, + "learning_rate": 0.00015902111241135187, + "loss": 3.6926, + "step": 2035 + }, + { + "epoch": 2.99, + "grad_norm": 9.536619186401367, + "learning_rate": 0.00015898324316611563, + "loss": 3.673, + "step": 2036 + }, + { + "epoch": 2.99, + "grad_norm": 5.059504508972168, + "learning_rate": 0.00015894536094496117, + "loss": 3.6625, + "step": 2037 + }, + { + "epoch": 3.0, + "grad_norm": 7.704652786254883, + "learning_rate": 0.00015890746575622231, + "loss": 3.6632, + "step": 2038 + }, + { + "epoch": 3.0, + "grad_norm": 4.4129557609558105, + "learning_rate": 0.00015886955760823578, + "loss": 3.6315, + "step": 2039 + }, + { + "epoch": 3.0, + "grad_norm": 7.341033935546875, + "learning_rate": 0.0001588316365093411, + "loss": 3.6763, + "step": 2040 + }, + { + "epoch": 3.0, + "grad_norm": 9.556234359741211, + "learning_rate": 0.00015879370246788067, + "loss": 3.6774, + "step": 2041 + }, + { + "epoch": 3.0, + "grad_norm": 3.8044204711914062, + "learning_rate": 0.00015875575549219973, + "loss": 3.6249, + "step": 2042 + }, + { + "epoch": 3.0, + "grad_norm": 7.746762275695801, + "learning_rate": 0.0001587177955906463, + "loss": 3.6366, + "step": 2043 + }, + { + "epoch": 3.0, + "grad_norm": 9.851776123046875, + "learning_rate": 0.0001586798227715714, + "loss": 3.6885, + "step": 2044 + }, + { + "epoch": 3.01, + "grad_norm": 5.628739833831787, + "learning_rate": 0.00015864183704332872, + "loss": 3.6248, + "step": 2045 + }, + { + "epoch": 3.01, + "grad_norm": 6.932154178619385, + "learning_rate": 0.00015860383841427485, + "loss": 3.6499, + "step": 2046 + }, + { + "epoch": 3.01, + "grad_norm": 10.925968170166016, + "learning_rate": 0.00015856582689276934, + "loss": 3.6793, + "step": 2047 + }, + { + "epoch": 3.01, + "grad_norm": 5.439330577850342, + "learning_rate": 0.00015852780248717442, + "loss": 3.654, + "step": 2048 + }, + { + "epoch": 3.01, + "grad_norm": 6.066278457641602, + "learning_rate": 0.0001584897652058551, + "loss": 3.6738, + "step": 2049 + }, + { + "epoch": 3.01, + "grad_norm": 8.065271377563477, + "learning_rate": 0.00015845171505717947, + "loss": 3.659, + "step": 2050 + }, + { + "epoch": 3.01, + "grad_norm": 3.8582706451416016, + "learning_rate": 0.00015841365204951826, + "loss": 3.6458, + "step": 2051 + }, + { + "epoch": 3.02, + "grad_norm": 9.716731071472168, + "learning_rate": 0.00015837557619124507, + "loss": 3.6313, + "step": 2052 + }, + { + "epoch": 3.0, + "grad_norm": 14.822004318237305, + "learning_rate": 0.00015833748749073632, + "loss": 3.6512, + "step": 2053 + }, + { + "epoch": 3.0, + "grad_norm": 8.674092292785645, + "learning_rate": 0.00015829938595637128, + "loss": 3.6251, + "step": 2054 + }, + { + "epoch": 3.0, + "grad_norm": 4.23799467086792, + "learning_rate": 0.00015826127159653203, + "loss": 3.6496, + "step": 2055 + }, + { + "epoch": 3.01, + "grad_norm": 9.582878112792969, + "learning_rate": 0.00015822314441960348, + "loss": 3.5982, + "step": 2056 + }, + { + "epoch": 3.01, + "grad_norm": 5.006598949432373, + "learning_rate": 0.00015818500443397337, + "loss": 3.6347, + "step": 2057 + }, + { + "epoch": 3.01, + "grad_norm": 4.559854984283447, + "learning_rate": 0.0001581468516480322, + "loss": 3.6414, + "step": 2058 + }, + { + "epoch": 3.01, + "grad_norm": 5.798408508300781, + "learning_rate": 0.00015810868607017337, + "loss": 3.6176, + "step": 2059 + }, + { + "epoch": 3.01, + "grad_norm": 1.9246889352798462, + "learning_rate": 0.00015807050770879304, + "loss": 3.6134, + "step": 2060 + }, + { + "epoch": 3.01, + "grad_norm": 1.124200701713562, + "learning_rate": 0.00015803231657229017, + "loss": 3.6295, + "step": 2061 + }, + { + "epoch": 3.01, + "grad_norm": 7.120904445648193, + "learning_rate": 0.00015799411266906657, + "loss": 3.6101, + "step": 2062 + }, + { + "epoch": 3.02, + "grad_norm": 5.709414482116699, + "learning_rate": 0.00015795589600752682, + "loss": 3.6255, + "step": 2063 + }, + { + "epoch": 3.02, + "grad_norm": 3.2987709045410156, + "learning_rate": 0.0001579176665960784, + "loss": 3.6194, + "step": 2064 + }, + { + "epoch": 3.02, + "grad_norm": 4.208385944366455, + "learning_rate": 0.00015787942444313145, + "loss": 3.6357, + "step": 2065 + }, + { + "epoch": 3.02, + "grad_norm": 2.485320568084717, + "learning_rate": 0.00015784116955709897, + "loss": 3.6233, + "step": 2066 + }, + { + "epoch": 3.02, + "grad_norm": 0.9779897332191467, + "learning_rate": 0.0001578029019463969, + "loss": 3.6382, + "step": 2067 + }, + { + "epoch": 3.02, + "grad_norm": 5.236642837524414, + "learning_rate": 0.0001577646216194437, + "loss": 3.5985, + "step": 2068 + }, + { + "epoch": 3.03, + "grad_norm": 4.276678562164307, + "learning_rate": 0.00015772632858466093, + "loss": 3.614, + "step": 2069 + }, + { + "epoch": 3.03, + "grad_norm": 4.390791893005371, + "learning_rate": 0.00015768802285047272, + "loss": 3.6166, + "step": 2070 + }, + { + "epoch": 3.03, + "grad_norm": 4.886763572692871, + "learning_rate": 0.00015764970442530609, + "loss": 3.5902, + "step": 2071 + }, + { + "epoch": 3.03, + "grad_norm": 3.569462776184082, + "learning_rate": 0.00015761137331759084, + "loss": 3.5779, + "step": 2072 + }, + { + "epoch": 3.03, + "grad_norm": 2.935269832611084, + "learning_rate": 0.00015757302953575954, + "loss": 3.6369, + "step": 2073 + }, + { + "epoch": 3.03, + "grad_norm": 4.911141872406006, + "learning_rate": 0.0001575346730882476, + "loss": 3.6034, + "step": 2074 + }, + { + "epoch": 3.03, + "grad_norm": 4.09962797164917, + "learning_rate": 0.00015749630398349316, + "loss": 3.6431, + "step": 2075 + }, + { + "epoch": 3.04, + "grad_norm": 4.247868537902832, + "learning_rate": 0.00015745792222993715, + "loss": 3.6258, + "step": 2076 + }, + { + "epoch": 3.04, + "grad_norm": 3.7722268104553223, + "learning_rate": 0.00015741952783602332, + "loss": 3.597, + "step": 2077 + }, + { + "epoch": 3.04, + "grad_norm": 5.005308151245117, + "learning_rate": 0.00015738112081019817, + "loss": 3.6146, + "step": 2078 + }, + { + "epoch": 3.04, + "grad_norm": 4.696826934814453, + "learning_rate": 0.000157342701160911, + "loss": 3.6477, + "step": 2079 + }, + { + "epoch": 3.04, + "grad_norm": 3.856837034225464, + "learning_rate": 0.00015730426889661386, + "loss": 3.6594, + "step": 2080 + }, + { + "epoch": 3.04, + "grad_norm": 3.4349594116210938, + "learning_rate": 0.0001572658240257616, + "loss": 3.5798, + "step": 2081 + }, + { + "epoch": 3.04, + "grad_norm": 4.7815842628479, + "learning_rate": 0.00015722736655681186, + "loss": 3.5797, + "step": 2082 + }, + { + "epoch": 3.05, + "grad_norm": 3.9778714179992676, + "learning_rate": 0.00015718889649822494, + "loss": 3.6173, + "step": 2083 + }, + { + "epoch": 3.05, + "grad_norm": 3.863435745239258, + "learning_rate": 0.0001571504138584641, + "loss": 3.5669, + "step": 2084 + }, + { + "epoch": 3.05, + "grad_norm": 3.9693548679351807, + "learning_rate": 0.0001571119186459952, + "loss": 3.6072, + "step": 2085 + }, + { + "epoch": 3.05, + "grad_norm": 4.673881530761719, + "learning_rate": 0.00015707341086928698, + "loss": 3.5794, + "step": 2086 + }, + { + "epoch": 3.05, + "grad_norm": 3.680610179901123, + "learning_rate": 0.0001570348905368108, + "loss": 3.5927, + "step": 2087 + }, + { + "epoch": 3.05, + "grad_norm": 4.484524726867676, + "learning_rate": 0.00015699635765704096, + "loss": 3.5588, + "step": 2088 + }, + { + "epoch": 3.05, + "grad_norm": 4.283873558044434, + "learning_rate": 0.00015695781223845441, + "loss": 3.6385, + "step": 2089 + }, + { + "epoch": 3.06, + "grad_norm": 3.3641796112060547, + "learning_rate": 0.00015691925428953089, + "loss": 3.6163, + "step": 2090 + }, + { + "epoch": 3.06, + "grad_norm": 2.774048089981079, + "learning_rate": 0.00015688068381875287, + "loss": 3.6696, + "step": 2091 + }, + { + "epoch": 3.06, + "grad_norm": 5.591458320617676, + "learning_rate": 0.0001568421008346056, + "loss": 3.54, + "step": 2092 + }, + { + "epoch": 3.06, + "grad_norm": 5.354968547821045, + "learning_rate": 0.0001568035053455771, + "loss": 3.5762, + "step": 2093 + }, + { + "epoch": 3.06, + "grad_norm": 2.219583749771118, + "learning_rate": 0.0001567648973601581, + "loss": 3.5527, + "step": 2094 + }, + { + "epoch": 3.06, + "grad_norm": 1.655458688735962, + "learning_rate": 0.0001567262768868421, + "loss": 3.5702, + "step": 2095 + }, + { + "epoch": 3.07, + "grad_norm": 5.801334381103516, + "learning_rate": 0.00015668764393412537, + "loss": 3.5317, + "step": 2096 + }, + { + "epoch": 3.07, + "grad_norm": 5.561006546020508, + "learning_rate": 0.00015664899851050684, + "loss": 3.5875, + "step": 2097 + }, + { + "epoch": 3.07, + "grad_norm": 2.8261959552764893, + "learning_rate": 0.0001566103406244883, + "loss": 3.5284, + "step": 2098 + }, + { + "epoch": 3.07, + "grad_norm": 2.3607285022735596, + "learning_rate": 0.00015657167028457423, + "loss": 3.5596, + "step": 2099 + }, + { + "epoch": 3.07, + "grad_norm": 6.529599666595459, + "learning_rate": 0.00015653298749927178, + "loss": 3.5603, + "step": 2100 + }, + { + "epoch": 3.07, + "grad_norm": 5.787454605102539, + "learning_rate": 0.00015649429227709098, + "loss": 3.5419, + "step": 2101 + }, + { + "epoch": 3.07, + "grad_norm": 1.2445347309112549, + "learning_rate": 0.00015645558462654447, + "loss": 3.5599, + "step": 2102 + }, + { + "epoch": 3.08, + "grad_norm": 1.686257243156433, + "learning_rate": 0.0001564168645561477, + "loss": 3.5051, + "step": 2103 + }, + { + "epoch": 3.08, + "grad_norm": 5.125936985015869, + "learning_rate": 0.00015637813207441882, + "loss": 3.4685, + "step": 2104 + }, + { + "epoch": 3.08, + "grad_norm": 4.658946514129639, + "learning_rate": 0.0001563393871898787, + "loss": 3.5291, + "step": 2105 + }, + { + "epoch": 3.08, + "grad_norm": 4.34171199798584, + "learning_rate": 0.00015630062991105098, + "loss": 3.5044, + "step": 2106 + }, + { + "epoch": 3.08, + "grad_norm": 5.2388458251953125, + "learning_rate": 0.00015626186024646197, + "loss": 3.4543, + "step": 2107 + }, + { + "epoch": 3.08, + "grad_norm": 3.39254093170166, + "learning_rate": 0.00015622307820464077, + "loss": 3.4678, + "step": 2108 + }, + { + "epoch": 3.08, + "grad_norm": 2.4865806102752686, + "learning_rate": 0.00015618428379411917, + "loss": 3.4326, + "step": 2109 + }, + { + "epoch": 3.09, + "grad_norm": 7.326157093048096, + "learning_rate": 0.00015614547702343166, + "loss": 3.4709, + "step": 2110 + }, + { + "epoch": 3.09, + "grad_norm": 7.048543930053711, + "learning_rate": 0.00015610665790111546, + "loss": 3.4498, + "step": 2111 + }, + { + "epoch": 3.09, + "grad_norm": 10.954419136047363, + "learning_rate": 0.00015606782643571055, + "loss": 3.3944, + "step": 2112 + }, + { + "epoch": 3.09, + "grad_norm": 11.594827651977539, + "learning_rate": 0.00015602898263575952, + "loss": 3.4224, + "step": 2113 + }, + { + "epoch": 3.09, + "grad_norm": 6.556940078735352, + "learning_rate": 0.0001559901265098079, + "loss": 3.4079, + "step": 2114 + }, + { + "epoch": 3.09, + "grad_norm": 3.0502734184265137, + "learning_rate": 0.00015595125806640357, + "loss": 3.409, + "step": 2115 + }, + { + "epoch": 3.09, + "grad_norm": 6.4218220710754395, + "learning_rate": 0.0001559123773140975, + "loss": 3.4382, + "step": 2116 + }, + { + "epoch": 3.1, + "grad_norm": 6.8841753005981445, + "learning_rate": 0.00015587348426144312, + "loss": 3.463, + "step": 2117 + }, + { + "epoch": 3.1, + "grad_norm": 35.60420227050781, + "learning_rate": 0.00015583457891699666, + "loss": 3.3878, + "step": 2118 + }, + { + "epoch": 3.1, + "grad_norm": 2.3054988384246826, + "learning_rate": 0.000155795661289317, + "loss": 3.382, + "step": 2119 + }, + { + "epoch": 3.1, + "grad_norm": 6.13869571685791, + "learning_rate": 0.00015575673138696578, + "loss": 3.4213, + "step": 2120 + }, + { + "epoch": 3.1, + "grad_norm": 5.998043537139893, + "learning_rate": 0.00015571778921850732, + "loss": 3.382, + "step": 2121 + }, + { + "epoch": 3.1, + "grad_norm": 1.6560511589050293, + "learning_rate": 0.00015567883479250863, + "loss": 3.4691, + "step": 2122 + }, + { + "epoch": 3.11, + "grad_norm": 2.120764970779419, + "learning_rate": 0.00015563986811753945, + "loss": 3.3711, + "step": 2123 + }, + { + "epoch": 3.11, + "grad_norm": 1.6065012216567993, + "learning_rate": 0.00015560088920217217, + "loss": 3.4219, + "step": 2124 + }, + { + "epoch": 3.11, + "grad_norm": 6.458485126495361, + "learning_rate": 0.00015556189805498188, + "loss": 3.4184, + "step": 2125 + }, + { + "epoch": 3.11, + "grad_norm": 5.5630202293396, + "learning_rate": 0.0001555228946845464, + "loss": 3.4185, + "step": 2126 + }, + { + "epoch": 3.11, + "grad_norm": 4.821037769317627, + "learning_rate": 0.00015548387909944614, + "loss": 3.4003, + "step": 2127 + }, + { + "epoch": 3.11, + "grad_norm": 13.251823425292969, + "learning_rate": 0.00015544485130826433, + "loss": 3.4081, + "step": 2128 + }, + { + "epoch": 3.11, + "grad_norm": 5.775404930114746, + "learning_rate": 0.00015540581131958682, + "loss": 3.3918, + "step": 2129 + }, + { + "epoch": 3.12, + "grad_norm": 6.066825866699219, + "learning_rate": 0.00015536675914200213, + "loss": 3.3863, + "step": 2130 + }, + { + "epoch": 3.12, + "grad_norm": 11.889984130859375, + "learning_rate": 0.0001553276947841015, + "loss": 3.4055, + "step": 2131 + }, + { + "epoch": 3.12, + "grad_norm": 3.267805576324463, + "learning_rate": 0.00015528861825447876, + "loss": 3.4031, + "step": 2132 + }, + { + "epoch": 3.12, + "grad_norm": 11.77675724029541, + "learning_rate": 0.00015524952956173056, + "loss": 3.4159, + "step": 2133 + }, + { + "epoch": 3.12, + "grad_norm": 16.67833137512207, + "learning_rate": 0.0001552104287144561, + "loss": 3.4706, + "step": 2134 + }, + { + "epoch": 3.12, + "grad_norm": 9.038000106811523, + "learning_rate": 0.00015517131572125736, + "loss": 3.4391, + "step": 2135 + }, + { + "epoch": 3.12, + "grad_norm": 7.349591255187988, + "learning_rate": 0.0001551321905907389, + "loss": 3.4181, + "step": 2136 + }, + { + "epoch": 3.13, + "grad_norm": 11.440853118896484, + "learning_rate": 0.00015509305333150797, + "loss": 3.4324, + "step": 2137 + }, + { + "epoch": 3.13, + "grad_norm": 9.080293655395508, + "learning_rate": 0.00015505390395217453, + "loss": 3.4841, + "step": 2138 + }, + { + "epoch": 3.13, + "grad_norm": 2.2852790355682373, + "learning_rate": 0.0001550147424613512, + "loss": 3.35, + "step": 2139 + }, + { + "epoch": 3.13, + "grad_norm": 17.961963653564453, + "learning_rate": 0.00015497556886765316, + "loss": 3.4265, + "step": 2140 + }, + { + "epoch": 3.13, + "grad_norm": 26.195865631103516, + "learning_rate": 0.00015493638317969844, + "loss": 3.6051, + "step": 2141 + }, + { + "epoch": 3.13, + "grad_norm": 19.95352554321289, + "learning_rate": 0.0001548971854061076, + "loss": 3.6381, + "step": 2142 + }, + { + "epoch": 3.13, + "grad_norm": 5.972691535949707, + "learning_rate": 0.00015485797555550386, + "loss": 3.5607, + "step": 2143 + }, + { + "epoch": 3.14, + "grad_norm": 13.078893661499023, + "learning_rate": 0.0001548187536365131, + "loss": 3.4778, + "step": 2144 + }, + { + "epoch": 3.14, + "grad_norm": 15.685088157653809, + "learning_rate": 0.000154779519657764, + "loss": 3.5011, + "step": 2145 + }, + { + "epoch": 3.14, + "grad_norm": 14.251028060913086, + "learning_rate": 0.00015474027362788763, + "loss": 3.6155, + "step": 2146 + }, + { + "epoch": 3.14, + "grad_norm": 10.396077156066895, + "learning_rate": 0.00015470101555551796, + "loss": 3.5646, + "step": 2147 + }, + { + "epoch": 3.14, + "grad_norm": 5.34552526473999, + "learning_rate": 0.0001546617454492914, + "loss": 3.4643, + "step": 2148 + }, + { + "epoch": 3.14, + "grad_norm": 18.76544189453125, + "learning_rate": 0.0001546224633178472, + "loss": 3.4686, + "step": 2149 + }, + { + "epoch": 3.15, + "grad_norm": 32.796051025390625, + "learning_rate": 0.00015458316916982715, + "loss": 3.6171, + "step": 2150 + }, + { + "epoch": 3.15, + "grad_norm": 31.727258682250977, + "learning_rate": 0.00015454386301387565, + "loss": 3.751, + "step": 2151 + }, + { + "epoch": 3.15, + "grad_norm": 18.45448112487793, + "learning_rate": 0.00015450454485863984, + "loss": 3.6405, + "step": 2152 + }, + { + "epoch": 3.15, + "grad_norm": 8.118375778198242, + "learning_rate": 0.00015446521471276946, + "loss": 3.6001, + "step": 2153 + }, + { + "epoch": 3.15, + "grad_norm": 16.14570426940918, + "learning_rate": 0.0001544258725849168, + "loss": 3.559, + "step": 2154 + }, + { + "epoch": 3.15, + "grad_norm": 16.983943939208984, + "learning_rate": 0.00015438651848373696, + "loss": 3.5108, + "step": 2155 + }, + { + "epoch": 3.15, + "grad_norm": 13.206223487854004, + "learning_rate": 0.00015434715241788752, + "loss": 3.4883, + "step": 2156 + }, + { + "epoch": 3.16, + "grad_norm": 8.664280891418457, + "learning_rate": 0.00015430777439602876, + "loss": 3.4346, + "step": 2157 + }, + { + "epoch": 3.16, + "grad_norm": 5.633321762084961, + "learning_rate": 0.00015426838442682362, + "loss": 3.4285, + "step": 2158 + }, + { + "epoch": 3.16, + "grad_norm": 14.557170867919922, + "learning_rate": 0.00015422898251893756, + "loss": 3.4706, + "step": 2159 + }, + { + "epoch": 3.16, + "grad_norm": 19.554941177368164, + "learning_rate": 0.00015418956868103878, + "loss": 3.4929, + "step": 2160 + }, + { + "epoch": 3.16, + "grad_norm": 15.579254150390625, + "learning_rate": 0.00015415014292179805, + "loss": 3.4144, + "step": 2161 + }, + { + "epoch": 3.16, + "grad_norm": 5.079093933105469, + "learning_rate": 0.0001541107052498888, + "loss": 3.3886, + "step": 2162 + }, + { + "epoch": 3.16, + "grad_norm": 10.48103141784668, + "learning_rate": 0.00015407125567398698, + "loss": 3.4392, + "step": 2163 + }, + { + "epoch": 3.17, + "grad_norm": 14.062845230102539, + "learning_rate": 0.00015403179420277131, + "loss": 3.4527, + "step": 2164 + }, + { + "epoch": 3.17, + "grad_norm": 13.237971305847168, + "learning_rate": 0.00015399232084492304, + "loss": 3.4518, + "step": 2165 + }, + { + "epoch": 3.17, + "grad_norm": 7.261989593505859, + "learning_rate": 0.000153952835609126, + "loss": 3.4716, + "step": 2166 + }, + { + "epoch": 3.17, + "grad_norm": 5.766734600067139, + "learning_rate": 0.0001539133385040667, + "loss": 3.3591, + "step": 2167 + }, + { + "epoch": 3.17, + "grad_norm": 10.160755157470703, + "learning_rate": 0.0001538738295384343, + "loss": 3.3638, + "step": 2168 + }, + { + "epoch": 3.17, + "grad_norm": 7.740986347198486, + "learning_rate": 0.00015383430872092037, + "loss": 3.4132, + "step": 2169 + }, + { + "epoch": 3.18, + "grad_norm": 2.3649628162384033, + "learning_rate": 0.00015379477606021933, + "loss": 3.4459, + "step": 2170 + }, + { + "epoch": 3.18, + "grad_norm": 4.513500213623047, + "learning_rate": 0.00015375523156502805, + "loss": 3.4318, + "step": 2171 + }, + { + "epoch": 3.18, + "grad_norm": 3.214402198791504, + "learning_rate": 0.00015371567524404613, + "loss": 3.3651, + "step": 2172 + }, + { + "epoch": 3.18, + "grad_norm": 3.8382275104522705, + "learning_rate": 0.0001536761071059756, + "loss": 3.4077, + "step": 2173 + }, + { + "epoch": 3.18, + "grad_norm": 1.4553401470184326, + "learning_rate": 0.0001536365271595212, + "loss": 3.3301, + "step": 2174 + }, + { + "epoch": 3.18, + "grad_norm": 4.466412544250488, + "learning_rate": 0.00015359693541339037, + "loss": 3.3988, + "step": 2175 + }, + { + "epoch": 3.18, + "grad_norm": 3.6478259563446045, + "learning_rate": 0.00015355733187629288, + "loss": 3.3779, + "step": 2176 + }, + { + "epoch": 3.19, + "grad_norm": 5.013825416564941, + "learning_rate": 0.0001535177165569413, + "loss": 3.413, + "step": 2177 + }, + { + "epoch": 3.19, + "grad_norm": 4.778167247772217, + "learning_rate": 0.00015347808946405075, + "loss": 3.3425, + "step": 2178 + }, + { + "epoch": 3.19, + "grad_norm": 5.024134159088135, + "learning_rate": 0.00015343845060633893, + "loss": 3.3676, + "step": 2179 + }, + { + "epoch": 3.19, + "grad_norm": 5.648217678070068, + "learning_rate": 0.00015339879999252606, + "loss": 3.3165, + "step": 2180 + }, + { + "epoch": 3.19, + "grad_norm": 2.347975015640259, + "learning_rate": 0.00015335913763133511, + "loss": 3.3588, + "step": 2181 + }, + { + "epoch": 3.19, + "grad_norm": 5.947681427001953, + "learning_rate": 0.00015331946353149144, + "loss": 3.3811, + "step": 2182 + }, + { + "epoch": 3.19, + "grad_norm": 5.794849395751953, + "learning_rate": 0.00015327977770172315, + "loss": 3.3751, + "step": 2183 + }, + { + "epoch": 3.2, + "grad_norm": 2.177245855331421, + "learning_rate": 0.0001532400801507608, + "loss": 3.3472, + "step": 2184 + }, + { + "epoch": 3.2, + "grad_norm": 4.221487045288086, + "learning_rate": 0.00015320037088733764, + "loss": 3.3766, + "step": 2185 + }, + { + "epoch": 3.2, + "grad_norm": 3.649068832397461, + "learning_rate": 0.00015316064992018945, + "loss": 3.3633, + "step": 2186 + }, + { + "epoch": 3.2, + "grad_norm": 3.780611038208008, + "learning_rate": 0.00015312091725805448, + "loss": 3.3371, + "step": 2187 + }, + { + "epoch": 3.2, + "grad_norm": 1.3155494928359985, + "learning_rate": 0.00015308117290967373, + "loss": 3.3083, + "step": 2188 + }, + { + "epoch": 3.2, + "grad_norm": 1.961712121963501, + "learning_rate": 0.0001530414168837907, + "loss": 3.3465, + "step": 2189 + }, + { + "epoch": 3.2, + "grad_norm": 2.569746494293213, + "learning_rate": 0.00015300164918915138, + "loss": 3.3853, + "step": 2190 + }, + { + "epoch": 3.21, + "grad_norm": 1.9574949741363525, + "learning_rate": 0.00015296186983450446, + "loss": 3.3444, + "step": 2191 + }, + { + "epoch": 3.21, + "grad_norm": 2.3811464309692383, + "learning_rate": 0.00015292207882860114, + "loss": 3.316, + "step": 2192 + }, + { + "epoch": 3.21, + "grad_norm": 1.6529579162597656, + "learning_rate": 0.00015288227618019513, + "loss": 3.3269, + "step": 2193 + }, + { + "epoch": 3.21, + "grad_norm": 1.4792758226394653, + "learning_rate": 0.00015284246189804272, + "loss": 3.3406, + "step": 2194 + }, + { + "epoch": 3.21, + "grad_norm": 1.3875869512557983, + "learning_rate": 0.0001528026359909029, + "loss": 3.3492, + "step": 2195 + }, + { + "epoch": 3.21, + "grad_norm": 3.4284157752990723, + "learning_rate": 0.00015276279846753696, + "loss": 3.3176, + "step": 2196 + }, + { + "epoch": 3.22, + "grad_norm": 1.68557608127594, + "learning_rate": 0.000152722949336709, + "loss": 3.3142, + "step": 2197 + }, + { + "epoch": 3.22, + "eval_loss": 3.1609973907470703, + "eval_runtime": 1163.3297, + "eval_samples_per_second": 14.735, + "eval_steps_per_second": 0.921, + "step": 2197 + }, + { + "epoch": 3.22, + "grad_norm": 4.416086673736572, + "learning_rate": 0.0001526830886071855, + "loss": 3.3143, + "step": 2198 + }, + { + "epoch": 3.22, + "grad_norm": 2.9269638061523438, + "learning_rate": 0.0001526432162877356, + "loss": 3.3417, + "step": 2199 + }, + { + "epoch": 3.22, + "grad_norm": 3.496257781982422, + "learning_rate": 0.00015260333238713088, + "loss": 3.3356, + "step": 2200 + }, + { + "epoch": 3.22, + "grad_norm": 3.9582598209381104, + "learning_rate": 0.00015256343691414558, + "loss": 3.3688, + "step": 2201 + }, + { + "epoch": 3.22, + "grad_norm": 3.0187695026397705, + "learning_rate": 0.0001525235298775564, + "loss": 3.3381, + "step": 2202 + }, + { + "epoch": 3.22, + "grad_norm": 1.4246793985366821, + "learning_rate": 0.0001524836112861427, + "loss": 3.3228, + "step": 2203 + }, + { + "epoch": 3.23, + "grad_norm": 2.6875085830688477, + "learning_rate": 0.00015244368114868625, + "loss": 3.2918, + "step": 2204 + }, + { + "epoch": 3.23, + "grad_norm": 2.1740024089813232, + "learning_rate": 0.00015240373947397133, + "loss": 3.2933, + "step": 2205 + }, + { + "epoch": 3.23, + "grad_norm": 5.561516284942627, + "learning_rate": 0.000152363786270785, + "loss": 3.3344, + "step": 2206 + }, + { + "epoch": 3.23, + "grad_norm": 2.453502655029297, + "learning_rate": 0.00015232382154791655, + "loss": 3.3204, + "step": 2207 + }, + { + "epoch": 3.23, + "grad_norm": 3.4560389518737793, + "learning_rate": 0.0001522838453141581, + "loss": 3.3054, + "step": 2208 + }, + { + "epoch": 3.23, + "grad_norm": 1.4101018905639648, + "learning_rate": 0.000152243857578304, + "loss": 3.3298, + "step": 2209 + }, + { + "epoch": 3.23, + "grad_norm": 1.3143606185913086, + "learning_rate": 0.0001522038583491514, + "loss": 3.3294, + "step": 2210 + }, + { + "epoch": 3.24, + "grad_norm": 2.1382877826690674, + "learning_rate": 0.00015216384763549977, + "loss": 3.2964, + "step": 2211 + }, + { + "epoch": 3.24, + "grad_norm": 3.5953447818756104, + "learning_rate": 0.00015212382544615126, + "loss": 3.3138, + "step": 2212 + }, + { + "epoch": 3.24, + "grad_norm": 2.013397216796875, + "learning_rate": 0.0001520837917899105, + "loss": 3.3318, + "step": 2213 + }, + { + "epoch": 3.24, + "grad_norm": 5.66317892074585, + "learning_rate": 0.0001520437466755846, + "loss": 3.3226, + "step": 2214 + }, + { + "epoch": 3.24, + "grad_norm": 5.973227024078369, + "learning_rate": 0.00015200369011198319, + "loss": 3.2917, + "step": 2215 + }, + { + "epoch": 3.24, + "grad_norm": 1.0823651552200317, + "learning_rate": 0.00015196362210791845, + "loss": 3.2837, + "step": 2216 + }, + { + "epoch": 3.24, + "grad_norm": 2.2828619480133057, + "learning_rate": 0.00015192354267220512, + "loss": 3.2994, + "step": 2217 + }, + { + "epoch": 3.25, + "grad_norm": 3.3102993965148926, + "learning_rate": 0.00015188345181366036, + "loss": 3.3674, + "step": 2218 + }, + { + "epoch": 3.25, + "grad_norm": 5.152593612670898, + "learning_rate": 0.00015184334954110394, + "loss": 3.3411, + "step": 2219 + }, + { + "epoch": 3.25, + "grad_norm": 6.140212535858154, + "learning_rate": 0.00015180323586335806, + "loss": 3.3188, + "step": 2220 + }, + { + "epoch": 3.25, + "grad_norm": 2.0337228775024414, + "learning_rate": 0.0001517631107892474, + "loss": 3.2678, + "step": 2221 + }, + { + "epoch": 3.25, + "grad_norm": 4.024622917175293, + "learning_rate": 0.0001517229743275993, + "loss": 3.2708, + "step": 2222 + }, + { + "epoch": 3.25, + "grad_norm": 2.378216505050659, + "learning_rate": 0.00015168282648724351, + "loss": 3.3462, + "step": 2223 + }, + { + "epoch": 3.26, + "grad_norm": 4.922878742218018, + "learning_rate": 0.0001516426672770122, + "loss": 3.291, + "step": 2224 + }, + { + "epoch": 3.26, + "grad_norm": 4.67563533782959, + "learning_rate": 0.00015160249670574026, + "loss": 3.2711, + "step": 2225 + }, + { + "epoch": 3.26, + "grad_norm": 4.26483154296875, + "learning_rate": 0.0001515623147822648, + "loss": 3.2862, + "step": 2226 + }, + { + "epoch": 3.26, + "grad_norm": 6.510219097137451, + "learning_rate": 0.00015152212151542567, + "loss": 3.3276, + "step": 2227 + }, + { + "epoch": 3.26, + "grad_norm": 4.469378471374512, + "learning_rate": 0.0001514819169140651, + "loss": 3.3276, + "step": 2228 + }, + { + "epoch": 3.26, + "grad_norm": 4.4898681640625, + "learning_rate": 0.0001514417009870278, + "loss": 3.2907, + "step": 2229 + }, + { + "epoch": 3.26, + "grad_norm": 5.1355485916137695, + "learning_rate": 0.00015140147374316106, + "loss": 3.3146, + "step": 2230 + }, + { + "epoch": 3.27, + "grad_norm": 3.954402208328247, + "learning_rate": 0.00015136123519131456, + "loss": 3.2888, + "step": 2231 + }, + { + "epoch": 3.27, + "grad_norm": 8.745506286621094, + "learning_rate": 0.00015132098534034052, + "loss": 3.3684, + "step": 2232 + }, + { + "epoch": 3.27, + "grad_norm": 7.295302391052246, + "learning_rate": 0.00015128072419909368, + "loss": 3.3378, + "step": 2233 + }, + { + "epoch": 3.27, + "grad_norm": 1.574733853340149, + "learning_rate": 0.00015124045177643112, + "loss": 3.3003, + "step": 2234 + }, + { + "epoch": 3.27, + "grad_norm": 6.663206100463867, + "learning_rate": 0.0001512001680812126, + "loss": 3.3466, + "step": 2235 + }, + { + "epoch": 3.27, + "grad_norm": 7.278547286987305, + "learning_rate": 0.00015115987312230026, + "loss": 3.2945, + "step": 2236 + }, + { + "epoch": 3.27, + "grad_norm": 3.385538101196289, + "learning_rate": 0.00015111956690855863, + "loss": 3.2958, + "step": 2237 + }, + { + "epoch": 3.28, + "grad_norm": 7.496275424957275, + "learning_rate": 0.00015107924944885486, + "loss": 3.3214, + "step": 2238 + }, + { + "epoch": 3.28, + "grad_norm": 25.01158332824707, + "learning_rate": 0.00015103892075205862, + "loss": 3.4204, + "step": 2239 + }, + { + "epoch": 3.28, + "grad_norm": 9.22580337524414, + "learning_rate": 0.00015099858082704182, + "loss": 3.2895, + "step": 2240 + }, + { + "epoch": 3.28, + "grad_norm": 1.5853315591812134, + "learning_rate": 0.00015095822968267896, + "loss": 3.3388, + "step": 2241 + }, + { + "epoch": 3.28, + "grad_norm": 8.119154930114746, + "learning_rate": 0.00015091786732784716, + "loss": 3.35, + "step": 2242 + }, + { + "epoch": 3.28, + "grad_norm": 10.604775428771973, + "learning_rate": 0.00015087749377142574, + "loss": 3.3851, + "step": 2243 + }, + { + "epoch": 3.28, + "grad_norm": 7.424951076507568, + "learning_rate": 0.00015083710902229664, + "loss": 3.3619, + "step": 2244 + }, + { + "epoch": 3.29, + "grad_norm": 5.7252373695373535, + "learning_rate": 0.0001507967130893443, + "loss": 3.342, + "step": 2245 + }, + { + "epoch": 3.29, + "grad_norm": 8.934366226196289, + "learning_rate": 0.00015075630598145547, + "loss": 3.3553, + "step": 2246 + }, + { + "epoch": 3.29, + "grad_norm": 6.990628719329834, + "learning_rate": 0.00015071588770751947, + "loss": 3.3318, + "step": 2247 + }, + { + "epoch": 3.29, + "grad_norm": 19.209877014160156, + "learning_rate": 0.00015067545827642808, + "loss": 3.3181, + "step": 2248 + }, + { + "epoch": 3.29, + "grad_norm": 6.441047191619873, + "learning_rate": 0.00015063501769707548, + "loss": 3.3515, + "step": 2249 + }, + { + "epoch": 3.29, + "grad_norm": 5.99086332321167, + "learning_rate": 0.00015059456597835835, + "loss": 3.334, + "step": 2250 + }, + { + "epoch": 3.3, + "grad_norm": 2.6446921825408936, + "learning_rate": 0.00015055410312917573, + "loss": 3.332, + "step": 2251 + }, + { + "epoch": 3.3, + "grad_norm": 266.819091796875, + "learning_rate": 0.00015051362915842923, + "loss": 3.3605, + "step": 2252 + }, + { + "epoch": 3.3, + "grad_norm": 16.00495719909668, + "learning_rate": 0.00015047314407502283, + "loss": 3.3521, + "step": 2253 + }, + { + "epoch": 3.3, + "grad_norm": 8.721566200256348, + "learning_rate": 0.00015043264788786305, + "loss": 3.3846, + "step": 2254 + }, + { + "epoch": 3.3, + "grad_norm": 3.7059578895568848, + "learning_rate": 0.00015039214060585864, + "loss": 3.39, + "step": 2255 + }, + { + "epoch": 3.3, + "grad_norm": 5.986879825592041, + "learning_rate": 0.00015035162223792106, + "loss": 3.3791, + "step": 2256 + }, + { + "epoch": 3.3, + "grad_norm": 4.337857723236084, + "learning_rate": 0.00015031109279296402, + "loss": 3.3309, + "step": 2257 + }, + { + "epoch": 3.31, + "grad_norm": 4.526264190673828, + "learning_rate": 0.00015027055227990373, + "loss": 3.3768, + "step": 2258 + }, + { + "epoch": 3.31, + "grad_norm": 7.14890718460083, + "learning_rate": 0.00015023000070765884, + "loss": 3.3677, + "step": 2259 + }, + { + "epoch": 3.31, + "grad_norm": 3.9721760749816895, + "learning_rate": 0.0001501894380851504, + "loss": 3.3136, + "step": 2260 + }, + { + "epoch": 3.31, + "grad_norm": 5.246433734893799, + "learning_rate": 0.00015014886442130197, + "loss": 3.3263, + "step": 2261 + }, + { + "epoch": 3.31, + "grad_norm": 7.421927452087402, + "learning_rate": 0.00015010827972503944, + "loss": 3.3386, + "step": 2262 + }, + { + "epoch": 3.31, + "grad_norm": 3.354271173477173, + "learning_rate": 0.00015006768400529117, + "loss": 3.3509, + "step": 2263 + }, + { + "epoch": 3.31, + "grad_norm": 5.732051849365234, + "learning_rate": 0.00015002707727098798, + "loss": 3.3885, + "step": 2264 + }, + { + "epoch": 3.32, + "grad_norm": 6.248003005981445, + "learning_rate": 0.00014998645953106303, + "loss": 3.3291, + "step": 2265 + }, + { + "epoch": 3.32, + "grad_norm": 2.3988747596740723, + "learning_rate": 0.000149945830794452, + "loss": 3.2951, + "step": 2266 + }, + { + "epoch": 3.32, + "grad_norm": 4.345824241638184, + "learning_rate": 0.00014990519107009292, + "loss": 3.3256, + "step": 2267 + }, + { + "epoch": 3.32, + "grad_norm": 4.299257278442383, + "learning_rate": 0.0001498645403669262, + "loss": 3.3154, + "step": 2268 + }, + { + "epoch": 3.32, + "grad_norm": 4.104548931121826, + "learning_rate": 0.00014982387869389485, + "loss": 3.3248, + "step": 2269 + }, + { + "epoch": 3.32, + "grad_norm": 3.4004967212677, + "learning_rate": 0.00014978320605994405, + "loss": 3.3147, + "step": 2270 + }, + { + "epoch": 3.33, + "grad_norm": 3.225266456604004, + "learning_rate": 0.00014974252247402156, + "loss": 3.2917, + "step": 2271 + }, + { + "epoch": 3.33, + "grad_norm": 3.6412038803100586, + "learning_rate": 0.00014970182794507748, + "loss": 3.3557, + "step": 2272 + }, + { + "epoch": 3.33, + "grad_norm": 3.6593449115753174, + "learning_rate": 0.00014966112248206436, + "loss": 3.3048, + "step": 2273 + }, + { + "epoch": 3.33, + "grad_norm": 3.3542096614837646, + "learning_rate": 0.0001496204060939371, + "loss": 3.2807, + "step": 2274 + }, + { + "epoch": 3.33, + "grad_norm": 3.2932636737823486, + "learning_rate": 0.000149579678789653, + "loss": 3.3457, + "step": 2275 + }, + { + "epoch": 3.33, + "grad_norm": 2.769178867340088, + "learning_rate": 0.00014953894057817188, + "loss": 3.3023, + "step": 2276 + }, + { + "epoch": 3.33, + "grad_norm": 3.352829694747925, + "learning_rate": 0.0001494981914684558, + "loss": 3.2836, + "step": 2277 + }, + { + "epoch": 3.34, + "grad_norm": 1.4692974090576172, + "learning_rate": 0.00014945743146946937, + "loss": 3.354, + "step": 2278 + }, + { + "epoch": 3.34, + "grad_norm": 5.136416912078857, + "learning_rate": 0.0001494166605901794, + "loss": 3.2915, + "step": 2279 + }, + { + "epoch": 3.34, + "grad_norm": 4.615678310394287, + "learning_rate": 0.00014937587883955534, + "loss": 3.2766, + "step": 2280 + }, + { + "epoch": 3.34, + "grad_norm": 1.0332159996032715, + "learning_rate": 0.00014933508622656878, + "loss": 3.3291, + "step": 2281 + }, + { + "epoch": 3.34, + "grad_norm": 1.1297160387039185, + "learning_rate": 0.00014929428276019388, + "loss": 3.3465, + "step": 2282 + }, + { + "epoch": 3.34, + "grad_norm": 3.5306038856506348, + "learning_rate": 0.00014925346844940718, + "loss": 3.2554, + "step": 2283 + }, + { + "epoch": 3.34, + "grad_norm": 9.56828498840332, + "learning_rate": 0.00014921264330318746, + "loss": 3.3203, + "step": 2284 + }, + { + "epoch": 3.35, + "grad_norm": 2.0608508586883545, + "learning_rate": 0.00014917180733051602, + "loss": 3.2899, + "step": 2285 + }, + { + "epoch": 3.35, + "grad_norm": 1.6151994466781616, + "learning_rate": 0.00014913096054037654, + "loss": 3.3209, + "step": 2286 + }, + { + "epoch": 3.35, + "grad_norm": 2.8889124393463135, + "learning_rate": 0.00014909010294175497, + "loss": 3.3144, + "step": 2287 + }, + { + "epoch": 3.35, + "grad_norm": 0.650810182094574, + "learning_rate": 0.00014904923454363975, + "loss": 3.2888, + "step": 2288 + }, + { + "epoch": 3.35, + "grad_norm": 2.751948833465576, + "learning_rate": 0.00014900835535502165, + "loss": 3.3317, + "step": 2289 + }, + { + "epoch": 3.35, + "grad_norm": 1.9524742364883423, + "learning_rate": 0.0001489674653848938, + "loss": 3.2798, + "step": 2290 + }, + { + "epoch": 3.35, + "grad_norm": 0.6142035126686096, + "learning_rate": 0.00014892656464225175, + "loss": 3.2485, + "step": 2291 + }, + { + "epoch": 3.36, + "grad_norm": 1.9580166339874268, + "learning_rate": 0.00014888565313609333, + "loss": 3.3353, + "step": 2292 + }, + { + "epoch": 3.36, + "grad_norm": 4.1013898849487305, + "learning_rate": 0.0001488447308754189, + "loss": 3.3178, + "step": 2293 + }, + { + "epoch": 3.36, + "grad_norm": 1.7551053762435913, + "learning_rate": 0.00014880379786923098, + "loss": 3.2874, + "step": 2294 + }, + { + "epoch": 3.36, + "grad_norm": 5.447327136993408, + "learning_rate": 0.00014876285412653456, + "loss": 3.2928, + "step": 2295 + }, + { + "epoch": 3.36, + "grad_norm": 5.519541263580322, + "learning_rate": 0.00014872189965633707, + "loss": 3.3418, + "step": 2296 + }, + { + "epoch": 3.36, + "grad_norm": 2.044977903366089, + "learning_rate": 0.0001486809344676482, + "loss": 3.3401, + "step": 2297 + }, + { + "epoch": 3.37, + "grad_norm": 5.334053993225098, + "learning_rate": 0.00014863995856947995, + "loss": 3.3156, + "step": 2298 + }, + { + "epoch": 3.37, + "grad_norm": 3.4951860904693604, + "learning_rate": 0.0001485989719708468, + "loss": 3.2812, + "step": 2299 + }, + { + "epoch": 3.37, + "grad_norm": 5.83620548248291, + "learning_rate": 0.00014855797468076552, + "loss": 3.3016, + "step": 2300 + }, + { + "epoch": 3.37, + "grad_norm": 6.08860969543457, + "learning_rate": 0.00014851696670825523, + "loss": 3.2941, + "step": 2301 + }, + { + "epoch": 3.37, + "grad_norm": 7.569385051727295, + "learning_rate": 0.00014847594806233737, + "loss": 3.3478, + "step": 2302 + }, + { + "epoch": 3.37, + "grad_norm": 5.372185230255127, + "learning_rate": 0.00014843491875203584, + "loss": 3.2692, + "step": 2303 + }, + { + "epoch": 3.37, + "grad_norm": 4.030975341796875, + "learning_rate": 0.0001483938787863768, + "loss": 3.2826, + "step": 2304 + }, + { + "epoch": 3.38, + "grad_norm": 6.85429048538208, + "learning_rate": 0.00014835282817438875, + "loss": 3.2932, + "step": 2305 + }, + { + "epoch": 3.38, + "grad_norm": 6.535613059997559, + "learning_rate": 0.0001483117669251025, + "loss": 3.3158, + "step": 2306 + }, + { + "epoch": 3.38, + "grad_norm": 3.962482452392578, + "learning_rate": 0.00014827069504755132, + "loss": 3.308, + "step": 2307 + }, + { + "epoch": 3.38, + "grad_norm": 6.860799312591553, + "learning_rate": 0.00014822961255077073, + "loss": 3.3145, + "step": 2308 + }, + { + "epoch": 3.38, + "grad_norm": 9.554657936096191, + "learning_rate": 0.0001481885194437986, + "loss": 3.3366, + "step": 2309 + }, + { + "epoch": 3.38, + "grad_norm": 4.180676460266113, + "learning_rate": 0.00014814741573567514, + "loss": 3.3096, + "step": 2310 + }, + { + "epoch": 3.38, + "grad_norm": 6.105428218841553, + "learning_rate": 0.0001481063014354429, + "loss": 3.2919, + "step": 2311 + }, + { + "epoch": 3.39, + "grad_norm": 7.584985733032227, + "learning_rate": 0.00014806517655214672, + "loss": 3.2756, + "step": 2312 + }, + { + "epoch": 3.39, + "grad_norm": 5.3676300048828125, + "learning_rate": 0.00014802404109483383, + "loss": 3.2969, + "step": 2313 + }, + { + "epoch": 3.39, + "grad_norm": 3.991481065750122, + "learning_rate": 0.00014798289507255373, + "loss": 3.2808, + "step": 2314 + }, + { + "epoch": 3.39, + "grad_norm": 6.212815761566162, + "learning_rate": 0.00014794173849435826, + "loss": 3.2809, + "step": 2315 + }, + { + "epoch": 3.39, + "grad_norm": 1.0110559463500977, + "learning_rate": 0.00014790057136930166, + "loss": 3.3016, + "step": 2316 + }, + { + "epoch": 3.39, + "grad_norm": 9.187312126159668, + "learning_rate": 0.00014785939370644036, + "loss": 3.275, + "step": 2317 + }, + { + "epoch": 3.39, + "grad_norm": 12.635149955749512, + "learning_rate": 0.00014781820551483315, + "loss": 3.3793, + "step": 2318 + }, + { + "epoch": 3.4, + "grad_norm": 9.69156551361084, + "learning_rate": 0.00014777700680354124, + "loss": 3.3063, + "step": 2319 + }, + { + "epoch": 3.4, + "grad_norm": 1.8201349973678589, + "learning_rate": 0.000147735797581628, + "loss": 3.3122, + "step": 2320 + }, + { + "epoch": 3.4, + "grad_norm": 11.608341217041016, + "learning_rate": 0.0001476945778581592, + "loss": 3.341, + "step": 2321 + }, + { + "epoch": 3.4, + "grad_norm": 13.871359825134277, + "learning_rate": 0.0001476533476422029, + "loss": 3.3459, + "step": 2322 + }, + { + "epoch": 3.4, + "grad_norm": 9.090968132019043, + "learning_rate": 0.00014761210694282947, + "loss": 3.369, + "step": 2323 + }, + { + "epoch": 3.4, + "grad_norm": 4.659329414367676, + "learning_rate": 0.0001475708557691116, + "loss": 3.417, + "step": 2324 + }, + { + "epoch": 3.41, + "grad_norm": 6.619305610656738, + "learning_rate": 0.00014752959413012428, + "loss": 3.3524, + "step": 2325 + }, + { + "epoch": 3.41, + "grad_norm": 9.944043159484863, + "learning_rate": 0.00014748832203494478, + "loss": 3.3326, + "step": 2326 + }, + { + "epoch": 3.41, + "grad_norm": 5.2585062980651855, + "learning_rate": 0.00014744703949265268, + "loss": 3.3292, + "step": 2327 + }, + { + "epoch": 3.41, + "grad_norm": 5.1725311279296875, + "learning_rate": 0.00014740574651232987, + "loss": 3.3172, + "step": 2328 + }, + { + "epoch": 3.41, + "grad_norm": 30.300024032592773, + "learning_rate": 0.00014736444310306052, + "loss": 3.3064, + "step": 2329 + }, + { + "epoch": 3.41, + "grad_norm": 9.257378578186035, + "learning_rate": 0.00014732312927393112, + "loss": 3.2872, + "step": 2330 + }, + { + "epoch": 3.41, + "grad_norm": 2.5596399307250977, + "learning_rate": 0.0001472818050340304, + "loss": 3.2914, + "step": 2331 + }, + { + "epoch": 3.42, + "grad_norm": 6.8813934326171875, + "learning_rate": 0.00014724047039244947, + "loss": 3.3698, + "step": 2332 + }, + { + "epoch": 3.42, + "grad_norm": 10.194779396057129, + "learning_rate": 0.00014719912535828166, + "loss": 3.2775, + "step": 2333 + }, + { + "epoch": 3.42, + "grad_norm": 5.2471489906311035, + "learning_rate": 0.00014715776994062254, + "loss": 3.3149, + "step": 2334 + }, + { + "epoch": 3.42, + "grad_norm": 3.753305196762085, + "learning_rate": 0.00014711640414857008, + "loss": 3.3024, + "step": 2335 + }, + { + "epoch": 3.42, + "grad_norm": 5.786775588989258, + "learning_rate": 0.00014707502799122447, + "loss": 3.2799, + "step": 2336 + }, + { + "epoch": 3.42, + "grad_norm": 1.7370816469192505, + "learning_rate": 0.00014703364147768822, + "loss": 3.2881, + "step": 2337 + }, + { + "epoch": 3.42, + "grad_norm": 7.590539932250977, + "learning_rate": 0.000146992244617066, + "loss": 3.3002, + "step": 2338 + }, + { + "epoch": 3.43, + "grad_norm": 9.67332935333252, + "learning_rate": 0.00014695083741846492, + "loss": 3.3037, + "step": 2339 + }, + { + "epoch": 3.43, + "grad_norm": 5.325073719024658, + "learning_rate": 0.00014690941989099426, + "loss": 3.3241, + "step": 2340 + }, + { + "epoch": 3.43, + "grad_norm": 7.251718044281006, + "learning_rate": 0.0001468679920437656, + "loss": 3.316, + "step": 2341 + }, + { + "epoch": 3.43, + "grad_norm": 8.890804290771484, + "learning_rate": 0.00014682655388589274, + "loss": 3.3177, + "step": 2342 + }, + { + "epoch": 3.43, + "grad_norm": 3.9834563732147217, + "learning_rate": 0.0001467851054264919, + "loss": 3.3002, + "step": 2343 + }, + { + "epoch": 3.43, + "grad_norm": 5.180391788482666, + "learning_rate": 0.0001467436466746814, + "loss": 3.3166, + "step": 2344 + }, + { + "epoch": 3.43, + "grad_norm": 9.531121253967285, + "learning_rate": 0.00014670217763958188, + "loss": 3.2482, + "step": 2345 + }, + { + "epoch": 3.44, + "grad_norm": 4.674380302429199, + "learning_rate": 0.00014666069833031624, + "loss": 3.2703, + "step": 2346 + }, + { + "epoch": 3.44, + "grad_norm": 3.9839441776275635, + "learning_rate": 0.00014661920875600973, + "loss": 3.2701, + "step": 2347 + }, + { + "epoch": 3.44, + "grad_norm": 4.408015251159668, + "learning_rate": 0.00014657770892578968, + "loss": 3.2791, + "step": 2348 + }, + { + "epoch": 3.44, + "grad_norm": 2.449519395828247, + "learning_rate": 0.0001465361988487858, + "loss": 3.2654, + "step": 2349 + }, + { + "epoch": 3.44, + "grad_norm": 1.4051802158355713, + "learning_rate": 0.00014649467853413008, + "loss": 3.3101, + "step": 2350 + }, + { + "epoch": 3.44, + "grad_norm": 6.514969825744629, + "learning_rate": 0.00014645314799095665, + "loss": 3.2282, + "step": 2351 + }, + { + "epoch": 3.45, + "grad_norm": 5.476304054260254, + "learning_rate": 0.000146411607228402, + "loss": 3.3043, + "step": 2352 + }, + { + "epoch": 3.45, + "grad_norm": 2.443798780441284, + "learning_rate": 0.00014637005625560477, + "loss": 3.3045, + "step": 2353 + }, + { + "epoch": 3.45, + "grad_norm": 6.676492214202881, + "learning_rate": 0.00014632849508170595, + "loss": 3.2794, + "step": 2354 + }, + { + "epoch": 3.45, + "grad_norm": 2.0837247371673584, + "learning_rate": 0.0001462869237158487, + "loss": 3.3134, + "step": 2355 + }, + { + "epoch": 3.45, + "grad_norm": 7.334787368774414, + "learning_rate": 0.00014624534216717841, + "loss": 3.2919, + "step": 2356 + }, + { + "epoch": 3.45, + "grad_norm": 8.694489479064941, + "learning_rate": 0.00014620375044484277, + "loss": 3.2639, + "step": 2357 + }, + { + "epoch": 3.45, + "grad_norm": 2.46984601020813, + "learning_rate": 0.00014616214855799167, + "loss": 3.2571, + "step": 2358 + }, + { + "epoch": 3.46, + "grad_norm": 7.231738567352295, + "learning_rate": 0.00014612053651577727, + "loss": 3.3468, + "step": 2359 + }, + { + "epoch": 3.46, + "grad_norm": 7.998284816741943, + "learning_rate": 0.0001460789143273539, + "loss": 3.3764, + "step": 2360 + }, + { + "epoch": 3.46, + "grad_norm": 4.452177047729492, + "learning_rate": 0.00014603728200187824, + "loss": 3.3293, + "step": 2361 + }, + { + "epoch": 3.46, + "grad_norm": 9.913199424743652, + "learning_rate": 0.00014599563954850901, + "loss": 3.2631, + "step": 2362 + }, + { + "epoch": 3.46, + "grad_norm": 15.90157413482666, + "learning_rate": 0.00014595398697640733, + "loss": 3.3812, + "step": 2363 + }, + { + "epoch": 3.46, + "grad_norm": 12.660530090332031, + "learning_rate": 0.0001459123242947365, + "loss": 3.371, + "step": 2364 + }, + { + "epoch": 3.46, + "grad_norm": 5.289348125457764, + "learning_rate": 0.00014587065151266203, + "loss": 3.3892, + "step": 2365 + }, + { + "epoch": 3.47, + "grad_norm": 9.648274421691895, + "learning_rate": 0.00014582896863935162, + "loss": 3.2931, + "step": 2366 + }, + { + "epoch": 3.47, + "eval_loss": 3.133899211883545, + "eval_runtime": 1162.2679, + "eval_samples_per_second": 14.749, + "eval_steps_per_second": 0.922, + "step": 2366 + }, + { + "epoch": 3.47, + "grad_norm": 9.562708854675293, + "learning_rate": 0.00014578727568397522, + "loss": 3.3485, + "step": 2367 + }, + { + "epoch": 3.47, + "grad_norm": 19.97218894958496, + "learning_rate": 0.00014574557265570505, + "loss": 3.3403, + "step": 2368 + }, + { + "epoch": 3.47, + "grad_norm": 6.222564220428467, + "learning_rate": 0.0001457038595637155, + "loss": 3.3743, + "step": 2369 + }, + { + "epoch": 3.47, + "grad_norm": 7.945571422576904, + "learning_rate": 0.00014566213641718307, + "loss": 3.2679, + "step": 2370 + }, + { + "epoch": 3.47, + "grad_norm": 8.883635520935059, + "learning_rate": 0.0001456204032252867, + "loss": 3.2295, + "step": 2371 + }, + { + "epoch": 3.48, + "grad_norm": 5.227753639221191, + "learning_rate": 0.00014557865999720735, + "loss": 3.2934, + "step": 2372 + }, + { + "epoch": 3.48, + "grad_norm": 8.865499496459961, + "learning_rate": 0.0001455369067421282, + "loss": 3.3526, + "step": 2373 + }, + { + "epoch": 3.48, + "grad_norm": 10.483139991760254, + "learning_rate": 0.00014549514346923477, + "loss": 3.3058, + "step": 2374 + }, + { + "epoch": 3.48, + "grad_norm": 12.922052383422852, + "learning_rate": 0.00014545337018771472, + "loss": 3.2366, + "step": 2375 + }, + { + "epoch": 3.48, + "grad_norm": 2.131599187850952, + "learning_rate": 0.00014541158690675778, + "loss": 3.2613, + "step": 2376 + }, + { + "epoch": 3.48, + "grad_norm": 10.010235786437988, + "learning_rate": 0.0001453697936355561, + "loss": 3.2698, + "step": 2377 + }, + { + "epoch": 3.48, + "grad_norm": 14.034173011779785, + "learning_rate": 0.00014532799038330385, + "loss": 3.3691, + "step": 2378 + }, + { + "epoch": 3.49, + "grad_norm": 9.930380821228027, + "learning_rate": 0.0001452861771591975, + "loss": 3.304, + "step": 2379 + }, + { + "epoch": 3.49, + "grad_norm": 7.6726555824279785, + "learning_rate": 0.00014524435397243567, + "loss": 3.2806, + "step": 2380 + }, + { + "epoch": 3.49, + "grad_norm": 8.018184661865234, + "learning_rate": 0.00014520252083221916, + "loss": 3.2673, + "step": 2381 + }, + { + "epoch": 3.49, + "grad_norm": 8.354979515075684, + "learning_rate": 0.00014516067774775101, + "loss": 3.295, + "step": 2382 + }, + { + "epoch": 3.49, + "grad_norm": 3.118138313293457, + "learning_rate": 0.00014511882472823643, + "loss": 3.2593, + "step": 2383 + }, + { + "epoch": 3.49, + "grad_norm": 9.12159538269043, + "learning_rate": 0.00014507696178288278, + "loss": 3.2893, + "step": 2384 + }, + { + "epoch": 3.49, + "grad_norm": 13.92984390258789, + "learning_rate": 0.0001450350889208996, + "loss": 3.3191, + "step": 2385 + }, + { + "epoch": 3.5, + "grad_norm": 9.545883178710938, + "learning_rate": 0.0001449932061514987, + "loss": 3.2753, + "step": 2386 + }, + { + "epoch": 3.5, + "grad_norm": 1.6055418252944946, + "learning_rate": 0.00014495131348389394, + "loss": 3.2373, + "step": 2387 + }, + { + "epoch": 3.5, + "grad_norm": 8.169912338256836, + "learning_rate": 0.00014490941092730152, + "loss": 3.2748, + "step": 2388 + }, + { + "epoch": 3.5, + "grad_norm": 8.912496566772461, + "learning_rate": 0.00014486749849093963, + "loss": 3.302, + "step": 2389 + }, + { + "epoch": 3.5, + "grad_norm": 6.068753242492676, + "learning_rate": 0.00014482557618402875, + "loss": 3.2805, + "step": 2390 + }, + { + "epoch": 3.5, + "grad_norm": 4.4562578201293945, + "learning_rate": 0.00014478364401579154, + "loss": 3.2488, + "step": 2391 + }, + { + "epoch": 3.5, + "grad_norm": 4.690038204193115, + "learning_rate": 0.00014474170199545275, + "loss": 3.2896, + "step": 2392 + }, + { + "epoch": 3.51, + "grad_norm": 3.260173797607422, + "learning_rate": 0.0001446997501322394, + "loss": 3.2497, + "step": 2393 + }, + { + "epoch": 3.51, + "grad_norm": 3.311274528503418, + "learning_rate": 0.0001446577884353806, + "loss": 3.2359, + "step": 2394 + }, + { + "epoch": 3.51, + "grad_norm": 2.5797505378723145, + "learning_rate": 0.00014461581691410758, + "loss": 3.3372, + "step": 2395 + }, + { + "epoch": 3.51, + "grad_norm": 1.9033737182617188, + "learning_rate": 0.00014457383557765386, + "loss": 3.2308, + "step": 2396 + }, + { + "epoch": 3.51, + "grad_norm": 3.9633514881134033, + "learning_rate": 0.000144531844435255, + "loss": 3.2787, + "step": 2397 + }, + { + "epoch": 3.51, + "grad_norm": 3.756439685821533, + "learning_rate": 0.00014448984349614886, + "loss": 3.3238, + "step": 2398 + }, + { + "epoch": 3.52, + "grad_norm": 3.5643348693847656, + "learning_rate": 0.00014444783276957524, + "loss": 3.2315, + "step": 2399 + }, + { + "epoch": 3.52, + "grad_norm": 4.421493053436279, + "learning_rate": 0.00014440581226477634, + "loss": 3.277, + "step": 2400 + }, + { + "epoch": 3.52, + "grad_norm": 4.980103969573975, + "learning_rate": 0.00014436378199099634, + "loss": 3.248, + "step": 2401 + }, + { + "epoch": 3.52, + "grad_norm": 7.962924003601074, + "learning_rate": 0.0001443217419574816, + "loss": 3.2628, + "step": 2402 + }, + { + "epoch": 3.52, + "grad_norm": 3.279404401779175, + "learning_rate": 0.00014427969217348065, + "loss": 3.2662, + "step": 2403 + }, + { + "epoch": 3.52, + "grad_norm": 6.002499580383301, + "learning_rate": 0.00014423763264824417, + "loss": 3.2498, + "step": 2404 + }, + { + "epoch": 3.52, + "grad_norm": 8.197328567504883, + "learning_rate": 0.00014419556339102503, + "loss": 3.3432, + "step": 2405 + }, + { + "epoch": 3.53, + "grad_norm": 3.395263910293579, + "learning_rate": 0.00014415348441107806, + "loss": 3.2308, + "step": 2406 + }, + { + "epoch": 3.53, + "grad_norm": 6.569970607757568, + "learning_rate": 0.00014411139571766042, + "loss": 3.2445, + "step": 2407 + }, + { + "epoch": 3.53, + "grad_norm": 9.0120849609375, + "learning_rate": 0.0001440692973200314, + "loss": 3.2689, + "step": 2408 + }, + { + "epoch": 3.53, + "grad_norm": 5.56130313873291, + "learning_rate": 0.00014402718922745228, + "loss": 3.2685, + "step": 2409 + }, + { + "epoch": 3.53, + "grad_norm": 4.367111682891846, + "learning_rate": 0.00014398507144918656, + "loss": 3.3052, + "step": 2410 + }, + { + "epoch": 3.53, + "grad_norm": 6.1577558517456055, + "learning_rate": 0.00014394294399449996, + "loss": 3.2687, + "step": 2411 + }, + { + "epoch": 3.53, + "grad_norm": 14.60132122039795, + "learning_rate": 0.00014390080687266013, + "loss": 3.2542, + "step": 2412 + }, + { + "epoch": 3.54, + "grad_norm": 19.244531631469727, + "learning_rate": 0.000143858660092937, + "loss": 3.2623, + "step": 2413 + }, + { + "epoch": 3.54, + "grad_norm": 5.000961780548096, + "learning_rate": 0.0001438165036646026, + "loss": 3.2887, + "step": 2414 + }, + { + "epoch": 3.54, + "grad_norm": 1.7813776731491089, + "learning_rate": 0.000143774337596931, + "loss": 3.2522, + "step": 2415 + }, + { + "epoch": 3.54, + "grad_norm": 11.567626953125, + "learning_rate": 0.00014373216189919851, + "loss": 3.3463, + "step": 2416 + }, + { + "epoch": 3.54, + "grad_norm": 17.07219696044922, + "learning_rate": 0.0001436899765806835, + "loss": 3.3653, + "step": 2417 + }, + { + "epoch": 3.54, + "grad_norm": 10.967621803283691, + "learning_rate": 0.00014364778165066642, + "loss": 3.339, + "step": 2418 + }, + { + "epoch": 3.54, + "grad_norm": 5.372880935668945, + "learning_rate": 0.00014360557711842991, + "loss": 3.3025, + "step": 2419 + }, + { + "epoch": 3.55, + "grad_norm": 8.31612777709961, + "learning_rate": 0.00014356336299325864, + "loss": 3.2886, + "step": 2420 + }, + { + "epoch": 3.55, + "grad_norm": 6.364083290100098, + "learning_rate": 0.00014352113928443947, + "loss": 3.3376, + "step": 2421 + }, + { + "epoch": 3.55, + "grad_norm": 39.3469123840332, + "learning_rate": 0.00014347890600126134, + "loss": 3.2798, + "step": 2422 + }, + { + "epoch": 3.55, + "grad_norm": 15.578664779663086, + "learning_rate": 0.00014343666315301526, + "loss": 3.3638, + "step": 2423 + }, + { + "epoch": 3.55, + "grad_norm": 15.445240020751953, + "learning_rate": 0.0001433944107489944, + "loss": 3.3672, + "step": 2424 + }, + { + "epoch": 3.55, + "grad_norm": 13.184659957885742, + "learning_rate": 0.00014335214879849394, + "loss": 3.3172, + "step": 2425 + }, + { + "epoch": 3.56, + "grad_norm": 4.225596904754639, + "learning_rate": 0.0001433098773108113, + "loss": 3.2598, + "step": 2426 + }, + { + "epoch": 3.56, + "grad_norm": 9.756654739379883, + "learning_rate": 0.00014326759629524594, + "loss": 3.3621, + "step": 2427 + }, + { + "epoch": 3.56, + "grad_norm": 14.042159080505371, + "learning_rate": 0.00014322530576109933, + "loss": 3.3233, + "step": 2428 + }, + { + "epoch": 3.56, + "grad_norm": 14.112271308898926, + "learning_rate": 0.00014318300571767513, + "loss": 3.3309, + "step": 2429 + }, + { + "epoch": 3.56, + "grad_norm": 9.754487037658691, + "learning_rate": 0.00014314069617427907, + "loss": 3.3182, + "step": 2430 + }, + { + "epoch": 3.56, + "grad_norm": 2.0633256435394287, + "learning_rate": 0.00014309837714021902, + "loss": 3.2493, + "step": 2431 + }, + { + "epoch": 3.56, + "grad_norm": 8.613134384155273, + "learning_rate": 0.00014305604862480478, + "loss": 3.3103, + "step": 2432 + }, + { + "epoch": 3.57, + "grad_norm": 9.88028335571289, + "learning_rate": 0.0001430137106373484, + "loss": 3.3556, + "step": 2433 + }, + { + "epoch": 3.57, + "grad_norm": 8.193672180175781, + "learning_rate": 0.00014297136318716395, + "loss": 3.3321, + "step": 2434 + }, + { + "epoch": 3.57, + "grad_norm": 4.98108434677124, + "learning_rate": 0.00014292900628356758, + "loss": 3.265, + "step": 2435 + }, + { + "epoch": 3.57, + "grad_norm": 5.972259521484375, + "learning_rate": 0.0001428866399358775, + "loss": 3.2759, + "step": 2436 + }, + { + "epoch": 3.57, + "grad_norm": 4.662635803222656, + "learning_rate": 0.00014284426415341408, + "loss": 3.3012, + "step": 2437 + }, + { + "epoch": 3.57, + "grad_norm": 4.476476192474365, + "learning_rate": 0.00014280187894549968, + "loss": 3.2919, + "step": 2438 + }, + { + "epoch": 3.57, + "grad_norm": 3.099634885787964, + "learning_rate": 0.00014275948432145876, + "loss": 3.3264, + "step": 2439 + }, + { + "epoch": 3.58, + "grad_norm": 6.200466632843018, + "learning_rate": 0.00014271708029061785, + "loss": 3.2982, + "step": 2440 + }, + { + "epoch": 3.58, + "grad_norm": 3.2733237743377686, + "learning_rate": 0.00014267466686230553, + "loss": 3.2932, + "step": 2441 + }, + { + "epoch": 3.58, + "grad_norm": 3.1728217601776123, + "learning_rate": 0.0001426322440458525, + "loss": 3.2532, + "step": 2442 + }, + { + "epoch": 3.58, + "grad_norm": 3.4622151851654053, + "learning_rate": 0.0001425898118505915, + "loss": 3.2794, + "step": 2443 + }, + { + "epoch": 3.58, + "grad_norm": 7.931596755981445, + "learning_rate": 0.0001425473702858573, + "loss": 3.2692, + "step": 2444 + }, + { + "epoch": 3.58, + "grad_norm": 3.586261749267578, + "learning_rate": 0.00014250491936098678, + "loss": 3.2432, + "step": 2445 + }, + { + "epoch": 3.58, + "grad_norm": 3.096931219100952, + "learning_rate": 0.00014246245908531882, + "loss": 3.2589, + "step": 2446 + }, + { + "epoch": 3.59, + "grad_norm": 3.957979202270508, + "learning_rate": 0.0001424199894681945, + "loss": 3.278, + "step": 2447 + }, + { + "epoch": 3.59, + "grad_norm": 0.8693408370018005, + "learning_rate": 0.00014237751051895676, + "loss": 3.2366, + "step": 2448 + }, + { + "epoch": 3.59, + "grad_norm": 2.3588359355926514, + "learning_rate": 0.0001423350222469507, + "loss": 3.2469, + "step": 2449 + }, + { + "epoch": 3.59, + "grad_norm": 1.2426791191101074, + "learning_rate": 0.00014229252466152345, + "loss": 3.2929, + "step": 2450 + }, + { + "epoch": 3.59, + "grad_norm": 2.7103374004364014, + "learning_rate": 0.00014225001777202424, + "loss": 3.248, + "step": 2451 + }, + { + "epoch": 3.59, + "grad_norm": 2.5343198776245117, + "learning_rate": 0.00014220750158780424, + "loss": 3.257, + "step": 2452 + }, + { + "epoch": 3.6, + "grad_norm": 1.3891563415527344, + "learning_rate": 0.00014216497611821676, + "loss": 3.2348, + "step": 2453 + }, + { + "epoch": 3.6, + "grad_norm": 3.7499053478240967, + "learning_rate": 0.0001421224413726171, + "loss": 3.2513, + "step": 2454 + }, + { + "epoch": 3.6, + "grad_norm": 1.5479190349578857, + "learning_rate": 0.00014207989736036267, + "loss": 3.2785, + "step": 2455 + }, + { + "epoch": 3.6, + "grad_norm": 2.783564567565918, + "learning_rate": 0.0001420373440908128, + "loss": 3.292, + "step": 2456 + }, + { + "epoch": 3.6, + "grad_norm": 3.4745991230010986, + "learning_rate": 0.00014199478157332895, + "loss": 3.2399, + "step": 2457 + }, + { + "epoch": 3.6, + "grad_norm": 2.5730254650115967, + "learning_rate": 0.0001419522098172746, + "loss": 3.257, + "step": 2458 + }, + { + "epoch": 3.6, + "grad_norm": 3.455461025238037, + "learning_rate": 0.00014190962883201523, + "loss": 3.225, + "step": 2459 + }, + { + "epoch": 3.61, + "grad_norm": 4.234341621398926, + "learning_rate": 0.0001418670386269184, + "loss": 3.2997, + "step": 2460 + }, + { + "epoch": 3.61, + "grad_norm": 9.0831937789917, + "learning_rate": 0.00014182443921135362, + "loss": 3.221, + "step": 2461 + }, + { + "epoch": 3.61, + "grad_norm": 3.933647871017456, + "learning_rate": 0.00014178183059469253, + "loss": 3.2355, + "step": 2462 + }, + { + "epoch": 3.61, + "grad_norm": 1.695810079574585, + "learning_rate": 0.00014173921278630872, + "loss": 3.2724, + "step": 2463 + }, + { + "epoch": 3.61, + "grad_norm": 2.3507513999938965, + "learning_rate": 0.0001416965857955778, + "loss": 3.273, + "step": 2464 + }, + { + "epoch": 3.61, + "grad_norm": 2.5299062728881836, + "learning_rate": 0.00014165394963187747, + "loss": 3.2366, + "step": 2465 + }, + { + "epoch": 3.61, + "grad_norm": 0.9136064052581787, + "learning_rate": 0.00014161130430458737, + "loss": 3.2318, + "step": 2466 + }, + { + "epoch": 3.62, + "grad_norm": 2.159097671508789, + "learning_rate": 0.00014156864982308917, + "loss": 3.2139, + "step": 2467 + }, + { + "epoch": 3.62, + "grad_norm": 2.3743743896484375, + "learning_rate": 0.0001415259861967666, + "loss": 3.2442, + "step": 2468 + }, + { + "epoch": 3.62, + "grad_norm": 1.7210527658462524, + "learning_rate": 0.00014148331343500538, + "loss": 3.2565, + "step": 2469 + }, + { + "epoch": 3.62, + "grad_norm": 4.262208938598633, + "learning_rate": 0.0001414406315471932, + "loss": 3.2542, + "step": 2470 + }, + { + "epoch": 3.62, + "grad_norm": 3.7595646381378174, + "learning_rate": 0.00014139794054271977, + "loss": 3.2048, + "step": 2471 + }, + { + "epoch": 3.62, + "grad_norm": 3.6839802265167236, + "learning_rate": 0.00014135524043097693, + "loss": 3.2873, + "step": 2472 + }, + { + "epoch": 3.63, + "grad_norm": 3.4619247913360596, + "learning_rate": 0.0001413125312213583, + "loss": 3.2562, + "step": 2473 + }, + { + "epoch": 3.63, + "grad_norm": 3.164567708969116, + "learning_rate": 0.00014126981292325965, + "loss": 3.2521, + "step": 2474 + }, + { + "epoch": 3.63, + "grad_norm": 3.066319227218628, + "learning_rate": 0.00014122708554607875, + "loss": 3.2671, + "step": 2475 + }, + { + "epoch": 3.63, + "grad_norm": 2.757352590560913, + "learning_rate": 0.00014118434909921534, + "loss": 3.2061, + "step": 2476 + }, + { + "epoch": 3.63, + "grad_norm": 1.9947091341018677, + "learning_rate": 0.0001411416035920711, + "loss": 3.2391, + "step": 2477 + }, + { + "epoch": 3.63, + "grad_norm": 4.6580986976623535, + "learning_rate": 0.0001410988490340498, + "loss": 3.252, + "step": 2478 + }, + { + "epoch": 3.63, + "grad_norm": 4.087027072906494, + "learning_rate": 0.00014105608543455716, + "loss": 3.2371, + "step": 2479 + }, + { + "epoch": 3.64, + "grad_norm": 3.322829008102417, + "learning_rate": 0.0001410133128030009, + "loss": 3.2261, + "step": 2480 + }, + { + "epoch": 3.64, + "grad_norm": 3.1107428073883057, + "learning_rate": 0.00014097053114879061, + "loss": 3.2418, + "step": 2481 + }, + { + "epoch": 3.64, + "grad_norm": 2.6654446125030518, + "learning_rate": 0.00014092774048133809, + "loss": 3.232, + "step": 2482 + }, + { + "epoch": 3.64, + "grad_norm": 1.8028963804244995, + "learning_rate": 0.00014088494081005695, + "loss": 3.2122, + "step": 2483 + }, + { + "epoch": 3.64, + "grad_norm": 9.199921607971191, + "learning_rate": 0.00014084213214436277, + "loss": 3.2409, + "step": 2484 + }, + { + "epoch": 3.64, + "grad_norm": 7.354410648345947, + "learning_rate": 0.00014079931449367328, + "loss": 3.2151, + "step": 2485 + }, + { + "epoch": 3.64, + "grad_norm": 2.6317131519317627, + "learning_rate": 0.00014075648786740806, + "loss": 3.2449, + "step": 2486 + }, + { + "epoch": 3.65, + "grad_norm": 6.4712042808532715, + "learning_rate": 0.00014071365227498857, + "loss": 3.2384, + "step": 2487 + }, + { + "epoch": 3.65, + "grad_norm": 7.566592216491699, + "learning_rate": 0.00014067080772583846, + "loss": 3.2208, + "step": 2488 + }, + { + "epoch": 3.65, + "grad_norm": 3.7115042209625244, + "learning_rate": 0.00014062795422938325, + "loss": 3.2403, + "step": 2489 + }, + { + "epoch": 3.65, + "grad_norm": 7.227043151855469, + "learning_rate": 0.00014058509179505036, + "loss": 3.2533, + "step": 2490 + }, + { + "epoch": 3.65, + "grad_norm": 10.836502075195312, + "learning_rate": 0.00014054222043226926, + "loss": 3.3034, + "step": 2491 + }, + { + "epoch": 3.65, + "grad_norm": 8.85554313659668, + "learning_rate": 0.00014049934015047136, + "loss": 3.2838, + "step": 2492 + }, + { + "epoch": 3.65, + "grad_norm": 7.335647106170654, + "learning_rate": 0.00014045645095909004, + "loss": 3.2506, + "step": 2493 + }, + { + "epoch": 3.66, + "grad_norm": 6.849621295928955, + "learning_rate": 0.00014041355286756066, + "loss": 3.266, + "step": 2494 + }, + { + "epoch": 3.66, + "grad_norm": 5.7214837074279785, + "learning_rate": 0.00014037064588532047, + "loss": 3.229, + "step": 2495 + }, + { + "epoch": 3.66, + "grad_norm": 1.1197165250778198, + "learning_rate": 0.00014032773002180875, + "loss": 3.189, + "step": 2496 + }, + { + "epoch": 3.66, + "grad_norm": 2.7683522701263428, + "learning_rate": 0.00014028480528646667, + "loss": 3.2357, + "step": 2497 + }, + { + "epoch": 3.66, + "grad_norm": 10.467253684997559, + "learning_rate": 0.0001402418716887374, + "loss": 3.2702, + "step": 2498 + }, + { + "epoch": 3.66, + "grad_norm": 5.283757209777832, + "learning_rate": 0.0001401989292380661, + "loss": 3.2404, + "step": 2499 + }, + { + "epoch": 3.67, + "grad_norm": 3.8525490760803223, + "learning_rate": 0.00014015597794389973, + "loss": 3.2352, + "step": 2500 + }, + { + "epoch": 3.67, + "grad_norm": 4.830957412719727, + "learning_rate": 0.0001401130178156873, + "loss": 3.2432, + "step": 2501 + }, + { + "epoch": 3.67, + "grad_norm": 28.55122184753418, + "learning_rate": 0.0001400700488628798, + "loss": 3.2176, + "step": 2502 + }, + { + "epoch": 3.67, + "grad_norm": 6.477649211883545, + "learning_rate": 0.0001400270710949301, + "loss": 3.1952, + "step": 2503 + }, + { + "epoch": 3.67, + "grad_norm": 4.75806188583374, + "learning_rate": 0.00013998408452129302, + "loss": 3.2384, + "step": 2504 + }, + { + "epoch": 3.67, + "grad_norm": 5.451016426086426, + "learning_rate": 0.00013994108915142526, + "loss": 3.2404, + "step": 2505 + }, + { + "epoch": 3.67, + "grad_norm": 2.510890245437622, + "learning_rate": 0.0001398980849947856, + "loss": 3.221, + "step": 2506 + }, + { + "epoch": 3.68, + "grad_norm": 6.08433198928833, + "learning_rate": 0.0001398550720608346, + "loss": 3.2386, + "step": 2507 + }, + { + "epoch": 3.68, + "grad_norm": 7.88702392578125, + "learning_rate": 0.00013981205035903485, + "loss": 3.2582, + "step": 2508 + }, + { + "epoch": 3.68, + "grad_norm": 3.3581769466400146, + "learning_rate": 0.00013976901989885086, + "loss": 3.2437, + "step": 2509 + }, + { + "epoch": 3.68, + "grad_norm": 5.539381980895996, + "learning_rate": 0.000139725980689749, + "loss": 3.2349, + "step": 2510 + }, + { + "epoch": 3.68, + "grad_norm": 6.059652328491211, + "learning_rate": 0.0001396829327411976, + "loss": 3.3016, + "step": 2511 + }, + { + "epoch": 3.68, + "grad_norm": 1.3593266010284424, + "learning_rate": 0.00013963987606266693, + "loss": 3.2587, + "step": 2512 + }, + { + "epoch": 3.68, + "grad_norm": 9.557854652404785, + "learning_rate": 0.0001395968106636292, + "loss": 3.2695, + "step": 2513 + }, + { + "epoch": 3.69, + "grad_norm": 26.98328971862793, + "learning_rate": 0.0001395537365535585, + "loss": 3.3668, + "step": 2514 + }, + { + "epoch": 3.69, + "grad_norm": 14.079690933227539, + "learning_rate": 0.0001395106537419308, + "loss": 3.2736, + "step": 2515 + }, + { + "epoch": 3.69, + "grad_norm": 5.151083946228027, + "learning_rate": 0.00013946756223822413, + "loss": 3.2461, + "step": 2516 + }, + { + "epoch": 3.69, + "grad_norm": 7.186163902282715, + "learning_rate": 0.00013942446205191824, + "loss": 3.2817, + "step": 2517 + }, + { + "epoch": 3.69, + "grad_norm": 9.277379035949707, + "learning_rate": 0.0001393813531924949, + "loss": 3.2008, + "step": 2518 + }, + { + "epoch": 3.69, + "grad_norm": 19.572731018066406, + "learning_rate": 0.00013933823566943777, + "loss": 3.2131, + "step": 2519 + }, + { + "epoch": 3.69, + "grad_norm": 4.6605329513549805, + "learning_rate": 0.00013929510949223245, + "loss": 3.2512, + "step": 2520 + }, + { + "epoch": 3.7, + "grad_norm": 151.2240753173828, + "learning_rate": 0.00013925197467036635, + "loss": 3.2621, + "step": 2521 + }, + { + "epoch": 3.7, + "grad_norm": 13.225665092468262, + "learning_rate": 0.0001392088312133289, + "loss": 3.2814, + "step": 2522 + }, + { + "epoch": 3.7, + "grad_norm": 14.404492378234863, + "learning_rate": 0.00013916567913061133, + "loss": 3.2903, + "step": 2523 + }, + { + "epoch": 3.7, + "grad_norm": 10.506006240844727, + "learning_rate": 0.00013912251843170687, + "loss": 3.226, + "step": 2524 + }, + { + "epoch": 3.7, + "grad_norm": 3.039416790008545, + "learning_rate": 0.0001390793491261105, + "loss": 3.26, + "step": 2525 + }, + { + "epoch": 3.7, + "grad_norm": 6.364565372467041, + "learning_rate": 0.00013903617122331923, + "loss": 3.1935, + "step": 2526 + }, + { + "epoch": 3.71, + "grad_norm": 3.801405668258667, + "learning_rate": 0.0001389929847328319, + "loss": 3.2204, + "step": 2527 + }, + { + "epoch": 3.71, + "grad_norm": 13.04414176940918, + "learning_rate": 0.00013894978966414926, + "loss": 3.2408, + "step": 2528 + }, + { + "epoch": 3.71, + "grad_norm": 14.216852188110352, + "learning_rate": 0.0001389065860267739, + "loss": 3.2739, + "step": 2529 + }, + { + "epoch": 3.71, + "grad_norm": 10.664838790893555, + "learning_rate": 0.0001388633738302104, + "loss": 3.2684, + "step": 2530 + }, + { + "epoch": 3.71, + "grad_norm": 6.671474456787109, + "learning_rate": 0.00013882015308396508, + "loss": 3.3327, + "step": 2531 + }, + { + "epoch": 3.71, + "grad_norm": 15.43130874633789, + "learning_rate": 0.00013877692379754626, + "loss": 3.2355, + "step": 2532 + }, + { + "epoch": 3.71, + "grad_norm": 44.657169342041016, + "learning_rate": 0.00013873368598046404, + "loss": 3.2309, + "step": 2533 + }, + { + "epoch": 3.72, + "grad_norm": 18.107580184936523, + "learning_rate": 0.00013869043964223051, + "loss": 3.2906, + "step": 2534 + }, + { + "epoch": 3.72, + "grad_norm": 11.789257049560547, + "learning_rate": 0.00013864718479235954, + "loss": 3.3045, + "step": 2535 + }, + { + "epoch": 3.72, + "eval_loss": 3.0709898471832275, + "eval_runtime": 1163.6634, + "eval_samples_per_second": 14.731, + "eval_steps_per_second": 0.921, + "step": 2535 + }, + { + "epoch": 3.72, + "grad_norm": 7.7072319984436035, + "learning_rate": 0.00013860392144036693, + "loss": 3.2131, + "step": 2536 + }, + { + "epoch": 3.72, + "grad_norm": 4.083733081817627, + "learning_rate": 0.0001385606495957703, + "loss": 3.2641, + "step": 2537 + }, + { + "epoch": 3.72, + "grad_norm": 11.785395622253418, + "learning_rate": 0.00013851736926808917, + "loss": 3.2106, + "step": 2538 + }, + { + "epoch": 3.72, + "grad_norm": 13.176704406738281, + "learning_rate": 0.00013847408046684497, + "loss": 3.2857, + "step": 2539 + }, + { + "epoch": 3.72, + "grad_norm": 5.579313278198242, + "learning_rate": 0.00013843078320156088, + "loss": 3.2861, + "step": 2540 + }, + { + "epoch": 3.73, + "grad_norm": 4.398528099060059, + "learning_rate": 0.0001383874774817621, + "loss": 3.225, + "step": 2541 + }, + { + "epoch": 3.73, + "grad_norm": 6.542925834655762, + "learning_rate": 0.00013834416331697544, + "loss": 3.204, + "step": 2542 + }, + { + "epoch": 3.73, + "grad_norm": 9.607292175292969, + "learning_rate": 0.00013830084071672985, + "loss": 3.2284, + "step": 2543 + }, + { + "epoch": 3.73, + "grad_norm": 2.1741347312927246, + "learning_rate": 0.00013825750969055604, + "loss": 3.2318, + "step": 2544 + }, + { + "epoch": 3.73, + "grad_norm": 2.4551427364349365, + "learning_rate": 0.00013821417024798642, + "loss": 3.2454, + "step": 2545 + }, + { + "epoch": 3.73, + "grad_norm": 1.8988637924194336, + "learning_rate": 0.0001381708223985555, + "loss": 3.2688, + "step": 2546 + }, + { + "epoch": 3.73, + "grad_norm": 1.2011829614639282, + "learning_rate": 0.00013812746615179944, + "loss": 3.2418, + "step": 2547 + }, + { + "epoch": 3.74, + "grad_norm": 2.4243390560150146, + "learning_rate": 0.0001380841015172563, + "loss": 3.2759, + "step": 2548 + }, + { + "epoch": 3.74, + "grad_norm": 0.9519268274307251, + "learning_rate": 0.00013804072850446612, + "loss": 3.2498, + "step": 2549 + }, + { + "epoch": 3.74, + "grad_norm": 5.196446418762207, + "learning_rate": 0.00013799734712297059, + "loss": 3.2358, + "step": 2550 + }, + { + "epoch": 3.74, + "grad_norm": 3.189202308654785, + "learning_rate": 0.00013795395738231333, + "loss": 3.2582, + "step": 2551 + }, + { + "epoch": 3.74, + "grad_norm": 5.616455078125, + "learning_rate": 0.0001379105592920398, + "loss": 3.2066, + "step": 2552 + }, + { + "epoch": 3.74, + "grad_norm": 9.704096794128418, + "learning_rate": 0.0001378671528616973, + "loss": 3.2245, + "step": 2553 + }, + { + "epoch": 3.75, + "grad_norm": 3.5445096492767334, + "learning_rate": 0.00013782373810083493, + "loss": 3.2097, + "step": 2554 + }, + { + "epoch": 3.75, + "grad_norm": 3.731013774871826, + "learning_rate": 0.00013778031501900367, + "loss": 3.2075, + "step": 2555 + }, + { + "epoch": 3.75, + "grad_norm": 8.907066345214844, + "learning_rate": 0.0001377368836257563, + "loss": 3.2249, + "step": 2556 + }, + { + "epoch": 3.75, + "grad_norm": 4.645308017730713, + "learning_rate": 0.00013769344393064743, + "loss": 3.2139, + "step": 2557 + }, + { + "epoch": 3.75, + "grad_norm": 4.715915679931641, + "learning_rate": 0.00013764999594323353, + "loss": 3.2613, + "step": 2558 + }, + { + "epoch": 3.75, + "grad_norm": 5.21357536315918, + "learning_rate": 0.0001376065396730728, + "loss": 3.2247, + "step": 2559 + }, + { + "epoch": 3.75, + "grad_norm": 1.811521053314209, + "learning_rate": 0.00013756307512972538, + "loss": 3.2714, + "step": 2560 + }, + { + "epoch": 3.76, + "grad_norm": 5.335707664489746, + "learning_rate": 0.0001375196023227532, + "loss": 3.2079, + "step": 2561 + }, + { + "epoch": 3.76, + "grad_norm": 3.844973087310791, + "learning_rate": 0.0001374761212617199, + "loss": 3.2411, + "step": 2562 + }, + { + "epoch": 3.76, + "grad_norm": 4.505429267883301, + "learning_rate": 0.00013743263195619115, + "loss": 3.2508, + "step": 2563 + }, + { + "epoch": 3.76, + "grad_norm": 5.144504547119141, + "learning_rate": 0.00013738913441573418, + "loss": 3.2112, + "step": 2564 + }, + { + "epoch": 3.76, + "grad_norm": 23.4456787109375, + "learning_rate": 0.00013734562864991823, + "loss": 3.2106, + "step": 2565 + }, + { + "epoch": 3.76, + "grad_norm": 4.499874591827393, + "learning_rate": 0.0001373021146683142, + "loss": 3.2324, + "step": 2566 + }, + { + "epoch": 3.76, + "grad_norm": 2.9865691661834717, + "learning_rate": 0.000137258592480495, + "loss": 3.2498, + "step": 2567 + }, + { + "epoch": 3.77, + "grad_norm": 3.308457374572754, + "learning_rate": 0.00013721506209603512, + "loss": 3.2132, + "step": 2568 + }, + { + "epoch": 3.77, + "grad_norm": 2.9479405879974365, + "learning_rate": 0.000137171523524511, + "loss": 3.2817, + "step": 2569 + }, + { + "epoch": 3.77, + "grad_norm": 2.7257778644561768, + "learning_rate": 0.0001371279767755008, + "loss": 3.185, + "step": 2570 + }, + { + "epoch": 3.77, + "grad_norm": 1.7737255096435547, + "learning_rate": 0.00013708442185858454, + "loss": 3.222, + "step": 2571 + }, + { + "epoch": 3.77, + "grad_norm": 3.827313184738159, + "learning_rate": 0.000137040858783344, + "loss": 3.1943, + "step": 2572 + }, + { + "epoch": 3.77, + "grad_norm": 2.6353366374969482, + "learning_rate": 0.00013699728755936277, + "loss": 3.2699, + "step": 2573 + }, + { + "epoch": 3.78, + "grad_norm": 4.652822017669678, + "learning_rate": 0.0001369537081962262, + "loss": 3.2313, + "step": 2574 + }, + { + "epoch": 3.78, + "grad_norm": 8.656730651855469, + "learning_rate": 0.0001369101207035215, + "loss": 3.1737, + "step": 2575 + }, + { + "epoch": 3.78, + "grad_norm": 1.588462471961975, + "learning_rate": 0.0001368665250908376, + "loss": 3.2057, + "step": 2576 + }, + { + "epoch": 3.78, + "grad_norm": 5.477829456329346, + "learning_rate": 0.00013682292136776524, + "loss": 3.2045, + "step": 2577 + }, + { + "epoch": 3.78, + "grad_norm": 6.158089637756348, + "learning_rate": 0.00013677930954389698, + "loss": 3.2183, + "step": 2578 + }, + { + "epoch": 3.78, + "grad_norm": 2.4769415855407715, + "learning_rate": 0.00013673568962882705, + "loss": 3.1601, + "step": 2579 + }, + { + "epoch": 3.78, + "grad_norm": 7.979514122009277, + "learning_rate": 0.00013669206163215164, + "loss": 3.2011, + "step": 2580 + }, + { + "epoch": 3.79, + "grad_norm": 11.770708084106445, + "learning_rate": 0.00013664842556346856, + "loss": 3.2215, + "step": 2581 + }, + { + "epoch": 3.79, + "grad_norm": 7.018968105316162, + "learning_rate": 0.00013660478143237746, + "loss": 3.2163, + "step": 2582 + }, + { + "epoch": 3.79, + "grad_norm": 4.448172569274902, + "learning_rate": 0.00013656112924847978, + "loss": 3.2027, + "step": 2583 + }, + { + "epoch": 3.79, + "grad_norm": 6.86191987991333, + "learning_rate": 0.00013651746902137868, + "loss": 3.1812, + "step": 2584 + }, + { + "epoch": 3.79, + "grad_norm": 4.056680202484131, + "learning_rate": 0.0001364738007606791, + "loss": 3.2857, + "step": 2585 + }, + { + "epoch": 3.79, + "grad_norm": 5.344080924987793, + "learning_rate": 0.0001364301244759878, + "loss": 3.255, + "step": 2586 + }, + { + "epoch": 3.79, + "grad_norm": 7.4425554275512695, + "learning_rate": 0.00013638644017691326, + "loss": 3.2032, + "step": 2587 + }, + { + "epoch": 3.8, + "grad_norm": 4.709530830383301, + "learning_rate": 0.00013634274787306577, + "loss": 3.25, + "step": 2588 + }, + { + "epoch": 3.8, + "grad_norm": 7.912161827087402, + "learning_rate": 0.00013629904757405723, + "loss": 3.2362, + "step": 2589 + }, + { + "epoch": 3.8, + "grad_norm": 24.812524795532227, + "learning_rate": 0.00013625533928950153, + "loss": 3.231, + "step": 2590 + }, + { + "epoch": 3.8, + "grad_norm": 9.660811424255371, + "learning_rate": 0.0001362116230290142, + "loss": 3.2175, + "step": 2591 + }, + { + "epoch": 3.8, + "grad_norm": 6.260350704193115, + "learning_rate": 0.00013616789880221242, + "loss": 3.2169, + "step": 2592 + }, + { + "epoch": 3.8, + "grad_norm": 2.6841564178466797, + "learning_rate": 0.00013612416661871533, + "loss": 3.2084, + "step": 2593 + }, + { + "epoch": 3.8, + "grad_norm": 6.153440475463867, + "learning_rate": 0.00013608042648814364, + "loss": 3.2029, + "step": 2594 + }, + { + "epoch": 3.81, + "grad_norm": 2.9339137077331543, + "learning_rate": 0.00013603667842011994, + "loss": 3.1789, + "step": 2595 + }, + { + "epoch": 3.81, + "grad_norm": 5.685370922088623, + "learning_rate": 0.00013599292242426851, + "loss": 3.2023, + "step": 2596 + }, + { + "epoch": 3.81, + "grad_norm": 12.19677734375, + "learning_rate": 0.00013594915851021532, + "loss": 3.2433, + "step": 2597 + }, + { + "epoch": 3.81, + "grad_norm": 5.506028652191162, + "learning_rate": 0.00013590538668758822, + "loss": 3.2088, + "step": 2598 + }, + { + "epoch": 3.81, + "grad_norm": 1.9152709245681763, + "learning_rate": 0.00013586160696601665, + "loss": 3.2067, + "step": 2599 + }, + { + "epoch": 3.81, + "grad_norm": 3.531026601791382, + "learning_rate": 0.00013581781935513188, + "loss": 3.2368, + "step": 2600 + }, + { + "epoch": 3.82, + "grad_norm": 1.68514883518219, + "learning_rate": 0.00013577402386456692, + "loss": 3.2567, + "step": 2601 + }, + { + "epoch": 3.82, + "grad_norm": 2.6391441822052, + "learning_rate": 0.00013573022050395643, + "loss": 3.1804, + "step": 2602 + }, + { + "epoch": 3.82, + "grad_norm": 2.4722888469696045, + "learning_rate": 0.0001356864092829369, + "loss": 3.2318, + "step": 2603 + }, + { + "epoch": 3.82, + "grad_norm": 1.5304458141326904, + "learning_rate": 0.0001356425902111465, + "loss": 3.2145, + "step": 2604 + }, + { + "epoch": 3.82, + "grad_norm": 4.6478705406188965, + "learning_rate": 0.0001355987632982251, + "loss": 3.2201, + "step": 2605 + }, + { + "epoch": 3.82, + "grad_norm": 3.80239200592041, + "learning_rate": 0.00013555492855381432, + "loss": 3.1819, + "step": 2606 + }, + { + "epoch": 3.82, + "grad_norm": 2.0871238708496094, + "learning_rate": 0.0001355110859875576, + "loss": 3.2169, + "step": 2607 + }, + { + "epoch": 3.83, + "grad_norm": 1.9589481353759766, + "learning_rate": 0.00013546723560909992, + "loss": 3.2077, + "step": 2608 + }, + { + "epoch": 3.83, + "grad_norm": 2.8086769580841064, + "learning_rate": 0.00013542337742808807, + "loss": 3.2386, + "step": 2609 + }, + { + "epoch": 3.83, + "grad_norm": 1.3195122480392456, + "learning_rate": 0.0001353795114541706, + "loss": 3.1651, + "step": 2610 + }, + { + "epoch": 3.83, + "grad_norm": 4.268999099731445, + "learning_rate": 0.00013533563769699774, + "loss": 3.2113, + "step": 2611 + }, + { + "epoch": 3.83, + "grad_norm": 2.7233781814575195, + "learning_rate": 0.00013529175616622133, + "loss": 3.2103, + "step": 2612 + }, + { + "epoch": 3.83, + "grad_norm": 3.8539392948150635, + "learning_rate": 0.00013524786687149512, + "loss": 3.2368, + "step": 2613 + }, + { + "epoch": 3.83, + "grad_norm": 3.89041805267334, + "learning_rate": 0.00013520396982247442, + "loss": 3.1865, + "step": 2614 + }, + { + "epoch": 3.84, + "grad_norm": 2.546916961669922, + "learning_rate": 0.0001351600650288163, + "loss": 3.192, + "step": 2615 + }, + { + "epoch": 3.84, + "grad_norm": 1.615480661392212, + "learning_rate": 0.0001351161525001795, + "loss": 3.2953, + "step": 2616 + }, + { + "epoch": 3.84, + "grad_norm": 3.4749913215637207, + "learning_rate": 0.00013507223224622445, + "loss": 3.1994, + "step": 2617 + }, + { + "epoch": 3.84, + "grad_norm": 1.6761186122894287, + "learning_rate": 0.00013502830427661338, + "loss": 3.1836, + "step": 2618 + }, + { + "epoch": 3.84, + "grad_norm": 6.091635704040527, + "learning_rate": 0.00013498436860101011, + "loss": 3.2183, + "step": 2619 + }, + { + "epoch": 3.84, + "grad_norm": 5.926030158996582, + "learning_rate": 0.00013494042522908022, + "loss": 3.2222, + "step": 2620 + }, + { + "epoch": 3.84, + "grad_norm": 12.00096321105957, + "learning_rate": 0.00013489647417049095, + "loss": 3.2073, + "step": 2621 + }, + { + "epoch": 3.85, + "grad_norm": 4.7124762535095215, + "learning_rate": 0.0001348525154349112, + "loss": 3.1859, + "step": 2622 + }, + { + "epoch": 3.85, + "grad_norm": 3.3024585247039795, + "learning_rate": 0.0001348085490320116, + "loss": 3.177, + "step": 2623 + }, + { + "epoch": 3.85, + "grad_norm": 4.006857395172119, + "learning_rate": 0.00013476457497146456, + "loss": 3.2317, + "step": 2624 + }, + { + "epoch": 3.85, + "grad_norm": 4.812809467315674, + "learning_rate": 0.00013472059326294396, + "loss": 3.2637, + "step": 2625 + }, + { + "epoch": 3.85, + "grad_norm": 1.2310017347335815, + "learning_rate": 0.00013467660391612555, + "loss": 3.2366, + "step": 2626 + }, + { + "epoch": 3.85, + "grad_norm": 2.896761894226074, + "learning_rate": 0.00013463260694068666, + "loss": 3.2069, + "step": 2627 + }, + { + "epoch": 3.86, + "grad_norm": 2.2707176208496094, + "learning_rate": 0.00013458860234630632, + "loss": 3.224, + "step": 2628 + }, + { + "epoch": 3.86, + "grad_norm": 2.2991745471954346, + "learning_rate": 0.00013454459014266526, + "loss": 3.2074, + "step": 2629 + }, + { + "epoch": 3.86, + "grad_norm": 2.838233470916748, + "learning_rate": 0.0001345005703394459, + "loss": 3.2019, + "step": 2630 + }, + { + "epoch": 3.86, + "grad_norm": 2.686645030975342, + "learning_rate": 0.0001344565429463323, + "loss": 3.1873, + "step": 2631 + }, + { + "epoch": 3.86, + "grad_norm": 3.610236883163452, + "learning_rate": 0.00013441250797301016, + "loss": 3.1667, + "step": 2632 + }, + { + "epoch": 3.86, + "grad_norm": 3.012873888015747, + "learning_rate": 0.00013436846542916686, + "loss": 3.2006, + "step": 2633 + }, + { + "epoch": 3.86, + "grad_norm": 2.223695755004883, + "learning_rate": 0.00013432441532449152, + "loss": 3.2004, + "step": 2634 + }, + { + "epoch": 3.87, + "grad_norm": 5.295112609863281, + "learning_rate": 0.00013428035766867484, + "loss": 3.1959, + "step": 2635 + }, + { + "epoch": 3.87, + "grad_norm": 1.9531642198562622, + "learning_rate": 0.00013423629247140919, + "loss": 3.1796, + "step": 2636 + }, + { + "epoch": 3.87, + "grad_norm": 11.508848190307617, + "learning_rate": 0.00013419221974238864, + "loss": 3.2325, + "step": 2637 + }, + { + "epoch": 3.87, + "grad_norm": 3.0934455394744873, + "learning_rate": 0.00013414813949130894, + "loss": 3.1904, + "step": 2638 + }, + { + "epoch": 3.87, + "grad_norm": 1.843153476715088, + "learning_rate": 0.00013410405172786734, + "loss": 3.177, + "step": 2639 + }, + { + "epoch": 3.87, + "grad_norm": 3.7216010093688965, + "learning_rate": 0.00013405995646176294, + "loss": 3.2278, + "step": 2640 + }, + { + "epoch": 3.87, + "grad_norm": 4.631261825561523, + "learning_rate": 0.00013401585370269637, + "loss": 3.2772, + "step": 2641 + }, + { + "epoch": 3.88, + "grad_norm": 2.928041696548462, + "learning_rate": 0.00013397174346036996, + "loss": 3.2489, + "step": 2642 + }, + { + "epoch": 3.88, + "grad_norm": 2.3705878257751465, + "learning_rate": 0.00013392762574448766, + "loss": 3.2577, + "step": 2643 + }, + { + "epoch": 3.88, + "grad_norm": 4.091426372528076, + "learning_rate": 0.00013388350056475504, + "loss": 3.1664, + "step": 2644 + }, + { + "epoch": 3.88, + "grad_norm": 1.9827549457550049, + "learning_rate": 0.0001338393679308794, + "loss": 3.1786, + "step": 2645 + }, + { + "epoch": 3.88, + "grad_norm": 4.402117729187012, + "learning_rate": 0.0001337952278525696, + "loss": 3.1966, + "step": 2646 + }, + { + "epoch": 3.88, + "grad_norm": 5.21814489364624, + "learning_rate": 0.00013375108033953614, + "loss": 3.2393, + "step": 2647 + }, + { + "epoch": 3.88, + "grad_norm": 2.4417850971221924, + "learning_rate": 0.0001337069254014912, + "loss": 3.1729, + "step": 2648 + }, + { + "epoch": 3.89, + "grad_norm": 3.521346092224121, + "learning_rate": 0.0001336627630481486, + "loss": 3.1997, + "step": 2649 + }, + { + "epoch": 3.89, + "grad_norm": 4.180556297302246, + "learning_rate": 0.0001336185932892237, + "loss": 3.1878, + "step": 2650 + }, + { + "epoch": 3.89, + "grad_norm": 2.869370460510254, + "learning_rate": 0.00013357441613443354, + "loss": 3.2052, + "step": 2651 + }, + { + "epoch": 3.89, + "grad_norm": 2.097275495529175, + "learning_rate": 0.0001335302315934969, + "loss": 3.1459, + "step": 2652 + }, + { + "epoch": 3.89, + "grad_norm": 3.70109486579895, + "learning_rate": 0.00013348603967613398, + "loss": 3.2187, + "step": 2653 + }, + { + "epoch": 3.89, + "grad_norm": 3.026033878326416, + "learning_rate": 0.00013344184039206677, + "loss": 3.2251, + "step": 2654 + }, + { + "epoch": 3.9, + "grad_norm": 2.825608015060425, + "learning_rate": 0.00013339763375101876, + "loss": 3.1845, + "step": 2655 + }, + { + "epoch": 3.9, + "grad_norm": 2.3940205574035645, + "learning_rate": 0.00013335341976271518, + "loss": 3.1909, + "step": 2656 + }, + { + "epoch": 3.9, + "grad_norm": 3.781881332397461, + "learning_rate": 0.00013330919843688275, + "loss": 3.1861, + "step": 2657 + }, + { + "epoch": 3.9, + "grad_norm": 3.5430831909179688, + "learning_rate": 0.0001332649697832499, + "loss": 3.2879, + "step": 2658 + }, + { + "epoch": 3.9, + "grad_norm": 3.327752113342285, + "learning_rate": 0.00013322073381154663, + "loss": 3.1407, + "step": 2659 + }, + { + "epoch": 3.9, + "grad_norm": 3.319768190383911, + "learning_rate": 0.00013317649053150456, + "loss": 3.2215, + "step": 2660 + }, + { + "epoch": 3.9, + "grad_norm": 2.317915201187134, + "learning_rate": 0.0001331322399528569, + "loss": 3.251, + "step": 2661 + }, + { + "epoch": 3.91, + "grad_norm": 2.999483108520508, + "learning_rate": 0.00013308798208533853, + "loss": 3.2182, + "step": 2662 + }, + { + "epoch": 3.91, + "grad_norm": 4.7933549880981445, + "learning_rate": 0.0001330437169386858, + "loss": 3.2272, + "step": 2663 + }, + { + "epoch": 3.91, + "grad_norm": 4.440158367156982, + "learning_rate": 0.00013299944452263682, + "loss": 3.202, + "step": 2664 + }, + { + "epoch": 3.91, + "grad_norm": 1.8913938999176025, + "learning_rate": 0.0001329551648469312, + "loss": 3.1705, + "step": 2665 + }, + { + "epoch": 3.91, + "grad_norm": 4.039271831512451, + "learning_rate": 0.00013291087792131017, + "loss": 3.213, + "step": 2666 + }, + { + "epoch": 3.91, + "grad_norm": 4.9916276931762695, + "learning_rate": 0.00013286658375551654, + "loss": 3.2316, + "step": 2667 + }, + { + "epoch": 3.91, + "grad_norm": 2.624239206314087, + "learning_rate": 0.00013282228235929474, + "loss": 3.1902, + "step": 2668 + }, + { + "epoch": 3.92, + "grad_norm": 2.3713390827178955, + "learning_rate": 0.0001327779737423908, + "loss": 3.1698, + "step": 2669 + }, + { + "epoch": 3.92, + "grad_norm": 3.9648921489715576, + "learning_rate": 0.0001327336579145523, + "loss": 3.1811, + "step": 2670 + }, + { + "epoch": 3.92, + "grad_norm": 3.6182374954223633, + "learning_rate": 0.00013268933488552845, + "loss": 3.2013, + "step": 2671 + }, + { + "epoch": 3.92, + "grad_norm": 1.5024638175964355, + "learning_rate": 0.00013264500466506998, + "loss": 3.1996, + "step": 2672 + }, + { + "epoch": 3.92, + "grad_norm": 1.957371473312378, + "learning_rate": 0.00013260066726292926, + "loss": 3.188, + "step": 2673 + }, + { + "epoch": 3.92, + "grad_norm": 1.9777625799179077, + "learning_rate": 0.00013255632268886023, + "loss": 3.2656, + "step": 2674 + }, + { + "epoch": 3.93, + "grad_norm": 1.5048279762268066, + "learning_rate": 0.0001325119709526184, + "loss": 3.2279, + "step": 2675 + }, + { + "epoch": 3.93, + "grad_norm": 35.774295806884766, + "learning_rate": 0.0001324676120639608, + "loss": 3.2366, + "step": 2676 + }, + { + "epoch": 3.93, + "grad_norm": 3.9443695545196533, + "learning_rate": 0.00013242324603264616, + "loss": 3.1658, + "step": 2677 + }, + { + "epoch": 3.93, + "grad_norm": 2.9231414794921875, + "learning_rate": 0.0001323788728684347, + "loss": 3.2058, + "step": 2678 + }, + { + "epoch": 3.93, + "grad_norm": 4.9291157722473145, + "learning_rate": 0.0001323344925810882, + "loss": 3.1528, + "step": 2679 + }, + { + "epoch": 3.93, + "grad_norm": 8.999177932739258, + "learning_rate": 0.00013229010518037002, + "loss": 3.2864, + "step": 2680 + }, + { + "epoch": 3.93, + "grad_norm": 4.351472854614258, + "learning_rate": 0.0001322457106760451, + "loss": 3.2048, + "step": 2681 + }, + { + "epoch": 3.94, + "grad_norm": 9.449458122253418, + "learning_rate": 0.00013220130907787994, + "loss": 3.2409, + "step": 2682 + }, + { + "epoch": 3.94, + "grad_norm": 2.7245359420776367, + "learning_rate": 0.00013215690039564257, + "loss": 3.1844, + "step": 2683 + }, + { + "epoch": 3.94, + "grad_norm": 1.7337415218353271, + "learning_rate": 0.00013211248463910262, + "loss": 3.1732, + "step": 2684 + }, + { + "epoch": 3.94, + "grad_norm": 2.845367431640625, + "learning_rate": 0.0001320680618180313, + "loss": 3.1608, + "step": 2685 + }, + { + "epoch": 3.94, + "grad_norm": 2.7872724533081055, + "learning_rate": 0.00013202363194220125, + "loss": 3.2057, + "step": 2686 + }, + { + "epoch": 3.94, + "grad_norm": 1.4545655250549316, + "learning_rate": 0.00013197919502138677, + "loss": 3.205, + "step": 2687 + }, + { + "epoch": 3.94, + "grad_norm": 3.010228157043457, + "learning_rate": 0.00013193475106536374, + "loss": 3.1844, + "step": 2688 + }, + { + "epoch": 3.95, + "grad_norm": 2.508854866027832, + "learning_rate": 0.00013189030008390945, + "loss": 3.2108, + "step": 2689 + }, + { + "epoch": 3.95, + "grad_norm": 3.1477272510528564, + "learning_rate": 0.0001318458420868029, + "loss": 3.2018, + "step": 2690 + }, + { + "epoch": 3.95, + "grad_norm": 2.543921947479248, + "learning_rate": 0.0001318013770838245, + "loss": 3.2235, + "step": 2691 + }, + { + "epoch": 3.95, + "grad_norm": 3.577669382095337, + "learning_rate": 0.00013175690508475626, + "loss": 3.2087, + "step": 2692 + }, + { + "epoch": 3.95, + "grad_norm": 2.2817633152008057, + "learning_rate": 0.00013171242609938175, + "loss": 3.1847, + "step": 2693 + }, + { + "epoch": 3.95, + "grad_norm": 7.939913272857666, + "learning_rate": 0.00013166794013748598, + "loss": 3.2269, + "step": 2694 + }, + { + "epoch": 3.95, + "grad_norm": 6.408430576324463, + "learning_rate": 0.00013162344720885565, + "loss": 3.1853, + "step": 2695 + }, + { + "epoch": 3.96, + "grad_norm": 95.63408660888672, + "learning_rate": 0.00013157894732327886, + "loss": 3.2441, + "step": 2696 + }, + { + "epoch": 3.96, + "grad_norm": 76.839111328125, + "learning_rate": 0.00013153444049054523, + "loss": 3.2624, + "step": 2697 + }, + { + "epoch": 3.96, + "grad_norm": 9.534677505493164, + "learning_rate": 0.00013148992672044608, + "loss": 3.2354, + "step": 2698 + }, + { + "epoch": 3.96, + "grad_norm": 29.769962310791016, + "learning_rate": 0.0001314454060227741, + "loss": 3.1857, + "step": 2699 + }, + { + "epoch": 3.96, + "grad_norm": 26.48052215576172, + "learning_rate": 0.00013140087840732355, + "loss": 3.2512, + "step": 2700 + }, + { + "epoch": 3.96, + "grad_norm": 12.384693145751953, + "learning_rate": 0.00013135634388389015, + "loss": 3.2054, + "step": 2701 + }, + { + "epoch": 3.97, + "grad_norm": 4.66050910949707, + "learning_rate": 0.00013131180246227125, + "loss": 3.2012, + "step": 2702 + }, + { + "epoch": 3.97, + "grad_norm": 28.54119300842285, + "learning_rate": 0.00013126725415226563, + "loss": 3.202, + "step": 2703 + }, + { + "epoch": 3.97, + "grad_norm": 69.08728790283203, + "learning_rate": 0.00013122269896367365, + "loss": 3.2423, + "step": 2704 + }, + { + "epoch": 3.97, + "eval_loss": 3.091963529586792, + "eval_runtime": 1165.9245, + "eval_samples_per_second": 14.702, + "eval_steps_per_second": 0.919, + "step": 2704 + }, + { + "epoch": 3.97, + "grad_norm": 22.657764434814453, + "learning_rate": 0.00013117813690629714, + "loss": 3.2586, + "step": 2705 + }, + { + "epoch": 3.97, + "grad_norm": 74.64742279052734, + "learning_rate": 0.00013113356798993947, + "loss": 3.2606, + "step": 2706 + }, + { + "epoch": 3.97, + "grad_norm": 14.374595642089844, + "learning_rate": 0.0001310889922244055, + "loss": 3.2525, + "step": 2707 + }, + { + "epoch": 3.97, + "grad_norm": 5.337686538696289, + "learning_rate": 0.00013104440961950154, + "loss": 3.2355, + "step": 2708 + }, + { + "epoch": 3.98, + "grad_norm": 3.205595016479492, + "learning_rate": 0.00013099982018503558, + "loss": 3.2323, + "step": 2709 + }, + { + "epoch": 3.98, + "grad_norm": 4.343560218811035, + "learning_rate": 0.0001309552239308169, + "loss": 3.2265, + "step": 2710 + }, + { + "epoch": 3.98, + "grad_norm": 68.85613250732422, + "learning_rate": 0.0001309106208666564, + "loss": 3.2412, + "step": 2711 + }, + { + "epoch": 3.98, + "grad_norm": 4.4929633140563965, + "learning_rate": 0.00013086601100236646, + "loss": 3.2243, + "step": 2712 + }, + { + "epoch": 3.98, + "grad_norm": 5.574833393096924, + "learning_rate": 0.000130821394347761, + "loss": 3.2, + "step": 2713 + }, + { + "epoch": 3.98, + "grad_norm": 2.189110040664673, + "learning_rate": 0.00013077677091265528, + "loss": 3.2177, + "step": 2714 + }, + { + "epoch": 3.98, + "grad_norm": 1.2807992696762085, + "learning_rate": 0.00013073214070686626, + "loss": 3.1971, + "step": 2715 + }, + { + "epoch": 3.99, + "grad_norm": 3.459458351135254, + "learning_rate": 0.0001306875037402122, + "loss": 3.186, + "step": 2716 + }, + { + "epoch": 3.99, + "grad_norm": 9.210981369018555, + "learning_rate": 0.000130642860022513, + "loss": 3.1966, + "step": 2717 + }, + { + "epoch": 3.99, + "grad_norm": 2.803443670272827, + "learning_rate": 0.00013059820956358998, + "loss": 3.1927, + "step": 2718 + }, + { + "epoch": 3.99, + "grad_norm": 2.078461170196533, + "learning_rate": 0.00013055355237326588, + "loss": 3.1776, + "step": 2719 + }, + { + "epoch": 3.99, + "grad_norm": 2.4591166973114014, + "learning_rate": 0.00013050888846136502, + "loss": 3.2126, + "step": 2720 + }, + { + "epoch": 3.99, + "grad_norm": 2.6859610080718994, + "learning_rate": 0.00013046421783771318, + "loss": 3.1956, + "step": 2721 + }, + { + "epoch": 3.99, + "grad_norm": 3.064992904663086, + "learning_rate": 0.00013041954051213757, + "loss": 3.25, + "step": 2722 + }, + { + "epoch": 4.0, + "grad_norm": 1.7159329652786255, + "learning_rate": 0.0001303748564944669, + "loss": 3.1734, + "step": 2723 + }, + { + "epoch": 4.0, + "grad_norm": 2.7244114875793457, + "learning_rate": 0.00013033016579453138, + "loss": 3.1743, + "step": 2724 + }, + { + "epoch": 4.0, + "grad_norm": 2.2309274673461914, + "learning_rate": 0.0001302854684221626, + "loss": 3.1751, + "step": 2725 + }, + { + "epoch": 4.0, + "grad_norm": 1.5704296827316284, + "learning_rate": 0.0001302407643871938, + "loss": 3.2018, + "step": 2726 + }, + { + "epoch": 4.0, + "grad_norm": 2.5336227416992188, + "learning_rate": 0.0001301960536994595, + "loss": 3.2139, + "step": 2727 + }, + { + "epoch": 4.0, + "grad_norm": 1.6835451126098633, + "learning_rate": 0.00013015133636879567, + "loss": 3.2198, + "step": 2728 + }, + { + "epoch": 4.01, + "grad_norm": 1.5716129541397095, + "learning_rate": 0.00013010661240503999, + "loss": 3.2108, + "step": 2729 + }, + { + "epoch": 4.01, + "grad_norm": 1.5545756816864014, + "learning_rate": 0.0001300618818180313, + "loss": 3.1592, + "step": 2730 + }, + { + "epoch": 4.01, + "grad_norm": 1.9438145160675049, + "learning_rate": 0.00013001714461761008, + "loss": 3.2145, + "step": 2731 + }, + { + "epoch": 4.01, + "grad_norm": 1.7333067655563354, + "learning_rate": 0.00012997240081361823, + "loss": 3.1362, + "step": 2732 + }, + { + "epoch": 4.01, + "grad_norm": 1.632812261581421, + "learning_rate": 0.00012992765041589904, + "loss": 3.2064, + "step": 2733 + }, + { + "epoch": 4.01, + "grad_norm": 1.1884607076644897, + "learning_rate": 0.00012988289343429733, + "loss": 3.1789, + "step": 2734 + }, + { + "epoch": 4.01, + "grad_norm": 2.9912478923797607, + "learning_rate": 0.00012983812987865935, + "loss": 3.1561, + "step": 2735 + }, + { + "epoch": 4.02, + "grad_norm": 1.3816310167312622, + "learning_rate": 0.00012979335975883276, + "loss": 3.1848, + "step": 2736 + }, + { + "epoch": 4.0, + "grad_norm": 19.01185417175293, + "learning_rate": 0.00012974858308466666, + "loss": 3.1801, + "step": 2737 + }, + { + "epoch": 4.0, + "grad_norm": 5.716503143310547, + "learning_rate": 0.0001297037998660117, + "loss": 3.2216, + "step": 2738 + }, + { + "epoch": 4.0, + "grad_norm": 2.985485792160034, + "learning_rate": 0.00012965901011271978, + "loss": 3.2136, + "step": 2739 + }, + { + "epoch": 4.01, + "grad_norm": 2.645477056503296, + "learning_rate": 0.00012961421383464445, + "loss": 3.176, + "step": 2740 + }, + { + "epoch": 4.01, + "grad_norm": 2.804504871368408, + "learning_rate": 0.0001295694110416405, + "loss": 3.1502, + "step": 2741 + }, + { + "epoch": 4.01, + "grad_norm": 2.257525682449341, + "learning_rate": 0.00012952460174356432, + "loss": 3.1909, + "step": 2742 + }, + { + "epoch": 4.01, + "grad_norm": 2.525448799133301, + "learning_rate": 0.0001294797859502736, + "loss": 3.2058, + "step": 2743 + }, + { + "epoch": 4.01, + "grad_norm": 2.589095115661621, + "learning_rate": 0.00012943496367162754, + "loss": 3.2137, + "step": 2744 + }, + { + "epoch": 4.01, + "grad_norm": 1.6353000402450562, + "learning_rate": 0.00012939013491748672, + "loss": 3.2001, + "step": 2745 + }, + { + "epoch": 4.01, + "grad_norm": 2.5494680404663086, + "learning_rate": 0.00012934529969771324, + "loss": 3.1529, + "step": 2746 + }, + { + "epoch": 4.02, + "grad_norm": 2.1689844131469727, + "learning_rate": 0.00012930045802217043, + "loss": 3.1931, + "step": 2747 + }, + { + "epoch": 4.02, + "grad_norm": 3.0633785724639893, + "learning_rate": 0.00012925560990072324, + "loss": 3.1941, + "step": 2748 + }, + { + "epoch": 4.02, + "grad_norm": 2.6445438861846924, + "learning_rate": 0.00012921075534323795, + "loss": 3.1673, + "step": 2749 + }, + { + "epoch": 4.02, + "grad_norm": 1.6328538656234741, + "learning_rate": 0.00012916589435958223, + "loss": 3.1533, + "step": 2750 + }, + { + "epoch": 4.02, + "grad_norm": 7.05519962310791, + "learning_rate": 0.0001291210269596252, + "loss": 3.1762, + "step": 2751 + }, + { + "epoch": 4.02, + "grad_norm": 10.450418472290039, + "learning_rate": 0.0001290761531532374, + "loss": 3.2032, + "step": 2752 + }, + { + "epoch": 4.02, + "grad_norm": 3.726336717605591, + "learning_rate": 0.00012903127295029078, + "loss": 3.1835, + "step": 2753 + }, + { + "epoch": 4.03, + "grad_norm": 2.4860239028930664, + "learning_rate": 0.00012898638636065868, + "loss": 3.1818, + "step": 2754 + }, + { + "epoch": 4.03, + "grad_norm": 2.166196584701538, + "learning_rate": 0.00012894149339421582, + "loss": 3.1609, + "step": 2755 + }, + { + "epoch": 4.03, + "grad_norm": 2.179659128189087, + "learning_rate": 0.00012889659406083837, + "loss": 3.2125, + "step": 2756 + }, + { + "epoch": 4.03, + "grad_norm": 2.2733445167541504, + "learning_rate": 0.0001288516883704039, + "loss": 3.2931, + "step": 2757 + }, + { + "epoch": 4.03, + "grad_norm": 1.3774925470352173, + "learning_rate": 0.00012880677633279132, + "loss": 3.1741, + "step": 2758 + }, + { + "epoch": 4.03, + "grad_norm": 4.090391159057617, + "learning_rate": 0.000128761857957881, + "loss": 3.2181, + "step": 2759 + }, + { + "epoch": 4.04, + "grad_norm": 3.762101411819458, + "learning_rate": 0.00012871693325555472, + "loss": 3.1991, + "step": 2760 + }, + { + "epoch": 4.04, + "grad_norm": 2.435084819793701, + "learning_rate": 0.00012867200223569555, + "loss": 3.1734, + "step": 2761 + }, + { + "epoch": 4.04, + "grad_norm": 11.30912971496582, + "learning_rate": 0.00012862706490818804, + "loss": 3.1755, + "step": 2762 + }, + { + "epoch": 4.04, + "grad_norm": 3.355214834213257, + "learning_rate": 0.0001285821212829181, + "loss": 3.1718, + "step": 2763 + }, + { + "epoch": 4.04, + "grad_norm": 2.4179513454437256, + "learning_rate": 0.00012853717136977305, + "loss": 3.1722, + "step": 2764 + }, + { + "epoch": 4.04, + "grad_norm": 2.18635630607605, + "learning_rate": 0.0001284922151786415, + "loss": 3.1982, + "step": 2765 + }, + { + "epoch": 4.04, + "grad_norm": 2.3029189109802246, + "learning_rate": 0.0001284472527194136, + "loss": 3.2427, + "step": 2766 + }, + { + "epoch": 4.05, + "grad_norm": 2.3949828147888184, + "learning_rate": 0.00012840228400198071, + "loss": 3.1944, + "step": 2767 + }, + { + "epoch": 4.05, + "grad_norm": 2.5206806659698486, + "learning_rate": 0.0001283573090362357, + "loss": 3.1899, + "step": 2768 + }, + { + "epoch": 4.05, + "grad_norm": 1.8850032091140747, + "learning_rate": 0.00012831232783207277, + "loss": 3.2212, + "step": 2769 + }, + { + "epoch": 4.05, + "grad_norm": 2.58017897605896, + "learning_rate": 0.0001282673403993874, + "loss": 3.2026, + "step": 2770 + }, + { + "epoch": 4.05, + "grad_norm": 2.990920305252075, + "learning_rate": 0.00012822234674807662, + "loss": 3.1852, + "step": 2771 + }, + { + "epoch": 4.05, + "grad_norm": 0.5992396473884583, + "learning_rate": 0.00012817734688803868, + "loss": 3.1736, + "step": 2772 + }, + { + "epoch": 4.05, + "grad_norm": 2.821469306945801, + "learning_rate": 0.0001281323408291733, + "loss": 3.1999, + "step": 2773 + }, + { + "epoch": 4.06, + "grad_norm": 15.83332633972168, + "learning_rate": 0.00012808732858138146, + "loss": 3.1844, + "step": 2774 + }, + { + "epoch": 4.06, + "grad_norm": 2.910820960998535, + "learning_rate": 0.0001280423101545655, + "loss": 3.1587, + "step": 2775 + }, + { + "epoch": 4.06, + "grad_norm": 17.933792114257812, + "learning_rate": 0.00012799728555862934, + "loss": 3.2201, + "step": 2776 + }, + { + "epoch": 4.06, + "grad_norm": 2.721127986907959, + "learning_rate": 0.0001279522548034779, + "loss": 3.1507, + "step": 2777 + }, + { + "epoch": 4.06, + "grad_norm": 1.5368844270706177, + "learning_rate": 0.00012790721789901775, + "loss": 3.1902, + "step": 2778 + }, + { + "epoch": 4.06, + "grad_norm": 3.0113532543182373, + "learning_rate": 0.00012786217485515674, + "loss": 3.149, + "step": 2779 + }, + { + "epoch": 4.06, + "grad_norm": 19.78055191040039, + "learning_rate": 0.00012781712568180393, + "loss": 3.1755, + "step": 2780 + }, + { + "epoch": 4.07, + "grad_norm": 2.508270025253296, + "learning_rate": 0.0001277720703888699, + "loss": 3.1915, + "step": 2781 + }, + { + "epoch": 4.07, + "grad_norm": 2.167954206466675, + "learning_rate": 0.00012772700898626652, + "loss": 3.2054, + "step": 2782 + }, + { + "epoch": 4.07, + "grad_norm": 4.016536235809326, + "learning_rate": 0.00012768194148390696, + "loss": 3.195, + "step": 2783 + }, + { + "epoch": 4.07, + "grad_norm": 3.441601276397705, + "learning_rate": 0.0001276368678917058, + "loss": 3.15, + "step": 2784 + }, + { + "epoch": 4.07, + "grad_norm": 7.370872974395752, + "learning_rate": 0.0001275917882195789, + "loss": 3.1633, + "step": 2785 + }, + { + "epoch": 4.07, + "grad_norm": 2.296081781387329, + "learning_rate": 0.00012754670247744354, + "loss": 3.2345, + "step": 2786 + }, + { + "epoch": 4.08, + "grad_norm": 2.009742259979248, + "learning_rate": 0.00012750161067521826, + "loss": 3.1938, + "step": 2787 + }, + { + "epoch": 4.08, + "grad_norm": 4.299360752105713, + "learning_rate": 0.00012745651282282288, + "loss": 3.2213, + "step": 2788 + }, + { + "epoch": 4.08, + "grad_norm": 3.636892080307007, + "learning_rate": 0.00012741140893017872, + "loss": 3.1713, + "step": 2789 + }, + { + "epoch": 4.08, + "grad_norm": 1.2036387920379639, + "learning_rate": 0.0001273662990072083, + "loss": 3.2107, + "step": 2790 + }, + { + "epoch": 4.08, + "grad_norm": 3.7308552265167236, + "learning_rate": 0.0001273211830638355, + "loss": 3.1955, + "step": 2791 + }, + { + "epoch": 4.08, + "grad_norm": 9.248310089111328, + "learning_rate": 0.00012727606110998549, + "loss": 3.1917, + "step": 2792 + }, + { + "epoch": 4.08, + "grad_norm": 15.789203643798828, + "learning_rate": 0.00012723093315558488, + "loss": 3.1611, + "step": 2793 + }, + { + "epoch": 4.09, + "grad_norm": 3.4315154552459717, + "learning_rate": 0.00012718579921056143, + "loss": 3.2364, + "step": 2794 + }, + { + "epoch": 4.09, + "grad_norm": 2.061835289001465, + "learning_rate": 0.00012714065928484436, + "loss": 3.2079, + "step": 2795 + }, + { + "epoch": 4.09, + "grad_norm": 2.4200713634490967, + "learning_rate": 0.00012709551338836413, + "loss": 3.1846, + "step": 2796 + }, + { + "epoch": 4.09, + "grad_norm": 3.561788558959961, + "learning_rate": 0.00012705036153105257, + "loss": 3.2151, + "step": 2797 + }, + { + "epoch": 4.09, + "grad_norm": 2.967207908630371, + "learning_rate": 0.00012700520372284273, + "loss": 3.2036, + "step": 2798 + }, + { + "epoch": 4.09, + "grad_norm": 3.490912437438965, + "learning_rate": 0.00012696003997366906, + "loss": 3.1893, + "step": 2799 + }, + { + "epoch": 4.09, + "grad_norm": 2.477809429168701, + "learning_rate": 0.00012691487029346728, + "loss": 3.1701, + "step": 2800 + }, + { + "epoch": 4.1, + "grad_norm": 3.558258295059204, + "learning_rate": 0.0001268696946921744, + "loss": 3.2343, + "step": 2801 + }, + { + "epoch": 4.1, + "grad_norm": 3.1721646785736084, + "learning_rate": 0.00012682451317972874, + "loss": 3.201, + "step": 2802 + }, + { + "epoch": 4.1, + "grad_norm": 2.562236785888672, + "learning_rate": 0.00012677932576606998, + "loss": 3.2291, + "step": 2803 + }, + { + "epoch": 4.1, + "grad_norm": 9.012580871582031, + "learning_rate": 0.00012673413246113905, + "loss": 3.2074, + "step": 2804 + }, + { + "epoch": 4.1, + "grad_norm": 2.1724753379821777, + "learning_rate": 0.00012668893327487805, + "loss": 3.2104, + "step": 2805 + }, + { + "epoch": 4.1, + "grad_norm": 1.8097708225250244, + "learning_rate": 0.00012664372821723064, + "loss": 3.1534, + "step": 2806 + }, + { + "epoch": 4.11, + "grad_norm": 2.409672498703003, + "learning_rate": 0.00012659851729814162, + "loss": 3.1748, + "step": 2807 + }, + { + "epoch": 4.11, + "grad_norm": 2.8095180988311768, + "learning_rate": 0.000126553300527557, + "loss": 3.1996, + "step": 2808 + }, + { + "epoch": 4.11, + "grad_norm": 2.353957176208496, + "learning_rate": 0.00012650807791542422, + "loss": 3.1945, + "step": 2809 + }, + { + "epoch": 4.11, + "grad_norm": 3.013678789138794, + "learning_rate": 0.00012646284947169195, + "loss": 3.1839, + "step": 2810 + }, + { + "epoch": 4.11, + "grad_norm": 2.154859781265259, + "learning_rate": 0.00012641761520631016, + "loss": 3.1541, + "step": 2811 + }, + { + "epoch": 4.11, + "grad_norm": 2.0929133892059326, + "learning_rate": 0.00012637237512923006, + "loss": 3.1655, + "step": 2812 + }, + { + "epoch": 4.11, + "grad_norm": 1.2277342081069946, + "learning_rate": 0.0001263271292504042, + "loss": 3.1933, + "step": 2813 + }, + { + "epoch": 4.12, + "grad_norm": 2.6938316822052, + "learning_rate": 0.0001262818775797863, + "loss": 3.1441, + "step": 2814 + }, + { + "epoch": 4.12, + "grad_norm": 1.9652117490768433, + "learning_rate": 0.00012623662012733146, + "loss": 3.1616, + "step": 2815 + }, + { + "epoch": 4.12, + "grad_norm": 1.6476244926452637, + "learning_rate": 0.00012619135690299603, + "loss": 3.1839, + "step": 2816 + }, + { + "epoch": 4.12, + "grad_norm": 1.534227967262268, + "learning_rate": 0.00012614608791673763, + "loss": 3.1646, + "step": 2817 + }, + { + "epoch": 4.12, + "grad_norm": 2.850703477859497, + "learning_rate": 0.00012610081317851511, + "loss": 3.207, + "step": 2818 + }, + { + "epoch": 4.12, + "grad_norm": 2.1084752082824707, + "learning_rate": 0.00012605553269828858, + "loss": 3.1688, + "step": 2819 + }, + { + "epoch": 4.12, + "grad_norm": 2.303504705429077, + "learning_rate": 0.0001260102464860195, + "loss": 3.229, + "step": 2820 + }, + { + "epoch": 4.13, + "grad_norm": 1.5470285415649414, + "learning_rate": 0.0001259649545516705, + "loss": 3.1814, + "step": 2821 + }, + { + "epoch": 4.13, + "grad_norm": 1.4991244077682495, + "learning_rate": 0.0001259196569052055, + "loss": 3.1927, + "step": 2822 + }, + { + "epoch": 4.13, + "grad_norm": 1.342314600944519, + "learning_rate": 0.0001258743535565897, + "loss": 3.1743, + "step": 2823 + }, + { + "epoch": 4.13, + "grad_norm": 2.662546396255493, + "learning_rate": 0.0001258290445157895, + "loss": 3.1884, + "step": 2824 + }, + { + "epoch": 4.13, + "grad_norm": 1.1698952913284302, + "learning_rate": 0.00012578372979277265, + "loss": 3.1841, + "step": 2825 + }, + { + "epoch": 4.13, + "grad_norm": 2.8144288063049316, + "learning_rate": 0.000125738409397508, + "loss": 3.1634, + "step": 2826 + }, + { + "epoch": 4.13, + "grad_norm": 1.0438013076782227, + "learning_rate": 0.00012569308333996578, + "loss": 3.1828, + "step": 2827 + }, + { + "epoch": 4.14, + "grad_norm": 5.697520732879639, + "learning_rate": 0.0001256477516301174, + "loss": 3.1897, + "step": 2828 + }, + { + "epoch": 4.14, + "grad_norm": 6.039661884307861, + "learning_rate": 0.00012560241427793555, + "loss": 3.2308, + "step": 2829 + }, + { + "epoch": 4.14, + "grad_norm": 51.954402923583984, + "learning_rate": 0.00012555707129339417, + "loss": 3.1693, + "step": 2830 + }, + { + "epoch": 4.14, + "grad_norm": 5.388270378112793, + "learning_rate": 0.00012551172268646834, + "loss": 3.2463, + "step": 2831 + }, + { + "epoch": 4.14, + "grad_norm": 4.189896583557129, + "learning_rate": 0.00012546636846713454, + "loss": 3.209, + "step": 2832 + }, + { + "epoch": 4.14, + "grad_norm": 2.421431064605713, + "learning_rate": 0.0001254210086453703, + "loss": 3.2312, + "step": 2833 + }, + { + "epoch": 4.15, + "grad_norm": 4.315710544586182, + "learning_rate": 0.00012537564323115454, + "loss": 3.2011, + "step": 2834 + }, + { + "epoch": 4.15, + "grad_norm": 1.7199636697769165, + "learning_rate": 0.00012533027223446733, + "loss": 3.1634, + "step": 2835 + }, + { + "epoch": 4.15, + "grad_norm": 4.567407608032227, + "learning_rate": 0.00012528489566528996, + "loss": 3.1999, + "step": 2836 + }, + { + "epoch": 4.15, + "grad_norm": 5.22501277923584, + "learning_rate": 0.00012523951353360504, + "loss": 3.1743, + "step": 2837 + }, + { + "epoch": 4.15, + "grad_norm": 2.197282314300537, + "learning_rate": 0.00012519412584939627, + "loss": 3.2332, + "step": 2838 + }, + { + "epoch": 4.15, + "grad_norm": 5.91029167175293, + "learning_rate": 0.00012514873262264864, + "loss": 3.1713, + "step": 2839 + }, + { + "epoch": 4.15, + "grad_norm": 6.8516716957092285, + "learning_rate": 0.0001251033338633484, + "loss": 3.1908, + "step": 2840 + }, + { + "epoch": 4.16, + "grad_norm": 3.321195125579834, + "learning_rate": 0.00012505792958148292, + "loss": 3.1693, + "step": 2841 + }, + { + "epoch": 4.16, + "grad_norm": 11.536995887756348, + "learning_rate": 0.00012501251978704088, + "loss": 3.1626, + "step": 2842 + }, + { + "epoch": 4.16, + "grad_norm": 9.267814636230469, + "learning_rate": 0.0001249671044900121, + "loss": 3.1807, + "step": 2843 + }, + { + "epoch": 4.16, + "grad_norm": 8.332723617553711, + "learning_rate": 0.00012492168370038765, + "loss": 3.1934, + "step": 2844 + }, + { + "epoch": 4.16, + "grad_norm": 3.4367361068725586, + "learning_rate": 0.00012487625742815985, + "loss": 3.1797, + "step": 2845 + }, + { + "epoch": 4.16, + "grad_norm": 6.636847019195557, + "learning_rate": 0.00012483082568332207, + "loss": 3.1851, + "step": 2846 + }, + { + "epoch": 4.16, + "grad_norm": 8.669164657592773, + "learning_rate": 0.00012478538847586907, + "loss": 3.2061, + "step": 2847 + }, + { + "epoch": 4.17, + "grad_norm": 6.34977912902832, + "learning_rate": 0.00012473994581579673, + "loss": 3.196, + "step": 2848 + }, + { + "epoch": 4.17, + "grad_norm": 4.2629780769348145, + "learning_rate": 0.00012469449771310208, + "loss": 3.2128, + "step": 2849 + }, + { + "epoch": 4.17, + "grad_norm": 5.895789623260498, + "learning_rate": 0.00012464904417778345, + "loss": 3.2769, + "step": 2850 + }, + { + "epoch": 4.17, + "grad_norm": 7.333654403686523, + "learning_rate": 0.0001246035852198403, + "loss": 3.2108, + "step": 2851 + }, + { + "epoch": 4.17, + "grad_norm": 2.965632915496826, + "learning_rate": 0.00012455812084927325, + "loss": 3.2012, + "step": 2852 + }, + { + "epoch": 4.17, + "grad_norm": 8.51198673248291, + "learning_rate": 0.0001245126510760842, + "loss": 3.2149, + "step": 2853 + }, + { + "epoch": 4.17, + "grad_norm": 7.021212100982666, + "learning_rate": 0.00012446717591027624, + "loss": 3.2624, + "step": 2854 + }, + { + "epoch": 4.18, + "grad_norm": 1.8178749084472656, + "learning_rate": 0.0001244216953618535, + "loss": 3.2099, + "step": 2855 + }, + { + "epoch": 4.18, + "grad_norm": 11.300747871398926, + "learning_rate": 0.00012437620944082145, + "loss": 3.1795, + "step": 2856 + }, + { + "epoch": 4.18, + "grad_norm": 9.994318962097168, + "learning_rate": 0.0001243307181571867, + "loss": 3.2285, + "step": 2857 + }, + { + "epoch": 4.18, + "grad_norm": 11.500404357910156, + "learning_rate": 0.00012428522152095703, + "loss": 3.1869, + "step": 2858 + }, + { + "epoch": 4.18, + "grad_norm": 5.326147556304932, + "learning_rate": 0.00012423971954214137, + "loss": 3.2226, + "step": 2859 + }, + { + "epoch": 4.18, + "grad_norm": 5.258158206939697, + "learning_rate": 0.00012419421223074982, + "loss": 3.1848, + "step": 2860 + }, + { + "epoch": 4.19, + "grad_norm": 7.557288646697998, + "learning_rate": 0.00012414869959679375, + "loss": 3.2594, + "step": 2861 + }, + { + "epoch": 4.19, + "grad_norm": 19.032909393310547, + "learning_rate": 0.0001241031816502856, + "loss": 3.2464, + "step": 2862 + }, + { + "epoch": 4.19, + "grad_norm": 9.38792896270752, + "learning_rate": 0.00012405765840123903, + "loss": 3.4169, + "step": 2863 + }, + { + "epoch": 4.19, + "grad_norm": 58.51255798339844, + "learning_rate": 0.0001240121298596688, + "loss": 3.3102, + "step": 2864 + }, + { + "epoch": 4.19, + "grad_norm": 9.871623992919922, + "learning_rate": 0.00012396659603559099, + "loss": 3.3069, + "step": 2865 + }, + { + "epoch": 4.19, + "grad_norm": 18.70452117919922, + "learning_rate": 0.00012392105693902262, + "loss": 3.2645, + "step": 2866 + }, + { + "epoch": 4.19, + "grad_norm": 6.495476245880127, + "learning_rate": 0.00012387551257998206, + "loss": 3.2188, + "step": 2867 + }, + { + "epoch": 4.2, + "grad_norm": 4.516813278198242, + "learning_rate": 0.00012382996296848875, + "loss": 3.2188, + "step": 2868 + }, + { + "epoch": 4.2, + "grad_norm": 9.024648666381836, + "learning_rate": 0.00012378440811456324, + "loss": 3.251, + "step": 2869 + }, + { + "epoch": 4.2, + "grad_norm": 10.055442810058594, + "learning_rate": 0.00012373884802822737, + "loss": 3.2687, + "step": 2870 + }, + { + "epoch": 4.2, + "grad_norm": 5.984691619873047, + "learning_rate": 0.00012369328271950405, + "loss": 3.2402, + "step": 2871 + }, + { + "epoch": 4.2, + "grad_norm": 4.42611026763916, + "learning_rate": 0.00012364771219841727, + "loss": 3.2241, + "step": 2872 + }, + { + "epoch": 4.2, + "grad_norm": 5.240525722503662, + "learning_rate": 0.0001236021364749923, + "loss": 3.2565, + "step": 2873 + }, + { + "epoch": 4.2, + "eval_loss": 3.031118154525757, + "eval_runtime": 1166.2283, + "eval_samples_per_second": 14.699, + "eval_steps_per_second": 0.919, + "step": 2873 + }, + { + "epoch": 4.2, + "grad_norm": 3.474372148513794, + "learning_rate": 0.00012355655555925548, + "loss": 3.1972, + "step": 2874 + }, + { + "epoch": 4.21, + "grad_norm": 3.6792542934417725, + "learning_rate": 0.00012351096946123431, + "loss": 3.1889, + "step": 2875 + }, + { + "epoch": 4.21, + "grad_norm": 3.1051828861236572, + "learning_rate": 0.00012346537819095743, + "loss": 3.206, + "step": 2876 + }, + { + "epoch": 4.21, + "grad_norm": 2.708695411682129, + "learning_rate": 0.00012341978175845462, + "loss": 3.2389, + "step": 2877 + }, + { + "epoch": 4.21, + "grad_norm": 2.4177584648132324, + "learning_rate": 0.00012337418017375676, + "loss": 3.1948, + "step": 2878 + }, + { + "epoch": 4.21, + "grad_norm": 3.8860323429107666, + "learning_rate": 0.00012332857344689587, + "loss": 3.2037, + "step": 2879 + }, + { + "epoch": 4.21, + "grad_norm": 4.779682159423828, + "learning_rate": 0.0001232829615879052, + "loss": 3.1901, + "step": 2880 + }, + { + "epoch": 4.21, + "grad_norm": 2.2922635078430176, + "learning_rate": 0.00012323734460681902, + "loss": 3.1367, + "step": 2881 + }, + { + "epoch": 4.22, + "grad_norm": 2.1851584911346436, + "learning_rate": 0.00012319172251367276, + "loss": 3.1782, + "step": 2882 + }, + { + "epoch": 4.22, + "grad_norm": 3.3458876609802246, + "learning_rate": 0.00012314609531850293, + "loss": 3.2361, + "step": 2883 + }, + { + "epoch": 4.22, + "grad_norm": 3.9705002307891846, + "learning_rate": 0.00012310046303134732, + "loss": 3.2227, + "step": 2884 + }, + { + "epoch": 4.22, + "grad_norm": 2.2800686359405518, + "learning_rate": 0.0001230548256622446, + "loss": 3.2052, + "step": 2885 + }, + { + "epoch": 4.22, + "grad_norm": 4.4462199211120605, + "learning_rate": 0.00012300918322123475, + "loss": 3.2142, + "step": 2886 + }, + { + "epoch": 4.22, + "grad_norm": 3.9958441257476807, + "learning_rate": 0.0001229635357183588, + "loss": 3.1989, + "step": 2887 + }, + { + "epoch": 4.23, + "grad_norm": 3.798068046569824, + "learning_rate": 0.00012291788316365888, + "loss": 3.1956, + "step": 2888 + }, + { + "epoch": 4.23, + "grad_norm": 5.135157108306885, + "learning_rate": 0.00012287222556717827, + "loss": 3.1854, + "step": 2889 + }, + { + "epoch": 4.23, + "grad_norm": 2.207172393798828, + "learning_rate": 0.0001228265629389613, + "loss": 3.1583, + "step": 2890 + }, + { + "epoch": 4.23, + "grad_norm": 8.646849632263184, + "learning_rate": 0.00012278089528905346, + "loss": 3.245, + "step": 2891 + }, + { + "epoch": 4.23, + "grad_norm": 7.601872444152832, + "learning_rate": 0.00012273522262750132, + "loss": 3.1968, + "step": 2892 + }, + { + "epoch": 4.23, + "grad_norm": 6.095008373260498, + "learning_rate": 0.00012268954496435258, + "loss": 3.1753, + "step": 2893 + }, + { + "epoch": 4.23, + "grad_norm": 3.311551094055176, + "learning_rate": 0.000122643862309656, + "loss": 3.2052, + "step": 2894 + }, + { + "epoch": 4.24, + "grad_norm": 6.019943714141846, + "learning_rate": 0.0001225981746734615, + "loss": 3.2064, + "step": 2895 + }, + { + "epoch": 4.24, + "grad_norm": 5.016195774078369, + "learning_rate": 0.00012255248206581996, + "loss": 3.2018, + "step": 2896 + }, + { + "epoch": 4.24, + "grad_norm": 2.2277472019195557, + "learning_rate": 0.00012250678449678354, + "loss": 3.1877, + "step": 2897 + }, + { + "epoch": 4.24, + "grad_norm": 5.388427734375, + "learning_rate": 0.00012246108197640538, + "loss": 3.2087, + "step": 2898 + }, + { + "epoch": 4.24, + "grad_norm": 3.7437267303466797, + "learning_rate": 0.0001224153745147397, + "loss": 3.1554, + "step": 2899 + }, + { + "epoch": 4.24, + "grad_norm": 2.8456854820251465, + "learning_rate": 0.0001223696621218418, + "loss": 3.1912, + "step": 2900 + }, + { + "epoch": 4.24, + "grad_norm": 2.9107625484466553, + "learning_rate": 0.0001223239448077682, + "loss": 3.1832, + "step": 2901 + }, + { + "epoch": 4.25, + "grad_norm": 2.9731972217559814, + "learning_rate": 0.00012227822258257636, + "loss": 3.1781, + "step": 2902 + }, + { + "epoch": 4.25, + "grad_norm": 3.2368147373199463, + "learning_rate": 0.00012223249545632482, + "loss": 3.1863, + "step": 2903 + }, + { + "epoch": 4.25, + "grad_norm": 1.3652539253234863, + "learning_rate": 0.00012218676343907328, + "loss": 3.1765, + "step": 2904 + }, + { + "epoch": 4.25, + "grad_norm": 6.007261753082275, + "learning_rate": 0.00012214102654088248, + "loss": 3.2073, + "step": 2905 + }, + { + "epoch": 4.25, + "grad_norm": 6.385366439819336, + "learning_rate": 0.00012209528477181422, + "loss": 3.1974, + "step": 2906 + }, + { + "epoch": 4.25, + "grad_norm": 1.5009405612945557, + "learning_rate": 0.00012204953814193139, + "loss": 3.1815, + "step": 2907 + }, + { + "epoch": 4.26, + "grad_norm": 4.717186450958252, + "learning_rate": 0.00012200378666129793, + "loss": 3.166, + "step": 2908 + }, + { + "epoch": 4.26, + "grad_norm": 4.571866035461426, + "learning_rate": 0.0001219580303399789, + "loss": 3.2043, + "step": 2909 + }, + { + "epoch": 4.26, + "grad_norm": 6.416868686676025, + "learning_rate": 0.0001219122691880403, + "loss": 3.1969, + "step": 2910 + }, + { + "epoch": 4.26, + "grad_norm": 4.813213348388672, + "learning_rate": 0.00012186650321554935, + "loss": 3.1706, + "step": 2911 + }, + { + "epoch": 4.26, + "grad_norm": 3.790310859680176, + "learning_rate": 0.00012182073243257426, + "loss": 3.178, + "step": 2912 + }, + { + "epoch": 4.26, + "grad_norm": 4.808967590332031, + "learning_rate": 0.00012177495684918423, + "loss": 3.2266, + "step": 2913 + }, + { + "epoch": 4.26, + "grad_norm": 8.275779724121094, + "learning_rate": 0.00012172917647544964, + "loss": 3.1736, + "step": 2914 + }, + { + "epoch": 4.27, + "grad_norm": 4.079070568084717, + "learning_rate": 0.00012168339132144186, + "loss": 3.1918, + "step": 2915 + }, + { + "epoch": 4.27, + "grad_norm": 2.108287811279297, + "learning_rate": 0.0001216376013972333, + "loss": 3.1543, + "step": 2916 + }, + { + "epoch": 4.27, + "grad_norm": 3.337580442428589, + "learning_rate": 0.00012159180671289743, + "loss": 3.1937, + "step": 2917 + }, + { + "epoch": 4.27, + "grad_norm": 2.6380503177642822, + "learning_rate": 0.0001215460072785088, + "loss": 3.1794, + "step": 2918 + }, + { + "epoch": 4.27, + "grad_norm": 2.371419906616211, + "learning_rate": 0.00012150020310414296, + "loss": 3.1682, + "step": 2919 + }, + { + "epoch": 4.27, + "grad_norm": 1.647721767425537, + "learning_rate": 0.00012145439419987652, + "loss": 3.1618, + "step": 2920 + }, + { + "epoch": 4.27, + "grad_norm": 0.846366822719574, + "learning_rate": 0.00012140858057578713, + "loss": 3.1883, + "step": 2921 + }, + { + "epoch": 4.28, + "grad_norm": 1.4289370775222778, + "learning_rate": 0.00012136276224195348, + "loss": 3.2119, + "step": 2922 + }, + { + "epoch": 4.28, + "grad_norm": 2.510512351989746, + "learning_rate": 0.00012131693920845533, + "loss": 3.1849, + "step": 2923 + }, + { + "epoch": 4.28, + "grad_norm": 1.3627164363861084, + "learning_rate": 0.0001212711114853734, + "loss": 3.1851, + "step": 2924 + }, + { + "epoch": 4.28, + "grad_norm": 2.383141279220581, + "learning_rate": 0.00012122527908278949, + "loss": 3.1599, + "step": 2925 + }, + { + "epoch": 4.28, + "grad_norm": 2.2744383811950684, + "learning_rate": 0.00012117944201078647, + "loss": 3.1814, + "step": 2926 + }, + { + "epoch": 4.28, + "grad_norm": 2.6775460243225098, + "learning_rate": 0.00012113360027944806, + "loss": 3.1495, + "step": 2927 + }, + { + "epoch": 4.28, + "grad_norm": 2.249178647994995, + "learning_rate": 0.00012108775389885929, + "loss": 3.1721, + "step": 2928 + }, + { + "epoch": 4.29, + "grad_norm": 1.3241461515426636, + "learning_rate": 0.00012104190287910597, + "loss": 3.1563, + "step": 2929 + }, + { + "epoch": 4.29, + "grad_norm": 3.977029323577881, + "learning_rate": 0.00012099604723027502, + "loss": 3.254, + "step": 2930 + }, + { + "epoch": 4.29, + "grad_norm": 3.9783012866973877, + "learning_rate": 0.00012095018696245441, + "loss": 3.1861, + "step": 2931 + }, + { + "epoch": 4.29, + "grad_norm": 1.687117338180542, + "learning_rate": 0.00012090432208573306, + "loss": 3.1259, + "step": 2932 + }, + { + "epoch": 4.29, + "grad_norm": 10.139452934265137, + "learning_rate": 0.00012085845261020095, + "loss": 3.1579, + "step": 2933 + }, + { + "epoch": 4.29, + "grad_norm": 4.6292500495910645, + "learning_rate": 0.00012081257854594906, + "loss": 3.2254, + "step": 2934 + }, + { + "epoch": 4.3, + "grad_norm": 3.5405800342559814, + "learning_rate": 0.00012076669990306934, + "loss": 3.1327, + "step": 2935 + }, + { + "epoch": 4.3, + "grad_norm": 1.5894381999969482, + "learning_rate": 0.00012072081669165482, + "loss": 3.1545, + "step": 2936 + }, + { + "epoch": 4.3, + "grad_norm": 3.921757221221924, + "learning_rate": 0.0001206749289217995, + "loss": 3.1646, + "step": 2937 + }, + { + "epoch": 4.3, + "grad_norm": 3.1364986896514893, + "learning_rate": 0.0001206290366035984, + "loss": 3.1835, + "step": 2938 + }, + { + "epoch": 4.3, + "grad_norm": 1.0131809711456299, + "learning_rate": 0.00012058313974714746, + "loss": 3.1446, + "step": 2939 + }, + { + "epoch": 4.3, + "grad_norm": 2.2197444438934326, + "learning_rate": 0.00012053723836254373, + "loss": 3.1805, + "step": 2940 + }, + { + "epoch": 4.3, + "grad_norm": 1.5680094957351685, + "learning_rate": 0.00012049133245988519, + "loss": 3.2005, + "step": 2941 + }, + { + "epoch": 4.31, + "grad_norm": 2.2786076068878174, + "learning_rate": 0.00012044542204927087, + "loss": 3.1663, + "step": 2942 + }, + { + "epoch": 4.31, + "grad_norm": 2.1755716800689697, + "learning_rate": 0.00012039950714080068, + "loss": 3.1709, + "step": 2943 + }, + { + "epoch": 4.31, + "grad_norm": 1.485142469406128, + "learning_rate": 0.00012035358774457563, + "loss": 3.2298, + "step": 2944 + }, + { + "epoch": 4.31, + "grad_norm": 1.9527829885482788, + "learning_rate": 0.00012030766387069771, + "loss": 3.1868, + "step": 2945 + }, + { + "epoch": 4.31, + "grad_norm": 1.5011379718780518, + "learning_rate": 0.00012026173552926981, + "loss": 3.1533, + "step": 2946 + }, + { + "epoch": 4.31, + "grad_norm": 1.8543486595153809, + "learning_rate": 0.00012021580273039585, + "loss": 3.1536, + "step": 2947 + }, + { + "epoch": 4.31, + "grad_norm": 2.353090763092041, + "learning_rate": 0.00012016986548418083, + "loss": 3.1675, + "step": 2948 + }, + { + "epoch": 4.32, + "grad_norm": 3.509500741958618, + "learning_rate": 0.00012012392380073056, + "loss": 3.1963, + "step": 2949 + }, + { + "epoch": 4.32, + "grad_norm": 4.018800258636475, + "learning_rate": 0.00012007797769015192, + "loss": 3.165, + "step": 2950 + }, + { + "epoch": 4.32, + "grad_norm": 2.620981454849243, + "learning_rate": 0.00012003202716255273, + "loss": 3.1379, + "step": 2951 + }, + { + "epoch": 4.32, + "grad_norm": 2501.735107421875, + "learning_rate": 0.00011998607222804181, + "loss": 3.2292, + "step": 2952 + }, + { + "epoch": 4.32, + "grad_norm": 7.561061859130859, + "learning_rate": 0.00011994011289672895, + "loss": 3.1922, + "step": 2953 + }, + { + "epoch": 4.32, + "grad_norm": 7.529016971588135, + "learning_rate": 0.00011989414917872488, + "loss": 3.1894, + "step": 2954 + }, + { + "epoch": 4.32, + "grad_norm": 2.2252986431121826, + "learning_rate": 0.00011984818108414133, + "loss": 3.1745, + "step": 2955 + }, + { + "epoch": 4.33, + "grad_norm": 6.323204040527344, + "learning_rate": 0.00011980220862309097, + "loss": 3.1978, + "step": 2956 + }, + { + "epoch": 4.33, + "grad_norm": 137.4775390625, + "learning_rate": 0.00011975623180568738, + "loss": 3.222, + "step": 2957 + }, + { + "epoch": 4.33, + "grad_norm": 9.54941177368164, + "learning_rate": 0.0001197102506420452, + "loss": 3.1733, + "step": 2958 + }, + { + "epoch": 4.33, + "grad_norm": 7.13044548034668, + "learning_rate": 0.00011966426514228, + "loss": 3.2191, + "step": 2959 + }, + { + "epoch": 4.33, + "grad_norm": 1.8224902153015137, + "learning_rate": 0.00011961827531650824, + "loss": 3.2066, + "step": 2960 + }, + { + "epoch": 4.33, + "grad_norm": 7.74959659576416, + "learning_rate": 0.00011957228117484736, + "loss": 3.2138, + "step": 2961 + }, + { + "epoch": 4.34, + "grad_norm": 8.487194061279297, + "learning_rate": 0.00011952628272741584, + "loss": 3.1951, + "step": 2962 + }, + { + "epoch": 4.34, + "grad_norm": 8.898509979248047, + "learning_rate": 0.00011948027998433296, + "loss": 3.1982, + "step": 2963 + }, + { + "epoch": 4.34, + "grad_norm": 11.637873649597168, + "learning_rate": 0.00011943427295571899, + "loss": 3.2156, + "step": 2964 + }, + { + "epoch": 4.34, + "grad_norm": 7.426854133605957, + "learning_rate": 0.00011938826165169525, + "loss": 3.2193, + "step": 2965 + }, + { + "epoch": 4.34, + "grad_norm": 12.111673355102539, + "learning_rate": 0.00011934224608238389, + "loss": 3.1818, + "step": 2966 + }, + { + "epoch": 4.34, + "grad_norm": 3.1284587383270264, + "learning_rate": 0.00011929622625790802, + "loss": 3.1583, + "step": 2967 + }, + { + "epoch": 4.34, + "grad_norm": 4.960999488830566, + "learning_rate": 0.00011925020218839167, + "loss": 3.1765, + "step": 2968 + }, + { + "epoch": 4.35, + "grad_norm": 7.320558071136475, + "learning_rate": 0.00011920417388395986, + "loss": 3.19, + "step": 2969 + }, + { + "epoch": 4.35, + "grad_norm": 4.303677082061768, + "learning_rate": 0.0001191581413547385, + "loss": 3.1875, + "step": 2970 + }, + { + "epoch": 4.35, + "grad_norm": 4.7796549797058105, + "learning_rate": 0.0001191121046108544, + "loss": 3.1843, + "step": 2971 + }, + { + "epoch": 4.35, + "grad_norm": 5.886425971984863, + "learning_rate": 0.0001190660636624354, + "loss": 3.1532, + "step": 2972 + }, + { + "epoch": 4.35, + "grad_norm": 2.2462821006774902, + "learning_rate": 0.00011902001851961017, + "loss": 3.1873, + "step": 2973 + }, + { + "epoch": 4.35, + "grad_norm": 6.365395545959473, + "learning_rate": 0.00011897396919250831, + "loss": 3.1509, + "step": 2974 + }, + { + "epoch": 4.35, + "grad_norm": 8.911288261413574, + "learning_rate": 0.0001189279156912604, + "loss": 3.1714, + "step": 2975 + }, + { + "epoch": 4.36, + "grad_norm": 5.056528091430664, + "learning_rate": 0.00011888185802599791, + "loss": 3.1708, + "step": 2976 + }, + { + "epoch": 4.36, + "grad_norm": 3.0920491218566895, + "learning_rate": 0.00011883579620685318, + "loss": 3.1315, + "step": 2977 + }, + { + "epoch": 4.36, + "grad_norm": 4.874855041503906, + "learning_rate": 0.00011878973024395952, + "loss": 3.1369, + "step": 2978 + }, + { + "epoch": 4.36, + "grad_norm": 2.600829601287842, + "learning_rate": 0.0001187436601474511, + "loss": 3.1673, + "step": 2979 + }, + { + "epoch": 4.36, + "grad_norm": 5.483844757080078, + "learning_rate": 0.00011869758592746309, + "loss": 3.1439, + "step": 2980 + }, + { + "epoch": 4.36, + "grad_norm": 7.73532772064209, + "learning_rate": 0.00011865150759413147, + "loss": 3.2182, + "step": 2981 + }, + { + "epoch": 4.36, + "grad_norm": 3.6422669887542725, + "learning_rate": 0.00011860542515759318, + "loss": 3.1551, + "step": 2982 + }, + { + "epoch": 4.37, + "grad_norm": 4.651947975158691, + "learning_rate": 0.00011855933862798605, + "loss": 3.1674, + "step": 2983 + }, + { + "epoch": 4.37, + "grad_norm": 6.519864559173584, + "learning_rate": 0.00011851324801544879, + "loss": 3.1561, + "step": 2984 + }, + { + "epoch": 4.37, + "grad_norm": 3.8000521659851074, + "learning_rate": 0.00011846715333012104, + "loss": 3.174, + "step": 2985 + }, + { + "epoch": 4.37, + "grad_norm": 4.087390899658203, + "learning_rate": 0.00011842105458214333, + "loss": 3.1715, + "step": 2986 + }, + { + "epoch": 4.37, + "grad_norm": 5.734311580657959, + "learning_rate": 0.00011837495178165706, + "loss": 3.1486, + "step": 2987 + }, + { + "epoch": 4.37, + "grad_norm": 2.4519381523132324, + "learning_rate": 0.00011832884493880452, + "loss": 3.1528, + "step": 2988 + }, + { + "epoch": 4.38, + "grad_norm": 4.925870895385742, + "learning_rate": 0.00011828273406372894, + "loss": 3.1785, + "step": 2989 + }, + { + "epoch": 4.38, + "grad_norm": 6.140622615814209, + "learning_rate": 0.0001182366191665744, + "loss": 3.207, + "step": 2990 + }, + { + "epoch": 4.38, + "grad_norm": 2.630859851837158, + "learning_rate": 0.00011819050025748583, + "loss": 3.1437, + "step": 2991 + }, + { + "epoch": 4.38, + "grad_norm": 6.897947311401367, + "learning_rate": 0.00011814437734660917, + "loss": 3.2066, + "step": 2992 + }, + { + "epoch": 4.38, + "grad_norm": 10.167567253112793, + "learning_rate": 0.00011809825044409106, + "loss": 3.1581, + "step": 2993 + }, + { + "epoch": 4.38, + "grad_norm": 13.395803451538086, + "learning_rate": 0.00011805211956007914, + "loss": 3.1348, + "step": 2994 + }, + { + "epoch": 4.38, + "grad_norm": 1.3696818351745605, + "learning_rate": 0.00011800598470472189, + "loss": 3.1391, + "step": 2995 + }, + { + "epoch": 4.39, + "grad_norm": 6.5317063331604, + "learning_rate": 0.00011795984588816869, + "loss": 3.1951, + "step": 2996 + }, + { + "epoch": 4.39, + "grad_norm": 63.888736724853516, + "learning_rate": 0.00011791370312056972, + "loss": 3.2487, + "step": 2997 + }, + { + "epoch": 4.39, + "grad_norm": 9.84652042388916, + "learning_rate": 0.00011786755641207614, + "loss": 3.1655, + "step": 2998 + }, + { + "epoch": 4.39, + "grad_norm": 7.442485809326172, + "learning_rate": 0.00011782140577283991, + "loss": 3.1921, + "step": 2999 + }, + { + "epoch": 4.39, + "grad_norm": 1.4993599653244019, + "learning_rate": 0.00011777525121301383, + "loss": 3.1366, + "step": 3000 + }, + { + "epoch": 4.39, + "grad_norm": 23.01314926147461, + "learning_rate": 0.00011772909274275159, + "loss": 3.1806, + "step": 3001 + }, + { + "epoch": 4.39, + "grad_norm": 13.442095756530762, + "learning_rate": 0.00011768293037220778, + "loss": 3.2192, + "step": 3002 + }, + { + "epoch": 4.4, + "grad_norm": 11.32374382019043, + "learning_rate": 0.0001176367641115378, + "loss": 3.1979, + "step": 3003 + }, + { + "epoch": 4.4, + "grad_norm": 4.027917385101318, + "learning_rate": 0.00011759059397089793, + "loss": 3.1963, + "step": 3004 + }, + { + "epoch": 4.4, + "grad_norm": 6.907668113708496, + "learning_rate": 0.00011754441996044523, + "loss": 3.168, + "step": 3005 + }, + { + "epoch": 4.4, + "grad_norm": 8.269980430603027, + "learning_rate": 0.00011749824209033775, + "loss": 3.1789, + "step": 3006 + }, + { + "epoch": 4.4, + "grad_norm": 6.633981227874756, + "learning_rate": 0.00011745206037073426, + "loss": 3.225, + "step": 3007 + }, + { + "epoch": 4.4, + "grad_norm": 5.4061994552612305, + "learning_rate": 0.00011740587481179442, + "loss": 3.2288, + "step": 3008 + }, + { + "epoch": 4.41, + "grad_norm": 5.786009311676025, + "learning_rate": 0.0001173596854236788, + "loss": 3.1826, + "step": 3009 + }, + { + "epoch": 4.41, + "grad_norm": 6.876216888427734, + "learning_rate": 0.00011731349221654875, + "loss": 3.1563, + "step": 3010 + }, + { + "epoch": 4.41, + "grad_norm": 6.093964099884033, + "learning_rate": 0.00011726729520056641, + "loss": 3.1879, + "step": 3011 + }, + { + "epoch": 4.41, + "grad_norm": 5.7057576179504395, + "learning_rate": 0.00011722109438589486, + "loss": 3.2176, + "step": 3012 + }, + { + "epoch": 4.41, + "grad_norm": 6.603635787963867, + "learning_rate": 0.00011717488978269795, + "loss": 3.1905, + "step": 3013 + }, + { + "epoch": 4.41, + "grad_norm": 4.3624067306518555, + "learning_rate": 0.00011712868140114037, + "loss": 3.1549, + "step": 3014 + }, + { + "epoch": 4.41, + "grad_norm": 3.682727098464966, + "learning_rate": 0.00011708246925138771, + "loss": 3.1376, + "step": 3015 + }, + { + "epoch": 4.42, + "grad_norm": 6.1669921875, + "learning_rate": 0.00011703625334360626, + "loss": 3.1513, + "step": 3016 + }, + { + "epoch": 4.42, + "grad_norm": 4.20526123046875, + "learning_rate": 0.00011699003368796328, + "loss": 3.1493, + "step": 3017 + }, + { + "epoch": 4.42, + "grad_norm": 2.534583568572998, + "learning_rate": 0.0001169438102946267, + "loss": 3.1625, + "step": 3018 + }, + { + "epoch": 4.42, + "grad_norm": 2.841315269470215, + "learning_rate": 0.00011689758317376547, + "loss": 3.2036, + "step": 3019 + }, + { + "epoch": 4.42, + "grad_norm": 14.288698196411133, + "learning_rate": 0.00011685135233554917, + "loss": 3.1499, + "step": 3020 + }, + { + "epoch": 4.42, + "grad_norm": 5.600762844085693, + "learning_rate": 0.00011680511779014827, + "loss": 3.1529, + "step": 3021 + }, + { + "epoch": 4.42, + "grad_norm": 2.945816993713379, + "learning_rate": 0.00011675887954773405, + "loss": 3.1832, + "step": 3022 + }, + { + "epoch": 4.43, + "grad_norm": 3.300783395767212, + "learning_rate": 0.0001167126376184787, + "loss": 3.16, + "step": 3023 + }, + { + "epoch": 4.43, + "grad_norm": 10.50096607208252, + "learning_rate": 0.00011666639201255506, + "loss": 3.128, + "step": 3024 + }, + { + "epoch": 4.43, + "grad_norm": 2.324991226196289, + "learning_rate": 0.00011662014274013684, + "loss": 3.1621, + "step": 3025 + }, + { + "epoch": 4.43, + "grad_norm": 3.7734105587005615, + "learning_rate": 0.00011657388981139865, + "loss": 3.1594, + "step": 3026 + }, + { + "epoch": 4.43, + "grad_norm": 3.324181079864502, + "learning_rate": 0.00011652763323651575, + "loss": 3.1495, + "step": 3027 + }, + { + "epoch": 4.43, + "grad_norm": 2.9058284759521484, + "learning_rate": 0.00011648137302566431, + "loss": 3.1342, + "step": 3028 + }, + { + "epoch": 4.43, + "grad_norm": 3.1446614265441895, + "learning_rate": 0.00011643510918902123, + "loss": 3.1408, + "step": 3029 + }, + { + "epoch": 4.44, + "grad_norm": 1.5861114263534546, + "learning_rate": 0.0001163888417367643, + "loss": 3.1833, + "step": 3030 + }, + { + "epoch": 4.44, + "grad_norm": 0.9234637022018433, + "learning_rate": 0.00011634257067907201, + "loss": 3.1692, + "step": 3031 + }, + { + "epoch": 4.44, + "grad_norm": 2.46486759185791, + "learning_rate": 0.00011629629602612368, + "loss": 3.1654, + "step": 3032 + }, + { + "epoch": 4.44, + "grad_norm": 1.57401442527771, + "learning_rate": 0.00011625001778809945, + "loss": 3.1469, + "step": 3033 + }, + { + "epoch": 4.44, + "grad_norm": 3.949265241622925, + "learning_rate": 0.00011620373597518024, + "loss": 3.2028, + "step": 3034 + }, + { + "epoch": 4.44, + "grad_norm": 4.033581256866455, + "learning_rate": 0.00011615745059754766, + "loss": 3.1827, + "step": 3035 + }, + { + "epoch": 4.45, + "grad_norm": 3.1215124130249023, + "learning_rate": 0.00011611116166538425, + "loss": 3.1908, + "step": 3036 + }, + { + "epoch": 4.45, + "grad_norm": 2.8856616020202637, + "learning_rate": 0.00011606486918887327, + "loss": 3.1631, + "step": 3037 + }, + { + "epoch": 4.45, + "grad_norm": 3.271087408065796, + "learning_rate": 0.00011601857317819871, + "loss": 3.1653, + "step": 3038 + }, + { + "epoch": 4.45, + "grad_norm": 3.887037515640259, + "learning_rate": 0.0001159722736435454, + "loss": 3.1656, + "step": 3039 + }, + { + "epoch": 4.45, + "grad_norm": 1.2171587944030762, + "learning_rate": 0.00011592597059509893, + "loss": 3.1612, + "step": 3040 + }, + { + "epoch": 4.45, + "grad_norm": 1.7865351438522339, + "learning_rate": 0.00011587966404304564, + "loss": 3.1736, + "step": 3041 + }, + { + "epoch": 4.45, + "grad_norm": 0.9533948302268982, + "learning_rate": 0.00011583335399757271, + "loss": 3.1167, + "step": 3042 + }, + { + "epoch": 4.45, + "eval_loss": 3.0038862228393555, + "eval_runtime": 1165.2502, + "eval_samples_per_second": 14.711, + "eval_steps_per_second": 0.92, + "step": 3042 + }, + { + "epoch": 4.46, + "grad_norm": 1.679763674736023, + "learning_rate": 0.000115787040468868, + "loss": 3.1827, + "step": 3043 + }, + { + "epoch": 4.46, + "grad_norm": 0.7595570683479309, + "learning_rate": 0.0001157407234671202, + "loss": 3.1615, + "step": 3044 + }, + { + "epoch": 4.46, + "grad_norm": 1.3758699893951416, + "learning_rate": 0.0001156944030025187, + "loss": 3.2312, + "step": 3045 + }, + { + "epoch": 4.46, + "grad_norm": 1.6547698974609375, + "learning_rate": 0.00011564807908525374, + "loss": 3.1552, + "step": 3046 + }, + { + "epoch": 4.46, + "grad_norm": 1.4767943620681763, + "learning_rate": 0.00011560175172551623, + "loss": 3.2091, + "step": 3047 + }, + { + "epoch": 4.46, + "grad_norm": 0.9520362615585327, + "learning_rate": 0.00011555542093349791, + "loss": 3.1592, + "step": 3048 + }, + { + "epoch": 4.46, + "grad_norm": 0.4594898223876953, + "learning_rate": 0.00011550908671939125, + "loss": 3.1426, + "step": 3049 + }, + { + "epoch": 4.47, + "grad_norm": 0.8685529232025146, + "learning_rate": 0.00011546274909338945, + "loss": 3.115, + "step": 3050 + }, + { + "epoch": 4.47, + "grad_norm": 0.986310601234436, + "learning_rate": 0.00011541640806568646, + "loss": 3.1239, + "step": 3051 + }, + { + "epoch": 4.47, + "grad_norm": 2.280745029449463, + "learning_rate": 0.000115370063646477, + "loss": 3.1833, + "step": 3052 + }, + { + "epoch": 4.47, + "grad_norm": 0.696453869342804, + "learning_rate": 0.00011532371584595656, + "loss": 3.1107, + "step": 3053 + }, + { + "epoch": 4.47, + "grad_norm": 3.4933245182037354, + "learning_rate": 0.00011527736467432133, + "loss": 3.1748, + "step": 3054 + }, + { + "epoch": 4.47, + "grad_norm": 2.052901268005371, + "learning_rate": 0.00011523101014176825, + "loss": 3.16, + "step": 3055 + }, + { + "epoch": 4.47, + "grad_norm": 3.9722251892089844, + "learning_rate": 0.00011518465225849502, + "loss": 3.1674, + "step": 3056 + }, + { + "epoch": 4.48, + "grad_norm": 3.952956438064575, + "learning_rate": 0.00011513829103470004, + "loss": 3.1506, + "step": 3057 + }, + { + "epoch": 4.48, + "grad_norm": 2.466555118560791, + "learning_rate": 0.00011509192648058249, + "loss": 3.1843, + "step": 3058 + }, + { + "epoch": 4.48, + "grad_norm": 2.9880712032318115, + "learning_rate": 0.00011504555860634225, + "loss": 3.2101, + "step": 3059 + }, + { + "epoch": 4.48, + "grad_norm": 3.0772719383239746, + "learning_rate": 0.00011499918742217996, + "loss": 3.1815, + "step": 3060 + }, + { + "epoch": 4.48, + "grad_norm": 3.1504437923431396, + "learning_rate": 0.00011495281293829697, + "loss": 3.1493, + "step": 3061 + }, + { + "epoch": 4.48, + "grad_norm": 1.549119472503662, + "learning_rate": 0.00011490643516489532, + "loss": 3.154, + "step": 3062 + }, + { + "epoch": 4.49, + "grad_norm": 5.081751346588135, + "learning_rate": 0.00011486005411217784, + "loss": 3.2329, + "step": 3063 + }, + { + "epoch": 4.49, + "grad_norm": 5.428029537200928, + "learning_rate": 0.00011481366979034806, + "loss": 3.1709, + "step": 3064 + }, + { + "epoch": 4.49, + "grad_norm": 1.1052792072296143, + "learning_rate": 0.00011476728220961015, + "loss": 3.1713, + "step": 3065 + }, + { + "epoch": 4.49, + "grad_norm": 4.415512561798096, + "learning_rate": 0.00011472089138016917, + "loss": 3.1753, + "step": 3066 + }, + { + "epoch": 4.49, + "grad_norm": 4.5699143409729, + "learning_rate": 0.00011467449731223075, + "loss": 3.209, + "step": 3067 + }, + { + "epoch": 4.49, + "grad_norm": 3.3557825088500977, + "learning_rate": 0.00011462810001600123, + "loss": 3.1721, + "step": 3068 + }, + { + "epoch": 4.49, + "grad_norm": 3.846703290939331, + "learning_rate": 0.00011458169950168774, + "loss": 3.1592, + "step": 3069 + }, + { + "epoch": 4.5, + "grad_norm": 3.0023891925811768, + "learning_rate": 0.0001145352957794981, + "loss": 3.1636, + "step": 3070 + }, + { + "epoch": 4.5, + "grad_norm": 6.419353485107422, + "learning_rate": 0.0001144888888596408, + "loss": 3.1472, + "step": 3071 + }, + { + "epoch": 4.5, + "grad_norm": 6.247819900512695, + "learning_rate": 0.00011444247875232503, + "loss": 3.1901, + "step": 3072 + }, + { + "epoch": 4.5, + "grad_norm": 4.539994239807129, + "learning_rate": 0.00011439606546776076, + "loss": 3.1452, + "step": 3073 + }, + { + "epoch": 4.5, + "grad_norm": 3.100914716720581, + "learning_rate": 0.00011434964901615852, + "loss": 3.1582, + "step": 3074 + }, + { + "epoch": 4.5, + "grad_norm": 8.092155456542969, + "learning_rate": 0.0001143032294077297, + "loss": 3.2045, + "step": 3075 + }, + { + "epoch": 4.5, + "grad_norm": 2.9788618087768555, + "learning_rate": 0.00011425680665268626, + "loss": 3.1545, + "step": 3076 + }, + { + "epoch": 4.51, + "grad_norm": 2.915862560272217, + "learning_rate": 0.00011421038076124089, + "loss": 3.1752, + "step": 3077 + }, + { + "epoch": 4.51, + "grad_norm": 3.0964643955230713, + "learning_rate": 0.000114163951743607, + "loss": 3.1225, + "step": 3078 + }, + { + "epoch": 4.51, + "grad_norm": 2.3280699253082275, + "learning_rate": 0.00011411751960999862, + "loss": 3.179, + "step": 3079 + }, + { + "epoch": 4.51, + "grad_norm": 1.9851235151290894, + "learning_rate": 0.00011407108437063057, + "loss": 3.1557, + "step": 3080 + }, + { + "epoch": 4.51, + "grad_norm": 2.6399571895599365, + "learning_rate": 0.00011402464603571825, + "loss": 3.131, + "step": 3081 + }, + { + "epoch": 4.51, + "grad_norm": 10.382826805114746, + "learning_rate": 0.00011397820461547777, + "loss": 3.1541, + "step": 3082 + }, + { + "epoch": 4.51, + "grad_norm": 1.8169869184494019, + "learning_rate": 0.00011393176012012596, + "loss": 3.2169, + "step": 3083 + }, + { + "epoch": 4.52, + "grad_norm": 3.3234314918518066, + "learning_rate": 0.00011388531255988033, + "loss": 3.1631, + "step": 3084 + }, + { + "epoch": 4.52, + "grad_norm": 1.7650176286697388, + "learning_rate": 0.00011383886194495896, + "loss": 3.1561, + "step": 3085 + }, + { + "epoch": 4.52, + "grad_norm": 5.759214878082275, + "learning_rate": 0.00011379240828558072, + "loss": 3.1424, + "step": 3086 + }, + { + "epoch": 4.52, + "grad_norm": 5.468089580535889, + "learning_rate": 0.00011374595159196506, + "loss": 3.1697, + "step": 3087 + }, + { + "epoch": 4.52, + "grad_norm": 25.312021255493164, + "learning_rate": 0.00011369949187433218, + "loss": 3.1632, + "step": 3088 + }, + { + "epoch": 4.52, + "grad_norm": 3.2354679107666016, + "learning_rate": 0.0001136530291429029, + "loss": 3.166, + "step": 3089 + }, + { + "epoch": 4.53, + "grad_norm": 11.292868614196777, + "learning_rate": 0.00011360656340789873, + "loss": 3.1495, + "step": 3090 + }, + { + "epoch": 4.53, + "grad_norm": 9.209776878356934, + "learning_rate": 0.00011356009467954176, + "loss": 3.189, + "step": 3091 + }, + { + "epoch": 4.53, + "grad_norm": 40.07894515991211, + "learning_rate": 0.00011351362296805485, + "loss": 3.1709, + "step": 3092 + }, + { + "epoch": 4.53, + "grad_norm": 4.195234298706055, + "learning_rate": 0.00011346714828366148, + "loss": 3.2041, + "step": 3093 + }, + { + "epoch": 4.53, + "grad_norm": 154.33792114257812, + "learning_rate": 0.0001134206706365857, + "loss": 3.1673, + "step": 3094 + }, + { + "epoch": 4.53, + "grad_norm": 3.392430305480957, + "learning_rate": 0.00011337419003705237, + "loss": 3.1747, + "step": 3095 + }, + { + "epoch": 4.53, + "grad_norm": 156.2117462158203, + "learning_rate": 0.00011332770649528681, + "loss": 3.1426, + "step": 3096 + }, + { + "epoch": 4.54, + "grad_norm": 3.3672966957092285, + "learning_rate": 0.00011328122002151516, + "loss": 3.1428, + "step": 3097 + }, + { + "epoch": 4.54, + "grad_norm": 1.9393489360809326, + "learning_rate": 0.00011323473062596413, + "loss": 3.1491, + "step": 3098 + }, + { + "epoch": 4.54, + "grad_norm": 4.762568473815918, + "learning_rate": 0.00011318823831886103, + "loss": 3.1408, + "step": 3099 + }, + { + "epoch": 4.54, + "grad_norm": 4.439011573791504, + "learning_rate": 0.00011314174311043391, + "loss": 3.149, + "step": 3100 + }, + { + "epoch": 4.54, + "grad_norm": 1099.63525390625, + "learning_rate": 0.00011309524501091135, + "loss": 3.1883, + "step": 3101 + }, + { + "epoch": 4.54, + "grad_norm": 14.831306457519531, + "learning_rate": 0.00011304874403052266, + "loss": 3.2264, + "step": 3102 + }, + { + "epoch": 4.54, + "grad_norm": 1.7882112264633179, + "learning_rate": 0.00011300224017949772, + "loss": 3.1552, + "step": 3103 + }, + { + "epoch": 4.55, + "grad_norm": 5.061419486999512, + "learning_rate": 0.00011295573346806709, + "loss": 3.1629, + "step": 3104 + }, + { + "epoch": 4.55, + "grad_norm": 2.1199474334716797, + "learning_rate": 0.00011290922390646192, + "loss": 3.1672, + "step": 3105 + }, + { + "epoch": 4.55, + "grad_norm": 2.093383312225342, + "learning_rate": 0.000112862711504914, + "loss": 3.1872, + "step": 3106 + }, + { + "epoch": 4.55, + "grad_norm": 1.5643643140792847, + "learning_rate": 0.00011281619627365576, + "loss": 3.2203, + "step": 3107 + }, + { + "epoch": 4.55, + "grad_norm": 3.1214048862457275, + "learning_rate": 0.00011276967822292024, + "loss": 3.1508, + "step": 3108 + }, + { + "epoch": 4.55, + "grad_norm": 17.56413459777832, + "learning_rate": 0.00011272315736294108, + "loss": 3.1855, + "step": 3109 + }, + { + "epoch": 4.56, + "grad_norm": 3.3824844360351562, + "learning_rate": 0.00011267663370395256, + "loss": 3.1449, + "step": 3110 + }, + { + "epoch": 4.56, + "grad_norm": 2.0639045238494873, + "learning_rate": 0.00011263010725618959, + "loss": 3.1227, + "step": 3111 + }, + { + "epoch": 4.56, + "grad_norm": 2.0494930744171143, + "learning_rate": 0.00011258357802988766, + "loss": 3.1386, + "step": 3112 + }, + { + "epoch": 4.56, + "grad_norm": 2.3310718536376953, + "learning_rate": 0.00011253704603528288, + "loss": 3.1649, + "step": 3113 + }, + { + "epoch": 4.56, + "grad_norm": 1.5938396453857422, + "learning_rate": 0.00011249051128261203, + "loss": 3.1524, + "step": 3114 + }, + { + "epoch": 4.56, + "grad_norm": 1.9962538480758667, + "learning_rate": 0.00011244397378211238, + "loss": 3.1949, + "step": 3115 + }, + { + "epoch": 4.56, + "grad_norm": 2.338400363922119, + "learning_rate": 0.00011239743354402188, + "loss": 3.1258, + "step": 3116 + }, + { + "epoch": 4.57, + "grad_norm": 1.9446300268173218, + "learning_rate": 0.00011235089057857911, + "loss": 3.1829, + "step": 3117 + }, + { + "epoch": 4.57, + "grad_norm": 2.5406713485717773, + "learning_rate": 0.00011230434489602317, + "loss": 3.1782, + "step": 3118 + }, + { + "epoch": 4.57, + "grad_norm": 1.6750293970108032, + "learning_rate": 0.00011225779650659383, + "loss": 3.139, + "step": 3119 + }, + { + "epoch": 4.57, + "grad_norm": 10.995355606079102, + "learning_rate": 0.00011221124542053139, + "loss": 3.152, + "step": 3120 + }, + { + "epoch": 4.57, + "grad_norm": 3.1163713932037354, + "learning_rate": 0.00011216469164807679, + "loss": 3.1542, + "step": 3121 + }, + { + "epoch": 4.57, + "grad_norm": 1.6949200630187988, + "learning_rate": 0.00011211813519947159, + "loss": 3.1438, + "step": 3122 + }, + { + "epoch": 4.57, + "grad_norm": 3.045480251312256, + "learning_rate": 0.00011207157608495779, + "loss": 3.1611, + "step": 3123 + }, + { + "epoch": 4.58, + "grad_norm": 2.5979998111724854, + "learning_rate": 0.0001120250143147782, + "loss": 3.1628, + "step": 3124 + }, + { + "epoch": 4.58, + "grad_norm": 2.419813632965088, + "learning_rate": 0.00011197844989917605, + "loss": 3.1876, + "step": 3125 + }, + { + "epoch": 4.58, + "grad_norm": 2.4058735370635986, + "learning_rate": 0.00011193188284839517, + "loss": 3.1617, + "step": 3126 + }, + { + "epoch": 4.58, + "grad_norm": 1.7974401712417603, + "learning_rate": 0.00011188531317268005, + "loss": 3.1276, + "step": 3127 + }, + { + "epoch": 4.58, + "grad_norm": 2.3527932167053223, + "learning_rate": 0.0001118387408822757, + "loss": 3.1843, + "step": 3128 + }, + { + "epoch": 4.58, + "grad_norm": 2.5821757316589355, + "learning_rate": 0.00011179216598742768, + "loss": 3.1492, + "step": 3129 + }, + { + "epoch": 4.58, + "grad_norm": 0.7473565936088562, + "learning_rate": 0.00011174558849838213, + "loss": 3.1384, + "step": 3130 + }, + { + "epoch": 4.59, + "grad_norm": 2.998114585876465, + "learning_rate": 0.00011169900842538588, + "loss": 3.173, + "step": 3131 + }, + { + "epoch": 4.59, + "grad_norm": 8.924659729003906, + "learning_rate": 0.00011165242577868616, + "loss": 3.1929, + "step": 3132 + }, + { + "epoch": 4.59, + "grad_norm": 13.762849807739258, + "learning_rate": 0.00011160584056853082, + "loss": 3.165, + "step": 3133 + }, + { + "epoch": 4.59, + "grad_norm": 2.115877151489258, + "learning_rate": 0.00011155925280516841, + "loss": 3.119, + "step": 3134 + }, + { + "epoch": 4.59, + "grad_norm": 3.3011577129364014, + "learning_rate": 0.00011151266249884783, + "loss": 3.1905, + "step": 3135 + }, + { + "epoch": 4.59, + "grad_norm": 26.584270477294922, + "learning_rate": 0.00011146606965981862, + "loss": 3.2361, + "step": 3136 + }, + { + "epoch": 4.6, + "grad_norm": 1.8690835237503052, + "learning_rate": 0.00011141947429833096, + "loss": 3.1527, + "step": 3137 + }, + { + "epoch": 4.6, + "grad_norm": 3.516594886779785, + "learning_rate": 0.00011137287642463545, + "loss": 3.1821, + "step": 3138 + }, + { + "epoch": 4.6, + "grad_norm": 2.7472705841064453, + "learning_rate": 0.00011132627604898338, + "loss": 3.1594, + "step": 3139 + }, + { + "epoch": 4.6, + "grad_norm": 1.927894115447998, + "learning_rate": 0.00011127967318162644, + "loss": 3.1446, + "step": 3140 + }, + { + "epoch": 4.6, + "grad_norm": 1.823498249053955, + "learning_rate": 0.00011123306783281701, + "loss": 3.1621, + "step": 3141 + }, + { + "epoch": 4.6, + "grad_norm": 2.379281759262085, + "learning_rate": 0.00011118646001280799, + "loss": 3.1441, + "step": 3142 + }, + { + "epoch": 4.6, + "grad_norm": 1.6748161315917969, + "learning_rate": 0.00011113984973185268, + "loss": 3.1687, + "step": 3143 + }, + { + "epoch": 4.61, + "grad_norm": 4.564174652099609, + "learning_rate": 0.00011109323700020508, + "loss": 3.1976, + "step": 3144 + }, + { + "epoch": 4.61, + "grad_norm": 4.5107550621032715, + "learning_rate": 0.00011104662182811974, + "loss": 3.1711, + "step": 3145 + }, + { + "epoch": 4.61, + "grad_norm": 1.1205596923828125, + "learning_rate": 0.00011100000422585158, + "loss": 3.1573, + "step": 3146 + }, + { + "epoch": 4.61, + "grad_norm": 2.339171886444092, + "learning_rate": 0.00011095338420365621, + "loss": 3.1498, + "step": 3147 + }, + { + "epoch": 4.61, + "grad_norm": 2.139634609222412, + "learning_rate": 0.00011090676177178973, + "loss": 3.1493, + "step": 3148 + }, + { + "epoch": 4.61, + "grad_norm": 0.9772034287452698, + "learning_rate": 0.00011086013694050876, + "loss": 3.1551, + "step": 3149 + }, + { + "epoch": 4.61, + "grad_norm": 3.0146749019622803, + "learning_rate": 0.00011081350972007043, + "loss": 3.1186, + "step": 3150 + }, + { + "epoch": 4.62, + "grad_norm": 3.702209711074829, + "learning_rate": 0.00011076688012073244, + "loss": 3.2039, + "step": 3151 + }, + { + "epoch": 4.62, + "grad_norm": 6.443127155303955, + "learning_rate": 0.000110720248152753, + "loss": 3.1794, + "step": 3152 + }, + { + "epoch": 4.62, + "grad_norm": 3.7430996894836426, + "learning_rate": 0.00011067361382639079, + "loss": 3.172, + "step": 3153 + }, + { + "epoch": 4.62, + "grad_norm": 0.7498925924301147, + "learning_rate": 0.00011062697715190507, + "loss": 3.1905, + "step": 3154 + }, + { + "epoch": 4.62, + "grad_norm": 5.21573543548584, + "learning_rate": 0.00011058033813955557, + "loss": 3.1483, + "step": 3155 + }, + { + "epoch": 4.62, + "grad_norm": 5.235692977905273, + "learning_rate": 0.00011053369679960264, + "loss": 3.1197, + "step": 3156 + }, + { + "epoch": 4.62, + "grad_norm": 2.265049934387207, + "learning_rate": 0.00011048705314230696, + "loss": 3.1359, + "step": 3157 + }, + { + "epoch": 4.63, + "grad_norm": 6.700308799743652, + "learning_rate": 0.00011044040717792989, + "loss": 3.141, + "step": 3158 + }, + { + "epoch": 4.63, + "grad_norm": 7.944544792175293, + "learning_rate": 0.00011039375891673322, + "loss": 3.2029, + "step": 3159 + }, + { + "epoch": 4.63, + "grad_norm": 3.51708984375, + "learning_rate": 0.00011034710836897921, + "loss": 3.1747, + "step": 3160 + }, + { + "epoch": 4.63, + "grad_norm": 4.683743000030518, + "learning_rate": 0.00011030045554493075, + "loss": 3.181, + "step": 3161 + }, + { + "epoch": 4.63, + "grad_norm": 7.184266090393066, + "learning_rate": 0.00011025380045485107, + "loss": 3.1523, + "step": 3162 + }, + { + "epoch": 4.63, + "grad_norm": 5.593520164489746, + "learning_rate": 0.00011020714310900401, + "loss": 3.1564, + "step": 3163 + }, + { + "epoch": 4.64, + "grad_norm": 1.7525691986083984, + "learning_rate": 0.0001101604835176539, + "loss": 3.1461, + "step": 3164 + }, + { + "epoch": 4.64, + "grad_norm": 6.076540470123291, + "learning_rate": 0.00011011382169106546, + "loss": 3.1734, + "step": 3165 + }, + { + "epoch": 4.64, + "grad_norm": 7.846095561981201, + "learning_rate": 0.00011006715763950406, + "loss": 3.1622, + "step": 3166 + }, + { + "epoch": 4.64, + "grad_norm": 4.10769510269165, + "learning_rate": 0.00011002049137323543, + "loss": 3.1823, + "step": 3167 + }, + { + "epoch": 4.64, + "grad_norm": 4.750815391540527, + "learning_rate": 0.00010997382290252585, + "loss": 3.1425, + "step": 3168 + }, + { + "epoch": 4.64, + "grad_norm": 4.68991231918335, + "learning_rate": 0.0001099271522376421, + "loss": 3.1231, + "step": 3169 + }, + { + "epoch": 4.64, + "grad_norm": 3.5092368125915527, + "learning_rate": 0.00010988047938885135, + "loss": 3.2153, + "step": 3170 + }, + { + "epoch": 4.65, + "grad_norm": 4.325351715087891, + "learning_rate": 0.00010983380436642136, + "loss": 3.1567, + "step": 3171 + }, + { + "epoch": 4.65, + "grad_norm": 3.288820505142212, + "learning_rate": 0.00010978712718062036, + "loss": 3.1142, + "step": 3172 + }, + { + "epoch": 4.65, + "grad_norm": 3.504354238510132, + "learning_rate": 0.00010974044784171693, + "loss": 3.143, + "step": 3173 + }, + { + "epoch": 4.65, + "grad_norm": 4.500937461853027, + "learning_rate": 0.00010969376635998023, + "loss": 3.1615, + "step": 3174 + }, + { + "epoch": 4.65, + "grad_norm": 39.75580596923828, + "learning_rate": 0.00010964708274567994, + "loss": 3.2307, + "step": 3175 + }, + { + "epoch": 4.65, + "grad_norm": 3.0470621585845947, + "learning_rate": 0.00010960039700908608, + "loss": 3.1374, + "step": 3176 + }, + { + "epoch": 4.65, + "grad_norm": 2.1173510551452637, + "learning_rate": 0.00010955370916046918, + "loss": 3.1752, + "step": 3177 + }, + { + "epoch": 4.66, + "grad_norm": 2.3193938732147217, + "learning_rate": 0.00010950701921010036, + "loss": 3.1474, + "step": 3178 + }, + { + "epoch": 4.66, + "grad_norm": 2.5212008953094482, + "learning_rate": 0.00010946032716825099, + "loss": 3.2541, + "step": 3179 + }, + { + "epoch": 4.66, + "grad_norm": 1.1363469362258911, + "learning_rate": 0.00010941363304519305, + "loss": 3.1424, + "step": 3180 + }, + { + "epoch": 4.66, + "grad_norm": 2.563216209411621, + "learning_rate": 0.00010936693685119892, + "loss": 3.1572, + "step": 3181 + }, + { + "epoch": 4.66, + "grad_norm": 1.520682454109192, + "learning_rate": 0.00010932023859654145, + "loss": 3.1507, + "step": 3182 + }, + { + "epoch": 4.66, + "grad_norm": 3.5463900566101074, + "learning_rate": 0.00010927353829149398, + "loss": 3.1672, + "step": 3183 + }, + { + "epoch": 4.66, + "grad_norm": 4.035703659057617, + "learning_rate": 0.00010922683594633021, + "loss": 3.2053, + "step": 3184 + }, + { + "epoch": 4.67, + "grad_norm": 3.0087568759918213, + "learning_rate": 0.00010918013157132436, + "loss": 3.1566, + "step": 3185 + }, + { + "epoch": 4.67, + "grad_norm": 2.657742500305176, + "learning_rate": 0.00010913342517675113, + "loss": 3.1382, + "step": 3186 + }, + { + "epoch": 4.67, + "grad_norm": 4.097416877746582, + "learning_rate": 0.00010908671677288552, + "loss": 3.1695, + "step": 3187 + }, + { + "epoch": 4.67, + "grad_norm": 3.2829132080078125, + "learning_rate": 0.00010904000637000313, + "loss": 3.2016, + "step": 3188 + }, + { + "epoch": 4.67, + "grad_norm": 0.8120042681694031, + "learning_rate": 0.00010899329397837996, + "loss": 3.1287, + "step": 3189 + }, + { + "epoch": 4.67, + "grad_norm": 1.4076851606369019, + "learning_rate": 0.00010894657960829235, + "loss": 3.1302, + "step": 3190 + }, + { + "epoch": 4.68, + "grad_norm": 2.486541986465454, + "learning_rate": 0.00010889986327001716, + "loss": 3.1751, + "step": 3191 + }, + { + "epoch": 4.68, + "grad_norm": 1.665196418762207, + "learning_rate": 0.00010885314497383175, + "loss": 3.1353, + "step": 3192 + }, + { + "epoch": 4.68, + "grad_norm": 3.312460422515869, + "learning_rate": 0.00010880642473001375, + "loss": 3.1851, + "step": 3193 + }, + { + "epoch": 4.68, + "grad_norm": 6.796298027038574, + "learning_rate": 0.0001087597025488413, + "loss": 3.1513, + "step": 3194 + }, + { + "epoch": 4.68, + "grad_norm": 1.5955018997192383, + "learning_rate": 0.00010871297844059304, + "loss": 3.1609, + "step": 3195 + }, + { + "epoch": 4.68, + "grad_norm": 4.358046531677246, + "learning_rate": 0.00010866625241554789, + "loss": 3.1224, + "step": 3196 + }, + { + "epoch": 4.68, + "grad_norm": 4.621119976043701, + "learning_rate": 0.00010861952448398529, + "loss": 3.1221, + "step": 3197 + }, + { + "epoch": 4.69, + "grad_norm": 1.6193621158599854, + "learning_rate": 0.00010857279465618508, + "loss": 3.1495, + "step": 3198 + }, + { + "epoch": 4.69, + "grad_norm": 3.5418941974639893, + "learning_rate": 0.00010852606294242749, + "loss": 3.182, + "step": 3199 + }, + { + "epoch": 4.69, + "grad_norm": 3.8759021759033203, + "learning_rate": 0.00010847932935299319, + "loss": 3.1323, + "step": 3200 + }, + { + "epoch": 4.69, + "grad_norm": 1.662946343421936, + "learning_rate": 0.00010843259389816325, + "loss": 3.1605, + "step": 3201 + }, + { + "epoch": 4.69, + "grad_norm": 3.667335271835327, + "learning_rate": 0.00010838585658821918, + "loss": 3.1713, + "step": 3202 + }, + { + "epoch": 4.69, + "grad_norm": 4.988526344299316, + "learning_rate": 0.00010833911743344287, + "loss": 3.1812, + "step": 3203 + }, + { + "epoch": 4.69, + "grad_norm": 3.444049596786499, + "learning_rate": 0.00010829237644411659, + "loss": 3.1978, + "step": 3204 + }, + { + "epoch": 4.7, + "grad_norm": 2.074106216430664, + "learning_rate": 0.00010824563363052307, + "loss": 3.1567, + "step": 3205 + }, + { + "epoch": 4.7, + "grad_norm": 5.682570934295654, + "learning_rate": 0.00010819888900294542, + "loss": 3.1777, + "step": 3206 + }, + { + "epoch": 4.7, + "grad_norm": 7.264477729797363, + "learning_rate": 0.00010815214257166712, + "loss": 3.2065, + "step": 3207 + }, + { + "epoch": 4.7, + "grad_norm": 4.073165416717529, + "learning_rate": 0.0001081053943469721, + "loss": 3.1564, + "step": 3208 + }, + { + "epoch": 4.7, + "grad_norm": 3.7732391357421875, + "learning_rate": 0.00010805864433914465, + "loss": 3.1654, + "step": 3209 + }, + { + "epoch": 4.7, + "grad_norm": 4.149001598358154, + "learning_rate": 0.00010801189255846944, + "loss": 3.1563, + "step": 3210 + }, + { + "epoch": 4.71, + "grad_norm": 4.86202335357666, + "learning_rate": 0.00010796513901523155, + "loss": 3.1624, + "step": 3211 + }, + { + "epoch": 4.71, + "eval_loss": 3.0107593536376953, + "eval_runtime": 1169.2042, + "eval_samples_per_second": 14.661, + "eval_steps_per_second": 0.917, + "step": 3211 + }, + { + "epoch": 4.71, + "grad_norm": 4.124433994293213, + "learning_rate": 0.00010791838371971648, + "loss": 3.1329, + "step": 3212 + }, + { + "epoch": 4.71, + "grad_norm": 1.8730932474136353, + "learning_rate": 0.00010787162668221004, + "loss": 3.1344, + "step": 3213 + }, + { + "epoch": 4.71, + "grad_norm": 3.890324831008911, + "learning_rate": 0.00010782486791299849, + "loss": 3.1527, + "step": 3214 + }, + { + "epoch": 4.71, + "grad_norm": 3.587141513824463, + "learning_rate": 0.00010777810742236844, + "loss": 3.1435, + "step": 3215 + }, + { + "epoch": 4.71, + "grad_norm": 1.505163311958313, + "learning_rate": 0.00010773134522060688, + "loss": 3.1474, + "step": 3216 + }, + { + "epoch": 4.71, + "grad_norm": 3.0347487926483154, + "learning_rate": 0.00010768458131800119, + "loss": 3.1363, + "step": 3217 + }, + { + "epoch": 4.72, + "grad_norm": 3.127628803253174, + "learning_rate": 0.0001076378157248391, + "loss": 3.118, + "step": 3218 + }, + { + "epoch": 4.72, + "grad_norm": 2.165717363357544, + "learning_rate": 0.00010759104845140877, + "loss": 3.1966, + "step": 3219 + }, + { + "epoch": 4.72, + "grad_norm": 2.3362936973571777, + "learning_rate": 0.00010754427950799861, + "loss": 3.1494, + "step": 3220 + }, + { + "epoch": 4.72, + "grad_norm": 2.43580961227417, + "learning_rate": 0.00010749750890489754, + "loss": 3.1281, + "step": 3221 + }, + { + "epoch": 4.72, + "grad_norm": 23.91058349609375, + "learning_rate": 0.00010745073665239476, + "loss": 3.1482, + "step": 3222 + }, + { + "epoch": 4.72, + "grad_norm": 3.7118659019470215, + "learning_rate": 0.00010740396276077985, + "loss": 3.1884, + "step": 3223 + }, + { + "epoch": 4.72, + "grad_norm": 3.4350130558013916, + "learning_rate": 0.00010735718724034273, + "loss": 3.1709, + "step": 3224 + }, + { + "epoch": 4.73, + "grad_norm": 0.7842398285865784, + "learning_rate": 0.00010731041010137375, + "loss": 3.162, + "step": 3225 + }, + { + "epoch": 4.73, + "grad_norm": 4.245161533355713, + "learning_rate": 0.00010726363135416351, + "loss": 3.1882, + "step": 3226 + }, + { + "epoch": 4.73, + "grad_norm": 6.0483622550964355, + "learning_rate": 0.00010721685100900308, + "loss": 3.1815, + "step": 3227 + }, + { + "epoch": 4.73, + "grad_norm": 4.542155742645264, + "learning_rate": 0.00010717006907618377, + "loss": 3.1413, + "step": 3228 + }, + { + "epoch": 4.73, + "grad_norm": 1.648128867149353, + "learning_rate": 0.00010712328556599733, + "loss": 3.1296, + "step": 3229 + }, + { + "epoch": 4.73, + "grad_norm": 5.657898902893066, + "learning_rate": 0.00010707650048873579, + "loss": 3.178, + "step": 3230 + }, + { + "epoch": 4.73, + "grad_norm": 5.276024341583252, + "learning_rate": 0.00010702971385469158, + "loss": 3.192, + "step": 3231 + }, + { + "epoch": 4.74, + "grad_norm": 4.145871639251709, + "learning_rate": 0.00010698292567415742, + "loss": 3.134, + "step": 3232 + }, + { + "epoch": 4.74, + "grad_norm": 12.922083854675293, + "learning_rate": 0.00010693613595742644, + "loss": 3.1523, + "step": 3233 + }, + { + "epoch": 4.74, + "grad_norm": 3.349857807159424, + "learning_rate": 0.00010688934471479199, + "loss": 3.1872, + "step": 3234 + }, + { + "epoch": 4.74, + "grad_norm": 2.6229217052459717, + "learning_rate": 0.00010684255195654788, + "loss": 3.1266, + "step": 3235 + }, + { + "epoch": 4.74, + "grad_norm": 3.148838520050049, + "learning_rate": 0.00010679575769298824, + "loss": 3.2267, + "step": 3236 + }, + { + "epoch": 4.74, + "grad_norm": 3.843184471130371, + "learning_rate": 0.00010674896193440742, + "loss": 3.1531, + "step": 3237 + }, + { + "epoch": 4.75, + "grad_norm": 2.107234239578247, + "learning_rate": 0.00010670216469110019, + "loss": 3.1549, + "step": 3238 + }, + { + "epoch": 4.75, + "grad_norm": 1.9575111865997314, + "learning_rate": 0.00010665536597336167, + "loss": 3.1334, + "step": 3239 + }, + { + "epoch": 4.75, + "grad_norm": 2.4358248710632324, + "learning_rate": 0.00010660856579148724, + "loss": 3.1869, + "step": 3240 + }, + { + "epoch": 4.75, + "grad_norm": 1.9233509302139282, + "learning_rate": 0.00010656176415577262, + "loss": 3.1206, + "step": 3241 + }, + { + "epoch": 4.75, + "grad_norm": 2.0048117637634277, + "learning_rate": 0.00010651496107651384, + "loss": 3.1099, + "step": 3242 + }, + { + "epoch": 4.75, + "grad_norm": 4.973776817321777, + "learning_rate": 0.00010646815656400731, + "loss": 3.1621, + "step": 3243 + }, + { + "epoch": 4.75, + "grad_norm": 3.1254444122314453, + "learning_rate": 0.00010642135062854965, + "loss": 3.1564, + "step": 3244 + }, + { + "epoch": 4.76, + "grad_norm": 2.404599666595459, + "learning_rate": 0.00010637454328043793, + "loss": 3.1772, + "step": 3245 + }, + { + "epoch": 4.76, + "grad_norm": 2.372962474822998, + "learning_rate": 0.00010632773452996936, + "loss": 3.1053, + "step": 3246 + }, + { + "epoch": 4.76, + "grad_norm": 1.7531036138534546, + "learning_rate": 0.00010628092438744167, + "loss": 3.1678, + "step": 3247 + }, + { + "epoch": 4.76, + "grad_norm": 15.43344783782959, + "learning_rate": 0.00010623411286315261, + "loss": 3.1616, + "step": 3248 + }, + { + "epoch": 4.76, + "grad_norm": 3.320650577545166, + "learning_rate": 0.00010618729996740056, + "loss": 3.2037, + "step": 3249 + }, + { + "epoch": 4.76, + "grad_norm": 14.557812690734863, + "learning_rate": 0.00010614048571048401, + "loss": 3.221, + "step": 3250 + }, + { + "epoch": 4.76, + "grad_norm": 1.5140577554702759, + "learning_rate": 0.00010609367010270168, + "loss": 3.1666, + "step": 3251 + }, + { + "epoch": 4.77, + "grad_norm": 2.14962100982666, + "learning_rate": 0.00010604685315435281, + "loss": 3.1691, + "step": 3252 + }, + { + "epoch": 4.77, + "grad_norm": 0.9101262092590332, + "learning_rate": 0.0001060000348757368, + "loss": 3.1835, + "step": 3253 + }, + { + "epoch": 4.77, + "grad_norm": 1.9691673517227173, + "learning_rate": 0.00010595321527715327, + "loss": 3.1425, + "step": 3254 + }, + { + "epoch": 4.77, + "grad_norm": 6.10009241104126, + "learning_rate": 0.0001059063943689023, + "loss": 3.163, + "step": 3255 + }, + { + "epoch": 4.77, + "grad_norm": 1.81471848487854, + "learning_rate": 0.00010585957216128415, + "loss": 3.1742, + "step": 3256 + }, + { + "epoch": 4.77, + "grad_norm": 1.8879374265670776, + "learning_rate": 0.0001058127486645994, + "loss": 3.1732, + "step": 3257 + }, + { + "epoch": 4.77, + "grad_norm": 0.9328217506408691, + "learning_rate": 0.0001057659238891489, + "loss": 3.1223, + "step": 3258 + }, + { + "epoch": 4.78, + "grad_norm": 1.236288070678711, + "learning_rate": 0.00010571909784523377, + "loss": 3.114, + "step": 3259 + }, + { + "epoch": 4.78, + "grad_norm": 2.6463773250579834, + "learning_rate": 0.00010567227054315545, + "loss": 3.1444, + "step": 3260 + }, + { + "epoch": 4.78, + "grad_norm": 1.5447609424591064, + "learning_rate": 0.00010562544199321562, + "loss": 3.1348, + "step": 3261 + }, + { + "epoch": 4.78, + "grad_norm": 3.4742603302001953, + "learning_rate": 0.00010557861220571625, + "loss": 3.1773, + "step": 3262 + }, + { + "epoch": 4.78, + "grad_norm": 2.8108527660369873, + "learning_rate": 0.0001055317811909596, + "loss": 3.1871, + "step": 3263 + }, + { + "epoch": 4.78, + "grad_norm": 2.6389920711517334, + "learning_rate": 0.00010548494895924816, + "loss": 3.1609, + "step": 3264 + }, + { + "epoch": 4.79, + "grad_norm": 2.764688491821289, + "learning_rate": 0.00010543811552088468, + "loss": 3.1937, + "step": 3265 + }, + { + "epoch": 4.79, + "grad_norm": 3.0502400398254395, + "learning_rate": 0.00010539128088617225, + "loss": 3.1669, + "step": 3266 + }, + { + "epoch": 4.79, + "grad_norm": 6.166628837585449, + "learning_rate": 0.00010534444506541417, + "loss": 3.142, + "step": 3267 + }, + { + "epoch": 4.79, + "grad_norm": 47.169883728027344, + "learning_rate": 0.00010529760806891395, + "loss": 3.1328, + "step": 3268 + }, + { + "epoch": 4.79, + "grad_norm": 3.0846803188323975, + "learning_rate": 0.0001052507699069755, + "loss": 3.1455, + "step": 3269 + }, + { + "epoch": 4.79, + "grad_norm": 2.2564151287078857, + "learning_rate": 0.00010520393058990285, + "loss": 3.1306, + "step": 3270 + }, + { + "epoch": 4.79, + "grad_norm": 1.4548720121383667, + "learning_rate": 0.00010515709012800033, + "loss": 3.1312, + "step": 3271 + }, + { + "epoch": 4.8, + "grad_norm": 1.8430453538894653, + "learning_rate": 0.00010511024853157255, + "loss": 3.1276, + "step": 3272 + }, + { + "epoch": 4.8, + "grad_norm": 1.2610652446746826, + "learning_rate": 0.00010506340581092433, + "loss": 3.1682, + "step": 3273 + }, + { + "epoch": 4.8, + "grad_norm": 0.8417081832885742, + "learning_rate": 0.00010501656197636078, + "loss": 3.1505, + "step": 3274 + }, + { + "epoch": 4.8, + "grad_norm": 1.3054739236831665, + "learning_rate": 0.00010496971703818721, + "loss": 3.1465, + "step": 3275 + }, + { + "epoch": 4.8, + "grad_norm": 2.4543936252593994, + "learning_rate": 0.00010492287100670919, + "loss": 3.2098, + "step": 3276 + }, + { + "epoch": 4.8, + "grad_norm": 1.5280994176864624, + "learning_rate": 0.00010487602389223257, + "loss": 3.1443, + "step": 3277 + }, + { + "epoch": 4.8, + "grad_norm": 1.9430744647979736, + "learning_rate": 0.00010482917570506334, + "loss": 3.1531, + "step": 3278 + }, + { + "epoch": 4.81, + "grad_norm": 1.8071426153182983, + "learning_rate": 0.00010478232645550782, + "loss": 3.171, + "step": 3279 + }, + { + "epoch": 4.81, + "grad_norm": 3.768815517425537, + "learning_rate": 0.00010473547615387256, + "loss": 3.1586, + "step": 3280 + }, + { + "epoch": 4.81, + "grad_norm": 2.8686270713806152, + "learning_rate": 0.00010468862481046427, + "loss": 3.112, + "step": 3281 + }, + { + "epoch": 4.81, + "grad_norm": 2.201878070831299, + "learning_rate": 0.0001046417724355899, + "loss": 3.1815, + "step": 3282 + }, + { + "epoch": 4.81, + "grad_norm": 1.5182592868804932, + "learning_rate": 0.00010459491903955676, + "loss": 3.1391, + "step": 3283 + }, + { + "epoch": 4.81, + "grad_norm": 23.839717864990234, + "learning_rate": 0.0001045480646326722, + "loss": 3.1331, + "step": 3284 + }, + { + "epoch": 4.81, + "grad_norm": 1.8520891666412354, + "learning_rate": 0.00010450120922524387, + "loss": 3.1622, + "step": 3285 + }, + { + "epoch": 4.82, + "grad_norm": 0.8244103789329529, + "learning_rate": 0.00010445435282757972, + "loss": 3.1308, + "step": 3286 + }, + { + "epoch": 4.82, + "grad_norm": 2.278554677963257, + "learning_rate": 0.00010440749544998778, + "loss": 3.1283, + "step": 3287 + }, + { + "epoch": 4.82, + "grad_norm": 3.664557933807373, + "learning_rate": 0.00010436063710277639, + "loss": 3.1424, + "step": 3288 + }, + { + "epoch": 4.82, + "grad_norm": 1.274968147277832, + "learning_rate": 0.00010431377779625407, + "loss": 3.1358, + "step": 3289 + }, + { + "epoch": 4.82, + "grad_norm": 0.9037652611732483, + "learning_rate": 0.00010426691754072952, + "loss": 3.1329, + "step": 3290 + }, + { + "epoch": 4.82, + "grad_norm": 2.134868860244751, + "learning_rate": 0.00010422005634651173, + "loss": 3.0978, + "step": 3291 + }, + { + "epoch": 4.83, + "grad_norm": 4.81277322769165, + "learning_rate": 0.00010417319422390979, + "loss": 3.118, + "step": 3292 + }, + { + "epoch": 4.83, + "grad_norm": 1.5502432584762573, + "learning_rate": 0.00010412633118323311, + "loss": 3.1258, + "step": 3293 + }, + { + "epoch": 4.83, + "grad_norm": 1.0471981763839722, + "learning_rate": 0.00010407946723479125, + "loss": 3.1163, + "step": 3294 + }, + { + "epoch": 4.83, + "grad_norm": 2.2549943923950195, + "learning_rate": 0.00010403260238889389, + "loss": 3.1441, + "step": 3295 + }, + { + "epoch": 4.83, + "grad_norm": 1.7782256603240967, + "learning_rate": 0.00010398573665585105, + "loss": 3.1251, + "step": 3296 + }, + { + "epoch": 4.83, + "grad_norm": 3.3199939727783203, + "learning_rate": 0.00010393887004597287, + "loss": 3.1731, + "step": 3297 + }, + { + "epoch": 4.83, + "grad_norm": 3.324904441833496, + "learning_rate": 0.00010389200256956968, + "loss": 3.1439, + "step": 3298 + }, + { + "epoch": 4.84, + "grad_norm": 2.5700957775115967, + "learning_rate": 0.00010384513423695196, + "loss": 3.138, + "step": 3299 + }, + { + "epoch": 4.84, + "grad_norm": 3.218172550201416, + "learning_rate": 0.00010379826505843054, + "loss": 3.1335, + "step": 3300 + }, + { + "epoch": 4.84, + "grad_norm": 1.4542341232299805, + "learning_rate": 0.00010375139504431622, + "loss": 3.1655, + "step": 3301 + }, + { + "epoch": 4.84, + "grad_norm": 21.637861251831055, + "learning_rate": 0.00010370452420492012, + "loss": 3.1785, + "step": 3302 + }, + { + "epoch": 4.84, + "grad_norm": 3.2379684448242188, + "learning_rate": 0.00010365765255055358, + "loss": 3.133, + "step": 3303 + }, + { + "epoch": 4.84, + "grad_norm": 1.356064796447754, + "learning_rate": 0.00010361078009152793, + "loss": 3.1157, + "step": 3304 + }, + { + "epoch": 4.84, + "grad_norm": 1.2905539274215698, + "learning_rate": 0.00010356390683815489, + "loss": 3.1529, + "step": 3305 + }, + { + "epoch": 4.85, + "grad_norm": 2.5484352111816406, + "learning_rate": 0.00010351703280074623, + "loss": 3.1459, + "step": 3306 + }, + { + "epoch": 4.85, + "grad_norm": 1.1913775205612183, + "learning_rate": 0.0001034701579896139, + "loss": 3.2115, + "step": 3307 + }, + { + "epoch": 4.85, + "grad_norm": 1.7160303592681885, + "learning_rate": 0.0001034232824150701, + "loss": 3.1572, + "step": 3308 + }, + { + "epoch": 4.85, + "grad_norm": 2.059537887573242, + "learning_rate": 0.0001033764060874271, + "loss": 3.1361, + "step": 3309 + }, + { + "epoch": 4.85, + "grad_norm": 1.258538007736206, + "learning_rate": 0.00010332952901699738, + "loss": 3.1483, + "step": 3310 + }, + { + "epoch": 4.85, + "grad_norm": 2.301053285598755, + "learning_rate": 0.00010328265121409361, + "loss": 3.214, + "step": 3311 + }, + { + "epoch": 4.86, + "grad_norm": 1.366959810256958, + "learning_rate": 0.00010323577268902853, + "loss": 3.1272, + "step": 3312 + }, + { + "epoch": 4.86, + "grad_norm": 1.7083238363265991, + "learning_rate": 0.00010318889345211518, + "loss": 3.1432, + "step": 3313 + }, + { + "epoch": 4.86, + "grad_norm": 1.0292158126831055, + "learning_rate": 0.00010314201351366666, + "loss": 3.16, + "step": 3314 + }, + { + "epoch": 4.86, + "grad_norm": 1.8825881481170654, + "learning_rate": 0.0001030951328839962, + "loss": 3.1021, + "step": 3315 + }, + { + "epoch": 4.86, + "grad_norm": 1.2715975046157837, + "learning_rate": 0.00010304825157341722, + "loss": 3.1305, + "step": 3316 + }, + { + "epoch": 4.86, + "grad_norm": 1.972051978111267, + "learning_rate": 0.00010300136959224335, + "loss": 3.1384, + "step": 3317 + }, + { + "epoch": 4.86, + "grad_norm": 2.031904697418213, + "learning_rate": 0.00010295448695078829, + "loss": 3.1444, + "step": 3318 + }, + { + "epoch": 4.87, + "grad_norm": 1.3081032037734985, + "learning_rate": 0.00010290760365936587, + "loss": 3.1418, + "step": 3319 + }, + { + "epoch": 4.87, + "grad_norm": 1.2461774349212646, + "learning_rate": 0.00010286071972829016, + "loss": 3.1776, + "step": 3320 + }, + { + "epoch": 4.87, + "grad_norm": 614.052978515625, + "learning_rate": 0.00010281383516787528, + "loss": 3.162, + "step": 3321 + }, + { + "epoch": 4.87, + "grad_norm": 16.35662269592285, + "learning_rate": 0.0001027669499884355, + "loss": 3.1498, + "step": 3322 + }, + { + "epoch": 4.87, + "grad_norm": 5.687806606292725, + "learning_rate": 0.00010272006420028528, + "loss": 3.1927, + "step": 3323 + }, + { + "epoch": 4.87, + "grad_norm": 4.659135341644287, + "learning_rate": 0.00010267317781373918, + "loss": 3.1678, + "step": 3324 + }, + { + "epoch": 4.87, + "grad_norm": 6.275466442108154, + "learning_rate": 0.00010262629083911186, + "loss": 3.1716, + "step": 3325 + }, + { + "epoch": 4.88, + "grad_norm": 3.4971659183502197, + "learning_rate": 0.00010257940328671813, + "loss": 3.1567, + "step": 3326 + }, + { + "epoch": 4.88, + "grad_norm": 4.331486225128174, + "learning_rate": 0.00010253251516687303, + "loss": 3.164, + "step": 3327 + }, + { + "epoch": 4.88, + "grad_norm": 2.695127010345459, + "learning_rate": 0.00010248562648989153, + "loss": 3.1222, + "step": 3328 + }, + { + "epoch": 4.88, + "grad_norm": 2.059220314025879, + "learning_rate": 0.00010243873726608881, + "loss": 3.1783, + "step": 3329 + }, + { + "epoch": 4.88, + "grad_norm": 2.7665822505950928, + "learning_rate": 0.0001023918475057803, + "loss": 3.1525, + "step": 3330 + }, + { + "epoch": 4.88, + "grad_norm": 1.4520095586776733, + "learning_rate": 0.00010234495721928137, + "loss": 3.1559, + "step": 3331 + }, + { + "epoch": 4.88, + "grad_norm": 3.7544808387756348, + "learning_rate": 0.00010229806641690754, + "loss": 3.196, + "step": 3332 + }, + { + "epoch": 4.89, + "grad_norm": 4.363980770111084, + "learning_rate": 0.0001022511751089745, + "loss": 3.1835, + "step": 3333 + }, + { + "epoch": 4.89, + "grad_norm": 2.252725124359131, + "learning_rate": 0.00010220428330579804, + "loss": 3.1515, + "step": 3334 + }, + { + "epoch": 4.89, + "grad_norm": 2.367949962615967, + "learning_rate": 0.00010215739101769399, + "loss": 3.1599, + "step": 3335 + }, + { + "epoch": 4.89, + "grad_norm": 2.9788756370544434, + "learning_rate": 0.00010211049825497841, + "loss": 3.147, + "step": 3336 + }, + { + "epoch": 4.89, + "grad_norm": 1.9200280904769897, + "learning_rate": 0.00010206360502796734, + "loss": 3.1723, + "step": 3337 + }, + { + "epoch": 4.89, + "grad_norm": 2.547455072402954, + "learning_rate": 0.00010201671134697699, + "loss": 3.1607, + "step": 3338 + }, + { + "epoch": 4.9, + "grad_norm": 3.103799343109131, + "learning_rate": 0.00010196981722232368, + "loss": 3.1501, + "step": 3339 + }, + { + "epoch": 4.9, + "grad_norm": 0.7216553092002869, + "learning_rate": 0.00010192292266432377, + "loss": 3.1397, + "step": 3340 + }, + { + "epoch": 4.9, + "grad_norm": 3.535027265548706, + "learning_rate": 0.00010187602768329378, + "loss": 3.1029, + "step": 3341 + }, + { + "epoch": 4.9, + "grad_norm": 3.598963975906372, + "learning_rate": 0.00010182913228955025, + "loss": 3.1537, + "step": 3342 + }, + { + "epoch": 4.9, + "grad_norm": 1.5128989219665527, + "learning_rate": 0.00010178223649340985, + "loss": 3.1342, + "step": 3343 + }, + { + "epoch": 4.9, + "grad_norm": 1.6262668371200562, + "learning_rate": 0.00010173534030518941, + "loss": 3.1654, + "step": 3344 + }, + { + "epoch": 4.9, + "grad_norm": 1.0745606422424316, + "learning_rate": 0.00010168844373520572, + "loss": 3.1105, + "step": 3345 + }, + { + "epoch": 4.91, + "grad_norm": 1.7129157781600952, + "learning_rate": 0.00010164154679377569, + "loss": 3.1317, + "step": 3346 + }, + { + "epoch": 4.91, + "grad_norm": 1.3464902639389038, + "learning_rate": 0.00010159464949121641, + "loss": 3.1503, + "step": 3347 + }, + { + "epoch": 4.91, + "grad_norm": 8.157245635986328, + "learning_rate": 0.00010154775183784493, + "loss": 3.1043, + "step": 3348 + }, + { + "epoch": 4.91, + "grad_norm": 4.697030544281006, + "learning_rate": 0.0001015008538439784, + "loss": 3.194, + "step": 3349 + }, + { + "epoch": 4.91, + "grad_norm": 34.21104049682617, + "learning_rate": 0.00010145395551993408, + "loss": 3.171, + "step": 3350 + }, + { + "epoch": 4.91, + "grad_norm": 3.0200319290161133, + "learning_rate": 0.00010140705687602929, + "loss": 3.1502, + "step": 3351 + }, + { + "epoch": 4.91, + "grad_norm": 0.9504959583282471, + "learning_rate": 0.00010136015792258142, + "loss": 3.1142, + "step": 3352 + }, + { + "epoch": 4.92, + "grad_norm": 2.7994866371154785, + "learning_rate": 0.00010131325866990793, + "loss": 3.1242, + "step": 3353 + }, + { + "epoch": 4.92, + "grad_norm": 2.5278077125549316, + "learning_rate": 0.00010126635912832631, + "loss": 3.0951, + "step": 3354 + }, + { + "epoch": 4.92, + "grad_norm": 3.453552007675171, + "learning_rate": 0.00010121945930815422, + "loss": 3.1458, + "step": 3355 + }, + { + "epoch": 4.92, + "grad_norm": 3.2152633666992188, + "learning_rate": 0.00010117255921970921, + "loss": 3.1589, + "step": 3356 + }, + { + "epoch": 4.92, + "grad_norm": 1.388268232345581, + "learning_rate": 0.00010112565887330903, + "loss": 3.1496, + "step": 3357 + }, + { + "epoch": 4.92, + "grad_norm": 3.23463773727417, + "learning_rate": 0.00010107875827927147, + "loss": 3.1184, + "step": 3358 + }, + { + "epoch": 4.92, + "grad_norm": 2.970151662826538, + "learning_rate": 0.00010103185744791431, + "loss": 3.0931, + "step": 3359 + }, + { + "epoch": 4.93, + "grad_norm": 7.270925998687744, + "learning_rate": 0.00010098495638955538, + "loss": 3.1356, + "step": 3360 + }, + { + "epoch": 4.93, + "grad_norm": 18.72118377685547, + "learning_rate": 0.00010093805511451271, + "loss": 3.1538, + "step": 3361 + }, + { + "epoch": 4.93, + "grad_norm": 52.44887161254883, + "learning_rate": 0.00010089115363310417, + "loss": 3.1924, + "step": 3362 + }, + { + "epoch": 4.93, + "grad_norm": 7.027931213378906, + "learning_rate": 0.00010084425195564782, + "loss": 3.1726, + "step": 3363 + }, + { + "epoch": 4.93, + "grad_norm": 17.143966674804688, + "learning_rate": 0.00010079735009246167, + "loss": 3.1921, + "step": 3364 + }, + { + "epoch": 4.93, + "grad_norm": 3.896998167037964, + "learning_rate": 0.00010075044805386386, + "loss": 3.1341, + "step": 3365 + }, + { + "epoch": 4.94, + "grad_norm": 4.068105697631836, + "learning_rate": 0.0001007035458501725, + "loss": 3.1548, + "step": 3366 + }, + { + "epoch": 4.94, + "grad_norm": 3.3968491554260254, + "learning_rate": 0.00010065664349170578, + "loss": 3.146, + "step": 3367 + }, + { + "epoch": 4.94, + "grad_norm": 3.594057083129883, + "learning_rate": 0.00010060974098878189, + "loss": 3.1631, + "step": 3368 + }, + { + "epoch": 4.94, + "grad_norm": 0.987125813961029, + "learning_rate": 0.00010056283835171906, + "loss": 3.1319, + "step": 3369 + }, + { + "epoch": 4.94, + "grad_norm": 4.898683547973633, + "learning_rate": 0.00010051593559083557, + "loss": 3.1316, + "step": 3370 + }, + { + "epoch": 4.94, + "grad_norm": 6.864829063415527, + "learning_rate": 0.0001004690327164497, + "loss": 3.2037, + "step": 3371 + }, + { + "epoch": 4.94, + "grad_norm": 5.136176109313965, + "learning_rate": 0.00010042212973887982, + "loss": 3.1921, + "step": 3372 + }, + { + "epoch": 4.95, + "grad_norm": 2.621640205383301, + "learning_rate": 0.00010037522666844415, + "loss": 3.1716, + "step": 3373 + }, + { + "epoch": 4.95, + "grad_norm": 4.9672746658325195, + "learning_rate": 0.00010032832351546117, + "loss": 3.1374, + "step": 3374 + }, + { + "epoch": 4.95, + "grad_norm": 6.860297203063965, + "learning_rate": 0.00010028142029024923, + "loss": 3.1863, + "step": 3375 + }, + { + "epoch": 4.95, + "grad_norm": 131.51553344726562, + "learning_rate": 0.00010023451700312669, + "loss": 3.1629, + "step": 3376 + }, + { + "epoch": 4.95, + "grad_norm": 4.443822860717773, + "learning_rate": 0.00010018761366441198, + "loss": 3.1513, + "step": 3377 + }, + { + "epoch": 4.95, + "grad_norm": 1.4408992528915405, + "learning_rate": 0.00010014071028442354, + "loss": 3.1317, + "step": 3378 + }, + { + "epoch": 4.95, + "grad_norm": 8.419514656066895, + "learning_rate": 0.00010009380687347975, + "loss": 3.1649, + "step": 3379 + }, + { + "epoch": 4.96, + "grad_norm": 19.037832260131836, + "learning_rate": 0.00010004690344189912, + "loss": 3.1697, + "step": 3380 + }, + { + "epoch": 4.96, + "eval_loss": 3.100813865661621, + "eval_runtime": 1176.1523, + "eval_samples_per_second": 14.575, + "eval_steps_per_second": 0.911, + "step": 3380 + }, + { + "epoch": 4.96, + "grad_norm": 11.52230167388916, + "learning_rate": 0.0001, + "loss": 3.2739, + "step": 3381 + }, + { + "epoch": 4.96, + "grad_norm": 8.755014419555664, + "learning_rate": 9.995309655810093e-05, + "loss": 3.2172, + "step": 3382 + }, + { + "epoch": 4.96, + "grad_norm": 5.496139049530029, + "learning_rate": 9.990619312652026e-05, + "loss": 3.1656, + "step": 3383 + }, + { + "epoch": 4.96, + "grad_norm": 3.737701416015625, + "learning_rate": 9.985928971557649e-05, + "loss": 3.2014, + "step": 3384 + }, + { + "epoch": 4.96, + "grad_norm": 6.020711898803711, + "learning_rate": 9.981238633558805e-05, + "loss": 3.2036, + "step": 3385 + }, + { + "epoch": 4.96, + "grad_norm": 7.874726295471191, + "learning_rate": 9.976548299687332e-05, + "loss": 3.1735, + "step": 3386 + }, + { + "epoch": 4.97, + "grad_norm": 7.075169563293457, + "learning_rate": 9.97185797097508e-05, + "loss": 3.1674, + "step": 3387 + }, + { + "epoch": 4.97, + "grad_norm": 3.235295057296753, + "learning_rate": 9.967167648453885e-05, + "loss": 3.1999, + "step": 3388 + }, + { + "epoch": 4.97, + "grad_norm": 3.599179744720459, + "learning_rate": 9.962477333155585e-05, + "loss": 3.1829, + "step": 3389 + }, + { + "epoch": 4.97, + "grad_norm": 6.6362528800964355, + "learning_rate": 9.95778702611202e-05, + "loss": 3.2165, + "step": 3390 + }, + { + "epoch": 4.97, + "grad_norm": 7.1437153816223145, + "learning_rate": 9.953096728355031e-05, + "loss": 3.1884, + "step": 3391 + }, + { + "epoch": 4.97, + "grad_norm": 4.677809715270996, + "learning_rate": 9.948406440916443e-05, + "loss": 3.1899, + "step": 3392 + }, + { + "epoch": 4.98, + "grad_norm": 1.904357671737671, + "learning_rate": 9.943716164828096e-05, + "loss": 3.1357, + "step": 3393 + }, + { + "epoch": 4.98, + "grad_norm": 5.046501636505127, + "learning_rate": 9.939025901121814e-05, + "loss": 3.1666, + "step": 3394 + }, + { + "epoch": 4.98, + "grad_norm": 6.0771074295043945, + "learning_rate": 9.934335650829425e-05, + "loss": 3.1961, + "step": 3395 + }, + { + "epoch": 4.98, + "grad_norm": 5.269542694091797, + "learning_rate": 9.92964541498275e-05, + "loss": 3.1937, + "step": 3396 + }, + { + "epoch": 4.98, + "grad_norm": 2.9679410457611084, + "learning_rate": 9.924955194613616e-05, + "loss": 3.1376, + "step": 3397 + }, + { + "epoch": 4.98, + "grad_norm": 3.366711139678955, + "learning_rate": 9.920264990753837e-05, + "loss": 3.1116, + "step": 3398 + }, + { + "epoch": 4.98, + "grad_norm": 6.38279390335083, + "learning_rate": 9.91557480443522e-05, + "loss": 3.1385, + "step": 3399 + }, + { + "epoch": 4.99, + "grad_norm": 5.471691608428955, + "learning_rate": 9.910884636689585e-05, + "loss": 3.1874, + "step": 3400 + }, + { + "epoch": 4.99, + "grad_norm": 2.8460099697113037, + "learning_rate": 9.906194488548734e-05, + "loss": 3.1898, + "step": 3401 + }, + { + "epoch": 4.99, + "grad_norm": 3.193862199783325, + "learning_rate": 9.901504361044461e-05, + "loss": 3.1727, + "step": 3402 + }, + { + "epoch": 4.99, + "grad_norm": 4.644516944885254, + "learning_rate": 9.896814255208571e-05, + "loss": 3.1828, + "step": 3403 + }, + { + "epoch": 4.99, + "grad_norm": 4.342918872833252, + "learning_rate": 9.892124172072857e-05, + "loss": 3.1714, + "step": 3404 + }, + { + "epoch": 4.99, + "grad_norm": 2.4274344444274902, + "learning_rate": 9.887434112669097e-05, + "loss": 3.1509, + "step": 3405 + }, + { + "epoch": 4.99, + "grad_norm": 3.326143264770508, + "learning_rate": 9.882744078029081e-05, + "loss": 3.1433, + "step": 3406 + }, + { + "epoch": 5.0, + "grad_norm": 5.951056003570557, + "learning_rate": 9.878054069184583e-05, + "loss": 3.1543, + "step": 3407 + }, + { + "epoch": 5.0, + "grad_norm": 5.0941972732543945, + "learning_rate": 9.873364087167367e-05, + "loss": 3.1558, + "step": 3408 + }, + { + "epoch": 5.0, + "grad_norm": 2.075986623764038, + "learning_rate": 9.868674133009209e-05, + "loss": 3.144, + "step": 3409 + }, + { + "epoch": 5.0, + "grad_norm": 3.3112316131591797, + "learning_rate": 9.863984207741859e-05, + "loss": 3.1444, + "step": 3410 + }, + { + "epoch": 5.0, + "grad_norm": 4.910267353057861, + "learning_rate": 9.85929431239707e-05, + "loss": 3.1823, + "step": 3411 + }, + { + "epoch": 5.0, + "grad_norm": 4.1945624351501465, + "learning_rate": 9.854604448006594e-05, + "loss": 3.1945, + "step": 3412 + }, + { + "epoch": 5.01, + "grad_norm": 1.877642273902893, + "learning_rate": 9.849914615602163e-05, + "loss": 3.1357, + "step": 3413 + }, + { + "epoch": 5.01, + "grad_norm": 3.460303783416748, + "learning_rate": 9.845224816215508e-05, + "loss": 3.1039, + "step": 3414 + }, + { + "epoch": 5.01, + "grad_norm": 5.127685070037842, + "learning_rate": 9.84053505087836e-05, + "loss": 3.1041, + "step": 3415 + }, + { + "epoch": 5.01, + "grad_norm": 3.685288906097412, + "learning_rate": 9.835845320622434e-05, + "loss": 3.1874, + "step": 3416 + }, + { + "epoch": 5.01, + "grad_norm": 1.9846138954162598, + "learning_rate": 9.83115562647943e-05, + "loss": 3.1365, + "step": 3417 + }, + { + "epoch": 5.01, + "grad_norm": 5.992250442504883, + "learning_rate": 9.826465969481061e-05, + "loss": 3.0993, + "step": 3418 + }, + { + "epoch": 5.01, + "grad_norm": 4.008816242218018, + "learning_rate": 9.821776350659017e-05, + "loss": 3.1471, + "step": 3419 + }, + { + "epoch": 5.0, + "grad_norm": 2.797140598297119, + "learning_rate": 9.81708677104498e-05, + "loss": 3.1822, + "step": 3420 + }, + { + "epoch": 5.0, + "grad_norm": 3.2682175636291504, + "learning_rate": 9.812397231670624e-05, + "loss": 3.1501, + "step": 3421 + }, + { + "epoch": 5.0, + "grad_norm": 4.154157638549805, + "learning_rate": 9.807707733567626e-05, + "loss": 3.1572, + "step": 3422 + }, + { + "epoch": 5.01, + "grad_norm": 2.315260648727417, + "learning_rate": 9.803018277767634e-05, + "loss": 3.1305, + "step": 3423 + }, + { + "epoch": 5.01, + "grad_norm": 4.300922870635986, + "learning_rate": 9.798328865302299e-05, + "loss": 3.2001, + "step": 3424 + }, + { + "epoch": 5.01, + "grad_norm": 4.946724891662598, + "learning_rate": 9.793639497203268e-05, + "loss": 3.1431, + "step": 3425 + }, + { + "epoch": 5.01, + "grad_norm": 1.1593897342681885, + "learning_rate": 9.788950174502164e-05, + "loss": 3.1631, + "step": 3426 + }, + { + "epoch": 5.01, + "grad_norm": 88.28909301757812, + "learning_rate": 9.784260898230603e-05, + "loss": 3.1888, + "step": 3427 + }, + { + "epoch": 5.01, + "grad_norm": 7.40468168258667, + "learning_rate": 9.7795716694202e-05, + "loss": 3.1605, + "step": 3428 + }, + { + "epoch": 5.01, + "grad_norm": 8.019392967224121, + "learning_rate": 9.774882489102555e-05, + "loss": 3.2278, + "step": 3429 + }, + { + "epoch": 5.02, + "grad_norm": 5.9179205894470215, + "learning_rate": 9.770193358309248e-05, + "loss": 3.1715, + "step": 3430 + }, + { + "epoch": 5.02, + "grad_norm": 2.9003238677978516, + "learning_rate": 9.765504278071866e-05, + "loss": 3.1651, + "step": 3431 + }, + { + "epoch": 5.02, + "grad_norm": 4.174776554107666, + "learning_rate": 9.760815249421973e-05, + "loss": 3.1573, + "step": 3432 + }, + { + "epoch": 5.02, + "grad_norm": 5.618264198303223, + "learning_rate": 9.756126273391118e-05, + "loss": 3.1799, + "step": 3433 + }, + { + "epoch": 5.02, + "grad_norm": 5.217837333679199, + "learning_rate": 9.75143735101085e-05, + "loss": 3.1674, + "step": 3434 + }, + { + "epoch": 5.02, + "grad_norm": 6.881947994232178, + "learning_rate": 9.746748483312702e-05, + "loss": 3.2233, + "step": 3435 + }, + { + "epoch": 5.02, + "grad_norm": 4.155002117156982, + "learning_rate": 9.742059671328185e-05, + "loss": 3.1715, + "step": 3436 + }, + { + "epoch": 5.03, + "grad_norm": 3.3155972957611084, + "learning_rate": 9.737370916088815e-05, + "loss": 3.1688, + "step": 3437 + }, + { + "epoch": 5.03, + "grad_norm": 4.904897689819336, + "learning_rate": 9.732682218626086e-05, + "loss": 3.1702, + "step": 3438 + }, + { + "epoch": 5.03, + "grad_norm": 27.04771614074707, + "learning_rate": 9.727993579971471e-05, + "loss": 3.1816, + "step": 3439 + }, + { + "epoch": 5.03, + "grad_norm": 3.3083443641662598, + "learning_rate": 9.723305001156452e-05, + "loss": 3.1635, + "step": 3440 + }, + { + "epoch": 5.03, + "grad_norm": 3.0965356826782227, + "learning_rate": 9.718616483212474e-05, + "loss": 3.1631, + "step": 3441 + }, + { + "epoch": 5.03, + "grad_norm": 2.470313310623169, + "learning_rate": 9.713928027170989e-05, + "loss": 3.1788, + "step": 3442 + }, + { + "epoch": 5.04, + "grad_norm": 1.455441951751709, + "learning_rate": 9.709239634063414e-05, + "loss": 3.1543, + "step": 3443 + }, + { + "epoch": 5.04, + "grad_norm": 3.4748833179473877, + "learning_rate": 9.704551304921175e-05, + "loss": 3.1228, + "step": 3444 + }, + { + "epoch": 5.04, + "grad_norm": 3.481116771697998, + "learning_rate": 9.69986304077567e-05, + "loss": 3.1595, + "step": 3445 + }, + { + "epoch": 5.04, + "grad_norm": 4.942490577697754, + "learning_rate": 9.69517484265828e-05, + "loss": 3.1443, + "step": 3446 + }, + { + "epoch": 5.04, + "grad_norm": 3.2956652641296387, + "learning_rate": 9.690486711600384e-05, + "loss": 3.1699, + "step": 3447 + }, + { + "epoch": 5.04, + "grad_norm": 2.4343531131744385, + "learning_rate": 9.68579864863334e-05, + "loss": 3.1488, + "step": 3448 + }, + { + "epoch": 5.04, + "grad_norm": 2.82985782623291, + "learning_rate": 9.681110654788482e-05, + "loss": 3.1032, + "step": 3449 + }, + { + "epoch": 5.05, + "grad_norm": 1.59348464012146, + "learning_rate": 9.676422731097149e-05, + "loss": 3.1256, + "step": 3450 + }, + { + "epoch": 5.05, + "grad_norm": 4.060868740081787, + "learning_rate": 9.671734878590644e-05, + "loss": 3.1592, + "step": 3451 + }, + { + "epoch": 5.05, + "grad_norm": 4.718034744262695, + "learning_rate": 9.667047098300262e-05, + "loss": 3.1666, + "step": 3452 + }, + { + "epoch": 5.05, + "grad_norm": 1.5934293270111084, + "learning_rate": 9.662359391257293e-05, + "loss": 3.1592, + "step": 3453 + }, + { + "epoch": 5.05, + "grad_norm": 4.770664215087891, + "learning_rate": 9.657671758492992e-05, + "loss": 3.1303, + "step": 3454 + }, + { + "epoch": 5.05, + "grad_norm": 5.743010520935059, + "learning_rate": 9.652984201038608e-05, + "loss": 3.1736, + "step": 3455 + }, + { + "epoch": 5.05, + "grad_norm": 22.09770393371582, + "learning_rate": 9.64829671992538e-05, + "loss": 3.1555, + "step": 3456 + }, + { + "epoch": 5.06, + "grad_norm": 4.085333347320557, + "learning_rate": 9.643609316184514e-05, + "loss": 3.1203, + "step": 3457 + }, + { + "epoch": 5.06, + "grad_norm": 3.675048589706421, + "learning_rate": 9.638921990847206e-05, + "loss": 3.128, + "step": 3458 + }, + { + "epoch": 5.06, + "grad_norm": 1.0433051586151123, + "learning_rate": 9.634234744944646e-05, + "loss": 3.1206, + "step": 3459 + }, + { + "epoch": 5.06, + "grad_norm": 4.096065521240234, + "learning_rate": 9.62954757950799e-05, + "loss": 3.1567, + "step": 3460 + }, + { + "epoch": 5.06, + "grad_norm": 5.4872870445251465, + "learning_rate": 9.624860495568379e-05, + "loss": 3.1482, + "step": 3461 + }, + { + "epoch": 5.06, + "grad_norm": 3.236421585083008, + "learning_rate": 9.620173494156949e-05, + "loss": 3.1172, + "step": 3462 + }, + { + "epoch": 5.06, + "grad_norm": 3.7397096157073975, + "learning_rate": 9.615486576304806e-05, + "loss": 3.1485, + "step": 3463 + }, + { + "epoch": 5.07, + "grad_norm": 2.640143632888794, + "learning_rate": 9.610799743043036e-05, + "loss": 3.1444, + "step": 3464 + }, + { + "epoch": 5.07, + "grad_norm": 2.277892589569092, + "learning_rate": 9.606112995402714e-05, + "loss": 3.117, + "step": 3465 + }, + { + "epoch": 5.07, + "grad_norm": 2.7552220821380615, + "learning_rate": 9.601426334414898e-05, + "loss": 3.1557, + "step": 3466 + }, + { + "epoch": 5.07, + "grad_norm": 3.1442368030548096, + "learning_rate": 9.596739761110613e-05, + "loss": 3.1161, + "step": 3467 + }, + { + "epoch": 5.07, + "grad_norm": 1.997355341911316, + "learning_rate": 9.592053276520877e-05, + "loss": 3.1332, + "step": 3468 + }, + { + "epoch": 5.07, + "grad_norm": 1.959526777267456, + "learning_rate": 9.58736688167669e-05, + "loss": 3.141, + "step": 3469 + }, + { + "epoch": 5.08, + "grad_norm": 2.0642502307891846, + "learning_rate": 9.582680577609022e-05, + "loss": 3.1532, + "step": 3470 + }, + { + "epoch": 5.08, + "grad_norm": 1.1112334728240967, + "learning_rate": 9.57799436534883e-05, + "loss": 3.1402, + "step": 3471 + }, + { + "epoch": 5.08, + "grad_norm": 2.348271131515503, + "learning_rate": 9.57330824592705e-05, + "loss": 3.162, + "step": 3472 + }, + { + "epoch": 5.08, + "grad_norm": 1.7725739479064941, + "learning_rate": 9.5686222203746e-05, + "loss": 3.159, + "step": 3473 + }, + { + "epoch": 5.08, + "grad_norm": 1.3065656423568726, + "learning_rate": 9.563936289722363e-05, + "loss": 3.155, + "step": 3474 + }, + { + "epoch": 5.08, + "grad_norm": 2.1432132720947266, + "learning_rate": 9.559250455001223e-05, + "loss": 3.0939, + "step": 3475 + }, + { + "epoch": 5.08, + "grad_norm": 1.2344911098480225, + "learning_rate": 9.554564717242032e-05, + "loss": 3.1094, + "step": 3476 + }, + { + "epoch": 5.09, + "grad_norm": 1.591896653175354, + "learning_rate": 9.549879077475613e-05, + "loss": 3.1632, + "step": 3477 + }, + { + "epoch": 5.09, + "grad_norm": 1.2728568315505981, + "learning_rate": 9.545193536732783e-05, + "loss": 3.1537, + "step": 3478 + }, + { + "epoch": 5.09, + "grad_norm": 0.9794215559959412, + "learning_rate": 9.540508096044329e-05, + "loss": 3.1012, + "step": 3479 + }, + { + "epoch": 5.09, + "grad_norm": 2.092595100402832, + "learning_rate": 9.535822756441011e-05, + "loss": 3.1352, + "step": 3480 + }, + { + "epoch": 5.09, + "grad_norm": 1.6915316581726074, + "learning_rate": 9.531137518953577e-05, + "loss": 3.1332, + "step": 3481 + }, + { + "epoch": 5.09, + "grad_norm": 6.075295925140381, + "learning_rate": 9.526452384612749e-05, + "loss": 3.1654, + "step": 3482 + }, + { + "epoch": 5.09, + "grad_norm": 2.9814789295196533, + "learning_rate": 9.521767354449219e-05, + "loss": 3.1196, + "step": 3483 + }, + { + "epoch": 5.1, + "grad_norm": 1.2538803815841675, + "learning_rate": 9.51708242949367e-05, + "loss": 3.1304, + "step": 3484 + }, + { + "epoch": 5.1, + "grad_norm": 1.274577260017395, + "learning_rate": 9.512397610776747e-05, + "loss": 3.1293, + "step": 3485 + }, + { + "epoch": 5.1, + "grad_norm": 10.467679023742676, + "learning_rate": 9.507712899329079e-05, + "loss": 3.1396, + "step": 3486 + }, + { + "epoch": 5.1, + "grad_norm": 3.3941073417663574, + "learning_rate": 9.503028296181281e-05, + "loss": 3.1455, + "step": 3487 + }, + { + "epoch": 5.1, + "grad_norm": 1.5310273170471191, + "learning_rate": 9.498343802363923e-05, + "loss": 3.141, + "step": 3488 + }, + { + "epoch": 5.1, + "grad_norm": 4.921885967254639, + "learning_rate": 9.493659418907566e-05, + "loss": 3.1257, + "step": 3489 + }, + { + "epoch": 5.1, + "grad_norm": 6.093456268310547, + "learning_rate": 9.488975146842746e-05, + "loss": 3.1542, + "step": 3490 + }, + { + "epoch": 5.11, + "grad_norm": 2.32529616355896, + "learning_rate": 9.484290987199968e-05, + "loss": 3.1527, + "step": 3491 + }, + { + "epoch": 5.11, + "grad_norm": 4.923595905303955, + "learning_rate": 9.47960694100972e-05, + "loss": 3.1667, + "step": 3492 + }, + { + "epoch": 5.11, + "grad_norm": 5.732090950012207, + "learning_rate": 9.474923009302452e-05, + "loss": 3.1218, + "step": 3493 + }, + { + "epoch": 5.11, + "grad_norm": 3.330801486968994, + "learning_rate": 9.470239193108607e-05, + "loss": 3.1252, + "step": 3494 + }, + { + "epoch": 5.11, + "grad_norm": 3.7033092975616455, + "learning_rate": 9.465555493458588e-05, + "loss": 3.1358, + "step": 3495 + }, + { + "epoch": 5.11, + "grad_norm": 6.471573352813721, + "learning_rate": 9.460871911382774e-05, + "loss": 3.1961, + "step": 3496 + }, + { + "epoch": 5.12, + "grad_norm": 6.723712921142578, + "learning_rate": 9.456188447911533e-05, + "loss": 3.0915, + "step": 3497 + }, + { + "epoch": 5.12, + "grad_norm": 1.7002429962158203, + "learning_rate": 9.451505104075188e-05, + "loss": 3.1452, + "step": 3498 + }, + { + "epoch": 5.12, + "grad_norm": 2.4126813411712646, + "learning_rate": 9.44682188090404e-05, + "loss": 3.1457, + "step": 3499 + }, + { + "epoch": 5.12, + "grad_norm": 1.8296704292297363, + "learning_rate": 9.442138779428376e-05, + "loss": 3.1215, + "step": 3500 + }, + { + "epoch": 5.12, + "grad_norm": 1.7919952869415283, + "learning_rate": 9.43745580067844e-05, + "loss": 3.179, + "step": 3501 + }, + { + "epoch": 5.12, + "grad_norm": 1.8230031728744507, + "learning_rate": 9.432772945684454e-05, + "loss": 3.1339, + "step": 3502 + }, + { + "epoch": 5.12, + "grad_norm": 1.5876286029815674, + "learning_rate": 9.428090215476625e-05, + "loss": 3.1248, + "step": 3503 + }, + { + "epoch": 5.13, + "grad_norm": 2.1340315341949463, + "learning_rate": 9.423407611085113e-05, + "loss": 3.1396, + "step": 3504 + }, + { + "epoch": 5.13, + "grad_norm": 1.6990450620651245, + "learning_rate": 9.41872513354006e-05, + "loss": 3.1119, + "step": 3505 + }, + { + "epoch": 5.13, + "grad_norm": 1.792682409286499, + "learning_rate": 9.414042783871586e-05, + "loss": 3.1476, + "step": 3506 + }, + { + "epoch": 5.13, + "grad_norm": 2.685699701309204, + "learning_rate": 9.409360563109773e-05, + "loss": 3.1468, + "step": 3507 + }, + { + "epoch": 5.13, + "grad_norm": 4.750230312347412, + "learning_rate": 9.404678472284673e-05, + "loss": 3.1361, + "step": 3508 + }, + { + "epoch": 5.13, + "grad_norm": 1.0667452812194824, + "learning_rate": 9.399996512426323e-05, + "loss": 3.1625, + "step": 3509 + }, + { + "epoch": 5.13, + "grad_norm": 2.092228889465332, + "learning_rate": 9.395314684564721e-05, + "loss": 3.1709, + "step": 3510 + }, + { + "epoch": 5.14, + "grad_norm": 1.991498589515686, + "learning_rate": 9.390632989729831e-05, + "loss": 3.1122, + "step": 3511 + }, + { + "epoch": 5.14, + "grad_norm": 10.466729164123535, + "learning_rate": 9.385951428951603e-05, + "loss": 3.1214, + "step": 3512 + }, + { + "epoch": 5.14, + "grad_norm": 3.6561026573181152, + "learning_rate": 9.381270003259945e-05, + "loss": 3.1363, + "step": 3513 + }, + { + "epoch": 5.14, + "grad_norm": 1.4475880861282349, + "learning_rate": 9.37658871368474e-05, + "loss": 3.1801, + "step": 3514 + }, + { + "epoch": 5.14, + "grad_norm": 5.58400821685791, + "learning_rate": 9.371907561255837e-05, + "loss": 3.1223, + "step": 3515 + }, + { + "epoch": 5.14, + "grad_norm": 9.835494041442871, + "learning_rate": 9.367226547003065e-05, + "loss": 3.2195, + "step": 3516 + }, + { + "epoch": 5.14, + "grad_norm": 6.080451011657715, + "learning_rate": 9.362545671956214e-05, + "loss": 3.1594, + "step": 3517 + }, + { + "epoch": 5.15, + "grad_norm": 1.3072457313537598, + "learning_rate": 9.357864937145036e-05, + "loss": 3.147, + "step": 3518 + }, + { + "epoch": 5.15, + "grad_norm": 5.248061180114746, + "learning_rate": 9.353184343599273e-05, + "loss": 3.1497, + "step": 3519 + }, + { + "epoch": 5.15, + "grad_norm": 5.6261420249938965, + "learning_rate": 9.34850389234862e-05, + "loss": 3.1824, + "step": 3520 + }, + { + "epoch": 5.15, + "grad_norm": 3.202202320098877, + "learning_rate": 9.343823584422742e-05, + "loss": 3.1609, + "step": 3521 + }, + { + "epoch": 5.15, + "grad_norm": 4.45296573638916, + "learning_rate": 9.339143420851279e-05, + "loss": 3.1792, + "step": 3522 + }, + { + "epoch": 5.15, + "grad_norm": 4.815643310546875, + "learning_rate": 9.334463402663837e-05, + "loss": 3.2226, + "step": 3523 + }, + { + "epoch": 5.16, + "grad_norm": 12.902685165405273, + "learning_rate": 9.32978353088998e-05, + "loss": 3.1462, + "step": 3524 + }, + { + "epoch": 5.16, + "grad_norm": 3.254309892654419, + "learning_rate": 9.32510380655926e-05, + "loss": 3.1306, + "step": 3525 + }, + { + "epoch": 5.16, + "grad_norm": 8.259904861450195, + "learning_rate": 9.320424230701179e-05, + "loss": 3.2094, + "step": 3526 + }, + { + "epoch": 5.16, + "grad_norm": 2.651756525039673, + "learning_rate": 9.31574480434521e-05, + "loss": 3.1366, + "step": 3527 + }, + { + "epoch": 5.16, + "grad_norm": 4.296948432922363, + "learning_rate": 9.311065528520802e-05, + "loss": 3.1398, + "step": 3528 + }, + { + "epoch": 5.16, + "grad_norm": 4.362511157989502, + "learning_rate": 9.30638640425736e-05, + "loss": 3.163, + "step": 3529 + }, + { + "epoch": 5.16, + "grad_norm": 1.9980674982070923, + "learning_rate": 9.301707432584257e-05, + "loss": 3.1884, + "step": 3530 + }, + { + "epoch": 5.17, + "grad_norm": 4.281768321990967, + "learning_rate": 9.297028614530844e-05, + "loss": 3.1488, + "step": 3531 + }, + { + "epoch": 5.17, + "grad_norm": 5.395291328430176, + "learning_rate": 9.292349951126424e-05, + "loss": 3.1816, + "step": 3532 + }, + { + "epoch": 5.17, + "grad_norm": 1.9070907831192017, + "learning_rate": 9.287671443400268e-05, + "loss": 3.1082, + "step": 3533 + }, + { + "epoch": 5.17, + "grad_norm": 5.979832649230957, + "learning_rate": 9.282993092381625e-05, + "loss": 3.1332, + "step": 3534 + }, + { + "epoch": 5.17, + "grad_norm": 7.1100382804870605, + "learning_rate": 9.278314899099694e-05, + "loss": 3.2056, + "step": 3535 + }, + { + "epoch": 5.17, + "grad_norm": 22.313732147216797, + "learning_rate": 9.273636864583647e-05, + "loss": 3.1634, + "step": 3536 + }, + { + "epoch": 5.17, + "grad_norm": 11.737483978271484, + "learning_rate": 9.268958989862628e-05, + "loss": 3.1481, + "step": 3537 + }, + { + "epoch": 5.18, + "grad_norm": 2.638352394104004, + "learning_rate": 9.264281275965728e-05, + "loss": 3.1341, + "step": 3538 + }, + { + "epoch": 5.18, + "grad_norm": 2.807657241821289, + "learning_rate": 9.25960372392202e-05, + "loss": 3.1505, + "step": 3539 + }, + { + "epoch": 5.18, + "grad_norm": 20.156600952148438, + "learning_rate": 9.254926334760526e-05, + "loss": 3.1956, + "step": 3540 + }, + { + "epoch": 5.18, + "grad_norm": 2.164813280105591, + "learning_rate": 9.25024910951025e-05, + "loss": 3.1844, + "step": 3541 + }, + { + "epoch": 5.18, + "grad_norm": 1.7954508066177368, + "learning_rate": 9.245572049200142e-05, + "loss": 3.222, + "step": 3542 + }, + { + "epoch": 5.18, + "grad_norm": 4.100772380828857, + "learning_rate": 9.240895154859125e-05, + "loss": 3.1239, + "step": 3543 + }, + { + "epoch": 5.19, + "grad_norm": 3.8500773906707764, + "learning_rate": 9.236218427516092e-05, + "loss": 3.1278, + "step": 3544 + }, + { + "epoch": 5.19, + "grad_norm": 2.076373815536499, + "learning_rate": 9.231541868199883e-05, + "loss": 3.164, + "step": 3545 + }, + { + "epoch": 5.19, + "grad_norm": 1.8136000633239746, + "learning_rate": 9.226865477939313e-05, + "loss": 3.1479, + "step": 3546 + }, + { + "epoch": 5.19, + "grad_norm": 128.6366424560547, + "learning_rate": 9.222189257763158e-05, + "loss": 3.1239, + "step": 3547 + }, + { + "epoch": 5.19, + "grad_norm": 3.7925548553466797, + "learning_rate": 9.217513208700154e-05, + "loss": 3.1402, + "step": 3548 + }, + { + "epoch": 5.19, + "grad_norm": 11.528712272644043, + "learning_rate": 9.212837331778997e-05, + "loss": 3.1434, + "step": 3549 + }, + { + "epoch": 5.19, + "eval_loss": 2.9914677143096924, + "eval_runtime": 1173.7001, + "eval_samples_per_second": 14.605, + "eval_steps_per_second": 0.913, + "step": 3549 + }, + { + "epoch": 5.19, + "grad_norm": 3.5686726570129395, + "learning_rate": 9.208161628028355e-05, + "loss": 3.1025, + "step": 3550 + }, + { + "epoch": 5.2, + "grad_norm": 19.285165786743164, + "learning_rate": 9.203486098476849e-05, + "loss": 3.1957, + "step": 3551 + }, + { + "epoch": 5.2, + "grad_norm": 7.824908256530762, + "learning_rate": 9.19881074415306e-05, + "loss": 3.1752, + "step": 3552 + }, + { + "epoch": 5.2, + "grad_norm": 3.9910645484924316, + "learning_rate": 9.194135566085537e-05, + "loss": 3.1751, + "step": 3553 + }, + { + "epoch": 5.2, + "grad_norm": 32.02030944824219, + "learning_rate": 9.189460565302794e-05, + "loss": 3.1539, + "step": 3554 + }, + { + "epoch": 5.2, + "grad_norm": 10.703150749206543, + "learning_rate": 9.18478574283329e-05, + "loss": 3.1623, + "step": 3555 + }, + { + "epoch": 5.2, + "grad_norm": 37.447715759277344, + "learning_rate": 9.180111099705459e-05, + "loss": 3.1512, + "step": 3556 + }, + { + "epoch": 5.2, + "grad_norm": 4.359159469604492, + "learning_rate": 9.175436636947697e-05, + "loss": 3.1754, + "step": 3557 + }, + { + "epoch": 5.21, + "grad_norm": 11.65587329864502, + "learning_rate": 9.170762355588342e-05, + "loss": 3.1365, + "step": 3558 + }, + { + "epoch": 5.21, + "grad_norm": 6.701794147491455, + "learning_rate": 9.166088256655715e-05, + "loss": 3.1705, + "step": 3559 + }, + { + "epoch": 5.21, + "grad_norm": 12.4017915725708, + "learning_rate": 9.161414341178084e-05, + "loss": 3.1724, + "step": 3560 + }, + { + "epoch": 5.21, + "grad_norm": 2.782519578933716, + "learning_rate": 9.156740610183674e-05, + "loss": 3.1809, + "step": 3561 + }, + { + "epoch": 5.21, + "grad_norm": 3.470179319381714, + "learning_rate": 9.152067064700683e-05, + "loss": 3.1531, + "step": 3562 + }, + { + "epoch": 5.21, + "grad_norm": 12.016058921813965, + "learning_rate": 9.147393705757254e-05, + "loss": 3.1714, + "step": 3563 + }, + { + "epoch": 5.21, + "grad_norm": 8.318673133850098, + "learning_rate": 9.142720534381497e-05, + "loss": 3.178, + "step": 3564 + }, + { + "epoch": 5.22, + "grad_norm": 26.07087516784668, + "learning_rate": 9.138047551601473e-05, + "loss": 3.1636, + "step": 3565 + }, + { + "epoch": 5.22, + "grad_norm": 55.82182312011719, + "learning_rate": 9.133374758445212e-05, + "loss": 3.1598, + "step": 3566 + }, + { + "epoch": 5.22, + "grad_norm": 3.5638935565948486, + "learning_rate": 9.1287021559407e-05, + "loss": 3.1711, + "step": 3567 + }, + { + "epoch": 5.22, + "grad_norm": 4.856345176696777, + "learning_rate": 9.12402974511587e-05, + "loss": 3.1686, + "step": 3568 + }, + { + "epoch": 5.22, + "grad_norm": 6.552631855010986, + "learning_rate": 9.119357526998628e-05, + "loss": 3.1437, + "step": 3569 + }, + { + "epoch": 5.22, + "grad_norm": 62.54035186767578, + "learning_rate": 9.11468550261683e-05, + "loss": 3.1711, + "step": 3570 + }, + { + "epoch": 5.23, + "grad_norm": 19.124649047851562, + "learning_rate": 9.110013672998284e-05, + "loss": 3.2002, + "step": 3571 + }, + { + "epoch": 5.23, + "grad_norm": 14.90050983428955, + "learning_rate": 9.105342039170768e-05, + "loss": 3.2217, + "step": 3572 + }, + { + "epoch": 5.23, + "grad_norm": 4.7844109535217285, + "learning_rate": 9.100670602162009e-05, + "loss": 3.1506, + "step": 3573 + }, + { + "epoch": 5.23, + "grad_norm": 1.9373784065246582, + "learning_rate": 9.095999362999687e-05, + "loss": 3.1288, + "step": 3574 + }, + { + "epoch": 5.23, + "grad_norm": 60.557308197021484, + "learning_rate": 9.091328322711449e-05, + "loss": 3.1368, + "step": 3575 + }, + { + "epoch": 5.23, + "grad_norm": 6.0116658210754395, + "learning_rate": 9.086657482324892e-05, + "loss": 3.1685, + "step": 3576 + }, + { + "epoch": 5.23, + "grad_norm": 7.286896228790283, + "learning_rate": 9.081986842867562e-05, + "loss": 3.1756, + "step": 3577 + }, + { + "epoch": 5.24, + "grad_norm": 2.469630479812622, + "learning_rate": 9.077316405366981e-05, + "loss": 3.1425, + "step": 3578 + }, + { + "epoch": 5.24, + "grad_norm": 3.226400136947632, + "learning_rate": 9.072646170850605e-05, + "loss": 3.1015, + "step": 3579 + }, + { + "epoch": 5.24, + "grad_norm": 19.016260147094727, + "learning_rate": 9.067976140345854e-05, + "loss": 3.1746, + "step": 3580 + }, + { + "epoch": 5.24, + "grad_norm": 5.787333011627197, + "learning_rate": 9.06330631488011e-05, + "loss": 3.1975, + "step": 3581 + }, + { + "epoch": 5.24, + "grad_norm": 5.138857841491699, + "learning_rate": 9.058636695480698e-05, + "loss": 3.181, + "step": 3582 + }, + { + "epoch": 5.24, + "grad_norm": 11.81260871887207, + "learning_rate": 9.053967283174901e-05, + "loss": 3.1447, + "step": 3583 + }, + { + "epoch": 5.24, + "grad_norm": 6.516455173492432, + "learning_rate": 9.049298078989967e-05, + "loss": 3.1348, + "step": 3584 + }, + { + "epoch": 5.25, + "grad_norm": 6.460428237915039, + "learning_rate": 9.044629083953082e-05, + "loss": 3.1878, + "step": 3585 + }, + { + "epoch": 5.25, + "grad_norm": 5.189371585845947, + "learning_rate": 9.039960299091397e-05, + "loss": 3.1585, + "step": 3586 + }, + { + "epoch": 5.25, + "grad_norm": 4.265483379364014, + "learning_rate": 9.035291725432008e-05, + "loss": 3.1688, + "step": 3587 + }, + { + "epoch": 5.25, + "grad_norm": 2.625492572784424, + "learning_rate": 9.030623364001979e-05, + "loss": 3.1438, + "step": 3588 + }, + { + "epoch": 5.25, + "grad_norm": 3.3770155906677246, + "learning_rate": 9.025955215828312e-05, + "loss": 3.1574, + "step": 3589 + }, + { + "epoch": 5.25, + "grad_norm": 21.14669418334961, + "learning_rate": 9.021287281937967e-05, + "loss": 3.1473, + "step": 3590 + }, + { + "epoch": 5.25, + "grad_norm": 30.993927001953125, + "learning_rate": 9.016619563357865e-05, + "loss": 3.2186, + "step": 3591 + }, + { + "epoch": 5.26, + "grad_norm": 5.043474197387695, + "learning_rate": 9.011952061114867e-05, + "loss": 3.1602, + "step": 3592 + }, + { + "epoch": 5.26, + "grad_norm": 17.649402618408203, + "learning_rate": 9.007284776235791e-05, + "loss": 3.139, + "step": 3593 + }, + { + "epoch": 5.26, + "grad_norm": 50.776214599609375, + "learning_rate": 9.002617709747416e-05, + "loss": 3.1279, + "step": 3594 + }, + { + "epoch": 5.26, + "grad_norm": 12.535754203796387, + "learning_rate": 8.997950862676462e-05, + "loss": 3.174, + "step": 3595 + }, + { + "epoch": 5.26, + "grad_norm": 4.332036972045898, + "learning_rate": 8.993284236049598e-05, + "loss": 3.106, + "step": 3596 + }, + { + "epoch": 5.26, + "grad_norm": 1.4636400938034058, + "learning_rate": 8.988617830893455e-05, + "loss": 3.1456, + "step": 3597 + }, + { + "epoch": 5.27, + "grad_norm": 4.1321024894714355, + "learning_rate": 8.983951648234617e-05, + "loss": 3.102, + "step": 3598 + }, + { + "epoch": 5.27, + "grad_norm": 4.079101085662842, + "learning_rate": 8.979285689099601e-05, + "loss": 3.1632, + "step": 3599 + }, + { + "epoch": 5.27, + "grad_norm": 7.1509690284729, + "learning_rate": 8.974619954514895e-05, + "loss": 3.1862, + "step": 3600 + }, + { + "epoch": 5.27, + "grad_norm": 3.1132075786590576, + "learning_rate": 8.96995444550693e-05, + "loss": 3.1587, + "step": 3601 + }, + { + "epoch": 5.27, + "grad_norm": 15.199933052062988, + "learning_rate": 8.965289163102078e-05, + "loss": 3.1385, + "step": 3602 + }, + { + "epoch": 5.27, + "grad_norm": 3.226987600326538, + "learning_rate": 8.960624108326679e-05, + "loss": 3.1793, + "step": 3603 + }, + { + "epoch": 5.27, + "grad_norm": 3.0855495929718018, + "learning_rate": 8.955959282207013e-05, + "loss": 3.121, + "step": 3604 + }, + { + "epoch": 5.28, + "grad_norm": 1.8328437805175781, + "learning_rate": 8.951294685769303e-05, + "loss": 3.1515, + "step": 3605 + }, + { + "epoch": 5.28, + "grad_norm": 2.6738593578338623, + "learning_rate": 8.946630320039737e-05, + "loss": 3.1412, + "step": 3606 + }, + { + "epoch": 5.28, + "grad_norm": 2.790065288543701, + "learning_rate": 8.941966186044444e-05, + "loss": 3.1307, + "step": 3607 + }, + { + "epoch": 5.28, + "grad_norm": 3.6638455390930176, + "learning_rate": 8.937302284809494e-05, + "loss": 3.1155, + "step": 3608 + }, + { + "epoch": 5.28, + "grad_norm": 3.6409106254577637, + "learning_rate": 8.932638617360925e-05, + "loss": 3.1242, + "step": 3609 + }, + { + "epoch": 5.28, + "grad_norm": 2.8414742946624756, + "learning_rate": 8.927975184724704e-05, + "loss": 3.1433, + "step": 3610 + }, + { + "epoch": 5.28, + "grad_norm": 4.518361568450928, + "learning_rate": 8.92331198792676e-05, + "loss": 3.1883, + "step": 3611 + }, + { + "epoch": 5.29, + "grad_norm": 7.363691806793213, + "learning_rate": 8.918649027992958e-05, + "loss": 3.1231, + "step": 3612 + }, + { + "epoch": 5.29, + "grad_norm": 2.302523136138916, + "learning_rate": 8.913986305949125e-05, + "loss": 3.0972, + "step": 3613 + }, + { + "epoch": 5.29, + "grad_norm": 1.6068569421768188, + "learning_rate": 8.909323822821031e-05, + "loss": 3.1498, + "step": 3614 + }, + { + "epoch": 5.29, + "grad_norm": 1.9950542449951172, + "learning_rate": 8.90466157963438e-05, + "loss": 3.1339, + "step": 3615 + }, + { + "epoch": 5.29, + "grad_norm": 1.7338231801986694, + "learning_rate": 8.899999577414845e-05, + "loss": 3.1382, + "step": 3616 + }, + { + "epoch": 5.29, + "grad_norm": 2.231691598892212, + "learning_rate": 8.895337817188031e-05, + "loss": 3.1417, + "step": 3617 + }, + { + "epoch": 5.29, + "grad_norm": 1.9115151166915894, + "learning_rate": 8.890676299979492e-05, + "loss": 3.1338, + "step": 3618 + }, + { + "epoch": 5.3, + "grad_norm": 2.1142683029174805, + "learning_rate": 8.886015026814736e-05, + "loss": 3.1768, + "step": 3619 + }, + { + "epoch": 5.3, + "grad_norm": 2.7301149368286133, + "learning_rate": 8.881353998719206e-05, + "loss": 3.1843, + "step": 3620 + }, + { + "epoch": 5.3, + "grad_norm": 3.0939254760742188, + "learning_rate": 8.876693216718297e-05, + "loss": 3.1582, + "step": 3621 + }, + { + "epoch": 5.3, + "grad_norm": 1.2294076681137085, + "learning_rate": 8.872032681837358e-05, + "loss": 3.1484, + "step": 3622 + }, + { + "epoch": 5.3, + "grad_norm": 2.869823694229126, + "learning_rate": 8.867372395101666e-05, + "loss": 3.1455, + "step": 3623 + }, + { + "epoch": 5.3, + "grad_norm": 3.04744291305542, + "learning_rate": 8.862712357536454e-05, + "loss": 3.151, + "step": 3624 + }, + { + "epoch": 5.31, + "grad_norm": 1.6261351108551025, + "learning_rate": 8.858052570166908e-05, + "loss": 3.1434, + "step": 3625 + }, + { + "epoch": 5.31, + "grad_norm": 2.8542559146881104, + "learning_rate": 8.85339303401814e-05, + "loss": 3.1384, + "step": 3626 + }, + { + "epoch": 5.31, + "grad_norm": 0.9066136479377747, + "learning_rate": 8.848733750115218e-05, + "loss": 3.1576, + "step": 3627 + }, + { + "epoch": 5.31, + "grad_norm": 2.466120719909668, + "learning_rate": 8.84407471948316e-05, + "loss": 3.1478, + "step": 3628 + }, + { + "epoch": 5.31, + "grad_norm": 1.9956951141357422, + "learning_rate": 8.839415943146919e-05, + "loss": 3.1492, + "step": 3629 + }, + { + "epoch": 5.31, + "grad_norm": 10.860211372375488, + "learning_rate": 8.834757422131385e-05, + "loss": 3.198, + "step": 3630 + }, + { + "epoch": 5.31, + "grad_norm": 4.2302327156066895, + "learning_rate": 8.830099157461413e-05, + "loss": 3.146, + "step": 3631 + }, + { + "epoch": 5.32, + "grad_norm": 13.344452857971191, + "learning_rate": 8.825441150161788e-05, + "loss": 3.1234, + "step": 3632 + }, + { + "epoch": 5.32, + "grad_norm": 3.983452558517456, + "learning_rate": 8.820783401257236e-05, + "loss": 3.1775, + "step": 3633 + }, + { + "epoch": 5.32, + "grad_norm": 2.76181960105896, + "learning_rate": 8.816125911772433e-05, + "loss": 3.1776, + "step": 3634 + }, + { + "epoch": 5.32, + "grad_norm": 5.169229030609131, + "learning_rate": 8.811468682731997e-05, + "loss": 3.1174, + "step": 3635 + }, + { + "epoch": 5.32, + "grad_norm": 2.876774549484253, + "learning_rate": 8.806811715160485e-05, + "loss": 3.1506, + "step": 3636 + }, + { + "epoch": 5.32, + "grad_norm": 1.740065336227417, + "learning_rate": 8.802155010082396e-05, + "loss": 3.1375, + "step": 3637 + }, + { + "epoch": 5.32, + "grad_norm": 2.2916555404663086, + "learning_rate": 8.797498568522182e-05, + "loss": 3.1736, + "step": 3638 + }, + { + "epoch": 5.33, + "grad_norm": 13.75604248046875, + "learning_rate": 8.792842391504222e-05, + "loss": 3.1382, + "step": 3639 + }, + { + "epoch": 5.33, + "grad_norm": 8.454766273498535, + "learning_rate": 8.788186480052844e-05, + "loss": 3.1663, + "step": 3640 + }, + { + "epoch": 5.33, + "grad_norm": 3.627807140350342, + "learning_rate": 8.783530835192322e-05, + "loss": 3.1633, + "step": 3641 + }, + { + "epoch": 5.33, + "grad_norm": 1.9736120700836182, + "learning_rate": 8.778875457946866e-05, + "loss": 3.2027, + "step": 3642 + }, + { + "epoch": 5.33, + "grad_norm": 4.044384479522705, + "learning_rate": 8.77422034934062e-05, + "loss": 3.1484, + "step": 3643 + }, + { + "epoch": 5.33, + "grad_norm": 4.4502763748168945, + "learning_rate": 8.769565510397684e-05, + "loss": 3.1589, + "step": 3644 + }, + { + "epoch": 5.34, + "grad_norm": 2.187070369720459, + "learning_rate": 8.764910942142093e-05, + "loss": 3.1593, + "step": 3645 + }, + { + "epoch": 5.34, + "grad_norm": 2.6429078578948975, + "learning_rate": 8.760256645597813e-05, + "loss": 3.1466, + "step": 3646 + }, + { + "epoch": 5.34, + "grad_norm": 1.6300479173660278, + "learning_rate": 8.755602621788764e-05, + "loss": 3.1133, + "step": 3647 + }, + { + "epoch": 5.34, + "grad_norm": 2.9914662837982178, + "learning_rate": 8.750948871738802e-05, + "loss": 3.114, + "step": 3648 + }, + { + "epoch": 5.34, + "grad_norm": 2.315000534057617, + "learning_rate": 8.746295396471711e-05, + "loss": 3.1427, + "step": 3649 + }, + { + "epoch": 5.34, + "grad_norm": 3.0107791423797607, + "learning_rate": 8.741642197011235e-05, + "loss": 3.1085, + "step": 3650 + }, + { + "epoch": 5.34, + "grad_norm": 11.682272911071777, + "learning_rate": 8.736989274381045e-05, + "loss": 3.1428, + "step": 3651 + }, + { + "epoch": 5.35, + "grad_norm": 2.8496906757354736, + "learning_rate": 8.732336629604745e-05, + "loss": 3.1269, + "step": 3652 + }, + { + "epoch": 5.35, + "grad_norm": 2.9771509170532227, + "learning_rate": 8.727684263705895e-05, + "loss": 3.147, + "step": 3653 + }, + { + "epoch": 5.35, + "grad_norm": 2.937957525253296, + "learning_rate": 8.723032177707978e-05, + "loss": 3.1731, + "step": 3654 + }, + { + "epoch": 5.35, + "grad_norm": 3.5484066009521484, + "learning_rate": 8.718380372634422e-05, + "loss": 3.1602, + "step": 3655 + }, + { + "epoch": 5.35, + "grad_norm": 9.137898445129395, + "learning_rate": 8.7137288495086e-05, + "loss": 3.1692, + "step": 3656 + }, + { + "epoch": 5.35, + "grad_norm": 10.832776069641113, + "learning_rate": 8.709077609353809e-05, + "loss": 3.1829, + "step": 3657 + }, + { + "epoch": 5.35, + "grad_norm": 3.5576894283294678, + "learning_rate": 8.704426653193289e-05, + "loss": 3.1439, + "step": 3658 + }, + { + "epoch": 5.36, + "grad_norm": 2.771862745285034, + "learning_rate": 8.699775982050229e-05, + "loss": 3.1435, + "step": 3659 + }, + { + "epoch": 5.36, + "grad_norm": 7.5768046379089355, + "learning_rate": 8.695125596947736e-05, + "loss": 3.1898, + "step": 3660 + }, + { + "epoch": 5.36, + "grad_norm": 3.332085132598877, + "learning_rate": 8.690475498908868e-05, + "loss": 3.1365, + "step": 3661 + }, + { + "epoch": 5.36, + "grad_norm": 8.380770683288574, + "learning_rate": 8.685825688956613e-05, + "loss": 3.1733, + "step": 3662 + }, + { + "epoch": 5.36, + "grad_norm": 9.0435791015625, + "learning_rate": 8.6811761681139e-05, + "loss": 3.1752, + "step": 3663 + }, + { + "epoch": 5.36, + "grad_norm": 12.012619018554688, + "learning_rate": 8.676526937403591e-05, + "loss": 3.1615, + "step": 3664 + }, + { + "epoch": 5.36, + "grad_norm": 6.051881313323975, + "learning_rate": 8.671877997848484e-05, + "loss": 3.1719, + "step": 3665 + }, + { + "epoch": 5.37, + "grad_norm": 5.059371471405029, + "learning_rate": 8.667229350471322e-05, + "loss": 3.1775, + "step": 3666 + }, + { + "epoch": 5.37, + "grad_norm": 9.862431526184082, + "learning_rate": 8.662580996294767e-05, + "loss": 3.1795, + "step": 3667 + }, + { + "epoch": 5.37, + "grad_norm": 15.149982452392578, + "learning_rate": 8.657932936341428e-05, + "loss": 3.2141, + "step": 3668 + }, + { + "epoch": 5.37, + "grad_norm": 9.565960884094238, + "learning_rate": 8.653285171633855e-05, + "loss": 3.1831, + "step": 3669 + }, + { + "epoch": 5.37, + "grad_norm": 19.23344612121582, + "learning_rate": 8.648637703194516e-05, + "loss": 3.1925, + "step": 3670 + }, + { + "epoch": 5.37, + "grad_norm": 6.268587589263916, + "learning_rate": 8.643990532045822e-05, + "loss": 3.1621, + "step": 3671 + }, + { + "epoch": 5.38, + "grad_norm": 56.983970642089844, + "learning_rate": 8.639343659210131e-05, + "loss": 3.1715, + "step": 3672 + }, + { + "epoch": 5.38, + "grad_norm": 50.29884719848633, + "learning_rate": 8.634697085709711e-05, + "loss": 3.1474, + "step": 3673 + }, + { + "epoch": 5.38, + "grad_norm": 13.747417449951172, + "learning_rate": 8.630050812566781e-05, + "loss": 3.1805, + "step": 3674 + }, + { + "epoch": 5.38, + "grad_norm": 20.72515106201172, + "learning_rate": 8.625404840803498e-05, + "loss": 3.1528, + "step": 3675 + }, + { + "epoch": 5.38, + "grad_norm": 12.049095153808594, + "learning_rate": 8.620759171441935e-05, + "loss": 3.181, + "step": 3676 + }, + { + "epoch": 5.38, + "grad_norm": 11.251856803894043, + "learning_rate": 8.616113805504106e-05, + "loss": 3.1962, + "step": 3677 + }, + { + "epoch": 5.38, + "grad_norm": 6.218480587005615, + "learning_rate": 8.61146874401197e-05, + "loss": 3.2132, + "step": 3678 + }, + { + "epoch": 5.39, + "grad_norm": 10.480876922607422, + "learning_rate": 8.606823987987405e-05, + "loss": 3.1913, + "step": 3679 + }, + { + "epoch": 5.39, + "grad_norm": 12.099661827087402, + "learning_rate": 8.602179538452222e-05, + "loss": 3.178, + "step": 3680 + }, + { + "epoch": 5.39, + "grad_norm": 23.720500946044922, + "learning_rate": 8.597535396428177e-05, + "loss": 3.1558, + "step": 3681 + }, + { + "epoch": 5.39, + "grad_norm": 16.246944427490234, + "learning_rate": 8.592891562936947e-05, + "loss": 3.1834, + "step": 3682 + }, + { + "epoch": 5.39, + "grad_norm": 37.614784240722656, + "learning_rate": 8.58824803900014e-05, + "loss": 3.174, + "step": 3683 + }, + { + "epoch": 5.39, + "grad_norm": 9.572596549987793, + "learning_rate": 8.583604825639302e-05, + "loss": 3.1871, + "step": 3684 + }, + { + "epoch": 5.39, + "grad_norm": 6.099634170532227, + "learning_rate": 8.578961923875915e-05, + "loss": 3.1663, + "step": 3685 + }, + { + "epoch": 5.4, + "grad_norm": 18.694616317749023, + "learning_rate": 8.57431933473138e-05, + "loss": 3.1668, + "step": 3686 + }, + { + "epoch": 5.4, + "grad_norm": 133.21104431152344, + "learning_rate": 8.569677059227032e-05, + "loss": 3.1881, + "step": 3687 + }, + { + "epoch": 5.4, + "grad_norm": 33.713504791259766, + "learning_rate": 8.565035098384149e-05, + "loss": 3.1648, + "step": 3688 + }, + { + "epoch": 5.4, + "grad_norm": 9.273075103759766, + "learning_rate": 8.56039345322393e-05, + "loss": 3.2306, + "step": 3689 + }, + { + "epoch": 5.4, + "grad_norm": 18.504985809326172, + "learning_rate": 8.555752124767498e-05, + "loss": 3.2082, + "step": 3690 + }, + { + "epoch": 5.4, + "grad_norm": 13.903569221496582, + "learning_rate": 8.551111114035922e-05, + "loss": 3.1519, + "step": 3691 + }, + { + "epoch": 5.4, + "grad_norm": 48.394412994384766, + "learning_rate": 8.546470422050193e-05, + "loss": 3.1922, + "step": 3692 + }, + { + "epoch": 5.41, + "grad_norm": 9.13818073272705, + "learning_rate": 8.541830049831227e-05, + "loss": 3.2432, + "step": 3693 + }, + { + "epoch": 5.41, + "grad_norm": 7.568233013153076, + "learning_rate": 8.537189998399879e-05, + "loss": 3.2007, + "step": 3694 + }, + { + "epoch": 5.41, + "grad_norm": 110.37165832519531, + "learning_rate": 8.53255026877693e-05, + "loss": 3.1837, + "step": 3695 + }, + { + "epoch": 5.41, + "grad_norm": 8.272992134094238, + "learning_rate": 8.527910861983084e-05, + "loss": 3.169, + "step": 3696 + }, + { + "epoch": 5.41, + "grad_norm": 29.426700592041016, + "learning_rate": 8.523271779038986e-05, + "loss": 3.1793, + "step": 3697 + }, + { + "epoch": 5.41, + "grad_norm": 9.58652400970459, + "learning_rate": 8.518633020965199e-05, + "loss": 3.1662, + "step": 3698 + }, + { + "epoch": 5.42, + "grad_norm": 5.326894283294678, + "learning_rate": 8.513994588782217e-05, + "loss": 3.1817, + "step": 3699 + }, + { + "epoch": 5.42, + "grad_norm": 8.72939395904541, + "learning_rate": 8.509356483510471e-05, + "loss": 3.1888, + "step": 3700 + }, + { + "epoch": 5.42, + "grad_norm": 9.870657920837402, + "learning_rate": 8.504718706170306e-05, + "loss": 3.1772, + "step": 3701 + }, + { + "epoch": 5.42, + "grad_norm": 39.32550811767578, + "learning_rate": 8.500081257782003e-05, + "loss": 3.203, + "step": 3702 + }, + { + "epoch": 5.42, + "grad_norm": 20.613840103149414, + "learning_rate": 8.495444139365776e-05, + "loss": 3.1721, + "step": 3703 + }, + { + "epoch": 5.42, + "grad_norm": 10.104787826538086, + "learning_rate": 8.490807351941753e-05, + "loss": 3.1791, + "step": 3704 + }, + { + "epoch": 5.42, + "grad_norm": 8.861245155334473, + "learning_rate": 8.486170896529996e-05, + "loss": 3.2222, + "step": 3705 + }, + { + "epoch": 5.43, + "grad_norm": 40.4407844543457, + "learning_rate": 8.481534774150502e-05, + "loss": 3.3014, + "step": 3706 + }, + { + "epoch": 5.43, + "grad_norm": 4.760523319244385, + "learning_rate": 8.476898985823177e-05, + "loss": 3.1825, + "step": 3707 + }, + { + "epoch": 5.43, + "grad_norm": 1924.5662841796875, + "learning_rate": 8.472263532567871e-05, + "loss": 3.1802, + "step": 3708 + }, + { + "epoch": 5.43, + "grad_norm": 5.515054702758789, + "learning_rate": 8.467628415404345e-05, + "loss": 3.1833, + "step": 3709 + }, + { + "epoch": 5.43, + "grad_norm": 2.0650458335876465, + "learning_rate": 8.462993635352303e-05, + "loss": 3.184, + "step": 3710 + }, + { + "epoch": 5.43, + "grad_norm": 4.477024078369141, + "learning_rate": 8.458359193431358e-05, + "loss": 3.1489, + "step": 3711 + }, + { + "epoch": 5.43, + "grad_norm": 8.2962646484375, + "learning_rate": 8.453725090661057e-05, + "loss": 3.1318, + "step": 3712 + }, + { + "epoch": 5.44, + "grad_norm": 4.55861234664917, + "learning_rate": 8.449091328060877e-05, + "loss": 3.185, + "step": 3713 + }, + { + "epoch": 5.44, + "grad_norm": 6.914697647094727, + "learning_rate": 8.44445790665021e-05, + "loss": 3.1659, + "step": 3714 + }, + { + "epoch": 5.44, + "grad_norm": 5.971372127532959, + "learning_rate": 8.439824827448376e-05, + "loss": 3.1691, + "step": 3715 + }, + { + "epoch": 5.44, + "grad_norm": 3.358804702758789, + "learning_rate": 8.435192091474628e-05, + "loss": 3.187, + "step": 3716 + }, + { + "epoch": 5.44, + "grad_norm": 50.52218246459961, + "learning_rate": 8.430559699748132e-05, + "loss": 3.1595, + "step": 3717 + }, + { + "epoch": 5.44, + "grad_norm": 4.3211774826049805, + "learning_rate": 8.425927653287981e-05, + "loss": 3.2301, + "step": 3718 + }, + { + "epoch": 5.44, + "eval_loss": 3.003319025039673, + "eval_runtime": 1168.0997, + "eval_samples_per_second": 14.675, + "eval_steps_per_second": 0.918, + "step": 3718 + }, + { + "epoch": 5.44, + "grad_norm": 3.5544116497039795, + "learning_rate": 8.421295953113201e-05, + "loss": 3.1628, + "step": 3719 + }, + { + "epoch": 5.45, + "grad_norm": 4.997739791870117, + "learning_rate": 8.416664600242733e-05, + "loss": 3.1741, + "step": 3720 + }, + { + "epoch": 5.45, + "grad_norm": 2.1352174282073975, + "learning_rate": 8.412033595695437e-05, + "loss": 3.1518, + "step": 3721 + }, + { + "epoch": 5.45, + "grad_norm": 7.084630489349365, + "learning_rate": 8.40740294049011e-05, + "loss": 3.1283, + "step": 3722 + }, + { + "epoch": 5.45, + "grad_norm": 2.8684113025665283, + "learning_rate": 8.402772635645465e-05, + "loss": 3.1823, + "step": 3723 + }, + { + "epoch": 5.45, + "grad_norm": 14.201617240905762, + "learning_rate": 8.398142682180131e-05, + "loss": 3.14, + "step": 3724 + }, + { + "epoch": 5.45, + "grad_norm": 4.209918975830078, + "learning_rate": 8.393513081112674e-05, + "loss": 3.1867, + "step": 3725 + }, + { + "epoch": 5.46, + "grad_norm": 2.907388687133789, + "learning_rate": 8.388883833461578e-05, + "loss": 3.2095, + "step": 3726 + }, + { + "epoch": 5.46, + "grad_norm": 11.771166801452637, + "learning_rate": 8.384254940245233e-05, + "loss": 3.1938, + "step": 3727 + }, + { + "epoch": 5.46, + "grad_norm": 15.321097373962402, + "learning_rate": 8.379626402481977e-05, + "loss": 3.1755, + "step": 3728 + }, + { + "epoch": 5.46, + "grad_norm": 3.204758882522583, + "learning_rate": 8.374998221190056e-05, + "loss": 3.1617, + "step": 3729 + }, + { + "epoch": 5.46, + "grad_norm": 7.3266754150390625, + "learning_rate": 8.37037039738763e-05, + "loss": 3.1647, + "step": 3730 + }, + { + "epoch": 5.46, + "grad_norm": 108.37910461425781, + "learning_rate": 8.365742932092801e-05, + "loss": 3.1807, + "step": 3731 + }, + { + "epoch": 5.46, + "grad_norm": 4.3307623863220215, + "learning_rate": 8.361115826323572e-05, + "loss": 3.2109, + "step": 3732 + }, + { + "epoch": 5.47, + "grad_norm": 1.9860690832138062, + "learning_rate": 8.35648908109788e-05, + "loss": 3.1484, + "step": 3733 + }, + { + "epoch": 5.47, + "grad_norm": 3.039022207260132, + "learning_rate": 8.351862697433573e-05, + "loss": 3.175, + "step": 3734 + }, + { + "epoch": 5.47, + "grad_norm": 7.283559799194336, + "learning_rate": 8.347236676348426e-05, + "loss": 3.219, + "step": 3735 + }, + { + "epoch": 5.47, + "grad_norm": 3.774641513824463, + "learning_rate": 8.342611018860139e-05, + "loss": 3.1932, + "step": 3736 + }, + { + "epoch": 5.47, + "grad_norm": 4.264026641845703, + "learning_rate": 8.337985725986314e-05, + "loss": 3.193, + "step": 3737 + }, + { + "epoch": 5.47, + "grad_norm": 1.858871340751648, + "learning_rate": 8.333360798744496e-05, + "loss": 3.122, + "step": 3738 + }, + { + "epoch": 5.47, + "grad_norm": 4.920361518859863, + "learning_rate": 8.328736238152133e-05, + "loss": 3.1425, + "step": 3739 + }, + { + "epoch": 5.48, + "grad_norm": 4.10701847076416, + "learning_rate": 8.324112045226593e-05, + "loss": 3.1595, + "step": 3740 + }, + { + "epoch": 5.48, + "grad_norm": 5.043720245361328, + "learning_rate": 8.319488220985174e-05, + "loss": 3.1407, + "step": 3741 + }, + { + "epoch": 5.48, + "grad_norm": 3.795624256134033, + "learning_rate": 8.314864766445088e-05, + "loss": 3.2385, + "step": 3742 + }, + { + "epoch": 5.48, + "grad_norm": 3.801236867904663, + "learning_rate": 8.310241682623454e-05, + "loss": 3.152, + "step": 3743 + }, + { + "epoch": 5.48, + "grad_norm": 2.491832971572876, + "learning_rate": 8.30561897053733e-05, + "loss": 3.1737, + "step": 3744 + }, + { + "epoch": 5.48, + "grad_norm": 2.1997487545013428, + "learning_rate": 8.300996631203675e-05, + "loss": 3.1273, + "step": 3745 + }, + { + "epoch": 5.49, + "grad_norm": 2.1604340076446533, + "learning_rate": 8.296374665639373e-05, + "loss": 3.2015, + "step": 3746 + }, + { + "epoch": 5.49, + "grad_norm": 14.758793830871582, + "learning_rate": 8.291753074861233e-05, + "loss": 3.1594, + "step": 3747 + }, + { + "epoch": 5.49, + "grad_norm": 1.7705718278884888, + "learning_rate": 8.287131859885964e-05, + "loss": 3.1615, + "step": 3748 + }, + { + "epoch": 5.49, + "grad_norm": 3.673743724822998, + "learning_rate": 8.282511021730206e-05, + "loss": 3.1665, + "step": 3749 + }, + { + "epoch": 5.49, + "grad_norm": 2.580552101135254, + "learning_rate": 8.277890561410518e-05, + "loss": 3.1972, + "step": 3750 + }, + { + "epoch": 5.49, + "grad_norm": 1.8429394960403442, + "learning_rate": 8.273270479943361e-05, + "loss": 3.1316, + "step": 3751 + }, + { + "epoch": 5.49, + "grad_norm": 1.4963501691818237, + "learning_rate": 8.268650778345126e-05, + "loss": 3.1498, + "step": 3752 + }, + { + "epoch": 5.5, + "grad_norm": 3.3234293460845947, + "learning_rate": 8.26403145763212e-05, + "loss": 3.1475, + "step": 3753 + }, + { + "epoch": 5.5, + "grad_norm": 3.2355213165283203, + "learning_rate": 8.25941251882056e-05, + "loss": 3.1404, + "step": 3754 + }, + { + "epoch": 5.5, + "grad_norm": 1.3712440729141235, + "learning_rate": 8.254793962926579e-05, + "loss": 3.1316, + "step": 3755 + }, + { + "epoch": 5.5, + "grad_norm": 1.9082093238830566, + "learning_rate": 8.250175790966228e-05, + "loss": 3.1388, + "step": 3756 + }, + { + "epoch": 5.5, + "grad_norm": 2.096512794494629, + "learning_rate": 8.245558003955481e-05, + "loss": 3.1753, + "step": 3757 + }, + { + "epoch": 5.5, + "grad_norm": 1.1812965869903564, + "learning_rate": 8.240940602910212e-05, + "loss": 3.1647, + "step": 3758 + }, + { + "epoch": 5.5, + "grad_norm": 2.799182415008545, + "learning_rate": 8.23632358884622e-05, + "loss": 3.1474, + "step": 3759 + }, + { + "epoch": 5.51, + "grad_norm": 2.4367105960845947, + "learning_rate": 8.231706962779222e-05, + "loss": 3.1312, + "step": 3760 + }, + { + "epoch": 5.51, + "grad_norm": 2.1872997283935547, + "learning_rate": 8.227090725724842e-05, + "loss": 3.0935, + "step": 3761 + }, + { + "epoch": 5.51, + "grad_norm": 3.3611397743225098, + "learning_rate": 8.222474878698617e-05, + "loss": 3.0987, + "step": 3762 + }, + { + "epoch": 5.51, + "grad_norm": 1.0651507377624512, + "learning_rate": 8.217859422716012e-05, + "loss": 3.183, + "step": 3763 + }, + { + "epoch": 5.51, + "grad_norm": 0.8456958532333374, + "learning_rate": 8.213244358792389e-05, + "loss": 3.123, + "step": 3764 + }, + { + "epoch": 5.51, + "grad_norm": 0.9663138389587402, + "learning_rate": 8.208629687943029e-05, + "loss": 3.16, + "step": 3765 + }, + { + "epoch": 5.51, + "grad_norm": 135.12335205078125, + "learning_rate": 8.204015411183135e-05, + "loss": 3.1066, + "step": 3766 + }, + { + "epoch": 5.52, + "grad_norm": 1.7855501174926758, + "learning_rate": 8.199401529527816e-05, + "loss": 3.0972, + "step": 3767 + }, + { + "epoch": 5.52, + "grad_norm": 1.187641978263855, + "learning_rate": 8.194788043992089e-05, + "loss": 3.1476, + "step": 3768 + }, + { + "epoch": 5.52, + "grad_norm": 6.5035834312438965, + "learning_rate": 8.190174955590897e-05, + "loss": 3.134, + "step": 3769 + }, + { + "epoch": 5.52, + "grad_norm": 3.347416400909424, + "learning_rate": 8.185562265339088e-05, + "loss": 3.1592, + "step": 3770 + }, + { + "epoch": 5.52, + "grad_norm": 2.2179319858551025, + "learning_rate": 8.180949974251416e-05, + "loss": 3.1656, + "step": 3771 + }, + { + "epoch": 5.52, + "grad_norm": 3.2867071628570557, + "learning_rate": 8.176338083342561e-05, + "loss": 3.1231, + "step": 3772 + }, + { + "epoch": 5.53, + "grad_norm": 4.15543794631958, + "learning_rate": 8.17172659362711e-05, + "loss": 3.1501, + "step": 3773 + }, + { + "epoch": 5.53, + "grad_norm": 3.2263450622558594, + "learning_rate": 8.167115506119549e-05, + "loss": 3.1269, + "step": 3774 + }, + { + "epoch": 5.53, + "grad_norm": 1.2961052656173706, + "learning_rate": 8.162504821834295e-05, + "loss": 3.1204, + "step": 3775 + }, + { + "epoch": 5.53, + "grad_norm": 2.9037585258483887, + "learning_rate": 8.157894541785671e-05, + "loss": 3.1801, + "step": 3776 + }, + { + "epoch": 5.53, + "grad_norm": 2.3933663368225098, + "learning_rate": 8.153284666987897e-05, + "loss": 3.1537, + "step": 3777 + }, + { + "epoch": 5.53, + "grad_norm": 1.6720260381698608, + "learning_rate": 8.148675198455123e-05, + "loss": 3.1501, + "step": 3778 + }, + { + "epoch": 5.53, + "grad_norm": 11.296658515930176, + "learning_rate": 8.144066137201398e-05, + "loss": 3.1364, + "step": 3779 + }, + { + "epoch": 5.54, + "grad_norm": 14.168936729431152, + "learning_rate": 8.139457484240687e-05, + "loss": 3.1629, + "step": 3780 + }, + { + "epoch": 5.54, + "grad_norm": 5.073040008544922, + "learning_rate": 8.134849240586854e-05, + "loss": 3.1765, + "step": 3781 + }, + { + "epoch": 5.54, + "grad_norm": 2.859703779220581, + "learning_rate": 8.130241407253693e-05, + "loss": 3.1252, + "step": 3782 + }, + { + "epoch": 5.54, + "grad_norm": 6.422140121459961, + "learning_rate": 8.125633985254893e-05, + "loss": 3.126, + "step": 3783 + }, + { + "epoch": 5.54, + "grad_norm": 16.3258056640625, + "learning_rate": 8.121026975604052e-05, + "loss": 3.1953, + "step": 3784 + }, + { + "epoch": 5.54, + "grad_norm": 7.476664066314697, + "learning_rate": 8.116420379314684e-05, + "loss": 3.1981, + "step": 3785 + }, + { + "epoch": 5.54, + "grad_norm": 57.822113037109375, + "learning_rate": 8.111814197400213e-05, + "loss": 3.1517, + "step": 3786 + }, + { + "epoch": 5.55, + "grad_norm": 6.158851146697998, + "learning_rate": 8.10720843087396e-05, + "loss": 3.1655, + "step": 3787 + }, + { + "epoch": 5.55, + "grad_norm": 7.291560173034668, + "learning_rate": 8.102603080749171e-05, + "loss": 3.113, + "step": 3788 + }, + { + "epoch": 5.55, + "grad_norm": 6.275900840759277, + "learning_rate": 8.097998148038985e-05, + "loss": 3.1209, + "step": 3789 + }, + { + "epoch": 5.55, + "grad_norm": 4.1805338859558105, + "learning_rate": 8.093393633756459e-05, + "loss": 3.1049, + "step": 3790 + }, + { + "epoch": 5.55, + "grad_norm": 6.27461051940918, + "learning_rate": 8.088789538914561e-05, + "loss": 3.169, + "step": 3791 + }, + { + "epoch": 5.55, + "grad_norm": 63.376373291015625, + "learning_rate": 8.084185864526153e-05, + "loss": 3.2346, + "step": 3792 + }, + { + "epoch": 5.55, + "grad_norm": 17.007471084594727, + "learning_rate": 8.079582611604015e-05, + "loss": 3.1799, + "step": 3793 + }, + { + "epoch": 5.56, + "grad_norm": 3.8963406085968018, + "learning_rate": 8.074979781160836e-05, + "loss": 3.1546, + "step": 3794 + }, + { + "epoch": 5.56, + "grad_norm": 1.5541036128997803, + "learning_rate": 8.070377374209202e-05, + "loss": 3.17, + "step": 3795 + }, + { + "epoch": 5.56, + "grad_norm": 4.511335372924805, + "learning_rate": 8.06577539176161e-05, + "loss": 3.1307, + "step": 3796 + }, + { + "epoch": 5.56, + "grad_norm": 3.2192275524139404, + "learning_rate": 8.061173834830476e-05, + "loss": 3.1262, + "step": 3797 + }, + { + "epoch": 5.56, + "grad_norm": 10.192057609558105, + "learning_rate": 8.056572704428103e-05, + "loss": 3.1341, + "step": 3798 + }, + { + "epoch": 5.56, + "grad_norm": 10.097148895263672, + "learning_rate": 8.051972001566708e-05, + "loss": 3.1627, + "step": 3799 + }, + { + "epoch": 5.57, + "grad_norm": 5.317000865936279, + "learning_rate": 8.04737172725842e-05, + "loss": 3.1695, + "step": 3800 + }, + { + "epoch": 5.57, + "grad_norm": 3.3502695560455322, + "learning_rate": 8.042771882515265e-05, + "loss": 3.1111, + "step": 3801 + }, + { + "epoch": 5.57, + "grad_norm": 3.3459572792053223, + "learning_rate": 8.038172468349178e-05, + "loss": 3.1737, + "step": 3802 + }, + { + "epoch": 5.57, + "grad_norm": 4.4290900230407715, + "learning_rate": 8.033573485772001e-05, + "loss": 3.1335, + "step": 3803 + }, + { + "epoch": 5.57, + "grad_norm": 3.444740056991577, + "learning_rate": 8.028974935795481e-05, + "loss": 3.1575, + "step": 3804 + }, + { + "epoch": 5.57, + "grad_norm": 6.389871120452881, + "learning_rate": 8.024376819431266e-05, + "loss": 3.1654, + "step": 3805 + }, + { + "epoch": 5.57, + "grad_norm": 4.629978179931641, + "learning_rate": 8.019779137690906e-05, + "loss": 3.1554, + "step": 3806 + }, + { + "epoch": 5.58, + "grad_norm": 2.2746245861053467, + "learning_rate": 8.015181891585871e-05, + "loss": 3.1065, + "step": 3807 + }, + { + "epoch": 5.58, + "grad_norm": 4.390033721923828, + "learning_rate": 8.010585082127513e-05, + "loss": 3.142, + "step": 3808 + }, + { + "epoch": 5.58, + "grad_norm": 8.950037956237793, + "learning_rate": 8.005988710327105e-05, + "loss": 3.1759, + "step": 3809 + }, + { + "epoch": 5.58, + "grad_norm": 7.786754608154297, + "learning_rate": 8.001392777195821e-05, + "loss": 3.1518, + "step": 3810 + }, + { + "epoch": 5.58, + "grad_norm": 1.803338646888733, + "learning_rate": 7.996797283744732e-05, + "loss": 3.1452, + "step": 3811 + }, + { + "epoch": 5.58, + "grad_norm": 5.436160087585449, + "learning_rate": 7.992202230984811e-05, + "loss": 3.1284, + "step": 3812 + }, + { + "epoch": 5.58, + "grad_norm": 5.9366912841796875, + "learning_rate": 7.987607619926946e-05, + "loss": 3.1547, + "step": 3813 + }, + { + "epoch": 5.59, + "grad_norm": 5.989798545837402, + "learning_rate": 7.98301345158192e-05, + "loss": 3.1264, + "step": 3814 + }, + { + "epoch": 5.59, + "grad_norm": 2.005276679992676, + "learning_rate": 7.978419726960413e-05, + "loss": 3.1387, + "step": 3815 + }, + { + "epoch": 5.59, + "grad_norm": 4.961897373199463, + "learning_rate": 7.97382644707302e-05, + "loss": 3.1506, + "step": 3816 + }, + { + "epoch": 5.59, + "grad_norm": 5.611112594604492, + "learning_rate": 7.969233612930234e-05, + "loss": 3.1237, + "step": 3817 + }, + { + "epoch": 5.59, + "grad_norm": 2.2314023971557617, + "learning_rate": 7.964641225542438e-05, + "loss": 3.1433, + "step": 3818 + }, + { + "epoch": 5.59, + "grad_norm": 27.2403564453125, + "learning_rate": 7.960049285919933e-05, + "loss": 3.1587, + "step": 3819 + }, + { + "epoch": 5.59, + "grad_norm": 7.609023571014404, + "learning_rate": 7.955457795072918e-05, + "loss": 3.2031, + "step": 3820 + }, + { + "epoch": 5.6, + "grad_norm": 6.942310810089111, + "learning_rate": 7.95086675401148e-05, + "loss": 3.1571, + "step": 3821 + }, + { + "epoch": 5.6, + "grad_norm": 5.55133056640625, + "learning_rate": 7.94627616374563e-05, + "loss": 3.1465, + "step": 3822 + }, + { + "epoch": 5.6, + "grad_norm": 2.6575374603271484, + "learning_rate": 7.941686025285257e-05, + "loss": 3.1384, + "step": 3823 + }, + { + "epoch": 5.6, + "grad_norm": 4.183603763580322, + "learning_rate": 7.937096339640162e-05, + "loss": 3.1551, + "step": 3824 + }, + { + "epoch": 5.6, + "grad_norm": 203.84231567382812, + "learning_rate": 7.932507107820051e-05, + "loss": 3.1756, + "step": 3825 + }, + { + "epoch": 5.6, + "grad_norm": 1.8645808696746826, + "learning_rate": 7.92791833083452e-05, + "loss": 3.1254, + "step": 3826 + }, + { + "epoch": 5.61, + "grad_norm": 3.2966809272766113, + "learning_rate": 7.923330009693071e-05, + "loss": 3.1409, + "step": 3827 + }, + { + "epoch": 5.61, + "grad_norm": 16.701486587524414, + "learning_rate": 7.918742145405098e-05, + "loss": 3.1687, + "step": 3828 + }, + { + "epoch": 5.61, + "grad_norm": 5.320707321166992, + "learning_rate": 7.914154738979908e-05, + "loss": 3.166, + "step": 3829 + }, + { + "epoch": 5.61, + "grad_norm": 3.770569086074829, + "learning_rate": 7.909567791426697e-05, + "loss": 3.129, + "step": 3830 + }, + { + "epoch": 5.61, + "grad_norm": 1.9080885648727417, + "learning_rate": 7.904981303754561e-05, + "loss": 3.1792, + "step": 3831 + }, + { + "epoch": 5.61, + "grad_norm": 2.2089335918426514, + "learning_rate": 7.900395276972502e-05, + "loss": 3.1417, + "step": 3832 + }, + { + "epoch": 5.61, + "grad_norm": 2.3139476776123047, + "learning_rate": 7.895809712089406e-05, + "loss": 3.1278, + "step": 3833 + }, + { + "epoch": 5.62, + "grad_norm": 3.4870378971099854, + "learning_rate": 7.891224610114072e-05, + "loss": 3.1562, + "step": 3834 + }, + { + "epoch": 5.62, + "grad_norm": 4.3670244216918945, + "learning_rate": 7.886639972055195e-05, + "loss": 3.1389, + "step": 3835 + }, + { + "epoch": 5.62, + "grad_norm": 5.9168596267700195, + "learning_rate": 7.882055798921358e-05, + "loss": 3.1104, + "step": 3836 + }, + { + "epoch": 5.62, + "grad_norm": 7.184929370880127, + "learning_rate": 7.87747209172105e-05, + "loss": 3.1555, + "step": 3837 + }, + { + "epoch": 5.62, + "grad_norm": 4.011050701141357, + "learning_rate": 7.872888851462663e-05, + "loss": 3.1192, + "step": 3838 + }, + { + "epoch": 5.62, + "grad_norm": 1.6208895444869995, + "learning_rate": 7.86830607915447e-05, + "loss": 3.1861, + "step": 3839 + }, + { + "epoch": 5.62, + "grad_norm": 1.3240761756896973, + "learning_rate": 7.863723775804651e-05, + "loss": 3.1656, + "step": 3840 + }, + { + "epoch": 5.63, + "grad_norm": 2.556436538696289, + "learning_rate": 7.85914194242129e-05, + "loss": 3.1392, + "step": 3841 + }, + { + "epoch": 5.63, + "grad_norm": 1.8541210889816284, + "learning_rate": 7.854560580012351e-05, + "loss": 3.1391, + "step": 3842 + }, + { + "epoch": 5.63, + "grad_norm": 2.840593099594116, + "learning_rate": 7.849979689585705e-05, + "loss": 3.0979, + "step": 3843 + }, + { + "epoch": 5.63, + "grad_norm": 1.50877845287323, + "learning_rate": 7.845399272149124e-05, + "loss": 3.1231, + "step": 3844 + }, + { + "epoch": 5.63, + "grad_norm": 2.6628737449645996, + "learning_rate": 7.84081932871026e-05, + "loss": 3.1347, + "step": 3845 + }, + { + "epoch": 5.63, + "grad_norm": 33.729000091552734, + "learning_rate": 7.836239860276671e-05, + "loss": 3.1319, + "step": 3846 + }, + { + "epoch": 5.64, + "grad_norm": 13.277280807495117, + "learning_rate": 7.831660867855816e-05, + "loss": 3.1439, + "step": 3847 + }, + { + "epoch": 5.64, + "grad_norm": 1.8134537935256958, + "learning_rate": 7.827082352455039e-05, + "loss": 3.1507, + "step": 3848 + }, + { + "epoch": 5.64, + "grad_norm": 2.101228952407837, + "learning_rate": 7.822504315081578e-05, + "loss": 3.1133, + "step": 3849 + }, + { + "epoch": 5.64, + "grad_norm": 2.533062696456909, + "learning_rate": 7.817926756742576e-05, + "loss": 3.1362, + "step": 3850 + }, + { + "epoch": 5.64, + "grad_norm": 4.869479656219482, + "learning_rate": 7.813349678445067e-05, + "loss": 3.1618, + "step": 3851 + }, + { + "epoch": 5.64, + "grad_norm": 2.6743147373199463, + "learning_rate": 7.808773081195972e-05, + "loss": 3.1278, + "step": 3852 + }, + { + "epoch": 5.64, + "grad_norm": 0.8585074543952942, + "learning_rate": 7.804196966002113e-05, + "loss": 3.1386, + "step": 3853 + }, + { + "epoch": 5.65, + "grad_norm": 7.510640621185303, + "learning_rate": 7.799621333870208e-05, + "loss": 3.1786, + "step": 3854 + }, + { + "epoch": 5.65, + "grad_norm": 2.670872449874878, + "learning_rate": 7.795046185806866e-05, + "loss": 3.1483, + "step": 3855 + }, + { + "epoch": 5.65, + "grad_norm": 22.928964614868164, + "learning_rate": 7.79047152281858e-05, + "loss": 3.1633, + "step": 3856 + }, + { + "epoch": 5.65, + "grad_norm": 7.530433177947998, + "learning_rate": 7.785897345911754e-05, + "loss": 3.1601, + "step": 3857 + }, + { + "epoch": 5.65, + "grad_norm": 7.938250541687012, + "learning_rate": 7.781323656092677e-05, + "loss": 3.1554, + "step": 3858 + }, + { + "epoch": 5.65, + "grad_norm": 2.2199909687042236, + "learning_rate": 7.776750454367521e-05, + "loss": 3.1512, + "step": 3859 + }, + { + "epoch": 5.65, + "grad_norm": 92.11273956298828, + "learning_rate": 7.772177741742369e-05, + "loss": 3.1275, + "step": 3860 + }, + { + "epoch": 5.66, + "grad_norm": 4.906170845031738, + "learning_rate": 7.767605519223183e-05, + "loss": 3.1245, + "step": 3861 + }, + { + "epoch": 5.66, + "grad_norm": 3.870858907699585, + "learning_rate": 7.76303378781582e-05, + "loss": 3.1105, + "step": 3862 + }, + { + "epoch": 5.66, + "grad_norm": 2.2505300045013428, + "learning_rate": 7.758462548526033e-05, + "loss": 3.1065, + "step": 3863 + }, + { + "epoch": 5.66, + "grad_norm": 2.947871208190918, + "learning_rate": 7.753891802359466e-05, + "loss": 3.1869, + "step": 3864 + }, + { + "epoch": 5.66, + "grad_norm": 3.654209613800049, + "learning_rate": 7.749321550321645e-05, + "loss": 3.1427, + "step": 3865 + }, + { + "epoch": 5.66, + "grad_norm": 3.259089469909668, + "learning_rate": 7.744751793418005e-05, + "loss": 3.1221, + "step": 3866 + }, + { + "epoch": 5.66, + "grad_norm": 2.3307595252990723, + "learning_rate": 7.740182532653853e-05, + "loss": 3.1556, + "step": 3867 + }, + { + "epoch": 5.67, + "grad_norm": 6.026763439178467, + "learning_rate": 7.735613769034399e-05, + "loss": 3.1646, + "step": 3868 + }, + { + "epoch": 5.67, + "grad_norm": 9.292561531066895, + "learning_rate": 7.731045503564743e-05, + "loss": 3.1498, + "step": 3869 + }, + { + "epoch": 5.67, + "grad_norm": 4.8578314781188965, + "learning_rate": 7.726477737249869e-05, + "loss": 3.1159, + "step": 3870 + }, + { + "epoch": 5.67, + "grad_norm": 5.385825157165527, + "learning_rate": 7.721910471094653e-05, + "loss": 3.1494, + "step": 3871 + }, + { + "epoch": 5.67, + "grad_norm": 3.9133341312408447, + "learning_rate": 7.717343706103873e-05, + "loss": 3.1402, + "step": 3872 + }, + { + "epoch": 5.67, + "grad_norm": 6.215272903442383, + "learning_rate": 7.712777443282175e-05, + "loss": 3.195, + "step": 3873 + }, + { + "epoch": 5.68, + "grad_norm": 3.6820201873779297, + "learning_rate": 7.708211683634112e-05, + "loss": 3.2287, + "step": 3874 + }, + { + "epoch": 5.68, + "grad_norm": 3.4176886081695557, + "learning_rate": 7.703646428164122e-05, + "loss": 3.1445, + "step": 3875 + }, + { + "epoch": 5.68, + "grad_norm": 8.820064544677734, + "learning_rate": 7.699081677876526e-05, + "loss": 3.2085, + "step": 3876 + }, + { + "epoch": 5.68, + "grad_norm": 17.15256118774414, + "learning_rate": 7.694517433775543e-05, + "loss": 3.1627, + "step": 3877 + }, + { + "epoch": 5.68, + "grad_norm": 1.7407811880111694, + "learning_rate": 7.689953696865271e-05, + "loss": 3.1868, + "step": 3878 + }, + { + "epoch": 5.68, + "grad_norm": 3.167860746383667, + "learning_rate": 7.685390468149707e-05, + "loss": 3.1838, + "step": 3879 + }, + { + "epoch": 5.68, + "grad_norm": 4.515937805175781, + "learning_rate": 7.680827748632728e-05, + "loss": 3.1624, + "step": 3880 + }, + { + "epoch": 5.69, + "grad_norm": 5.7442755699157715, + "learning_rate": 7.676265539318098e-05, + "loss": 3.1957, + "step": 3881 + }, + { + "epoch": 5.69, + "grad_norm": 4.945896625518799, + "learning_rate": 7.671703841209481e-05, + "loss": 3.1714, + "step": 3882 + }, + { + "epoch": 5.69, + "grad_norm": 8.167930603027344, + "learning_rate": 7.667142655310414e-05, + "loss": 3.1944, + "step": 3883 + }, + { + "epoch": 5.69, + "grad_norm": 17.595746994018555, + "learning_rate": 7.662581982624326e-05, + "loss": 3.1853, + "step": 3884 + }, + { + "epoch": 5.69, + "grad_norm": 2.751532554626465, + "learning_rate": 7.658021824154542e-05, + "loss": 3.1348, + "step": 3885 + }, + { + "epoch": 5.69, + "grad_norm": 6.661686897277832, + "learning_rate": 7.653462180904258e-05, + "loss": 3.2022, + "step": 3886 + }, + { + "epoch": 5.69, + "grad_norm": 9.880109786987305, + "learning_rate": 7.648903053876568e-05, + "loss": 3.1686, + "step": 3887 + }, + { + "epoch": 5.69, + "eval_loss": 2.989283800125122, + "eval_runtime": 1170.3187, + "eval_samples_per_second": 14.647, + "eval_steps_per_second": 0.916, + "step": 3887 + }, + { + "epoch": 5.7, + "grad_norm": 3.414788246154785, + "learning_rate": 7.644344444074453e-05, + "loss": 3.1669, + "step": 3888 + }, + { + "epoch": 5.7, + "grad_norm": 3.5142674446105957, + "learning_rate": 7.639786352500773e-05, + "loss": 3.1526, + "step": 3889 + }, + { + "epoch": 5.7, + "grad_norm": 7.385360240936279, + "learning_rate": 7.635228780158274e-05, + "loss": 3.1528, + "step": 3890 + }, + { + "epoch": 5.7, + "grad_norm": 7.439982891082764, + "learning_rate": 7.630671728049598e-05, + "loss": 3.1417, + "step": 3891 + }, + { + "epoch": 5.7, + "grad_norm": 4.842582702636719, + "learning_rate": 7.626115197177266e-05, + "loss": 3.2058, + "step": 3892 + }, + { + "epoch": 5.7, + "grad_norm": 6.158822536468506, + "learning_rate": 7.621559188543677e-05, + "loss": 3.2051, + "step": 3893 + }, + { + "epoch": 5.7, + "grad_norm": 4.262391090393066, + "learning_rate": 7.617003703151129e-05, + "loss": 3.2001, + "step": 3894 + }, + { + "epoch": 5.71, + "grad_norm": 22.497093200683594, + "learning_rate": 7.612448742001797e-05, + "loss": 3.1808, + "step": 3895 + }, + { + "epoch": 5.71, + "grad_norm": 3.1336076259613037, + "learning_rate": 7.607894306097739e-05, + "loss": 3.2113, + "step": 3896 + }, + { + "epoch": 5.71, + "grad_norm": 4.773746967315674, + "learning_rate": 7.603340396440902e-05, + "loss": 3.1906, + "step": 3897 + }, + { + "epoch": 5.71, + "grad_norm": 5.188856601715088, + "learning_rate": 7.59878701403312e-05, + "loss": 3.1663, + "step": 3898 + }, + { + "epoch": 5.71, + "grad_norm": 7.35548734664917, + "learning_rate": 7.594234159876098e-05, + "loss": 3.1467, + "step": 3899 + }, + { + "epoch": 5.71, + "grad_norm": 7.549004077911377, + "learning_rate": 7.589681834971443e-05, + "loss": 3.1928, + "step": 3900 + }, + { + "epoch": 5.72, + "grad_norm": 6.5583086013793945, + "learning_rate": 7.585130040320627e-05, + "loss": 3.2067, + "step": 3901 + }, + { + "epoch": 5.72, + "grad_norm": 14.854689598083496, + "learning_rate": 7.580578776925022e-05, + "loss": 3.2208, + "step": 3902 + }, + { + "epoch": 5.72, + "grad_norm": 11.917444229125977, + "learning_rate": 7.576028045785868e-05, + "loss": 3.1828, + "step": 3903 + }, + { + "epoch": 5.72, + "grad_norm": 6.152881145477295, + "learning_rate": 7.5714778479043e-05, + "loss": 3.204, + "step": 3904 + }, + { + "epoch": 5.72, + "grad_norm": 4.3145294189453125, + "learning_rate": 7.566928184281334e-05, + "loss": 3.2282, + "step": 3905 + }, + { + "epoch": 5.72, + "grad_norm": 3.392911434173584, + "learning_rate": 7.562379055917856e-05, + "loss": 3.1795, + "step": 3906 + }, + { + "epoch": 5.72, + "grad_norm": 5.250206470489502, + "learning_rate": 7.557830463814652e-05, + "loss": 3.1574, + "step": 3907 + }, + { + "epoch": 5.73, + "grad_norm": 11.659444808959961, + "learning_rate": 7.553282408972382e-05, + "loss": 3.1975, + "step": 3908 + }, + { + "epoch": 5.73, + "grad_norm": 6.504120826721191, + "learning_rate": 7.54873489239158e-05, + "loss": 3.2318, + "step": 3909 + }, + { + "epoch": 5.73, + "grad_norm": 3.6852314472198486, + "learning_rate": 7.544187915072676e-05, + "loss": 3.1501, + "step": 3910 + }, + { + "epoch": 5.73, + "grad_norm": 4.429998874664307, + "learning_rate": 7.539641478015976e-05, + "loss": 3.1623, + "step": 3911 + }, + { + "epoch": 5.73, + "grad_norm": 2.792065382003784, + "learning_rate": 7.535095582221656e-05, + "loss": 3.1624, + "step": 3912 + }, + { + "epoch": 5.73, + "grad_norm": 21.118961334228516, + "learning_rate": 7.530550228689794e-05, + "loss": 3.2112, + "step": 3913 + }, + { + "epoch": 5.73, + "grad_norm": 18.30959701538086, + "learning_rate": 7.526005418420332e-05, + "loss": 3.2394, + "step": 3914 + }, + { + "epoch": 5.74, + "grad_norm": 8.257102012634277, + "learning_rate": 7.521461152413092e-05, + "loss": 3.2588, + "step": 3915 + }, + { + "epoch": 5.74, + "grad_norm": 3.5304644107818604, + "learning_rate": 7.516917431667795e-05, + "loss": 3.2331, + "step": 3916 + }, + { + "epoch": 5.74, + "grad_norm": 24.953046798706055, + "learning_rate": 7.512374257184019e-05, + "loss": 3.1763, + "step": 3917 + }, + { + "epoch": 5.74, + "grad_norm": 5.283435821533203, + "learning_rate": 7.507831629961234e-05, + "loss": 3.1631, + "step": 3918 + }, + { + "epoch": 5.74, + "grad_norm": 7.136661052703857, + "learning_rate": 7.503289550998791e-05, + "loss": 3.2551, + "step": 3919 + }, + { + "epoch": 5.74, + "grad_norm": 3.8187716007232666, + "learning_rate": 7.498748021295914e-05, + "loss": 3.1214, + "step": 3920 + }, + { + "epoch": 5.74, + "grad_norm": 10.967880249023438, + "learning_rate": 7.494207041851708e-05, + "loss": 3.144, + "step": 3921 + }, + { + "epoch": 5.75, + "grad_norm": 30.177751541137695, + "learning_rate": 7.489666613665163e-05, + "loss": 3.1466, + "step": 3922 + }, + { + "epoch": 5.75, + "grad_norm": 6.267055511474609, + "learning_rate": 7.48512673773514e-05, + "loss": 3.2192, + "step": 3923 + }, + { + "epoch": 5.75, + "grad_norm": 3.4286611080169678, + "learning_rate": 7.480587415060378e-05, + "loss": 3.1519, + "step": 3924 + }, + { + "epoch": 5.75, + "grad_norm": 9.199522972106934, + "learning_rate": 7.476048646639498e-05, + "loss": 3.1738, + "step": 3925 + }, + { + "epoch": 5.75, + "grad_norm": 9.091236114501953, + "learning_rate": 7.471510433471005e-05, + "loss": 3.1704, + "step": 3926 + }, + { + "epoch": 5.75, + "grad_norm": 7.052969455718994, + "learning_rate": 7.46697277655327e-05, + "loss": 3.1835, + "step": 3927 + }, + { + "epoch": 5.76, + "grad_norm": 30.890663146972656, + "learning_rate": 7.462435676884547e-05, + "loss": 3.1883, + "step": 3928 + }, + { + "epoch": 5.76, + "grad_norm": 3.2805261611938477, + "learning_rate": 7.457899135462974e-05, + "loss": 3.2009, + "step": 3929 + }, + { + "epoch": 5.76, + "grad_norm": 11.584432601928711, + "learning_rate": 7.45336315328655e-05, + "loss": 3.2384, + "step": 3930 + }, + { + "epoch": 5.76, + "grad_norm": 6.337993144989014, + "learning_rate": 7.448827731353166e-05, + "loss": 3.1973, + "step": 3931 + }, + { + "epoch": 5.76, + "grad_norm": 13.461885452270508, + "learning_rate": 7.444292870660586e-05, + "loss": 3.2435, + "step": 3932 + }, + { + "epoch": 5.76, + "grad_norm": 9.56443977355957, + "learning_rate": 7.439758572206448e-05, + "loss": 3.1465, + "step": 3933 + }, + { + "epoch": 5.76, + "grad_norm": 7.9625115394592285, + "learning_rate": 7.43522483698826e-05, + "loss": 3.1635, + "step": 3934 + }, + { + "epoch": 5.77, + "grad_norm": 6.776467800140381, + "learning_rate": 7.430691666003424e-05, + "loss": 3.2405, + "step": 3935 + }, + { + "epoch": 5.77, + "grad_norm": 43.86566162109375, + "learning_rate": 7.426159060249204e-05, + "loss": 3.192, + "step": 3936 + }, + { + "epoch": 5.77, + "grad_norm": 12.768895149230957, + "learning_rate": 7.421627020722738e-05, + "loss": 3.1741, + "step": 3937 + }, + { + "epoch": 5.77, + "grad_norm": 6.572139263153076, + "learning_rate": 7.41709554842105e-05, + "loss": 3.1637, + "step": 3938 + }, + { + "epoch": 5.77, + "grad_norm": 5.105015277862549, + "learning_rate": 7.412564644341033e-05, + "loss": 3.2329, + "step": 3939 + }, + { + "epoch": 5.77, + "grad_norm": 4.994419574737549, + "learning_rate": 7.40803430947945e-05, + "loss": 3.2124, + "step": 3940 + }, + { + "epoch": 5.77, + "grad_norm": 13.615537643432617, + "learning_rate": 7.403504544832951e-05, + "loss": 3.2148, + "step": 3941 + }, + { + "epoch": 5.78, + "grad_norm": 6.730780124664307, + "learning_rate": 7.398975351398053e-05, + "loss": 3.2086, + "step": 3942 + }, + { + "epoch": 5.78, + "grad_norm": 8.399843215942383, + "learning_rate": 7.394446730171141e-05, + "loss": 3.2455, + "step": 3943 + }, + { + "epoch": 5.78, + "grad_norm": 7.892696857452393, + "learning_rate": 7.389918682148491e-05, + "loss": 3.263, + "step": 3944 + }, + { + "epoch": 5.78, + "grad_norm": 3.2410852909088135, + "learning_rate": 7.38539120832624e-05, + "loss": 3.2023, + "step": 3945 + }, + { + "epoch": 5.78, + "grad_norm": 43.567543029785156, + "learning_rate": 7.380864309700396e-05, + "loss": 3.2505, + "step": 3946 + }, + { + "epoch": 5.78, + "grad_norm": 8.303853988647461, + "learning_rate": 7.376337987266856e-05, + "loss": 3.2253, + "step": 3947 + }, + { + "epoch": 5.79, + "grad_norm": 5.137606620788574, + "learning_rate": 7.371812242021374e-05, + "loss": 3.2614, + "step": 3948 + }, + { + "epoch": 5.79, + "grad_norm": 7.571026802062988, + "learning_rate": 7.367287074959586e-05, + "loss": 3.3035, + "step": 3949 + }, + { + "epoch": 5.79, + "grad_norm": 5.098783016204834, + "learning_rate": 7.362762487076995e-05, + "loss": 3.24, + "step": 3950 + }, + { + "epoch": 5.79, + "grad_norm": 5.123687744140625, + "learning_rate": 7.358238479368985e-05, + "loss": 3.2333, + "step": 3951 + }, + { + "epoch": 5.79, + "grad_norm": 9.861023902893066, + "learning_rate": 7.353715052830808e-05, + "loss": 3.2665, + "step": 3952 + }, + { + "epoch": 5.79, + "grad_norm": 30.322845458984375, + "learning_rate": 7.349192208457579e-05, + "loss": 3.2337, + "step": 3953 + }, + { + "epoch": 5.79, + "grad_norm": 10.311440467834473, + "learning_rate": 7.344669947244302e-05, + "loss": 3.2307, + "step": 3954 + }, + { + "epoch": 5.8, + "grad_norm": 13.122282981872559, + "learning_rate": 7.340148270185842e-05, + "loss": 3.2456, + "step": 3955 + }, + { + "epoch": 5.8, + "grad_norm": 16.757612228393555, + "learning_rate": 7.335627178276937e-05, + "loss": 3.3208, + "step": 3956 + }, + { + "epoch": 5.8, + "grad_norm": 36.39468002319336, + "learning_rate": 7.331106672512197e-05, + "loss": 3.2313, + "step": 3957 + }, + { + "epoch": 5.8, + "grad_norm": 39.07672882080078, + "learning_rate": 7.326586753886102e-05, + "loss": 3.3304, + "step": 3958 + }, + { + "epoch": 5.8, + "grad_norm": 67.7435302734375, + "learning_rate": 7.322067423393002e-05, + "loss": 3.3623, + "step": 3959 + }, + { + "epoch": 5.8, + "grad_norm": 102.08399963378906, + "learning_rate": 7.317548682027127e-05, + "loss": 3.3498, + "step": 3960 + }, + { + "epoch": 5.8, + "grad_norm": 68.23622131347656, + "learning_rate": 7.313030530782564e-05, + "loss": 3.4284, + "step": 3961 + }, + { + "epoch": 5.81, + "grad_norm": 54.879310607910156, + "learning_rate": 7.308512970653274e-05, + "loss": 3.4101, + "step": 3962 + }, + { + "epoch": 5.81, + "grad_norm": 154.8038787841797, + "learning_rate": 7.303996002633096e-05, + "loss": 3.5803, + "step": 3963 + }, + { + "epoch": 5.81, + "grad_norm": 228.81173706054688, + "learning_rate": 7.29947962771573e-05, + "loss": 3.489, + "step": 3964 + }, + { + "epoch": 5.81, + "grad_norm": 350.7447509765625, + "learning_rate": 7.294963846894744e-05, + "loss": 3.6323, + "step": 3965 + }, + { + "epoch": 5.81, + "grad_norm": 858.3496704101562, + "learning_rate": 7.290448661163587e-05, + "loss": 4.0812, + "step": 3966 + }, + { + "epoch": 5.81, + "grad_norm": 730.7702026367188, + "learning_rate": 7.285934071515568e-05, + "loss": 4.0206, + "step": 3967 + }, + { + "epoch": 5.81, + "grad_norm": 658.6021118164062, + "learning_rate": 7.281420078943855e-05, + "loss": 4.3312, + "step": 3968 + }, + { + "epoch": 5.82, + "grad_norm": 475.8223876953125, + "learning_rate": 7.276906684441514e-05, + "loss": 4.5484, + "step": 3969 + }, + { + "epoch": 5.82, + "grad_norm": 309.2330322265625, + "learning_rate": 7.272393889001454e-05, + "loss": 4.6918, + "step": 3970 + }, + { + "epoch": 5.82, + "grad_norm": 195.97572326660156, + "learning_rate": 7.267881693616452e-05, + "loss": 4.7661, + "step": 3971 + }, + { + "epoch": 5.82, + "grad_norm": 134.8620147705078, + "learning_rate": 7.263370099279172e-05, + "loss": 4.6478, + "step": 3972 + }, + { + "epoch": 5.82, + "grad_norm": 108.79113006591797, + "learning_rate": 7.25885910698213e-05, + "loss": 4.6325, + "step": 3973 + }, + { + "epoch": 5.82, + "grad_norm": 149.6044921875, + "learning_rate": 7.254348717717714e-05, + "loss": 4.8721, + "step": 3974 + }, + { + "epoch": 5.83, + "grad_norm": 117.6059341430664, + "learning_rate": 7.249838932478177e-05, + "loss": 4.8513, + "step": 3975 + }, + { + "epoch": 5.83, + "grad_norm": 56.039024353027344, + "learning_rate": 7.245329752255647e-05, + "loss": 4.8064, + "step": 3976 + }, + { + "epoch": 5.83, + "grad_norm": 100.5042953491211, + "learning_rate": 7.24082117804211e-05, + "loss": 4.8077, + "step": 3977 + }, + { + "epoch": 5.83, + "grad_norm": 101.55089569091797, + "learning_rate": 7.236313210829419e-05, + "loss": 4.9091, + "step": 3978 + }, + { + "epoch": 5.83, + "grad_norm": 54.18492889404297, + "learning_rate": 7.231805851609306e-05, + "loss": 4.9624, + "step": 3979 + }, + { + "epoch": 5.83, + "grad_norm": 90.85334014892578, + "learning_rate": 7.227299101373352e-05, + "loss": 4.8436, + "step": 3980 + }, + { + "epoch": 5.83, + "grad_norm": 108.68850708007812, + "learning_rate": 7.222792961113012e-05, + "loss": 4.7235, + "step": 3981 + }, + { + "epoch": 5.84, + "grad_norm": 46.6330451965332, + "learning_rate": 7.21828743181961e-05, + "loss": 4.7477, + "step": 3982 + }, + { + "epoch": 5.84, + "grad_norm": 116.49347686767578, + "learning_rate": 7.213782514484333e-05, + "loss": 4.6439, + "step": 3983 + }, + { + "epoch": 5.84, + "grad_norm": 46.60487747192383, + "learning_rate": 7.209278210098224e-05, + "loss": 4.6174, + "step": 3984 + }, + { + "epoch": 5.84, + "grad_norm": 74.72257995605469, + "learning_rate": 7.204774519652211e-05, + "loss": 4.6026, + "step": 3985 + }, + { + "epoch": 5.84, + "grad_norm": 117.54887390136719, + "learning_rate": 7.200271444137072e-05, + "loss": 4.3403, + "step": 3986 + }, + { + "epoch": 5.84, + "grad_norm": 66.1172866821289, + "learning_rate": 7.195768984543448e-05, + "loss": 4.2326, + "step": 3987 + }, + { + "epoch": 5.84, + "grad_norm": 23.685178756713867, + "learning_rate": 7.191267141861858e-05, + "loss": 4.2885, + "step": 3988 + }, + { + "epoch": 5.85, + "grad_norm": 33.490779876708984, + "learning_rate": 7.186765917082674e-05, + "loss": 4.2881, + "step": 3989 + }, + { + "epoch": 5.85, + "grad_norm": 49.65608215332031, + "learning_rate": 7.18226531119613e-05, + "loss": 4.0779, + "step": 3990 + }, + { + "epoch": 5.85, + "grad_norm": 21.95450782775879, + "learning_rate": 7.177765325192338e-05, + "loss": 4.0829, + "step": 3991 + }, + { + "epoch": 5.85, + "grad_norm": 36.09683609008789, + "learning_rate": 7.173265960061261e-05, + "loss": 4.1138, + "step": 3992 + }, + { + "epoch": 5.85, + "grad_norm": 135.8228302001953, + "learning_rate": 7.168767216792724e-05, + "loss": 4.0337, + "step": 3993 + }, + { + "epoch": 5.85, + "grad_norm": 42.24320602416992, + "learning_rate": 7.164269096376432e-05, + "loss": 4.0038, + "step": 3994 + }, + { + "epoch": 5.85, + "grad_norm": 90.61548614501953, + "learning_rate": 7.15977159980193e-05, + "loss": 3.9393, + "step": 3995 + }, + { + "epoch": 5.86, + "grad_norm": 83.08573150634766, + "learning_rate": 7.155274728058645e-05, + "loss": 4.085, + "step": 3996 + }, + { + "epoch": 5.86, + "grad_norm": 51.404388427734375, + "learning_rate": 7.150778482135851e-05, + "loss": 4.025, + "step": 3997 + }, + { + "epoch": 5.86, + "grad_norm": 133.36476135253906, + "learning_rate": 7.146282863022699e-05, + "loss": 4.0341, + "step": 3998 + }, + { + "epoch": 5.86, + "grad_norm": 68.87841033935547, + "learning_rate": 7.141787871708194e-05, + "loss": 4.1013, + "step": 3999 + }, + { + "epoch": 5.86, + "grad_norm": 83.18514251708984, + "learning_rate": 7.137293509181197e-05, + "loss": 4.0575, + "step": 4000 + }, + { + "epoch": 5.86, + "grad_norm": 63.65642166137695, + "learning_rate": 7.132799776430446e-05, + "loss": 4.0338, + "step": 4001 + }, + { + "epoch": 5.87, + "grad_norm": 84.51449584960938, + "learning_rate": 7.128306674444531e-05, + "loss": 4.0289, + "step": 4002 + }, + { + "epoch": 5.87, + "grad_norm": 1112.17431640625, + "learning_rate": 7.123814204211899e-05, + "loss": 3.9574, + "step": 4003 + }, + { + "epoch": 5.87, + "grad_norm": 113.361572265625, + "learning_rate": 7.11932236672087e-05, + "loss": 3.9683, + "step": 4004 + }, + { + "epoch": 5.87, + "grad_norm": 94.9935073852539, + "learning_rate": 7.114831162959614e-05, + "loss": 3.867, + "step": 4005 + }, + { + "epoch": 5.87, + "grad_norm": 99.41561126708984, + "learning_rate": 7.110340593916164e-05, + "loss": 4.0003, + "step": 4006 + }, + { + "epoch": 5.87, + "grad_norm": 68.38880157470703, + "learning_rate": 7.105850660578422e-05, + "loss": 3.8946, + "step": 4007 + }, + { + "epoch": 5.87, + "grad_norm": 48.9935188293457, + "learning_rate": 7.101361363934136e-05, + "loss": 3.9, + "step": 4008 + }, + { + "epoch": 5.88, + "grad_norm": 75.62947082519531, + "learning_rate": 7.096872704970923e-05, + "loss": 3.9334, + "step": 4009 + }, + { + "epoch": 5.88, + "grad_norm": 259.18511962890625, + "learning_rate": 7.092384684676262e-05, + "loss": 3.9539, + "step": 4010 + }, + { + "epoch": 5.88, + "grad_norm": 60.03557586669922, + "learning_rate": 7.087897304037483e-05, + "loss": 3.9378, + "step": 4011 + }, + { + "epoch": 5.88, + "grad_norm": 60.427616119384766, + "learning_rate": 7.083410564041779e-05, + "loss": 3.922, + "step": 4012 + }, + { + "epoch": 5.88, + "grad_norm": 86.16722869873047, + "learning_rate": 7.078924465676208e-05, + "loss": 3.9123, + "step": 4013 + }, + { + "epoch": 5.88, + "grad_norm": 92.25879669189453, + "learning_rate": 7.07443900992768e-05, + "loss": 3.9653, + "step": 4014 + }, + { + "epoch": 5.88, + "grad_norm": 63.97455596923828, + "learning_rate": 7.069954197782958e-05, + "loss": 3.9709, + "step": 4015 + }, + { + "epoch": 5.89, + "grad_norm": 85.49566650390625, + "learning_rate": 7.065470030228679e-05, + "loss": 3.9614, + "step": 4016 + }, + { + "epoch": 5.89, + "grad_norm": 200.93997192382812, + "learning_rate": 7.060986508251329e-05, + "loss": 3.9715, + "step": 4017 + }, + { + "epoch": 5.89, + "grad_norm": 47.623600006103516, + "learning_rate": 7.056503632837247e-05, + "loss": 3.9969, + "step": 4018 + }, + { + "epoch": 5.89, + "grad_norm": 57.71437454223633, + "learning_rate": 7.052021404972641e-05, + "loss": 4.0892, + "step": 4019 + }, + { + "epoch": 5.89, + "grad_norm": 101.89254760742188, + "learning_rate": 7.047539825643572e-05, + "loss": 4.1731, + "step": 4020 + }, + { + "epoch": 5.89, + "grad_norm": 56.065486907958984, + "learning_rate": 7.043058895835953e-05, + "loss": 4.1648, + "step": 4021 + }, + { + "epoch": 5.89, + "grad_norm": 98.78327178955078, + "learning_rate": 7.038578616535556e-05, + "loss": 4.3178, + "step": 4022 + }, + { + "epoch": 5.9, + "grad_norm": 383.0876770019531, + "learning_rate": 7.034098988728024e-05, + "loss": 4.3457, + "step": 4023 + }, + { + "epoch": 5.9, + "grad_norm": 162.98101806640625, + "learning_rate": 7.029620013398837e-05, + "loss": 4.3242, + "step": 4024 + }, + { + "epoch": 5.9, + "grad_norm": 149.046142578125, + "learning_rate": 7.025141691533334e-05, + "loss": 4.4494, + "step": 4025 + }, + { + "epoch": 5.9, + "grad_norm": 135.7252655029297, + "learning_rate": 7.020664024116726e-05, + "loss": 4.487, + "step": 4026 + }, + { + "epoch": 5.9, + "grad_norm": 103.65221405029297, + "learning_rate": 7.016187012134068e-05, + "loss": 4.544, + "step": 4027 + }, + { + "epoch": 5.9, + "grad_norm": 95.76521301269531, + "learning_rate": 7.011710656570267e-05, + "loss": 4.5937, + "step": 4028 + }, + { + "epoch": 5.91, + "grad_norm": 287.5489196777344, + "learning_rate": 7.007234958410098e-05, + "loss": 4.6397, + "step": 4029 + }, + { + "epoch": 5.91, + "grad_norm": 55.86167526245117, + "learning_rate": 7.002759918638183e-05, + "loss": 4.6566, + "step": 4030 + }, + { + "epoch": 5.91, + "grad_norm": 50.3218879699707, + "learning_rate": 6.998285538238993e-05, + "loss": 4.6248, + "step": 4031 + }, + { + "epoch": 5.91, + "grad_norm": 414.169921875, + "learning_rate": 6.993811818196872e-05, + "loss": 4.6006, + "step": 4032 + }, + { + "epoch": 5.91, + "grad_norm": 112.764404296875, + "learning_rate": 6.989338759496006e-05, + "loss": 4.6415, + "step": 4033 + }, + { + "epoch": 5.91, + "grad_norm": 53.906734466552734, + "learning_rate": 6.984866363120432e-05, + "loss": 4.6018, + "step": 4034 + }, + { + "epoch": 5.91, + "grad_norm": 45.21807098388672, + "learning_rate": 6.980394630054054e-05, + "loss": 4.5761, + "step": 4035 + }, + { + "epoch": 5.92, + "grad_norm": 123.97179412841797, + "learning_rate": 6.975923561280624e-05, + "loss": 4.4143, + "step": 4036 + }, + { + "epoch": 5.92, + "grad_norm": 39.991905212402344, + "learning_rate": 6.971453157783739e-05, + "loss": 4.3971, + "step": 4037 + }, + { + "epoch": 5.92, + "grad_norm": 245.3745574951172, + "learning_rate": 6.966983420546866e-05, + "loss": 4.3387, + "step": 4038 + }, + { + "epoch": 5.92, + "grad_norm": 535.9085083007812, + "learning_rate": 6.962514350553313e-05, + "loss": 4.2919, + "step": 4039 + }, + { + "epoch": 5.92, + "grad_norm": 31.66384506225586, + "learning_rate": 6.958045948786245e-05, + "loss": 4.2878, + "step": 4040 + }, + { + "epoch": 5.92, + "grad_norm": 104.09449768066406, + "learning_rate": 6.953578216228685e-05, + "loss": 4.2645, + "step": 4041 + }, + { + "epoch": 5.92, + "grad_norm": 73.07312774658203, + "learning_rate": 6.9491111538635e-05, + "loss": 4.2232, + "step": 4042 + }, + { + "epoch": 5.93, + "grad_norm": 55.163421630859375, + "learning_rate": 6.944644762673413e-05, + "loss": 4.2499, + "step": 4043 + }, + { + "epoch": 5.93, + "grad_norm": 97.98957061767578, + "learning_rate": 6.940179043641005e-05, + "loss": 4.2195, + "step": 4044 + }, + { + "epoch": 5.93, + "grad_norm": 35.54196548461914, + "learning_rate": 6.9357139977487e-05, + "loss": 4.2314, + "step": 4045 + }, + { + "epoch": 5.93, + "grad_norm": 36.159202575683594, + "learning_rate": 6.931249625978782e-05, + "loss": 4.1943, + "step": 4046 + }, + { + "epoch": 5.93, + "grad_norm": 119.87337493896484, + "learning_rate": 6.926785929313376e-05, + "loss": 4.1162, + "step": 4047 + }, + { + "epoch": 5.93, + "grad_norm": 7.6323442459106445, + "learning_rate": 6.922322908734474e-05, + "loss": 4.1434, + "step": 4048 + }, + { + "epoch": 5.94, + "grad_norm": 13.205349922180176, + "learning_rate": 6.917860565223905e-05, + "loss": 4.1026, + "step": 4049 + }, + { + "epoch": 5.94, + "grad_norm": 29.02250099182129, + "learning_rate": 6.913398899763352e-05, + "loss": 4.0863, + "step": 4050 + }, + { + "epoch": 5.94, + "grad_norm": 18.842418670654297, + "learning_rate": 6.908937913334362e-05, + "loss": 4.1056, + "step": 4051 + }, + { + "epoch": 5.94, + "grad_norm": 26.442873001098633, + "learning_rate": 6.904477606918313e-05, + "loss": 4.1195, + "step": 4052 + }, + { + "epoch": 5.94, + "grad_norm": 146.73545837402344, + "learning_rate": 6.900017981496443e-05, + "loss": 4.0448, + "step": 4053 + }, + { + "epoch": 5.94, + "grad_norm": 18.860145568847656, + "learning_rate": 6.895559038049846e-05, + "loss": 4.0507, + "step": 4054 + }, + { + "epoch": 5.94, + "grad_norm": 7.4111504554748535, + "learning_rate": 6.891100777559454e-05, + "loss": 3.9955, + "step": 4055 + }, + { + "epoch": 5.95, + "grad_norm": 2.2253613471984863, + "learning_rate": 6.886643201006052e-05, + "loss": 3.9959, + "step": 4056 + }, + { + "epoch": 5.95, + "eval_loss": 3.7561120986938477, + "eval_runtime": 1174.8961, + "eval_samples_per_second": 14.59, + "eval_steps_per_second": 0.912, + "step": 4056 + }, + { + "epoch": 5.95, + "grad_norm": 2.182352066040039, + "learning_rate": 6.882186309370287e-05, + "loss": 3.9472, + "step": 4057 + }, + { + "epoch": 5.95, + "grad_norm": 1.8885817527770996, + "learning_rate": 6.877730103632638e-05, + "loss": 3.9685, + "step": 4058 + }, + { + "epoch": 5.95, + "grad_norm": 1.6018338203430176, + "learning_rate": 6.87327458477344e-05, + "loss": 3.8923, + "step": 4059 + }, + { + "epoch": 5.95, + "grad_norm": 1.4974128007888794, + "learning_rate": 6.868819753772879e-05, + "loss": 3.8199, + "step": 4060 + }, + { + "epoch": 5.95, + "grad_norm": 1.9076157808303833, + "learning_rate": 6.86436561161099e-05, + "loss": 3.8143, + "step": 4061 + }, + { + "epoch": 5.95, + "grad_norm": 1.2935985326766968, + "learning_rate": 6.859912159267648e-05, + "loss": 3.8232, + "step": 4062 + }, + { + "epoch": 5.96, + "grad_norm": 1.5463148355484009, + "learning_rate": 6.85545939772259e-05, + "loss": 3.7682, + "step": 4063 + }, + { + "epoch": 5.96, + "grad_norm": 1.2979236841201782, + "learning_rate": 6.851007327955393e-05, + "loss": 3.812, + "step": 4064 + }, + { + "epoch": 5.96, + "grad_norm": 1.235023856163025, + "learning_rate": 6.846555950945475e-05, + "loss": 3.6794, + "step": 4065 + }, + { + "epoch": 5.96, + "grad_norm": 1.587446689605713, + "learning_rate": 6.842105267672117e-05, + "loss": 3.6736, + "step": 4066 + }, + { + "epoch": 5.96, + "grad_norm": 1.6601265668869019, + "learning_rate": 6.837655279114439e-05, + "loss": 3.7298, + "step": 4067 + }, + { + "epoch": 5.96, + "grad_norm": 1.1815412044525146, + "learning_rate": 6.833205986251404e-05, + "loss": 3.6491, + "step": 4068 + }, + { + "epoch": 5.96, + "grad_norm": 2.2124860286712646, + "learning_rate": 6.828757390061828e-05, + "loss": 3.6432, + "step": 4069 + }, + { + "epoch": 5.97, + "grad_norm": 1.2953498363494873, + "learning_rate": 6.824309491524377e-05, + "loss": 3.6124, + "step": 4070 + }, + { + "epoch": 5.97, + "grad_norm": 1.5017386674880981, + "learning_rate": 6.819862291617555e-05, + "loss": 3.6377, + "step": 4071 + }, + { + "epoch": 5.97, + "grad_norm": 2.5354573726654053, + "learning_rate": 6.815415791319712e-05, + "loss": 3.6211, + "step": 4072 + }, + { + "epoch": 5.97, + "grad_norm": 2.2560806274414062, + "learning_rate": 6.810969991609055e-05, + "loss": 3.5855, + "step": 4073 + }, + { + "epoch": 5.97, + "grad_norm": 2.3098514080047607, + "learning_rate": 6.806524893463631e-05, + "loss": 3.5944, + "step": 4074 + }, + { + "epoch": 5.97, + "grad_norm": 3.807551860809326, + "learning_rate": 6.802080497861326e-05, + "loss": 3.582, + "step": 4075 + }, + { + "epoch": 5.98, + "grad_norm": 0.8520615100860596, + "learning_rate": 6.797636805779878e-05, + "loss": 3.5581, + "step": 4076 + }, + { + "epoch": 5.98, + "grad_norm": 2.77345871925354, + "learning_rate": 6.793193818196876e-05, + "loss": 3.5948, + "step": 4077 + }, + { + "epoch": 5.98, + "grad_norm": 1.2732380628585815, + "learning_rate": 6.788751536089739e-05, + "loss": 3.5607, + "step": 4078 + }, + { + "epoch": 5.98, + "grad_norm": 4.257369518280029, + "learning_rate": 6.784309960435744e-05, + "loss": 3.588, + "step": 4079 + }, + { + "epoch": 5.98, + "grad_norm": 4.027552604675293, + "learning_rate": 6.77986909221201e-05, + "loss": 3.5961, + "step": 4080 + }, + { + "epoch": 5.98, + "grad_norm": 1.3005938529968262, + "learning_rate": 6.77542893239549e-05, + "loss": 3.52, + "step": 4081 + }, + { + "epoch": 5.98, + "grad_norm": 3.389254570007324, + "learning_rate": 6.770989481963001e-05, + "loss": 3.5313, + "step": 4082 + }, + { + "epoch": 5.99, + "grad_norm": 1.3640910387039185, + "learning_rate": 6.766550741891182e-05, + "loss": 3.5291, + "step": 4083 + }, + { + "epoch": 5.99, + "grad_norm": 5.111477375030518, + "learning_rate": 6.76211271315653e-05, + "loss": 3.5717, + "step": 4084 + }, + { + "epoch": 5.99, + "grad_norm": 5.650928974151611, + "learning_rate": 6.757675396735386e-05, + "loss": 3.5892, + "step": 4085 + }, + { + "epoch": 5.99, + "grad_norm": 2.029646158218384, + "learning_rate": 6.753238793603923e-05, + "loss": 3.5179, + "step": 4086 + }, + { + "epoch": 5.99, + "grad_norm": 5.816488742828369, + "learning_rate": 6.748802904738163e-05, + "loss": 3.5582, + "step": 4087 + }, + { + "epoch": 5.99, + "grad_norm": 7.348140716552734, + "learning_rate": 6.74436773111398e-05, + "loss": 3.5371, + "step": 4088 + }, + { + "epoch": 5.99, + "grad_norm": 6.118741512298584, + "learning_rate": 6.739933273707078e-05, + "loss": 3.5298, + "step": 4089 + }, + { + "epoch": 6.0, + "grad_norm": 0.9806273579597473, + "learning_rate": 6.735499533493003e-05, + "loss": 3.5438, + "step": 4090 + }, + { + "epoch": 6.0, + "grad_norm": 9.882466316223145, + "learning_rate": 6.731066511447159e-05, + "loss": 3.5713, + "step": 4091 + }, + { + "epoch": 6.0, + "grad_norm": 14.419783592224121, + "learning_rate": 6.726634208544773e-05, + "loss": 3.569, + "step": 4092 + }, + { + "epoch": 6.0, + "grad_norm": 11.746726036071777, + "learning_rate": 6.722202625760923e-05, + "loss": 3.6069, + "step": 4093 + }, + { + "epoch": 6.0, + "grad_norm": 5.577601432800293, + "learning_rate": 6.717771764070527e-05, + "loss": 3.5287, + "step": 4094 + }, + { + "epoch": 6.0, + "grad_norm": 5.144962310791016, + "learning_rate": 6.713341624448351e-05, + "loss": 3.5013, + "step": 4095 + }, + { + "epoch": 6.0, + "grad_norm": 8.980729103088379, + "learning_rate": 6.708912207868987e-05, + "loss": 3.5225, + "step": 4096 + }, + { + "epoch": 6.01, + "grad_norm": 9.171212196350098, + "learning_rate": 6.70448351530688e-05, + "loss": 3.537, + "step": 4097 + }, + { + "epoch": 6.01, + "grad_norm": 6.954055309295654, + "learning_rate": 6.70005554773632e-05, + "loss": 3.5512, + "step": 4098 + }, + { + "epoch": 6.01, + "grad_norm": 25.35405731201172, + "learning_rate": 6.695628306131422e-05, + "loss": 3.525, + "step": 4099 + }, + { + "epoch": 6.01, + "grad_norm": 7.4213948249816895, + "learning_rate": 6.691201791466148e-05, + "loss": 3.5411, + "step": 4100 + }, + { + "epoch": 6.01, + "grad_norm": 10.75389289855957, + "learning_rate": 6.68677600471431e-05, + "loss": 3.5158, + "step": 4101 + }, + { + "epoch": 6.01, + "grad_norm": 9.7072114944458, + "learning_rate": 6.682350946849546e-05, + "loss": 3.5728, + "step": 4102 + }, + { + "epoch": 6.02, + "grad_norm": 4.010187149047852, + "learning_rate": 6.677926618845337e-05, + "loss": 3.4962, + "step": 4103 + }, + { + "epoch": 6.0, + "grad_norm": 5.2085371017456055, + "learning_rate": 6.673503021675011e-05, + "loss": 3.5387, + "step": 4104 + }, + { + "epoch": 6.0, + "grad_norm": 7.690248012542725, + "learning_rate": 6.669080156311729e-05, + "loss": 3.5012, + "step": 4105 + }, + { + "epoch": 6.0, + "grad_norm": 7.818577766418457, + "learning_rate": 6.664658023728485e-05, + "loss": 3.5594, + "step": 4106 + }, + { + "epoch": 6.01, + "grad_norm": 4.1020355224609375, + "learning_rate": 6.660236624898124e-05, + "loss": 3.4641, + "step": 4107 + }, + { + "epoch": 6.01, + "grad_norm": 3.3743233680725098, + "learning_rate": 6.655815960793328e-05, + "loss": 3.5123, + "step": 4108 + }, + { + "epoch": 6.01, + "grad_norm": 5.35082483291626, + "learning_rate": 6.651396032386603e-05, + "loss": 3.5047, + "step": 4109 + }, + { + "epoch": 6.01, + "grad_norm": 2.8813364505767822, + "learning_rate": 6.646976840650312e-05, + "loss": 3.4699, + "step": 4110 + }, + { + "epoch": 6.01, + "grad_norm": 3.898869037628174, + "learning_rate": 6.642558386556648e-05, + "loss": 3.4723, + "step": 4111 + }, + { + "epoch": 6.01, + "grad_norm": 3.897127389907837, + "learning_rate": 6.638140671077633e-05, + "loss": 3.4887, + "step": 4112 + }, + { + "epoch": 6.01, + "grad_norm": 0.8849645256996155, + "learning_rate": 6.633723695185141e-05, + "loss": 3.4564, + "step": 4113 + }, + { + "epoch": 6.02, + "grad_norm": 4.971652507781982, + "learning_rate": 6.62930745985088e-05, + "loss": 3.5218, + "step": 4114 + }, + { + "epoch": 6.02, + "grad_norm": 4.998004913330078, + "learning_rate": 6.624891966046385e-05, + "loss": 3.5211, + "step": 4115 + }, + { + "epoch": 6.02, + "grad_norm": 1.827019214630127, + "learning_rate": 6.620477214743042e-05, + "loss": 3.4959, + "step": 4116 + }, + { + "epoch": 6.02, + "grad_norm": 4.872849464416504, + "learning_rate": 6.616063206912063e-05, + "loss": 3.4849, + "step": 4117 + }, + { + "epoch": 6.02, + "grad_norm": 5.961864948272705, + "learning_rate": 6.611649943524498e-05, + "loss": 3.4685, + "step": 4118 + }, + { + "epoch": 6.02, + "grad_norm": 3.4859068393707275, + "learning_rate": 6.607237425551236e-05, + "loss": 3.4841, + "step": 4119 + }, + { + "epoch": 6.03, + "grad_norm": 4.598318099975586, + "learning_rate": 6.602825653963006e-05, + "loss": 3.5138, + "step": 4120 + }, + { + "epoch": 6.03, + "grad_norm": 8.418424606323242, + "learning_rate": 6.598414629730367e-05, + "loss": 3.4861, + "step": 4121 + }, + { + "epoch": 6.03, + "grad_norm": 6.949596881866455, + "learning_rate": 6.594004353823708e-05, + "loss": 3.4545, + "step": 4122 + }, + { + "epoch": 6.03, + "grad_norm": 1.4345308542251587, + "learning_rate": 6.589594827213268e-05, + "loss": 3.4615, + "step": 4123 + }, + { + "epoch": 6.03, + "grad_norm": 5.833983898162842, + "learning_rate": 6.585186050869112e-05, + "loss": 3.4469, + "step": 4124 + }, + { + "epoch": 6.03, + "grad_norm": 8.324283599853516, + "learning_rate": 6.580778025761136e-05, + "loss": 3.5428, + "step": 4125 + }, + { + "epoch": 6.03, + "grad_norm": 7.164588451385498, + "learning_rate": 6.576370752859085e-05, + "loss": 3.44, + "step": 4126 + }, + { + "epoch": 6.04, + "grad_norm": 4.524144172668457, + "learning_rate": 6.57196423313252e-05, + "loss": 3.4798, + "step": 4127 + }, + { + "epoch": 6.04, + "grad_norm": 3.6542019844055176, + "learning_rate": 6.56755846755085e-05, + "loss": 3.4431, + "step": 4128 + }, + { + "epoch": 6.04, + "grad_norm": 6.415909290313721, + "learning_rate": 6.563153457083315e-05, + "loss": 3.4509, + "step": 4129 + }, + { + "epoch": 6.04, + "grad_norm": 4.0182623863220215, + "learning_rate": 6.558749202698988e-05, + "loss": 3.4182, + "step": 4130 + }, + { + "epoch": 6.04, + "grad_norm": 22.318021774291992, + "learning_rate": 6.55434570536677e-05, + "loss": 3.5704, + "step": 4131 + }, + { + "epoch": 6.04, + "grad_norm": 6.184370040893555, + "learning_rate": 6.54994296605541e-05, + "loss": 3.3746, + "step": 4132 + }, + { + "epoch": 6.04, + "grad_norm": 7.076589584350586, + "learning_rate": 6.545540985733473e-05, + "loss": 3.4426, + "step": 4133 + }, + { + "epoch": 6.05, + "grad_norm": 3.5989532470703125, + "learning_rate": 6.541139765369368e-05, + "loss": 3.4139, + "step": 4134 + }, + { + "epoch": 6.05, + "grad_norm": 4.95648717880249, + "learning_rate": 6.536739305931338e-05, + "loss": 3.392, + "step": 4135 + }, + { + "epoch": 6.05, + "grad_norm": 8.156216621398926, + "learning_rate": 6.53233960838745e-05, + "loss": 3.4219, + "step": 4136 + }, + { + "epoch": 6.05, + "grad_norm": 6.330258846282959, + "learning_rate": 6.527940673705605e-05, + "loss": 3.4077, + "step": 4137 + }, + { + "epoch": 6.05, + "grad_norm": 1.3029074668884277, + "learning_rate": 6.523542502853547e-05, + "loss": 3.422, + "step": 4138 + }, + { + "epoch": 6.05, + "grad_norm": 5.435974597930908, + "learning_rate": 6.519145096798842e-05, + "loss": 3.4704, + "step": 4139 + }, + { + "epoch": 6.05, + "grad_norm": 5.678780555725098, + "learning_rate": 6.514748456508885e-05, + "loss": 3.419, + "step": 4140 + }, + { + "epoch": 6.06, + "grad_norm": 3.2500386238098145, + "learning_rate": 6.510352582950909e-05, + "loss": 3.4207, + "step": 4141 + }, + { + "epoch": 6.06, + "grad_norm": 4.012285232543945, + "learning_rate": 6.505957477091981e-05, + "loss": 3.3755, + "step": 4142 + }, + { + "epoch": 6.06, + "grad_norm": 6.388498783111572, + "learning_rate": 6.501563139898991e-05, + "loss": 3.4086, + "step": 4143 + }, + { + "epoch": 6.06, + "grad_norm": 3.951239824295044, + "learning_rate": 6.497169572338661e-05, + "loss": 3.3432, + "step": 4144 + }, + { + "epoch": 6.06, + "grad_norm": 4.527097702026367, + "learning_rate": 6.492776775377556e-05, + "loss": 3.3961, + "step": 4145 + }, + { + "epoch": 6.06, + "grad_norm": 5.017838954925537, + "learning_rate": 6.488384749982053e-05, + "loss": 3.3483, + "step": 4146 + }, + { + "epoch": 6.07, + "grad_norm": 4.438302516937256, + "learning_rate": 6.483993497118371e-05, + "loss": 3.456, + "step": 4147 + }, + { + "epoch": 6.07, + "grad_norm": 0.9947657585144043, + "learning_rate": 6.479603017752559e-05, + "loss": 3.3646, + "step": 4148 + }, + { + "epoch": 6.07, + "grad_norm": 2.749354600906372, + "learning_rate": 6.47521331285049e-05, + "loss": 3.4993, + "step": 4149 + }, + { + "epoch": 6.07, + "grad_norm": 2.3780393600463867, + "learning_rate": 6.470824383377867e-05, + "loss": 3.3783, + "step": 4150 + }, + { + "epoch": 6.07, + "grad_norm": 2.840714454650879, + "learning_rate": 6.46643623030023e-05, + "loss": 3.3901, + "step": 4151 + }, + { + "epoch": 6.07, + "grad_norm": 2.8188705444335938, + "learning_rate": 6.462048854582944e-05, + "loss": 3.3313, + "step": 4152 + }, + { + "epoch": 6.07, + "grad_norm": 0.6622484922409058, + "learning_rate": 6.457662257191195e-05, + "loss": 3.3428, + "step": 4153 + }, + { + "epoch": 6.08, + "grad_norm": 2.970710515975952, + "learning_rate": 6.453276439090012e-05, + "loss": 3.3343, + "step": 4154 + }, + { + "epoch": 6.08, + "grad_norm": 2.88561749458313, + "learning_rate": 6.448891401244244e-05, + "loss": 3.3591, + "step": 4155 + }, + { + "epoch": 6.08, + "grad_norm": 2.3622708320617676, + "learning_rate": 6.444507144618566e-05, + "loss": 3.39, + "step": 4156 + }, + { + "epoch": 6.08, + "grad_norm": 1.4716769456863403, + "learning_rate": 6.440123670177492e-05, + "loss": 3.4068, + "step": 4157 + }, + { + "epoch": 6.08, + "grad_norm": 10.4777193069458, + "learning_rate": 6.435740978885353e-05, + "loss": 3.352, + "step": 4158 + }, + { + "epoch": 6.08, + "grad_norm": 6.331012725830078, + "learning_rate": 6.43135907170631e-05, + "loss": 3.3506, + "step": 4159 + }, + { + "epoch": 6.08, + "grad_norm": 4.355317115783691, + "learning_rate": 6.426977949604358e-05, + "loss": 3.3904, + "step": 4160 + }, + { + "epoch": 6.09, + "grad_norm": 1.3198426961898804, + "learning_rate": 6.422597613543309e-05, + "loss": 3.3279, + "step": 4161 + }, + { + "epoch": 6.09, + "grad_norm": 3.2531580924987793, + "learning_rate": 6.41821806448681e-05, + "loss": 3.427, + "step": 4162 + }, + { + "epoch": 6.09, + "grad_norm": 1.780542254447937, + "learning_rate": 6.413839303398337e-05, + "loss": 3.382, + "step": 4163 + }, + { + "epoch": 6.09, + "grad_norm": 4.514572620391846, + "learning_rate": 6.40946133124118e-05, + "loss": 3.3181, + "step": 4164 + }, + { + "epoch": 6.09, + "grad_norm": 4.696477890014648, + "learning_rate": 6.405084148978471e-05, + "loss": 3.3693, + "step": 4165 + }, + { + "epoch": 6.09, + "grad_norm": 1.9454889297485352, + "learning_rate": 6.400707757573152e-05, + "loss": 3.3362, + "step": 4166 + }, + { + "epoch": 6.1, + "grad_norm": 4.424253463745117, + "learning_rate": 6.396332157988007e-05, + "loss": 3.3042, + "step": 4167 + }, + { + "epoch": 6.1, + "grad_norm": 6.185297012329102, + "learning_rate": 6.39195735118564e-05, + "loss": 3.3837, + "step": 4168 + }, + { + "epoch": 6.1, + "grad_norm": 4.592702865600586, + "learning_rate": 6.387583338128471e-05, + "loss": 3.35, + "step": 4169 + }, + { + "epoch": 6.1, + "grad_norm": 1.6287086009979248, + "learning_rate": 6.38321011977876e-05, + "loss": 3.3304, + "step": 4170 + }, + { + "epoch": 6.1, + "grad_norm": 4.277772903442383, + "learning_rate": 6.378837697098586e-05, + "loss": 3.321, + "step": 4171 + }, + { + "epoch": 6.1, + "grad_norm": 2.2683122158050537, + "learning_rate": 6.374466071049847e-05, + "loss": 3.2938, + "step": 4172 + }, + { + "epoch": 6.1, + "grad_norm": 2.8517701625823975, + "learning_rate": 6.370095242594279e-05, + "loss": 3.338, + "step": 4173 + }, + { + "epoch": 6.11, + "grad_norm": 3.4475059509277344, + "learning_rate": 6.365725212693428e-05, + "loss": 3.3719, + "step": 4174 + }, + { + "epoch": 6.11, + "grad_norm": 1.0694806575775146, + "learning_rate": 6.361355982308675e-05, + "loss": 3.3108, + "step": 4175 + }, + { + "epoch": 6.11, + "grad_norm": 4.983922958374023, + "learning_rate": 6.356987552401222e-05, + "loss": 3.326, + "step": 4176 + }, + { + "epoch": 6.11, + "grad_norm": 5.746216773986816, + "learning_rate": 6.352619923932093e-05, + "loss": 3.3831, + "step": 4177 + }, + { + "epoch": 6.11, + "grad_norm": 2.469836473464966, + "learning_rate": 6.348253097862134e-05, + "loss": 3.3593, + "step": 4178 + }, + { + "epoch": 6.11, + "grad_norm": 3.8957202434539795, + "learning_rate": 6.343887075152024e-05, + "loss": 3.3594, + "step": 4179 + }, + { + "epoch": 6.11, + "grad_norm": 5.375302791595459, + "learning_rate": 6.339521856762254e-05, + "loss": 3.3357, + "step": 4180 + }, + { + "epoch": 6.12, + "grad_norm": 4.205276966094971, + "learning_rate": 6.335157443653142e-05, + "loss": 3.3987, + "step": 4181 + }, + { + "epoch": 6.12, + "grad_norm": 1.7369492053985596, + "learning_rate": 6.330793836784836e-05, + "loss": 3.3606, + "step": 4182 + }, + { + "epoch": 6.12, + "grad_norm": 2.639575481414795, + "learning_rate": 6.326431037117296e-05, + "loss": 3.4109, + "step": 4183 + }, + { + "epoch": 6.12, + "grad_norm": 0.6227443218231201, + "learning_rate": 6.322069045610305e-05, + "loss": 3.3122, + "step": 4184 + }, + { + "epoch": 6.12, + "grad_norm": 2.4166953563690186, + "learning_rate": 6.317707863223477e-05, + "loss": 3.339, + "step": 4185 + }, + { + "epoch": 6.12, + "grad_norm": 1.3330849409103394, + "learning_rate": 6.313347490916243e-05, + "loss": 3.2775, + "step": 4186 + }, + { + "epoch": 6.12, + "grad_norm": 3.1274631023406982, + "learning_rate": 6.308987929647852e-05, + "loss": 3.2931, + "step": 4187 + }, + { + "epoch": 6.13, + "grad_norm": 2.9381067752838135, + "learning_rate": 6.30462918037738e-05, + "loss": 3.347, + "step": 4188 + }, + { + "epoch": 6.13, + "grad_norm": 0.6513715982437134, + "learning_rate": 6.300271244063726e-05, + "loss": 3.3084, + "step": 4189 + }, + { + "epoch": 6.13, + "grad_norm": 1.9384881258010864, + "learning_rate": 6.295914121665602e-05, + "loss": 3.3508, + "step": 4190 + }, + { + "epoch": 6.13, + "grad_norm": 1.33826744556427, + "learning_rate": 6.291557814141547e-05, + "loss": 3.3385, + "step": 4191 + }, + { + "epoch": 6.13, + "grad_norm": 1.159730076789856, + "learning_rate": 6.287202322449922e-05, + "loss": 3.3025, + "step": 4192 + }, + { + "epoch": 6.13, + "grad_norm": 15.805403709411621, + "learning_rate": 6.282847647548904e-05, + "loss": 3.3027, + "step": 4193 + }, + { + "epoch": 6.14, + "grad_norm": 4.59617280960083, + "learning_rate": 6.278493790396489e-05, + "loss": 3.3538, + "step": 4194 + }, + { + "epoch": 6.14, + "grad_norm": 2.9616920948028564, + "learning_rate": 6.274140751950501e-05, + "loss": 3.3287, + "step": 4195 + }, + { + "epoch": 6.14, + "grad_norm": 2.299812078475952, + "learning_rate": 6.269788533168581e-05, + "loss": 3.3451, + "step": 4196 + }, + { + "epoch": 6.14, + "grad_norm": 2.9177210330963135, + "learning_rate": 6.26543713500818e-05, + "loss": 3.3583, + "step": 4197 + }, + { + "epoch": 6.14, + "grad_norm": 0.9003336429595947, + "learning_rate": 6.261086558426586e-05, + "loss": 3.2672, + "step": 4198 + }, + { + "epoch": 6.14, + "grad_norm": 6.0708184242248535, + "learning_rate": 6.25673680438089e-05, + "loss": 3.2929, + "step": 4199 + }, + { + "epoch": 6.14, + "grad_norm": 6.111046314239502, + "learning_rate": 6.25238787382801e-05, + "loss": 3.3031, + "step": 4200 + }, + { + "epoch": 6.15, + "grad_norm": 4.454471588134766, + "learning_rate": 6.248039767724683e-05, + "loss": 3.3293, + "step": 4201 + }, + { + "epoch": 6.15, + "grad_norm": 0.9689366221427917, + "learning_rate": 6.243692487027464e-05, + "loss": 3.2844, + "step": 4202 + }, + { + "epoch": 6.15, + "grad_norm": 4.885942459106445, + "learning_rate": 6.239346032692721e-05, + "loss": 3.3346, + "step": 4203 + }, + { + "epoch": 6.15, + "grad_norm": 3.6569879055023193, + "learning_rate": 6.23500040567665e-05, + "loss": 3.2936, + "step": 4204 + }, + { + "epoch": 6.15, + "grad_norm": 1.7212425470352173, + "learning_rate": 6.23065560693526e-05, + "loss": 3.3124, + "step": 4205 + }, + { + "epoch": 6.15, + "grad_norm": 2.9631359577178955, + "learning_rate": 6.226311637424371e-05, + "loss": 3.3716, + "step": 4206 + }, + { + "epoch": 6.15, + "grad_norm": 1.4877088069915771, + "learning_rate": 6.221968498099636e-05, + "loss": 3.2881, + "step": 4207 + }, + { + "epoch": 6.16, + "grad_norm": 5.8026533126831055, + "learning_rate": 6.21762618991651e-05, + "loss": 3.2705, + "step": 4208 + }, + { + "epoch": 6.16, + "grad_norm": 6.365312576293945, + "learning_rate": 6.213284713830271e-05, + "loss": 3.3174, + "step": 4209 + }, + { + "epoch": 6.16, + "grad_norm": 5.353916645050049, + "learning_rate": 6.208944070796022e-05, + "loss": 3.3226, + "step": 4210 + }, + { + "epoch": 6.16, + "grad_norm": 1.4095226526260376, + "learning_rate": 6.204604261768669e-05, + "loss": 3.3037, + "step": 4211 + }, + { + "epoch": 6.16, + "grad_norm": 2.51399827003479, + "learning_rate": 6.200265287702942e-05, + "loss": 3.3118, + "step": 4212 + }, + { + "epoch": 6.16, + "grad_norm": 1.4615494012832642, + "learning_rate": 6.19592714955339e-05, + "loss": 3.2963, + "step": 4213 + }, + { + "epoch": 6.16, + "grad_norm": 2.4555823802948, + "learning_rate": 6.191589848274368e-05, + "loss": 3.275, + "step": 4214 + }, + { + "epoch": 6.17, + "grad_norm": 3.085233211517334, + "learning_rate": 6.187253384820061e-05, + "loss": 3.2726, + "step": 4215 + }, + { + "epoch": 6.17, + "grad_norm": 0.8324146270751953, + "learning_rate": 6.182917760144453e-05, + "loss": 3.3158, + "step": 4216 + }, + { + "epoch": 6.17, + "grad_norm": 1.7136584520339966, + "learning_rate": 6.178582975201359e-05, + "loss": 3.2768, + "step": 4217 + }, + { + "epoch": 6.17, + "grad_norm": 1.937159538269043, + "learning_rate": 6.1742490309444e-05, + "loss": 3.3338, + "step": 4218 + }, + { + "epoch": 6.17, + "grad_norm": 1.2840466499328613, + "learning_rate": 6.169915928327012e-05, + "loss": 3.2964, + "step": 4219 + }, + { + "epoch": 6.17, + "grad_norm": 2.5718352794647217, + "learning_rate": 6.165583668302457e-05, + "loss": 3.287, + "step": 4220 + }, + { + "epoch": 6.18, + "grad_norm": 4.60259485244751, + "learning_rate": 6.161252251823795e-05, + "loss": 3.2494, + "step": 4221 + }, + { + "epoch": 6.18, + "grad_norm": 1.160286545753479, + "learning_rate": 6.15692167984391e-05, + "loss": 3.3453, + "step": 4222 + }, + { + "epoch": 6.18, + "grad_norm": 2.802889823913574, + "learning_rate": 6.152591953315505e-05, + "loss": 3.3133, + "step": 4223 + }, + { + "epoch": 6.18, + "grad_norm": 1.775070071220398, + "learning_rate": 6.148263073191082e-05, + "loss": 3.2936, + "step": 4224 + }, + { + "epoch": 6.18, + "grad_norm": 1.9516983032226562, + "learning_rate": 6.14393504042297e-05, + "loss": 3.3066, + "step": 4225 + }, + { + "epoch": 6.18, + "eval_loss": 3.107570171356201, + "eval_runtime": 1177.648, + "eval_samples_per_second": 14.556, + "eval_steps_per_second": 0.91, + "step": 4225 + }, + { + "epoch": 6.18, + "grad_norm": 1.7126461267471313, + "learning_rate": 6.13960785596331e-05, + "loss": 3.2474, + "step": 4226 + }, + { + "epoch": 6.18, + "grad_norm": 1.2491967678070068, + "learning_rate": 6.135281520764049e-05, + "loss": 3.3243, + "step": 4227 + }, + { + "epoch": 6.19, + "grad_norm": 1.4934719800949097, + "learning_rate": 6.130956035776951e-05, + "loss": 3.3423, + "step": 4228 + }, + { + "epoch": 6.19, + "grad_norm": 1.9269001483917236, + "learning_rate": 6.126631401953598e-05, + "loss": 3.2633, + "step": 4229 + }, + { + "epoch": 6.19, + "grad_norm": 1.3831140995025635, + "learning_rate": 6.12230762024538e-05, + "loss": 3.286, + "step": 4230 + }, + { + "epoch": 6.19, + "grad_norm": 2.75412917137146, + "learning_rate": 6.117984691603493e-05, + "loss": 3.2788, + "step": 4231 + }, + { + "epoch": 6.19, + "grad_norm": 3.342359781265259, + "learning_rate": 6.113662616978961e-05, + "loss": 3.2737, + "step": 4232 + }, + { + "epoch": 6.19, + "grad_norm": 0.9600224494934082, + "learning_rate": 6.109341397322611e-05, + "loss": 3.2899, + "step": 4233 + }, + { + "epoch": 6.19, + "grad_norm": 2.350534200668335, + "learning_rate": 6.105021033585074e-05, + "loss": 3.2768, + "step": 4234 + }, + { + "epoch": 6.2, + "grad_norm": 1.1939469575881958, + "learning_rate": 6.10070152671681e-05, + "loss": 3.2706, + "step": 4235 + }, + { + "epoch": 6.2, + "grad_norm": 4.221333980560303, + "learning_rate": 6.096382877668079e-05, + "loss": 3.2853, + "step": 4236 + }, + { + "epoch": 6.2, + "grad_norm": 4.897353649139404, + "learning_rate": 6.092065087388953e-05, + "loss": 3.3084, + "step": 4237 + }, + { + "epoch": 6.2, + "grad_norm": 2.3200554847717285, + "learning_rate": 6.087748156829316e-05, + "loss": 3.2422, + "step": 4238 + }, + { + "epoch": 6.2, + "grad_norm": 4.042237758636475, + "learning_rate": 6.08343208693887e-05, + "loss": 3.2441, + "step": 4239 + }, + { + "epoch": 6.2, + "grad_norm": 4.729053974151611, + "learning_rate": 6.079116878667115e-05, + "loss": 3.2911, + "step": 4240 + }, + { + "epoch": 6.2, + "grad_norm": 3.419166088104248, + "learning_rate": 6.074802532963368e-05, + "loss": 3.2984, + "step": 4241 + }, + { + "epoch": 6.21, + "grad_norm": 2.1573331356048584, + "learning_rate": 6.0704890507767597e-05, + "loss": 3.2494, + "step": 4242 + }, + { + "epoch": 6.21, + "grad_norm": 3.604609489440918, + "learning_rate": 6.0661764330562276e-05, + "loss": 3.2961, + "step": 4243 + }, + { + "epoch": 6.21, + "grad_norm": 1.5325900316238403, + "learning_rate": 6.061864680750513e-05, + "loss": 3.2401, + "step": 4244 + }, + { + "epoch": 6.21, + "grad_norm": 3.4049322605133057, + "learning_rate": 6.0575537948081793e-05, + "loss": 3.292, + "step": 4245 + }, + { + "epoch": 6.21, + "grad_norm": 4.517602920532227, + "learning_rate": 6.053243776177591e-05, + "loss": 3.2811, + "step": 4246 + }, + { + "epoch": 6.21, + "grad_norm": 2.2732179164886475, + "learning_rate": 6.0489346258069177e-05, + "loss": 3.2883, + "step": 4247 + }, + { + "epoch": 6.22, + "grad_norm": 5.137287139892578, + "learning_rate": 6.044626344644151e-05, + "loss": 3.2465, + "step": 4248 + }, + { + "epoch": 6.22, + "grad_norm": 5.510692596435547, + "learning_rate": 6.040318933637081e-05, + "loss": 3.3609, + "step": 4249 + }, + { + "epoch": 6.22, + "grad_norm": 4.384718894958496, + "learning_rate": 6.036012393733306e-05, + "loss": 3.2768, + "step": 4250 + }, + { + "epoch": 6.22, + "grad_norm": 0.5632131695747375, + "learning_rate": 6.031706725880243e-05, + "loss": 3.2637, + "step": 4251 + }, + { + "epoch": 6.22, + "grad_norm": 4.083331108093262, + "learning_rate": 6.027401931025103e-05, + "loss": 3.2579, + "step": 4252 + }, + { + "epoch": 6.22, + "grad_norm": 4.434952735900879, + "learning_rate": 6.0230980101149136e-05, + "loss": 3.2995, + "step": 4253 + }, + { + "epoch": 6.22, + "grad_norm": 1.9891043901443481, + "learning_rate": 6.0187949640965146e-05, + "loss": 3.2652, + "step": 4254 + }, + { + "epoch": 6.23, + "grad_norm": 3.985672950744629, + "learning_rate": 6.014492793916539e-05, + "loss": 3.256, + "step": 4255 + }, + { + "epoch": 6.23, + "grad_norm": 5.051290988922119, + "learning_rate": 6.010191500521439e-05, + "loss": 3.2773, + "step": 4256 + }, + { + "epoch": 6.23, + "grad_norm": 3.2940456867218018, + "learning_rate": 6.005891084857473e-05, + "loss": 3.3033, + "step": 4257 + }, + { + "epoch": 6.23, + "grad_norm": 2.565002202987671, + "learning_rate": 6.001591547870701e-05, + "loss": 3.2838, + "step": 4258 + }, + { + "epoch": 6.23, + "grad_norm": 3.5721094608306885, + "learning_rate": 5.997292890506989e-05, + "loss": 3.2499, + "step": 4259 + }, + { + "epoch": 6.23, + "grad_norm": 1.7328767776489258, + "learning_rate": 5.992995113712019e-05, + "loss": 3.2925, + "step": 4260 + }, + { + "epoch": 6.23, + "grad_norm": 3.9454033374786377, + "learning_rate": 5.9886982184312725e-05, + "loss": 3.2744, + "step": 4261 + }, + { + "epoch": 6.24, + "grad_norm": 4.74951171875, + "learning_rate": 5.984402205610031e-05, + "loss": 3.2632, + "step": 4262 + }, + { + "epoch": 6.24, + "grad_norm": 2.6633002758026123, + "learning_rate": 5.980107076193393e-05, + "loss": 3.2585, + "step": 4263 + }, + { + "epoch": 6.24, + "grad_norm": 2.6646337509155273, + "learning_rate": 5.975812831126261e-05, + "loss": 3.3006, + "step": 4264 + }, + { + "epoch": 6.24, + "grad_norm": 3.7945520877838135, + "learning_rate": 5.971519471353335e-05, + "loss": 3.2882, + "step": 4265 + }, + { + "epoch": 6.24, + "grad_norm": 1.7113827466964722, + "learning_rate": 5.9672269978191266e-05, + "loss": 3.2554, + "step": 4266 + }, + { + "epoch": 6.24, + "grad_norm": 3.288426399230957, + "learning_rate": 5.9629354114679556e-05, + "loss": 3.329, + "step": 4267 + }, + { + "epoch": 6.25, + "grad_norm": 5.059687614440918, + "learning_rate": 5.958644713243937e-05, + "loss": 3.2254, + "step": 4268 + }, + { + "epoch": 6.25, + "grad_norm": 2.444850206375122, + "learning_rate": 5.9543549040909954e-05, + "loss": 3.2804, + "step": 4269 + }, + { + "epoch": 6.25, + "grad_norm": 2.6846628189086914, + "learning_rate": 5.9500659849528664e-05, + "loss": 3.2151, + "step": 4270 + }, + { + "epoch": 6.25, + "grad_norm": 3.8976359367370605, + "learning_rate": 5.945777956773077e-05, + "loss": 3.3193, + "step": 4271 + }, + { + "epoch": 6.25, + "grad_norm": 2.4183270931243896, + "learning_rate": 5.941490820494965e-05, + "loss": 3.261, + "step": 4272 + }, + { + "epoch": 6.25, + "grad_norm": 3.022622585296631, + "learning_rate": 5.937204577061677e-05, + "loss": 3.2569, + "step": 4273 + }, + { + "epoch": 6.25, + "grad_norm": 3.768277168273926, + "learning_rate": 5.932919227416156e-05, + "loss": 3.2582, + "step": 4274 + }, + { + "epoch": 6.26, + "grad_norm": 2.1014842987060547, + "learning_rate": 5.9286347725011424e-05, + "loss": 3.236, + "step": 4275 + }, + { + "epoch": 6.26, + "grad_norm": 2.9817419052124023, + "learning_rate": 5.9243512132591985e-05, + "loss": 3.2667, + "step": 4276 + }, + { + "epoch": 6.26, + "grad_norm": 4.514947891235352, + "learning_rate": 5.9200685506326734e-05, + "loss": 3.3089, + "step": 4277 + }, + { + "epoch": 6.26, + "grad_norm": 2.3111038208007812, + "learning_rate": 5.915786785563722e-05, + "loss": 3.253, + "step": 4278 + }, + { + "epoch": 6.26, + "grad_norm": 2.7890591621398926, + "learning_rate": 5.911505918994309e-05, + "loss": 3.2643, + "step": 4279 + }, + { + "epoch": 6.26, + "grad_norm": 3.4461700916290283, + "learning_rate": 5.907225951866195e-05, + "loss": 3.2625, + "step": 4280 + }, + { + "epoch": 6.26, + "grad_norm": 2.0314648151397705, + "learning_rate": 5.902946885120939e-05, + "loss": 3.2712, + "step": 4281 + }, + { + "epoch": 6.27, + "grad_norm": 2.992061138153076, + "learning_rate": 5.8986687196999135e-05, + "loss": 3.2365, + "step": 4282 + }, + { + "epoch": 6.27, + "grad_norm": 4.312194347381592, + "learning_rate": 5.8943914565442856e-05, + "loss": 3.2867, + "step": 4283 + }, + { + "epoch": 6.27, + "grad_norm": 2.6370973587036133, + "learning_rate": 5.8901150965950194e-05, + "loss": 3.239, + "step": 4284 + }, + { + "epoch": 6.27, + "grad_norm": 3.0366334915161133, + "learning_rate": 5.885839640792892e-05, + "loss": 3.2651, + "step": 4285 + }, + { + "epoch": 6.27, + "grad_norm": 4.975083827972412, + "learning_rate": 5.881565090078469e-05, + "loss": 3.2896, + "step": 4286 + }, + { + "epoch": 6.27, + "grad_norm": 2.4393575191497803, + "learning_rate": 5.8772914453921276e-05, + "loss": 3.2273, + "step": 4287 + }, + { + "epoch": 6.27, + "grad_norm": 2.5122718811035156, + "learning_rate": 5.873018707674036e-05, + "loss": 3.2617, + "step": 4288 + }, + { + "epoch": 6.28, + "grad_norm": 3.912677526473999, + "learning_rate": 5.868746877864173e-05, + "loss": 3.3112, + "step": 4289 + }, + { + "epoch": 6.28, + "grad_norm": 1.8555430173873901, + "learning_rate": 5.864475956902312e-05, + "loss": 3.322, + "step": 4290 + }, + { + "epoch": 6.28, + "grad_norm": 11.267170906066895, + "learning_rate": 5.860205945728021e-05, + "loss": 3.2759, + "step": 4291 + }, + { + "epoch": 6.28, + "grad_norm": 7.718640327453613, + "learning_rate": 5.855936845280682e-05, + "loss": 3.2407, + "step": 4292 + }, + { + "epoch": 6.28, + "grad_norm": 8.495302200317383, + "learning_rate": 5.851668656499465e-05, + "loss": 3.3493, + "step": 4293 + }, + { + "epoch": 6.28, + "grad_norm": 4.235414981842041, + "learning_rate": 5.8474013803233394e-05, + "loss": 3.2929, + "step": 4294 + }, + { + "epoch": 6.29, + "grad_norm": 2.0188095569610596, + "learning_rate": 5.843135017691085e-05, + "loss": 3.2134, + "step": 4295 + }, + { + "epoch": 6.29, + "grad_norm": 4.253573417663574, + "learning_rate": 5.8388695695412666e-05, + "loss": 3.2903, + "step": 4296 + }, + { + "epoch": 6.29, + "grad_norm": 2.978238582611084, + "learning_rate": 5.834605036812255e-05, + "loss": 3.2783, + "step": 4297 + }, + { + "epoch": 6.29, + "grad_norm": 1.3776922225952148, + "learning_rate": 5.830341420442221e-05, + "loss": 3.2773, + "step": 4298 + }, + { + "epoch": 6.29, + "grad_norm": 2.41166353225708, + "learning_rate": 5.82607872136913e-05, + "loss": 3.2214, + "step": 4299 + }, + { + "epoch": 6.29, + "grad_norm": 0.7983728647232056, + "learning_rate": 5.821816940530749e-05, + "loss": 3.3103, + "step": 4300 + }, + { + "epoch": 6.29, + "grad_norm": 1.7378346920013428, + "learning_rate": 5.8175560788646414e-05, + "loss": 3.2548, + "step": 4301 + }, + { + "epoch": 6.3, + "grad_norm": 0.7058059573173523, + "learning_rate": 5.813296137308164e-05, + "loss": 3.2865, + "step": 4302 + }, + { + "epoch": 6.3, + "grad_norm": 2.132636070251465, + "learning_rate": 5.809037116798479e-05, + "loss": 3.2757, + "step": 4303 + }, + { + "epoch": 6.3, + "grad_norm": 1.359208345413208, + "learning_rate": 5.804779018272544e-05, + "loss": 3.2997, + "step": 4304 + }, + { + "epoch": 6.3, + "grad_norm": 2.7147774696350098, + "learning_rate": 5.800521842667107e-05, + "loss": 3.2789, + "step": 4305 + }, + { + "epoch": 6.3, + "grad_norm": 1.9574331045150757, + "learning_rate": 5.7962655909187226e-05, + "loss": 3.2634, + "step": 4306 + }, + { + "epoch": 6.3, + "grad_norm": 1.1457573175430298, + "learning_rate": 5.792010263963735e-05, + "loss": 3.2802, + "step": 4307 + }, + { + "epoch": 6.3, + "grad_norm": 0.9981921315193176, + "learning_rate": 5.7877558627382886e-05, + "loss": 3.3431, + "step": 4308 + }, + { + "epoch": 6.31, + "grad_norm": 1.708836317062378, + "learning_rate": 5.783502388178328e-05, + "loss": 3.2667, + "step": 4309 + }, + { + "epoch": 6.31, + "grad_norm": 1.0355262756347656, + "learning_rate": 5.7792498412195764e-05, + "loss": 3.2917, + "step": 4310 + }, + { + "epoch": 6.31, + "grad_norm": 3.3587546348571777, + "learning_rate": 5.774998222797577e-05, + "loss": 3.2507, + "step": 4311 + }, + { + "epoch": 6.31, + "grad_norm": 3.278871774673462, + "learning_rate": 5.770747533847658e-05, + "loss": 3.2427, + "step": 4312 + }, + { + "epoch": 6.31, + "grad_norm": 0.9090314507484436, + "learning_rate": 5.766497775304931e-05, + "loss": 3.2401, + "step": 4313 + }, + { + "epoch": 6.31, + "grad_norm": 3.0170810222625732, + "learning_rate": 5.7622489481043254e-05, + "loss": 3.2708, + "step": 4314 + }, + { + "epoch": 6.31, + "grad_norm": 3.1690783500671387, + "learning_rate": 5.758001053180555e-05, + "loss": 3.2354, + "step": 4315 + }, + { + "epoch": 6.32, + "grad_norm": 0.4450177550315857, + "learning_rate": 5.753754091468115e-05, + "loss": 3.2435, + "step": 4316 + }, + { + "epoch": 6.32, + "grad_norm": 3.6252622604370117, + "learning_rate": 5.749508063901328e-05, + "loss": 3.2594, + "step": 4317 + }, + { + "epoch": 6.32, + "grad_norm": 5.237750053405762, + "learning_rate": 5.745262971414276e-05, + "loss": 3.2698, + "step": 4318 + }, + { + "epoch": 6.32, + "grad_norm": 1.0414494276046753, + "learning_rate": 5.741018814940853e-05, + "loss": 3.2509, + "step": 4319 + }, + { + "epoch": 6.32, + "grad_norm": 2.6525533199310303, + "learning_rate": 5.736775595414756e-05, + "loss": 3.2609, + "step": 4320 + }, + { + "epoch": 6.32, + "grad_norm": 3.908628225326538, + "learning_rate": 5.7325333137694525e-05, + "loss": 3.2479, + "step": 4321 + }, + { + "epoch": 6.33, + "grad_norm": 1.5807950496673584, + "learning_rate": 5.728291970938218e-05, + "loss": 3.2861, + "step": 4322 + }, + { + "epoch": 6.33, + "grad_norm": 3.6872596740722656, + "learning_rate": 5.72405156785413e-05, + "loss": 3.2316, + "step": 4323 + }, + { + "epoch": 6.33, + "grad_norm": 4.621198654174805, + "learning_rate": 5.719812105450035e-05, + "loss": 3.245, + "step": 4324 + }, + { + "epoch": 6.33, + "grad_norm": 1.5106016397476196, + "learning_rate": 5.715573584658589e-05, + "loss": 3.215, + "step": 4325 + }, + { + "epoch": 6.33, + "grad_norm": 3.5040335655212402, + "learning_rate": 5.711336006412251e-05, + "loss": 3.2626, + "step": 4326 + }, + { + "epoch": 6.33, + "grad_norm": 4.461084842681885, + "learning_rate": 5.707099371643245e-05, + "loss": 3.2804, + "step": 4327 + }, + { + "epoch": 6.33, + "grad_norm": 2.81062912940979, + "learning_rate": 5.7028636812836035e-05, + "loss": 3.2476, + "step": 4328 + }, + { + "epoch": 6.34, + "grad_norm": 2.1609275341033936, + "learning_rate": 5.6986289362651624e-05, + "loss": 3.2922, + "step": 4329 + }, + { + "epoch": 6.34, + "grad_norm": 39.97700881958008, + "learning_rate": 5.694395137519525e-05, + "loss": 3.262, + "step": 4330 + }, + { + "epoch": 6.34, + "grad_norm": 4.451782703399658, + "learning_rate": 5.690162285978102e-05, + "loss": 3.3011, + "step": 4331 + }, + { + "epoch": 6.34, + "grad_norm": 1.8545382022857666, + "learning_rate": 5.6859303825720935e-05, + "loss": 3.2684, + "step": 4332 + }, + { + "epoch": 6.34, + "grad_norm": 3.6993982791900635, + "learning_rate": 5.6816994282324875e-05, + "loss": 3.2664, + "step": 4333 + }, + { + "epoch": 6.34, + "grad_norm": 4.563552379608154, + "learning_rate": 5.67746942389007e-05, + "loss": 3.2004, + "step": 4334 + }, + { + "epoch": 6.34, + "grad_norm": 3.6278369426727295, + "learning_rate": 5.6732403704754086e-05, + "loss": 3.1965, + "step": 4335 + }, + { + "epoch": 6.35, + "grad_norm": 1.3190866708755493, + "learning_rate": 5.66901226891887e-05, + "loss": 3.2505, + "step": 4336 + }, + { + "epoch": 6.35, + "grad_norm": 3.1047582626342773, + "learning_rate": 5.664785120150607e-05, + "loss": 3.3744, + "step": 4337 + }, + { + "epoch": 6.35, + "grad_norm": 1.4010859727859497, + "learning_rate": 5.660558925100565e-05, + "loss": 3.2555, + "step": 4338 + }, + { + "epoch": 6.35, + "grad_norm": 2.454122304916382, + "learning_rate": 5.656333684698476e-05, + "loss": 3.2401, + "step": 4339 + }, + { + "epoch": 6.35, + "grad_norm": 2.7071664333343506, + "learning_rate": 5.652109399873873e-05, + "loss": 3.2436, + "step": 4340 + }, + { + "epoch": 6.35, + "grad_norm": 0.4050593674182892, + "learning_rate": 5.647886071556054e-05, + "loss": 3.2843, + "step": 4341 + }, + { + "epoch": 6.35, + "grad_norm": 3.2467408180236816, + "learning_rate": 5.6436637006741375e-05, + "loss": 3.2526, + "step": 4342 + }, + { + "epoch": 6.36, + "grad_norm": 3.083259344100952, + "learning_rate": 5.639442288157015e-05, + "loss": 3.2539, + "step": 4343 + }, + { + "epoch": 6.36, + "grad_norm": 1.2270174026489258, + "learning_rate": 5.635221834933359e-05, + "loss": 3.2435, + "step": 4344 + }, + { + "epoch": 6.36, + "grad_norm": 1.6782840490341187, + "learning_rate": 5.63100234193165e-05, + "loss": 3.2838, + "step": 4345 + }, + { + "epoch": 6.36, + "grad_norm": 1.8353314399719238, + "learning_rate": 5.6267838100801516e-05, + "loss": 3.2793, + "step": 4346 + }, + { + "epoch": 6.36, + "grad_norm": 1.642630934715271, + "learning_rate": 5.622566240306899e-05, + "loss": 3.261, + "step": 4347 + }, + { + "epoch": 6.36, + "grad_norm": 1.191734790802002, + "learning_rate": 5.618349633539741e-05, + "loss": 3.2822, + "step": 4348 + }, + { + "epoch": 6.37, + "grad_norm": 0.7267574071884155, + "learning_rate": 5.614133990706304e-05, + "loss": 3.2482, + "step": 4349 + }, + { + "epoch": 6.37, + "grad_norm": 0.9115816950798035, + "learning_rate": 5.6099193127339864e-05, + "loss": 3.2347, + "step": 4350 + }, + { + "epoch": 6.37, + "grad_norm": 1.4531807899475098, + "learning_rate": 5.6057056005500085e-05, + "loss": 3.2186, + "step": 4351 + }, + { + "epoch": 6.37, + "grad_norm": 1.0633835792541504, + "learning_rate": 5.601492855081345e-05, + "loss": 3.2526, + "step": 4352 + }, + { + "epoch": 6.37, + "grad_norm": 0.716886579990387, + "learning_rate": 5.5972810772547724e-05, + "loss": 3.2448, + "step": 4353 + }, + { + "epoch": 6.37, + "grad_norm": 0.9413835406303406, + "learning_rate": 5.5930702679968636e-05, + "loss": 3.2125, + "step": 4354 + }, + { + "epoch": 6.37, + "grad_norm": 1.2338863611221313, + "learning_rate": 5.58886042823396e-05, + "loss": 3.2228, + "step": 4355 + }, + { + "epoch": 6.38, + "grad_norm": 2.2373838424682617, + "learning_rate": 5.584651558892193e-05, + "loss": 3.2777, + "step": 4356 + }, + { + "epoch": 6.38, + "grad_norm": 1.3574821949005127, + "learning_rate": 5.580443660897503e-05, + "loss": 3.2371, + "step": 4357 + }, + { + "epoch": 6.38, + "grad_norm": 1.0699307918548584, + "learning_rate": 5.576236735175584e-05, + "loss": 3.2889, + "step": 4358 + }, + { + "epoch": 6.38, + "grad_norm": 0.737332284450531, + "learning_rate": 5.572030782651938e-05, + "loss": 3.2533, + "step": 4359 + }, + { + "epoch": 6.38, + "grad_norm": 1.1307657957077026, + "learning_rate": 5.567825804251843e-05, + "loss": 3.2832, + "step": 4360 + }, + { + "epoch": 6.38, + "grad_norm": 0.6265879273414612, + "learning_rate": 5.563621800900369e-05, + "loss": 3.2663, + "step": 4361 + }, + { + "epoch": 6.38, + "grad_norm": 2.2958106994628906, + "learning_rate": 5.559418773522367e-05, + "loss": 3.2426, + "step": 4362 + }, + { + "epoch": 6.39, + "grad_norm": 1.0793524980545044, + "learning_rate": 5.555216723042476e-05, + "loss": 3.2336, + "step": 4363 + }, + { + "epoch": 6.39, + "grad_norm": 1.8671746253967285, + "learning_rate": 5.551015650385119e-05, + "loss": 3.2449, + "step": 4364 + }, + { + "epoch": 6.39, + "grad_norm": 1.5960359573364258, + "learning_rate": 5.5468155564745e-05, + "loss": 3.2135, + "step": 4365 + }, + { + "epoch": 6.39, + "grad_norm": 2.3406100273132324, + "learning_rate": 5.542616442234618e-05, + "loss": 3.2034, + "step": 4366 + }, + { + "epoch": 6.39, + "grad_norm": 1.8202327489852905, + "learning_rate": 5.538418308589246e-05, + "loss": 3.2608, + "step": 4367 + }, + { + "epoch": 6.39, + "grad_norm": 1.2679892778396606, + "learning_rate": 5.534221156461945e-05, + "loss": 3.2336, + "step": 4368 + }, + { + "epoch": 6.4, + "grad_norm": 1.0765639543533325, + "learning_rate": 5.530024986776061e-05, + "loss": 3.2034, + "step": 4369 + }, + { + "epoch": 6.4, + "grad_norm": 2.065032720565796, + "learning_rate": 5.5258298004547246e-05, + "loss": 3.2323, + "step": 4370 + }, + { + "epoch": 6.4, + "grad_norm": 1.7677899599075317, + "learning_rate": 5.521635598420847e-05, + "loss": 3.2416, + "step": 4371 + }, + { + "epoch": 6.4, + "grad_norm": 1.2142525911331177, + "learning_rate": 5.5174423815971254e-05, + "loss": 3.2783, + "step": 4372 + }, + { + "epoch": 6.4, + "grad_norm": 1.2158989906311035, + "learning_rate": 5.513250150906039e-05, + "loss": 3.2203, + "step": 4373 + }, + { + "epoch": 6.4, + "grad_norm": 2.0104503631591797, + "learning_rate": 5.509058907269854e-05, + "loss": 3.2627, + "step": 4374 + }, + { + "epoch": 6.4, + "grad_norm": 1.794409155845642, + "learning_rate": 5.504868651610604e-05, + "loss": 3.2247, + "step": 4375 + }, + { + "epoch": 6.41, + "grad_norm": 1.241929531097412, + "learning_rate": 5.5006793848501314e-05, + "loss": 3.2352, + "step": 4376 + }, + { + "epoch": 6.41, + "grad_norm": 1.3682582378387451, + "learning_rate": 5.4964911079100446e-05, + "loss": 3.2834, + "step": 4377 + }, + { + "epoch": 6.41, + "grad_norm": 1.42366623878479, + "learning_rate": 5.492303821711724e-05, + "loss": 3.2356, + "step": 4378 + }, + { + "epoch": 6.41, + "grad_norm": 0.9967645406723022, + "learning_rate": 5.488117527176357e-05, + "loss": 3.2623, + "step": 4379 + }, + { + "epoch": 6.41, + "grad_norm": 1.6339219808578491, + "learning_rate": 5.4839322252249016e-05, + "loss": 3.2162, + "step": 4380 + }, + { + "epoch": 6.41, + "grad_norm": 0.9269053936004639, + "learning_rate": 5.479747916778088e-05, + "loss": 3.2236, + "step": 4381 + }, + { + "epoch": 6.41, + "grad_norm": 3.181241989135742, + "learning_rate": 5.4755646027564336e-05, + "loss": 3.2464, + "step": 4382 + }, + { + "epoch": 6.42, + "grad_norm": 3.343838930130005, + "learning_rate": 5.4713822840802544e-05, + "loss": 3.227, + "step": 4383 + }, + { + "epoch": 6.42, + "grad_norm": 0.8503923416137695, + "learning_rate": 5.467200961669619e-05, + "loss": 3.2975, + "step": 4384 + }, + { + "epoch": 6.42, + "grad_norm": 2.970435380935669, + "learning_rate": 5.4630206364443936e-05, + "loss": 3.2325, + "step": 4385 + }, + { + "epoch": 6.42, + "grad_norm": 3.7299089431762695, + "learning_rate": 5.4588413093242244e-05, + "loss": 3.1893, + "step": 4386 + }, + { + "epoch": 6.42, + "grad_norm": 1.3558520078659058, + "learning_rate": 5.454662981228532e-05, + "loss": 3.2106, + "step": 4387 + }, + { + "epoch": 6.42, + "grad_norm": 3.116187572479248, + "learning_rate": 5.450485653076524e-05, + "loss": 3.2292, + "step": 4388 + }, + { + "epoch": 6.42, + "grad_norm": 3.9433794021606445, + "learning_rate": 5.446309325787181e-05, + "loss": 3.2971, + "step": 4389 + }, + { + "epoch": 6.43, + "grad_norm": 2.001657247543335, + "learning_rate": 5.44213400027927e-05, + "loss": 3.2795, + "step": 4390 + }, + { + "epoch": 6.43, + "grad_norm": 2.557114601135254, + "learning_rate": 5.4379596774713334e-05, + "loss": 3.2177, + "step": 4391 + }, + { + "epoch": 6.43, + "grad_norm": 3.371882677078247, + "learning_rate": 5.4337863582816936e-05, + "loss": 3.2478, + "step": 4392 + }, + { + "epoch": 6.43, + "grad_norm": 1.5067156553268433, + "learning_rate": 5.4296140436284546e-05, + "loss": 3.2362, + "step": 4393 + }, + { + "epoch": 6.43, + "grad_norm": 2.2881274223327637, + "learning_rate": 5.425442734429495e-05, + "loss": 3.2567, + "step": 4394 + }, + { + "epoch": 6.43, + "eval_loss": 3.067939043045044, + "eval_runtime": 1166.4557, + "eval_samples_per_second": 14.696, + "eval_steps_per_second": 0.919, + "step": 4394 + }, + { + "epoch": 6.43, + "grad_norm": 2.7140753269195557, + "learning_rate": 5.4212724316024776e-05, + "loss": 3.1976, + "step": 4395 + }, + { + "epoch": 6.44, + "grad_norm": 1.5862021446228027, + "learning_rate": 5.4171031360648406e-05, + "loss": 3.2628, + "step": 4396 + }, + { + "epoch": 6.44, + "grad_norm": 1.6570472717285156, + "learning_rate": 5.4129348487337995e-05, + "loss": 3.2399, + "step": 4397 + }, + { + "epoch": 6.44, + "grad_norm": 1.0379539728164673, + "learning_rate": 5.408767570526352e-05, + "loss": 3.2495, + "step": 4398 + }, + { + "epoch": 6.44, + "grad_norm": 0.9085617065429688, + "learning_rate": 5.404601302359269e-05, + "loss": 3.2073, + "step": 4399 + }, + { + "epoch": 6.44, + "grad_norm": 1.3575804233551025, + "learning_rate": 5.400436045149102e-05, + "loss": 3.2191, + "step": 4400 + }, + { + "epoch": 6.44, + "grad_norm": 1.4507752656936646, + "learning_rate": 5.39627179981218e-05, + "loss": 3.2083, + "step": 4401 + }, + { + "epoch": 6.44, + "grad_norm": 1.4040168523788452, + "learning_rate": 5.392108567264609e-05, + "loss": 3.2492, + "step": 4402 + }, + { + "epoch": 6.45, + "grad_norm": 1.684368371963501, + "learning_rate": 5.387946348422274e-05, + "loss": 3.2433, + "step": 4403 + }, + { + "epoch": 6.45, + "grad_norm": 1.438843846321106, + "learning_rate": 5.383785144200832e-05, + "loss": 3.2335, + "step": 4404 + }, + { + "epoch": 6.45, + "grad_norm": 1.311325192451477, + "learning_rate": 5.379624955515723e-05, + "loss": 3.2418, + "step": 4405 + }, + { + "epoch": 6.45, + "grad_norm": 1.0702279806137085, + "learning_rate": 5.375465783282163e-05, + "loss": 3.2534, + "step": 4406 + }, + { + "epoch": 6.45, + "grad_norm": 1.4737073183059692, + "learning_rate": 5.371307628415131e-05, + "loss": 3.2167, + "step": 4407 + }, + { + "epoch": 6.45, + "grad_norm": 1.0920416116714478, + "learning_rate": 5.367150491829408e-05, + "loss": 3.2418, + "step": 4408 + }, + { + "epoch": 6.45, + "grad_norm": 2.5086584091186523, + "learning_rate": 5.362994374439525e-05, + "loss": 3.2499, + "step": 4409 + }, + { + "epoch": 6.46, + "grad_norm": 1.8857651948928833, + "learning_rate": 5.3588392771598e-05, + "loss": 3.2232, + "step": 4410 + }, + { + "epoch": 6.46, + "grad_norm": 1.4724599123001099, + "learning_rate": 5.354685200904337e-05, + "loss": 3.2534, + "step": 4411 + }, + { + "epoch": 6.46, + "grad_norm": 1.5440292358398438, + "learning_rate": 5.350532146586996e-05, + "loss": 3.2061, + "step": 4412 + }, + { + "epoch": 6.46, + "grad_norm": 1.186196208000183, + "learning_rate": 5.346380115121419e-05, + "loss": 3.2526, + "step": 4413 + }, + { + "epoch": 6.46, + "grad_norm": 1.4283154010772705, + "learning_rate": 5.342229107421037e-05, + "loss": 3.2498, + "step": 4414 + }, + { + "epoch": 6.46, + "grad_norm": 1.7490146160125732, + "learning_rate": 5.3380791243990325e-05, + "loss": 3.2403, + "step": 4415 + }, + { + "epoch": 6.46, + "grad_norm": 1.5179111957550049, + "learning_rate": 5.333930166968374e-05, + "loss": 3.2239, + "step": 4416 + }, + { + "epoch": 6.47, + "grad_norm": 1.3260568380355835, + "learning_rate": 5.329782236041816e-05, + "loss": 3.2357, + "step": 4417 + }, + { + "epoch": 6.47, + "grad_norm": 0.9309088587760925, + "learning_rate": 5.325635332531864e-05, + "loss": 3.2212, + "step": 4418 + }, + { + "epoch": 6.47, + "grad_norm": 1.8849228620529175, + "learning_rate": 5.321489457350812e-05, + "loss": 3.2158, + "step": 4419 + }, + { + "epoch": 6.47, + "grad_norm": 2.031489610671997, + "learning_rate": 5.3173446114107265e-05, + "loss": 3.229, + "step": 4420 + }, + { + "epoch": 6.47, + "grad_norm": 1.7095611095428467, + "learning_rate": 5.313200795623443e-05, + "loss": 3.2323, + "step": 4421 + }, + { + "epoch": 6.47, + "grad_norm": 2.060396671295166, + "learning_rate": 5.309058010900576e-05, + "loss": 3.2519, + "step": 4422 + }, + { + "epoch": 6.48, + "grad_norm": 0.750653088092804, + "learning_rate": 5.3049162581535094e-05, + "loss": 3.2344, + "step": 4423 + }, + { + "epoch": 6.48, + "grad_norm": 1.6587752103805542, + "learning_rate": 5.300775538293401e-05, + "loss": 3.2506, + "step": 4424 + }, + { + "epoch": 6.48, + "grad_norm": 0.8414655923843384, + "learning_rate": 5.296635852231182e-05, + "loss": 3.2065, + "step": 4425 + }, + { + "epoch": 6.48, + "grad_norm": 0.4840925335884094, + "learning_rate": 5.292497200877553e-05, + "loss": 3.2054, + "step": 4426 + }, + { + "epoch": 6.48, + "grad_norm": 0.6098539233207703, + "learning_rate": 5.288359585142993e-05, + "loss": 3.2024, + "step": 4427 + }, + { + "epoch": 6.48, + "grad_norm": 0.46651723980903625, + "learning_rate": 5.2842230059377486e-05, + "loss": 3.2161, + "step": 4428 + }, + { + "epoch": 6.48, + "grad_norm": 0.508057177066803, + "learning_rate": 5.280087464171838e-05, + "loss": 3.1871, + "step": 4429 + }, + { + "epoch": 6.49, + "grad_norm": 0.5408399701118469, + "learning_rate": 5.275952960755055e-05, + "loss": 3.2624, + "step": 4430 + }, + { + "epoch": 6.49, + "grad_norm": 1.0045307874679565, + "learning_rate": 5.271819496596964e-05, + "loss": 3.2303, + "step": 4431 + }, + { + "epoch": 6.49, + "grad_norm": 0.6911265850067139, + "learning_rate": 5.2676870726068906e-05, + "loss": 3.2087, + "step": 4432 + }, + { + "epoch": 6.49, + "grad_norm": 0.6002917289733887, + "learning_rate": 5.2635556896939486e-05, + "loss": 3.1978, + "step": 4433 + }, + { + "epoch": 6.49, + "grad_norm": 0.9630332589149475, + "learning_rate": 5.259425348767018e-05, + "loss": 3.2238, + "step": 4434 + }, + { + "epoch": 6.49, + "grad_norm": 0.7534378170967102, + "learning_rate": 5.2552960507347325e-05, + "loss": 3.2308, + "step": 4435 + }, + { + "epoch": 6.49, + "grad_norm": 0.9733850359916687, + "learning_rate": 5.251167796505523e-05, + "loss": 3.2392, + "step": 4436 + }, + { + "epoch": 6.5, + "grad_norm": 1.1052712202072144, + "learning_rate": 5.247040586987575e-05, + "loss": 3.2257, + "step": 4437 + }, + { + "epoch": 6.5, + "grad_norm": 0.8000559210777283, + "learning_rate": 5.2429144230888384e-05, + "loss": 3.2837, + "step": 4438 + }, + { + "epoch": 6.5, + "grad_norm": 0.7555025219917297, + "learning_rate": 5.238789305717053e-05, + "loss": 3.2463, + "step": 4439 + }, + { + "epoch": 6.5, + "grad_norm": 1.0577791929244995, + "learning_rate": 5.2346652357797146e-05, + "loss": 3.2328, + "step": 4440 + }, + { + "epoch": 6.5, + "grad_norm": 0.6421762704849243, + "learning_rate": 5.230542214184081e-05, + "loss": 3.2186, + "step": 4441 + }, + { + "epoch": 6.5, + "grad_norm": 1.1943224668502808, + "learning_rate": 5.226420241837204e-05, + "loss": 3.2487, + "step": 4442 + }, + { + "epoch": 6.5, + "grad_norm": 0.6367546916007996, + "learning_rate": 5.22229931964588e-05, + "loss": 3.2938, + "step": 4443 + }, + { + "epoch": 6.51, + "grad_norm": 1.3476903438568115, + "learning_rate": 5.2181794485166824e-05, + "loss": 3.2098, + "step": 4444 + }, + { + "epoch": 6.51, + "grad_norm": 1.686813235282898, + "learning_rate": 5.214060629355968e-05, + "loss": 3.2456, + "step": 4445 + }, + { + "epoch": 6.51, + "grad_norm": 0.49779945611953735, + "learning_rate": 5.209942863069837e-05, + "loss": 3.2537, + "step": 4446 + }, + { + "epoch": 6.51, + "grad_norm": 1.0947519540786743, + "learning_rate": 5.20582615056417e-05, + "loss": 3.3028, + "step": 4447 + }, + { + "epoch": 6.51, + "grad_norm": 0.6501612663269043, + "learning_rate": 5.201710492744629e-05, + "loss": 3.2317, + "step": 4448 + }, + { + "epoch": 6.51, + "grad_norm": 0.4535963535308838, + "learning_rate": 5.19759589051662e-05, + "loss": 3.23, + "step": 4449 + }, + { + "epoch": 6.52, + "grad_norm": 0.5250782370567322, + "learning_rate": 5.193482344785327e-05, + "loss": 3.2435, + "step": 4450 + }, + { + "epoch": 6.52, + "grad_norm": 0.400921493768692, + "learning_rate": 5.1893698564557124e-05, + "loss": 3.2188, + "step": 4451 + }, + { + "epoch": 6.52, + "grad_norm": 0.902998149394989, + "learning_rate": 5.1852584264324866e-05, + "loss": 3.2495, + "step": 4452 + }, + { + "epoch": 6.52, + "grad_norm": 0.7799638509750366, + "learning_rate": 5.1811480556201375e-05, + "loss": 3.234, + "step": 4453 + }, + { + "epoch": 6.52, + "grad_norm": 1.8399096727371216, + "learning_rate": 5.1770387449229276e-05, + "loss": 3.3015, + "step": 4454 + }, + { + "epoch": 6.52, + "grad_norm": 0.94755619764328, + "learning_rate": 5.172930495244869e-05, + "loss": 3.2394, + "step": 4455 + }, + { + "epoch": 6.52, + "grad_norm": 2.66386079788208, + "learning_rate": 5.168823307489751e-05, + "loss": 3.2519, + "step": 4456 + }, + { + "epoch": 6.53, + "grad_norm": 1.893293857574463, + "learning_rate": 5.1647171825611284e-05, + "loss": 3.1926, + "step": 4457 + }, + { + "epoch": 6.53, + "grad_norm": 0.6333227753639221, + "learning_rate": 5.160612121362322e-05, + "loss": 3.2169, + "step": 4458 + }, + { + "epoch": 6.53, + "grad_norm": 1.2479146718978882, + "learning_rate": 5.156508124796415e-05, + "loss": 3.2308, + "step": 4459 + }, + { + "epoch": 6.53, + "grad_norm": 0.3292587399482727, + "learning_rate": 5.152405193766263e-05, + "loss": 3.2359, + "step": 4460 + }, + { + "epoch": 6.53, + "grad_norm": 1.253524899482727, + "learning_rate": 5.1483033291744797e-05, + "loss": 3.2176, + "step": 4461 + }, + { + "epoch": 6.53, + "grad_norm": 0.4894808232784271, + "learning_rate": 5.14420253192345e-05, + "loss": 3.2433, + "step": 4462 + }, + { + "epoch": 6.53, + "grad_norm": 1.1630624532699585, + "learning_rate": 5.140102802915321e-05, + "loss": 3.213, + "step": 4463 + }, + { + "epoch": 6.54, + "grad_norm": 0.8647369742393494, + "learning_rate": 5.136004143052007e-05, + "loss": 3.2736, + "step": 4464 + }, + { + "epoch": 6.54, + "grad_norm": 0.9852306842803955, + "learning_rate": 5.1319065532351865e-05, + "loss": 3.2649, + "step": 4465 + }, + { + "epoch": 6.54, + "grad_norm": 0.7359192371368408, + "learning_rate": 5.1278100343662916e-05, + "loss": 3.2311, + "step": 4466 + }, + { + "epoch": 6.54, + "grad_norm": 0.38752880692481995, + "learning_rate": 5.123714587346543e-05, + "loss": 3.2308, + "step": 4467 + }, + { + "epoch": 6.54, + "grad_norm": 1.2246233224868774, + "learning_rate": 5.119620213076909e-05, + "loss": 3.1778, + "step": 4468 + }, + { + "epoch": 6.54, + "grad_norm": 0.7400122284889221, + "learning_rate": 5.115526912458113e-05, + "loss": 3.2127, + "step": 4469 + }, + { + "epoch": 6.55, + "grad_norm": 0.9233433604240417, + "learning_rate": 5.111434686390666e-05, + "loss": 3.184, + "step": 4470 + }, + { + "epoch": 6.55, + "grad_norm": 1.7230321168899536, + "learning_rate": 5.107343535774831e-05, + "loss": 3.2116, + "step": 4471 + }, + { + "epoch": 6.55, + "grad_norm": 1.6479278802871704, + "learning_rate": 5.1032534615106195e-05, + "loss": 3.2049, + "step": 4472 + }, + { + "epoch": 6.55, + "grad_norm": 2.626024007797241, + "learning_rate": 5.0991644644978356e-05, + "loss": 3.2434, + "step": 4473 + }, + { + "epoch": 6.55, + "grad_norm": 3.1591696739196777, + "learning_rate": 5.095076545636028e-05, + "loss": 3.2116, + "step": 4474 + }, + { + "epoch": 6.55, + "grad_norm": 0.8667658567428589, + "learning_rate": 5.0909897058245025e-05, + "loss": 3.1994, + "step": 4475 + }, + { + "epoch": 6.55, + "grad_norm": 3.626915454864502, + "learning_rate": 5.08690394596235e-05, + "loss": 3.221, + "step": 4476 + }, + { + "epoch": 6.56, + "grad_norm": 4.862388610839844, + "learning_rate": 5.0828192669484e-05, + "loss": 3.2172, + "step": 4477 + }, + { + "epoch": 6.56, + "grad_norm": 2.2208385467529297, + "learning_rate": 5.078735669681257e-05, + "loss": 3.2153, + "step": 4478 + }, + { + "epoch": 6.56, + "grad_norm": 2.200000286102295, + "learning_rate": 5.074653155059286e-05, + "loss": 3.2971, + "step": 4479 + }, + { + "epoch": 6.56, + "grad_norm": 2.9845163822174072, + "learning_rate": 5.0705717239806125e-05, + "loss": 3.2522, + "step": 4480 + }, + { + "epoch": 6.56, + "grad_norm": 1.4984869956970215, + "learning_rate": 5.066491377343126e-05, + "loss": 3.2009, + "step": 4481 + }, + { + "epoch": 6.56, + "grad_norm": 2.6040592193603516, + "learning_rate": 5.0624121160444716e-05, + "loss": 3.1945, + "step": 4482 + }, + { + "epoch": 6.56, + "grad_norm": 3.3462085723876953, + "learning_rate": 5.0583339409820616e-05, + "loss": 3.1927, + "step": 4483 + }, + { + "epoch": 6.57, + "grad_norm": 1.2761520147323608, + "learning_rate": 5.0542568530530676e-05, + "loss": 3.1981, + "step": 4484 + }, + { + "epoch": 6.57, + "grad_norm": 2.611973285675049, + "learning_rate": 5.05018085315442e-05, + "loss": 3.1832, + "step": 4485 + }, + { + "epoch": 6.57, + "grad_norm": 3.3917155265808105, + "learning_rate": 5.046105942182815e-05, + "loss": 3.2319, + "step": 4486 + }, + { + "epoch": 6.57, + "grad_norm": 2.2328946590423584, + "learning_rate": 5.042032121034701e-05, + "loss": 3.2443, + "step": 4487 + }, + { + "epoch": 6.57, + "grad_norm": 2.1951651573181152, + "learning_rate": 5.037959390606294e-05, + "loss": 3.2073, + "step": 4488 + }, + { + "epoch": 6.57, + "grad_norm": 3.5785024166107178, + "learning_rate": 5.033887751793567e-05, + "loss": 3.264, + "step": 4489 + }, + { + "epoch": 6.57, + "grad_norm": 1.865713119506836, + "learning_rate": 5.0298172054922534e-05, + "loss": 3.2455, + "step": 4490 + }, + { + "epoch": 6.58, + "grad_norm": 2.268017292022705, + "learning_rate": 5.0257477525978446e-05, + "loss": 3.2089, + "step": 4491 + }, + { + "epoch": 6.58, + "grad_norm": 2.626833200454712, + "learning_rate": 5.021679394005596e-05, + "loss": 3.2474, + "step": 4492 + }, + { + "epoch": 6.58, + "grad_norm": 1.4748424291610718, + "learning_rate": 5.0176121306105164e-05, + "loss": 3.2171, + "step": 4493 + }, + { + "epoch": 6.58, + "grad_norm": 1.906065583229065, + "learning_rate": 5.0135459633073776e-05, + "loss": 3.2625, + "step": 4494 + }, + { + "epoch": 6.58, + "grad_norm": 2.797215700149536, + "learning_rate": 5.00948089299071e-05, + "loss": 3.2316, + "step": 4495 + }, + { + "epoch": 6.58, + "grad_norm": 0.824268639087677, + "learning_rate": 5.0054169205548016e-05, + "loss": 3.211, + "step": 4496 + }, + { + "epoch": 6.59, + "grad_norm": 2.206386089324951, + "learning_rate": 5.001354046893697e-05, + "loss": 3.2278, + "step": 4497 + }, + { + "epoch": 6.59, + "grad_norm": 2.08886981010437, + "learning_rate": 4.997292272901203e-05, + "loss": 3.2376, + "step": 4498 + }, + { + "epoch": 6.59, + "grad_norm": 0.7808924913406372, + "learning_rate": 4.993231599470887e-05, + "loss": 3.2411, + "step": 4499 + }, + { + "epoch": 6.59, + "grad_norm": 1.8568205833435059, + "learning_rate": 4.9891720274960566e-05, + "loss": 3.2718, + "step": 4500 + }, + { + "epoch": 6.59, + "grad_norm": 1.4308000802993774, + "learning_rate": 4.985113557869803e-05, + "loss": 3.2299, + "step": 4501 + }, + { + "epoch": 6.59, + "grad_norm": 1.8029470443725586, + "learning_rate": 4.981056191484962e-05, + "loss": 3.2458, + "step": 4502 + }, + { + "epoch": 6.59, + "grad_norm": 1.7437525987625122, + "learning_rate": 4.9769999292341194e-05, + "loss": 3.2255, + "step": 4503 + }, + { + "epoch": 6.6, + "grad_norm": 1.1221601963043213, + "learning_rate": 4.9729447720096265e-05, + "loss": 3.1949, + "step": 4504 + }, + { + "epoch": 6.6, + "grad_norm": 1.0662968158721924, + "learning_rate": 4.968890720703602e-05, + "loss": 3.1963, + "step": 4505 + }, + { + "epoch": 6.6, + "grad_norm": 1.716160535812378, + "learning_rate": 4.9648377762078966e-05, + "loss": 3.2037, + "step": 4506 + }, + { + "epoch": 6.6, + "grad_norm": 1.2884470224380493, + "learning_rate": 4.960785939414134e-05, + "loss": 3.2482, + "step": 4507 + }, + { + "epoch": 6.6, + "grad_norm": 2.104020595550537, + "learning_rate": 4.956735211213701e-05, + "loss": 3.2097, + "step": 4508 + }, + { + "epoch": 6.6, + "grad_norm": 2.3521175384521484, + "learning_rate": 4.9526855924977175e-05, + "loss": 3.2618, + "step": 4509 + }, + { + "epoch": 6.6, + "grad_norm": 1.780421257019043, + "learning_rate": 4.9486370841570803e-05, + "loss": 3.2177, + "step": 4510 + }, + { + "epoch": 6.61, + "grad_norm": 1.345507264137268, + "learning_rate": 4.94458968708243e-05, + "loss": 3.1907, + "step": 4511 + }, + { + "epoch": 6.61, + "grad_norm": 1.3987131118774414, + "learning_rate": 4.9405434021641705e-05, + "loss": 3.2361, + "step": 4512 + }, + { + "epoch": 6.61, + "grad_norm": 1.295686960220337, + "learning_rate": 4.9364982302924545e-05, + "loss": 3.2055, + "step": 4513 + }, + { + "epoch": 6.61, + "grad_norm": 1.0267897844314575, + "learning_rate": 4.9324541723571936e-05, + "loss": 3.2089, + "step": 4514 + }, + { + "epoch": 6.61, + "grad_norm": 1.2543052434921265, + "learning_rate": 4.9284112292480535e-05, + "loss": 3.2271, + "step": 4515 + }, + { + "epoch": 6.61, + "grad_norm": 1.0035556554794312, + "learning_rate": 4.924369401854455e-05, + "loss": 3.2686, + "step": 4516 + }, + { + "epoch": 6.61, + "grad_norm": 1.4459877014160156, + "learning_rate": 4.920328691065573e-05, + "loss": 3.2149, + "step": 4517 + }, + { + "epoch": 6.62, + "grad_norm": 1.3623418807983398, + "learning_rate": 4.916289097770337e-05, + "loss": 3.2456, + "step": 4518 + }, + { + "epoch": 6.62, + "grad_norm": 1.6200556755065918, + "learning_rate": 4.9122506228574286e-05, + "loss": 3.1883, + "step": 4519 + }, + { + "epoch": 6.62, + "grad_norm": 1.0919500589370728, + "learning_rate": 4.908213267215287e-05, + "loss": 3.1899, + "step": 4520 + }, + { + "epoch": 6.62, + "grad_norm": 1.162226676940918, + "learning_rate": 4.904177031732103e-05, + "loss": 3.2402, + "step": 4521 + }, + { + "epoch": 6.62, + "grad_norm": 1.1378471851348877, + "learning_rate": 4.900141917295822e-05, + "loss": 3.3377, + "step": 4522 + }, + { + "epoch": 6.62, + "grad_norm": 1.183754563331604, + "learning_rate": 4.896107924794141e-05, + "loss": 3.2368, + "step": 4523 + }, + { + "epoch": 6.63, + "grad_norm": 1.1688711643218994, + "learning_rate": 4.892075055114511e-05, + "loss": 3.2358, + "step": 4524 + }, + { + "epoch": 6.63, + "grad_norm": 0.9402367472648621, + "learning_rate": 4.888043309144138e-05, + "loss": 3.1947, + "step": 4525 + }, + { + "epoch": 6.63, + "grad_norm": 0.9889912009239197, + "learning_rate": 4.8840126877699787e-05, + "loss": 3.2409, + "step": 4526 + }, + { + "epoch": 6.63, + "grad_norm": 1.3844680786132812, + "learning_rate": 4.879983191878742e-05, + "loss": 3.2658, + "step": 4527 + }, + { + "epoch": 6.63, + "grad_norm": 0.8452149033546448, + "learning_rate": 4.875954822356893e-05, + "loss": 3.1865, + "step": 4528 + }, + { + "epoch": 6.63, + "grad_norm": 0.9102594256401062, + "learning_rate": 4.871927580090636e-05, + "loss": 3.2226, + "step": 4529 + }, + { + "epoch": 6.63, + "grad_norm": 0.7820479869842529, + "learning_rate": 4.867901465965949e-05, + "loss": 3.2125, + "step": 4530 + }, + { + "epoch": 6.64, + "grad_norm": 1.0344125032424927, + "learning_rate": 4.863876480868549e-05, + "loss": 3.1863, + "step": 4531 + }, + { + "epoch": 6.64, + "grad_norm": 1.1749505996704102, + "learning_rate": 4.8598526256838947e-05, + "loss": 3.2641, + "step": 4532 + }, + { + "epoch": 6.64, + "grad_norm": 1.117828130722046, + "learning_rate": 4.855829901297223e-05, + "loss": 3.1992, + "step": 4533 + }, + { + "epoch": 6.64, + "grad_norm": 1.255987286567688, + "learning_rate": 4.8518083085934954e-05, + "loss": 3.1951, + "step": 4534 + }, + { + "epoch": 6.64, + "grad_norm": 0.6159536838531494, + "learning_rate": 4.847787848457433e-05, + "loss": 3.2087, + "step": 4535 + }, + { + "epoch": 6.64, + "grad_norm": 1.0251574516296387, + "learning_rate": 4.843768521773523e-05, + "loss": 3.202, + "step": 4536 + }, + { + "epoch": 6.64, + "grad_norm": 1.4189834594726562, + "learning_rate": 4.8397503294259795e-05, + "loss": 3.2329, + "step": 4537 + }, + { + "epoch": 6.65, + "grad_norm": 1.0889549255371094, + "learning_rate": 4.835733272298777e-05, + "loss": 3.2087, + "step": 4538 + }, + { + "epoch": 6.65, + "grad_norm": 1.1167737245559692, + "learning_rate": 4.831717351275653e-05, + "loss": 3.2098, + "step": 4539 + }, + { + "epoch": 6.65, + "grad_norm": 1.141223430633545, + "learning_rate": 4.827702567240072e-05, + "loss": 3.2151, + "step": 4540 + }, + { + "epoch": 6.65, + "grad_norm": 1.1248409748077393, + "learning_rate": 4.823688921075258e-05, + "loss": 3.2147, + "step": 4541 + }, + { + "epoch": 6.65, + "grad_norm": 1.471689224243164, + "learning_rate": 4.8196764136642e-05, + "loss": 3.1923, + "step": 4542 + }, + { + "epoch": 6.65, + "grad_norm": 1.6416572332382202, + "learning_rate": 4.81566504588961e-05, + "loss": 3.2319, + "step": 4543 + }, + { + "epoch": 6.65, + "grad_norm": 1.1048762798309326, + "learning_rate": 4.811654818633966e-05, + "loss": 3.2002, + "step": 4544 + }, + { + "epoch": 6.66, + "grad_norm": 1.3798350095748901, + "learning_rate": 4.807645732779491e-05, + "loss": 3.1903, + "step": 4545 + }, + { + "epoch": 6.66, + "grad_norm": 1.5549540519714355, + "learning_rate": 4.8036377892081565e-05, + "loss": 3.181, + "step": 4546 + }, + { + "epoch": 6.66, + "grad_norm": 0.42195671796798706, + "learning_rate": 4.7996309888016844e-05, + "loss": 3.1872, + "step": 4547 + }, + { + "epoch": 6.66, + "grad_norm": 1.59599769115448, + "learning_rate": 4.795625332441545e-05, + "loss": 3.2153, + "step": 4548 + }, + { + "epoch": 6.66, + "grad_norm": 1.7859076261520386, + "learning_rate": 4.7916208210089516e-05, + "loss": 3.246, + "step": 4549 + }, + { + "epoch": 6.66, + "grad_norm": 1.5909998416900635, + "learning_rate": 4.787617455384874e-05, + "loss": 3.1916, + "step": 4550 + }, + { + "epoch": 6.67, + "grad_norm": 2.1160545349121094, + "learning_rate": 4.783615236450024e-05, + "loss": 3.1881, + "step": 4551 + }, + { + "epoch": 6.67, + "grad_norm": 0.9319918751716614, + "learning_rate": 4.779614165084864e-05, + "loss": 3.2286, + "step": 4552 + }, + { + "epoch": 6.67, + "grad_norm": 2.0976734161376953, + "learning_rate": 4.775614242169606e-05, + "loss": 3.2103, + "step": 4553 + }, + { + "epoch": 6.67, + "grad_norm": 1.695967197418213, + "learning_rate": 4.7716154685841944e-05, + "loss": 3.1638, + "step": 4554 + }, + { + "epoch": 6.67, + "grad_norm": 2.1299004554748535, + "learning_rate": 4.767617845208344e-05, + "loss": 3.21, + "step": 4555 + }, + { + "epoch": 6.67, + "grad_norm": 1.7198915481567383, + "learning_rate": 4.763621372921506e-05, + "loss": 3.2789, + "step": 4556 + }, + { + "epoch": 6.67, + "grad_norm": 0.6392155885696411, + "learning_rate": 4.759626052602866e-05, + "loss": 3.2074, + "step": 4557 + }, + { + "epoch": 6.68, + "grad_norm": 1.7603440284729004, + "learning_rate": 4.7556318851313784e-05, + "loss": 3.1682, + "step": 4558 + }, + { + "epoch": 6.68, + "grad_norm": 0.6075698733329773, + "learning_rate": 4.7516388713857344e-05, + "loss": 3.2082, + "step": 4559 + }, + { + "epoch": 6.68, + "grad_norm": 1.1561111211776733, + "learning_rate": 4.7476470122443576e-05, + "loss": 3.2178, + "step": 4560 + }, + { + "epoch": 6.68, + "grad_norm": 0.6579719185829163, + "learning_rate": 4.743656308585441e-05, + "loss": 3.2283, + "step": 4561 + }, + { + "epoch": 6.68, + "grad_norm": 1.3691036701202393, + "learning_rate": 4.739666761286915e-05, + "loss": 3.2209, + "step": 4562 + }, + { + "epoch": 6.68, + "grad_norm": 0.4905560314655304, + "learning_rate": 4.735678371226441e-05, + "loss": 3.1764, + "step": 4563 + }, + { + "epoch": 6.68, + "eval_loss": 3.0459163188934326, + "eval_runtime": 1166.1122, + "eval_samples_per_second": 14.7, + "eval_steps_per_second": 0.919, + "step": 4563 + }, + { + "epoch": 6.68, + "grad_norm": 0.34124380350112915, + "learning_rate": 4.731691139281449e-05, + "loss": 3.2197, + "step": 4564 + }, + { + "epoch": 6.69, + "grad_norm": 0.7039169073104858, + "learning_rate": 4.7277050663291035e-05, + "loss": 3.2013, + "step": 4565 + }, + { + "epoch": 6.69, + "grad_norm": 1.3341939449310303, + "learning_rate": 4.723720153246303e-05, + "loss": 3.2084, + "step": 4566 + }, + { + "epoch": 6.69, + "grad_norm": 0.4564155638217926, + "learning_rate": 4.719736400909716e-05, + "loss": 3.1865, + "step": 4567 + }, + { + "epoch": 6.69, + "grad_norm": 0.9328960180282593, + "learning_rate": 4.715753810195729e-05, + "loss": 3.2035, + "step": 4568 + }, + { + "epoch": 6.69, + "grad_norm": 0.6796619892120361, + "learning_rate": 4.7117723819804884e-05, + "loss": 3.2024, + "step": 4569 + }, + { + "epoch": 6.69, + "grad_norm": 0.4047408699989319, + "learning_rate": 4.7077921171398895e-05, + "loss": 3.1935, + "step": 4570 + }, + { + "epoch": 6.7, + "grad_norm": 0.6333791017532349, + "learning_rate": 4.7038130165495555e-05, + "loss": 3.1794, + "step": 4571 + }, + { + "epoch": 6.7, + "grad_norm": 0.6958187818527222, + "learning_rate": 4.6998350810848605e-05, + "loss": 3.2401, + "step": 4572 + }, + { + "epoch": 6.7, + "grad_norm": 1.0862935781478882, + "learning_rate": 4.6958583116209344e-05, + "loss": 3.2027, + "step": 4573 + }, + { + "epoch": 6.7, + "grad_norm": 0.5394841432571411, + "learning_rate": 4.691882709032629e-05, + "loss": 3.2092, + "step": 4574 + }, + { + "epoch": 6.7, + "grad_norm": 1.7290090322494507, + "learning_rate": 4.687908274194555e-05, + "loss": 3.2199, + "step": 4575 + }, + { + "epoch": 6.7, + "grad_norm": 1.2097941637039185, + "learning_rate": 4.683935007981061e-05, + "loss": 3.2319, + "step": 4576 + }, + { + "epoch": 6.7, + "grad_norm": 1.8092429637908936, + "learning_rate": 4.6799629112662366e-05, + "loss": 3.1968, + "step": 4577 + }, + { + "epoch": 6.71, + "grad_norm": 1.7616616487503052, + "learning_rate": 4.675991984923921e-05, + "loss": 3.2206, + "step": 4578 + }, + { + "epoch": 6.71, + "grad_norm": 0.6763386726379395, + "learning_rate": 4.672022229827687e-05, + "loss": 3.1857, + "step": 4579 + }, + { + "epoch": 6.71, + "grad_norm": 1.4232481718063354, + "learning_rate": 4.668053646850856e-05, + "loss": 3.184, + "step": 4580 + }, + { + "epoch": 6.71, + "grad_norm": 1.7722804546356201, + "learning_rate": 4.664086236866491e-05, + "loss": 3.225, + "step": 4581 + }, + { + "epoch": 6.71, + "grad_norm": 1.0724995136260986, + "learning_rate": 4.660120000747393e-05, + "loss": 3.1924, + "step": 4582 + }, + { + "epoch": 6.71, + "grad_norm": 0.3242776393890381, + "learning_rate": 4.656154939366109e-05, + "loss": 3.2068, + "step": 4583 + }, + { + "epoch": 6.71, + "grad_norm": 1.4377954006195068, + "learning_rate": 4.652191053594924e-05, + "loss": 3.193, + "step": 4584 + }, + { + "epoch": 6.72, + "grad_norm": 4.462525844573975, + "learning_rate": 4.64822834430587e-05, + "loss": 3.275, + "step": 4585 + }, + { + "epoch": 6.72, + "grad_norm": 1.4812723398208618, + "learning_rate": 4.644266812370714e-05, + "loss": 3.2099, + "step": 4586 + }, + { + "epoch": 6.72, + "grad_norm": 0.6171543002128601, + "learning_rate": 4.640306458660966e-05, + "loss": 3.178, + "step": 4587 + }, + { + "epoch": 6.72, + "grad_norm": 1.2330536842346191, + "learning_rate": 4.636347284047877e-05, + "loss": 3.2013, + "step": 4588 + }, + { + "epoch": 6.72, + "grad_norm": 0.5005393624305725, + "learning_rate": 4.6323892894024414e-05, + "loss": 3.2114, + "step": 4589 + }, + { + "epoch": 6.72, + "grad_norm": 1.613756775856018, + "learning_rate": 4.628432475595393e-05, + "loss": 3.1852, + "step": 4590 + }, + { + "epoch": 6.72, + "grad_norm": 0.8560778498649597, + "learning_rate": 4.6244768434971944e-05, + "loss": 3.2414, + "step": 4591 + }, + { + "epoch": 6.73, + "grad_norm": 1.844510555267334, + "learning_rate": 4.620522393978068e-05, + "loss": 3.1791, + "step": 4592 + }, + { + "epoch": 6.73, + "grad_norm": 1.6803126335144043, + "learning_rate": 4.616569127907968e-05, + "loss": 3.2298, + "step": 4593 + }, + { + "epoch": 6.73, + "grad_norm": 1.2029790878295898, + "learning_rate": 4.6126170461565746e-05, + "loss": 3.2044, + "step": 4594 + }, + { + "epoch": 6.73, + "grad_norm": 0.8809365630149841, + "learning_rate": 4.608666149593328e-05, + "loss": 3.1935, + "step": 4595 + }, + { + "epoch": 6.73, + "grad_norm": 0.6382874846458435, + "learning_rate": 4.604716439087403e-05, + "loss": 3.1965, + "step": 4596 + }, + { + "epoch": 6.73, + "grad_norm": 1.0096112489700317, + "learning_rate": 4.600767915507695e-05, + "loss": 3.1924, + "step": 4597 + }, + { + "epoch": 6.74, + "grad_norm": 0.8474308252334595, + "learning_rate": 4.596820579722867e-05, + "loss": 3.1994, + "step": 4598 + }, + { + "epoch": 6.74, + "grad_norm": 1.0024635791778564, + "learning_rate": 4.592874432601304e-05, + "loss": 3.1834, + "step": 4599 + }, + { + "epoch": 6.74, + "grad_norm": 0.9615989923477173, + "learning_rate": 4.588929475011125e-05, + "loss": 3.1855, + "step": 4600 + }, + { + "epoch": 6.74, + "grad_norm": 0.4216245114803314, + "learning_rate": 4.584985707820198e-05, + "loss": 3.2381, + "step": 4601 + }, + { + "epoch": 6.74, + "grad_norm": 0.5493209362030029, + "learning_rate": 4.581043131896126e-05, + "loss": 3.1567, + "step": 4602 + }, + { + "epoch": 6.74, + "grad_norm": 0.6222708225250244, + "learning_rate": 4.5771017481062475e-05, + "loss": 3.2149, + "step": 4603 + }, + { + "epoch": 6.74, + "grad_norm": 0.5318418741226196, + "learning_rate": 4.573161557317642e-05, + "loss": 3.2158, + "step": 4604 + }, + { + "epoch": 6.75, + "grad_norm": 0.7692480087280273, + "learning_rate": 4.5692225603971264e-05, + "loss": 3.2096, + "step": 4605 + }, + { + "epoch": 6.75, + "grad_norm": 0.5248206257820129, + "learning_rate": 4.56528475821125e-05, + "loss": 3.1933, + "step": 4606 + }, + { + "epoch": 6.75, + "grad_norm": 1.0218887329101562, + "learning_rate": 4.561348151626306e-05, + "loss": 3.1819, + "step": 4607 + }, + { + "epoch": 6.75, + "grad_norm": 2.8102357387542725, + "learning_rate": 4.55741274150832e-05, + "loss": 3.2286, + "step": 4608 + }, + { + "epoch": 6.75, + "grad_norm": 0.5493364334106445, + "learning_rate": 4.5534785287230575e-05, + "loss": 3.1981, + "step": 4609 + }, + { + "epoch": 6.75, + "grad_norm": 0.438005656003952, + "learning_rate": 4.5495455141360165e-05, + "loss": 3.1831, + "step": 4610 + }, + { + "epoch": 6.75, + "grad_norm": 1.3687691688537598, + "learning_rate": 4.5456136986124355e-05, + "loss": 3.1993, + "step": 4611 + }, + { + "epoch": 6.76, + "grad_norm": 0.7376412749290466, + "learning_rate": 4.5416830830172876e-05, + "loss": 3.2093, + "step": 4612 + }, + { + "epoch": 6.76, + "grad_norm": 0.710468590259552, + "learning_rate": 4.53775366821528e-05, + "loss": 3.1834, + "step": 4613 + }, + { + "epoch": 6.76, + "grad_norm": 0.6461208462715149, + "learning_rate": 4.53382545507086e-05, + "loss": 3.2279, + "step": 4614 + }, + { + "epoch": 6.76, + "grad_norm": 0.4084911644458771, + "learning_rate": 4.5298984444482075e-05, + "loss": 3.1921, + "step": 4615 + }, + { + "epoch": 6.76, + "grad_norm": 0.5108605027198792, + "learning_rate": 4.5259726372112376e-05, + "loss": 3.2015, + "step": 4616 + }, + { + "epoch": 6.76, + "grad_norm": 0.36462754011154175, + "learning_rate": 4.522048034223603e-05, + "loss": 3.2099, + "step": 4617 + }, + { + "epoch": 6.76, + "grad_norm": 0.5964351892471313, + "learning_rate": 4.518124636348688e-05, + "loss": 3.223, + "step": 4618 + }, + { + "epoch": 6.77, + "grad_norm": 0.532785177230835, + "learning_rate": 4.514202444449615e-05, + "loss": 3.1739, + "step": 4619 + }, + { + "epoch": 6.77, + "grad_norm": 0.7253513336181641, + "learning_rate": 4.510281459389242e-05, + "loss": 3.202, + "step": 4620 + }, + { + "epoch": 6.77, + "grad_norm": 1.3560603857040405, + "learning_rate": 4.506361682030156e-05, + "loss": 3.2175, + "step": 4621 + }, + { + "epoch": 6.77, + "grad_norm": 0.6363286375999451, + "learning_rate": 4.502443113234688e-05, + "loss": 3.2485, + "step": 4622 + }, + { + "epoch": 6.77, + "grad_norm": 1.2093783617019653, + "learning_rate": 4.498525753864884e-05, + "loss": 3.2012, + "step": 4623 + }, + { + "epoch": 6.77, + "grad_norm": 1.1001776456832886, + "learning_rate": 4.4946096047825515e-05, + "loss": 3.1938, + "step": 4624 + }, + { + "epoch": 6.78, + "grad_norm": 0.4927830398082733, + "learning_rate": 4.490694666849208e-05, + "loss": 3.234, + "step": 4625 + }, + { + "epoch": 6.78, + "grad_norm": 1.6655173301696777, + "learning_rate": 4.486780940926112e-05, + "loss": 3.1661, + "step": 4626 + }, + { + "epoch": 6.78, + "grad_norm": 1.3249220848083496, + "learning_rate": 4.482868427874268e-05, + "loss": 3.2283, + "step": 4627 + }, + { + "epoch": 6.78, + "grad_norm": 2.2860875129699707, + "learning_rate": 4.478957128554392e-05, + "loss": 3.2115, + "step": 4628 + }, + { + "epoch": 6.78, + "grad_norm": 2.1470510959625244, + "learning_rate": 4.4750470438269446e-05, + "loss": 3.2151, + "step": 4629 + }, + { + "epoch": 6.78, + "grad_norm": 1.3575111627578735, + "learning_rate": 4.4711381745521275e-05, + "loss": 3.1938, + "step": 4630 + }, + { + "epoch": 6.78, + "grad_norm": 1.7453752756118774, + "learning_rate": 4.467230521589856e-05, + "loss": 3.1969, + "step": 4631 + }, + { + "epoch": 6.79, + "grad_norm": 1.7523752450942993, + "learning_rate": 4.463324085799786e-05, + "loss": 3.2266, + "step": 4632 + }, + { + "epoch": 6.79, + "grad_norm": 1.433594822883606, + "learning_rate": 4.4594188680413206e-05, + "loss": 3.1962, + "step": 4633 + }, + { + "epoch": 6.79, + "grad_norm": 1.0330843925476074, + "learning_rate": 4.4555148691735685e-05, + "loss": 3.2325, + "step": 4634 + }, + { + "epoch": 6.79, + "grad_norm": 0.9396421313285828, + "learning_rate": 4.451612090055388e-05, + "loss": 3.2274, + "step": 4635 + }, + { + "epoch": 6.79, + "grad_norm": 1.1527962684631348, + "learning_rate": 4.447710531545364e-05, + "loss": 3.2075, + "step": 4636 + }, + { + "epoch": 6.79, + "grad_norm": 0.8900932669639587, + "learning_rate": 4.443810194501813e-05, + "loss": 3.1952, + "step": 4637 + }, + { + "epoch": 6.79, + "grad_norm": 1.3143221139907837, + "learning_rate": 4.4399110797827835e-05, + "loss": 3.2086, + "step": 4638 + }, + { + "epoch": 6.8, + "grad_norm": 1.1299476623535156, + "learning_rate": 4.4360131882460555e-05, + "loss": 3.2, + "step": 4639 + }, + { + "epoch": 6.8, + "grad_norm": 0.8985263109207153, + "learning_rate": 4.432116520749137e-05, + "loss": 3.15, + "step": 4640 + }, + { + "epoch": 6.8, + "grad_norm": 1.0726135969161987, + "learning_rate": 4.4282210781492684e-05, + "loss": 3.2016, + "step": 4641 + }, + { + "epoch": 6.8, + "grad_norm": 0.556115984916687, + "learning_rate": 4.424326861303424e-05, + "loss": 3.2035, + "step": 4642 + }, + { + "epoch": 6.8, + "grad_norm": 1.1372523307800293, + "learning_rate": 4.420433871068302e-05, + "loss": 3.1998, + "step": 4643 + }, + { + "epoch": 6.8, + "grad_norm": 0.9902729988098145, + "learning_rate": 4.416542108300338e-05, + "loss": 3.1661, + "step": 4644 + }, + { + "epoch": 6.8, + "grad_norm": 0.6972740292549133, + "learning_rate": 4.412651573855689e-05, + "loss": 3.191, + "step": 4645 + }, + { + "epoch": 6.81, + "grad_norm": 0.7196264863014221, + "learning_rate": 4.408762268590251e-05, + "loss": 3.1772, + "step": 4646 + }, + { + "epoch": 6.81, + "grad_norm": 0.6130596995353699, + "learning_rate": 4.404874193359646e-05, + "loss": 3.1859, + "step": 4647 + }, + { + "epoch": 6.81, + "grad_norm": 0.8541386723518372, + "learning_rate": 4.4009873490192135e-05, + "loss": 3.1891, + "step": 4648 + }, + { + "epoch": 6.81, + "grad_norm": 0.4068596065044403, + "learning_rate": 4.397101736424046e-05, + "loss": 3.2386, + "step": 4649 + }, + { + "epoch": 6.81, + "grad_norm": 0.35903048515319824, + "learning_rate": 4.3932173564289506e-05, + "loss": 3.2046, + "step": 4650 + }, + { + "epoch": 6.81, + "grad_norm": 1.0451300144195557, + "learning_rate": 4.389334209888455e-05, + "loss": 3.2677, + "step": 4651 + }, + { + "epoch": 6.82, + "grad_norm": 0.3824341595172882, + "learning_rate": 4.385452297656836e-05, + "loss": 3.1955, + "step": 4652 + }, + { + "epoch": 6.82, + "grad_norm": 0.9694157838821411, + "learning_rate": 4.381571620588088e-05, + "loss": 3.2113, + "step": 4653 + }, + { + "epoch": 6.82, + "grad_norm": 0.6231606602668762, + "learning_rate": 4.3776921795359216e-05, + "loss": 3.1968, + "step": 4654 + }, + { + "epoch": 6.82, + "grad_norm": 0.39795351028442383, + "learning_rate": 4.373813975353802e-05, + "loss": 3.1859, + "step": 4655 + }, + { + "epoch": 6.82, + "grad_norm": 0.3494223654270172, + "learning_rate": 4.3699370088949066e-05, + "loss": 3.2194, + "step": 4656 + }, + { + "epoch": 6.82, + "grad_norm": 0.7372296452522278, + "learning_rate": 4.366061281012129e-05, + "loss": 3.2062, + "step": 4657 + }, + { + "epoch": 6.82, + "grad_norm": 1.144978642463684, + "learning_rate": 4.362186792558122e-05, + "loss": 3.1948, + "step": 4658 + }, + { + "epoch": 6.83, + "grad_norm": 0.5863081812858582, + "learning_rate": 4.358313544385233e-05, + "loss": 3.1734, + "step": 4659 + }, + { + "epoch": 6.83, + "grad_norm": 1.334794282913208, + "learning_rate": 4.354441537345553e-05, + "loss": 3.187, + "step": 4660 + }, + { + "epoch": 6.83, + "grad_norm": 0.4816846251487732, + "learning_rate": 4.350570772290906e-05, + "loss": 3.1916, + "step": 4661 + }, + { + "epoch": 6.83, + "grad_norm": 0.8258687853813171, + "learning_rate": 4.3467012500728245e-05, + "loss": 3.2171, + "step": 4662 + }, + { + "epoch": 6.83, + "grad_norm": 0.9091712236404419, + "learning_rate": 4.342832971542578e-05, + "loss": 3.1695, + "step": 4663 + }, + { + "epoch": 6.83, + "grad_norm": 1.3142671585083008, + "learning_rate": 4.3389659375511725e-05, + "loss": 3.1681, + "step": 4664 + }, + { + "epoch": 6.83, + "grad_norm": 2.6121108531951904, + "learning_rate": 4.335100148949318e-05, + "loss": 3.2161, + "step": 4665 + }, + { + "epoch": 6.84, + "grad_norm": 0.6220512986183167, + "learning_rate": 4.3312356065874635e-05, + "loss": 3.1868, + "step": 4666 + }, + { + "epoch": 6.84, + "grad_norm": 1.4593570232391357, + "learning_rate": 4.327372311315792e-05, + "loss": 3.1968, + "step": 4667 + }, + { + "epoch": 6.84, + "grad_norm": 1.4904062747955322, + "learning_rate": 4.323510263984192e-05, + "loss": 3.2013, + "step": 4668 + }, + { + "epoch": 6.84, + "grad_norm": 0.8087517023086548, + "learning_rate": 4.3196494654422916e-05, + "loss": 3.2106, + "step": 4669 + }, + { + "epoch": 6.84, + "grad_norm": 0.5217124223709106, + "learning_rate": 4.315789916539441e-05, + "loss": 3.2188, + "step": 4670 + }, + { + "epoch": 6.84, + "grad_norm": 0.7983100414276123, + "learning_rate": 4.311931618124715e-05, + "loss": 3.217, + "step": 4671 + }, + { + "epoch": 6.84, + "grad_norm": 0.49825963377952576, + "learning_rate": 4.308074571046913e-05, + "loss": 3.2089, + "step": 4672 + }, + { + "epoch": 6.85, + "grad_norm": 0.9408275485038757, + "learning_rate": 4.30421877615456e-05, + "loss": 3.1924, + "step": 4673 + }, + { + "epoch": 6.85, + "grad_norm": 0.6962942481040955, + "learning_rate": 4.300364234295905e-05, + "loss": 3.2022, + "step": 4674 + }, + { + "epoch": 6.85, + "grad_norm": 0.5625227093696594, + "learning_rate": 4.296510946318921e-05, + "loss": 3.2056, + "step": 4675 + }, + { + "epoch": 6.85, + "grad_norm": 0.6208386421203613, + "learning_rate": 4.292658913071306e-05, + "loss": 3.1973, + "step": 4676 + }, + { + "epoch": 6.85, + "grad_norm": 0.5512813925743103, + "learning_rate": 4.2888081354004805e-05, + "loss": 3.2122, + "step": 4677 + }, + { + "epoch": 6.85, + "grad_norm": 1.9432053565979004, + "learning_rate": 4.2849586141535945e-05, + "loss": 3.2239, + "step": 4678 + }, + { + "epoch": 6.86, + "grad_norm": 4.5366291999816895, + "learning_rate": 4.281110350177506e-05, + "loss": 3.1618, + "step": 4679 + }, + { + "epoch": 6.86, + "grad_norm": 0.9245926737785339, + "learning_rate": 4.277263344318817e-05, + "loss": 3.2309, + "step": 4680 + }, + { + "epoch": 6.86, + "grad_norm": 0.3581990897655487, + "learning_rate": 4.2734175974238444e-05, + "loss": 3.1872, + "step": 4681 + }, + { + "epoch": 6.86, + "grad_norm": 0.9905257821083069, + "learning_rate": 4.269573110338615e-05, + "loss": 3.2077, + "step": 4682 + }, + { + "epoch": 6.86, + "grad_norm": 1.2098913192749023, + "learning_rate": 4.265729883908902e-05, + "loss": 3.2249, + "step": 4683 + }, + { + "epoch": 6.86, + "grad_norm": 0.711704671382904, + "learning_rate": 4.2618879189801885e-05, + "loss": 3.1625, + "step": 4684 + }, + { + "epoch": 6.86, + "grad_norm": 2.870743989944458, + "learning_rate": 4.25804721639767e-05, + "loss": 3.1953, + "step": 4685 + }, + { + "epoch": 6.87, + "grad_norm": 1.7720706462860107, + "learning_rate": 4.254207777006287e-05, + "loss": 3.1671, + "step": 4686 + }, + { + "epoch": 6.87, + "grad_norm": 0.8603296875953674, + "learning_rate": 4.25036960165069e-05, + "loss": 3.166, + "step": 4687 + }, + { + "epoch": 6.87, + "grad_norm": 2.4739768505096436, + "learning_rate": 4.246532691175241e-05, + "loss": 3.1946, + "step": 4688 + }, + { + "epoch": 6.87, + "grad_norm": 2.0245025157928467, + "learning_rate": 4.242697046424046e-05, + "loss": 3.2304, + "step": 4689 + }, + { + "epoch": 6.87, + "grad_norm": 0.7027145028114319, + "learning_rate": 4.2388626682409194e-05, + "loss": 3.2526, + "step": 4690 + }, + { + "epoch": 6.87, + "grad_norm": 2.3779094219207764, + "learning_rate": 4.23502955746939e-05, + "loss": 3.2621, + "step": 4691 + }, + { + "epoch": 6.87, + "grad_norm": 1.9404739141464233, + "learning_rate": 4.23119771495273e-05, + "loss": 3.1891, + "step": 4692 + }, + { + "epoch": 6.88, + "grad_norm": 0.6578683257102966, + "learning_rate": 4.227367141533909e-05, + "loss": 3.2023, + "step": 4693 + }, + { + "epoch": 6.88, + "grad_norm": 5.048638820648193, + "learning_rate": 4.2235378380556255e-05, + "loss": 3.136, + "step": 4694 + }, + { + "epoch": 6.88, + "grad_norm": 2.4565985202789307, + "learning_rate": 4.2197098053603136e-05, + "loss": 3.1868, + "step": 4695 + }, + { + "epoch": 6.88, + "grad_norm": 0.770892858505249, + "learning_rate": 4.215883044290103e-05, + "loss": 3.1455, + "step": 4696 + }, + { + "epoch": 6.88, + "grad_norm": 2.0807790756225586, + "learning_rate": 4.2120575556868594e-05, + "loss": 3.1894, + "step": 4697 + }, + { + "epoch": 6.88, + "grad_norm": 2.7063095569610596, + "learning_rate": 4.208233340392164e-05, + "loss": 3.1982, + "step": 4698 + }, + { + "epoch": 6.89, + "grad_norm": 1.1912204027175903, + "learning_rate": 4.2044103992473195e-05, + "loss": 3.1818, + "step": 4699 + }, + { + "epoch": 6.89, + "grad_norm": 2.9936110973358154, + "learning_rate": 4.2005887330933457e-05, + "loss": 3.2109, + "step": 4700 + }, + { + "epoch": 6.89, + "grad_norm": 3.9401204586029053, + "learning_rate": 4.196768342770986e-05, + "loss": 3.2246, + "step": 4701 + }, + { + "epoch": 6.89, + "grad_norm": 2.341156482696533, + "learning_rate": 4.1929492291206983e-05, + "loss": 3.1819, + "step": 4702 + }, + { + "epoch": 6.89, + "grad_norm": 0.7375439405441284, + "learning_rate": 4.1891313929826635e-05, + "loss": 3.2047, + "step": 4703 + }, + { + "epoch": 6.89, + "grad_norm": 1.9721698760986328, + "learning_rate": 4.185314835196781e-05, + "loss": 3.2131, + "step": 4704 + }, + { + "epoch": 6.89, + "grad_norm": 0.8218457102775574, + "learning_rate": 4.181499556602665e-05, + "loss": 3.1891, + "step": 4705 + }, + { + "epoch": 6.9, + "grad_norm": 1.2695006132125854, + "learning_rate": 4.177685558039652e-05, + "loss": 3.1774, + "step": 4706 + }, + { + "epoch": 6.9, + "grad_norm": 1.4132906198501587, + "learning_rate": 4.1738728403467974e-05, + "loss": 3.1856, + "step": 4707 + }, + { + "epoch": 6.9, + "grad_norm": 0.5927780866622925, + "learning_rate": 4.170061404362874e-05, + "loss": 3.2209, + "step": 4708 + }, + { + "epoch": 6.9, + "grad_norm": 1.482879877090454, + "learning_rate": 4.166251250926371e-05, + "loss": 3.1621, + "step": 4709 + }, + { + "epoch": 6.9, + "grad_norm": 0.4204716384410858, + "learning_rate": 4.1624423808754956e-05, + "loss": 3.171, + "step": 4710 + }, + { + "epoch": 6.9, + "grad_norm": 1.5936245918273926, + "learning_rate": 4.158634795048175e-05, + "loss": 3.1746, + "step": 4711 + }, + { + "epoch": 6.9, + "grad_norm": 1.1268365383148193, + "learning_rate": 4.154828494282056e-05, + "loss": 3.1932, + "step": 4712 + }, + { + "epoch": 6.91, + "grad_norm": 1.2134582996368408, + "learning_rate": 4.151023479414489e-05, + "loss": 3.2064, + "step": 4713 + }, + { + "epoch": 6.91, + "grad_norm": 1.2098625898361206, + "learning_rate": 4.147219751282562e-05, + "loss": 3.2025, + "step": 4714 + }, + { + "epoch": 6.91, + "grad_norm": 4.699478626251221, + "learning_rate": 4.143417310723069e-05, + "loss": 3.1962, + "step": 4715 + }, + { + "epoch": 6.91, + "grad_norm": 1.845900297164917, + "learning_rate": 4.139616158572512e-05, + "loss": 3.1635, + "step": 4716 + }, + { + "epoch": 6.91, + "grad_norm": 1.1032308340072632, + "learning_rate": 4.1358162956671295e-05, + "loss": 3.1781, + "step": 4717 + }, + { + "epoch": 6.91, + "grad_norm": 1.5497393608093262, + "learning_rate": 4.1320177228428646e-05, + "loss": 3.1825, + "step": 4718 + }, + { + "epoch": 6.91, + "grad_norm": 2.2472078800201416, + "learning_rate": 4.128220440935372e-05, + "loss": 3.1866, + "step": 4719 + }, + { + "epoch": 6.92, + "grad_norm": 92.44196319580078, + "learning_rate": 4.124424450780029e-05, + "loss": 3.2326, + "step": 4720 + }, + { + "epoch": 6.92, + "grad_norm": 1.094028115272522, + "learning_rate": 4.120629753211936e-05, + "loss": 3.2104, + "step": 4721 + }, + { + "epoch": 6.92, + "grad_norm": 0.9184134602546692, + "learning_rate": 4.116836349065892e-05, + "loss": 3.2233, + "step": 4722 + }, + { + "epoch": 6.92, + "grad_norm": 1.4922984838485718, + "learning_rate": 4.1130442391764214e-05, + "loss": 3.1623, + "step": 4723 + }, + { + "epoch": 6.92, + "grad_norm": 0.9635801911354065, + "learning_rate": 4.109253424377772e-05, + "loss": 3.1788, + "step": 4724 + }, + { + "epoch": 6.92, + "grad_norm": 1.1866681575775146, + "learning_rate": 4.105463905503888e-05, + "loss": 3.2056, + "step": 4725 + }, + { + "epoch": 6.93, + "grad_norm": 0.8471019268035889, + "learning_rate": 4.101675683388442e-05, + "loss": 3.1978, + "step": 4726 + }, + { + "epoch": 6.93, + "grad_norm": 0.5931624174118042, + "learning_rate": 4.0978887588648155e-05, + "loss": 3.2222, + "step": 4727 + }, + { + "epoch": 6.93, + "grad_norm": 0.9211606383323669, + "learning_rate": 4.094103132766109e-05, + "loss": 3.1993, + "step": 4728 + }, + { + "epoch": 6.93, + "grad_norm": 0.7005136013031006, + "learning_rate": 4.090318805925133e-05, + "loss": 3.2065, + "step": 4729 + }, + { + "epoch": 6.93, + "grad_norm": 0.9550567269325256, + "learning_rate": 4.086535779174415e-05, + "loss": 3.1617, + "step": 4730 + }, + { + "epoch": 6.93, + "grad_norm": 0.9668344259262085, + "learning_rate": 4.082754053346195e-05, + "loss": 3.2148, + "step": 4731 + }, + { + "epoch": 6.93, + "grad_norm": 0.6807712912559509, + "learning_rate": 4.078973629272428e-05, + "loss": 3.1848, + "step": 4732 + }, + { + "epoch": 6.93, + "eval_loss": 3.0342421531677246, + "eval_runtime": 1171.9875, + "eval_samples_per_second": 14.626, + "eval_steps_per_second": 0.915, + "step": 4732 + }, + { + "epoch": 6.94, + "grad_norm": 1.207279920578003, + "learning_rate": 4.075194507784782e-05, + "loss": 3.1595, + "step": 4733 + }, + { + "epoch": 6.94, + "grad_norm": 1.3139106035232544, + "learning_rate": 4.071416689714637e-05, + "loss": 3.1985, + "step": 4734 + }, + { + "epoch": 6.94, + "grad_norm": 1.053685188293457, + "learning_rate": 4.0676401758930894e-05, + "loss": 3.2048, + "step": 4735 + }, + { + "epoch": 6.94, + "grad_norm": 1.043835997581482, + "learning_rate": 4.063864967150945e-05, + "loss": 3.1763, + "step": 4736 + }, + { + "epoch": 6.94, + "grad_norm": 0.4672507047653198, + "learning_rate": 4.060091064318727e-05, + "loss": 3.1722, + "step": 4737 + }, + { + "epoch": 6.94, + "grad_norm": 1.6286187171936035, + "learning_rate": 4.0563184682266676e-05, + "loss": 3.2824, + "step": 4738 + }, + { + "epoch": 6.94, + "grad_norm": 0.8605678081512451, + "learning_rate": 4.052547179704711e-05, + "loss": 3.198, + "step": 4739 + }, + { + "epoch": 6.95, + "grad_norm": 1.1343073844909668, + "learning_rate": 4.048777199582518e-05, + "loss": 3.1726, + "step": 4740 + }, + { + "epoch": 6.95, + "grad_norm": 0.6983288526535034, + "learning_rate": 4.0450085286894565e-05, + "loss": 3.2139, + "step": 4741 + }, + { + "epoch": 6.95, + "grad_norm": 0.8509522080421448, + "learning_rate": 4.041241167854612e-05, + "loss": 3.176, + "step": 4742 + }, + { + "epoch": 6.95, + "grad_norm": 0.8853251934051514, + "learning_rate": 4.037475117906775e-05, + "loss": 3.1622, + "step": 4743 + }, + { + "epoch": 6.95, + "grad_norm": 1.2624071836471558, + "learning_rate": 4.033710379674458e-05, + "loss": 3.1633, + "step": 4744 + }, + { + "epoch": 6.95, + "grad_norm": 1.0235035419464111, + "learning_rate": 4.029946953985865e-05, + "loss": 3.1794, + "step": 4745 + }, + { + "epoch": 6.95, + "grad_norm": 1.0335735082626343, + "learning_rate": 4.0261848416689394e-05, + "loss": 3.2398, + "step": 4746 + }, + { + "epoch": 6.96, + "grad_norm": 1.2208091020584106, + "learning_rate": 4.022424043551312e-05, + "loss": 3.2424, + "step": 4747 + }, + { + "epoch": 6.96, + "grad_norm": 0.8216298818588257, + "learning_rate": 4.0186645604603304e-05, + "loss": 3.1948, + "step": 4748 + }, + { + "epoch": 6.96, + "grad_norm": 1.2754603624343872, + "learning_rate": 4.014906393223067e-05, + "loss": 3.1608, + "step": 4749 + }, + { + "epoch": 6.96, + "grad_norm": 1.283899188041687, + "learning_rate": 4.011149542666284e-05, + "loss": 3.2403, + "step": 4750 + }, + { + "epoch": 6.96, + "grad_norm": 0.569480836391449, + "learning_rate": 4.007394009616462e-05, + "loss": 3.2395, + "step": 4751 + }, + { + "epoch": 6.96, + "grad_norm": 0.7259752750396729, + "learning_rate": 4.003639794899804e-05, + "loss": 3.1733, + "step": 4752 + }, + { + "epoch": 6.97, + "grad_norm": 0.3225840628147125, + "learning_rate": 3.9998868993422026e-05, + "loss": 3.2326, + "step": 4753 + }, + { + "epoch": 6.97, + "grad_norm": 1.8098729848861694, + "learning_rate": 3.996135323769269e-05, + "loss": 3.2116, + "step": 4754 + }, + { + "epoch": 6.97, + "grad_norm": 0.893907904624939, + "learning_rate": 3.9923850690063345e-05, + "loss": 3.1768, + "step": 4755 + }, + { + "epoch": 6.97, + "grad_norm": 0.552627444267273, + "learning_rate": 3.988636135878421e-05, + "loss": 3.2011, + "step": 4756 + }, + { + "epoch": 6.97, + "grad_norm": 0.809914231300354, + "learning_rate": 3.984888525210266e-05, + "loss": 3.1777, + "step": 4757 + }, + { + "epoch": 6.97, + "grad_norm": 0.4486781060695648, + "learning_rate": 3.981142237826332e-05, + "loss": 3.193, + "step": 4758 + }, + { + "epoch": 6.97, + "grad_norm": 0.7444935441017151, + "learning_rate": 3.977397274550766e-05, + "loss": 3.1806, + "step": 4759 + }, + { + "epoch": 6.98, + "grad_norm": 0.49755245447158813, + "learning_rate": 3.973653636207437e-05, + "loss": 3.1676, + "step": 4760 + }, + { + "epoch": 6.98, + "grad_norm": 1.1105209589004517, + "learning_rate": 3.9699113236199226e-05, + "loss": 3.1885, + "step": 4761 + }, + { + "epoch": 6.98, + "grad_norm": 0.8656307458877563, + "learning_rate": 3.9661703376115034e-05, + "loss": 3.2385, + "step": 4762 + }, + { + "epoch": 6.98, + "grad_norm": 1.250411033630371, + "learning_rate": 3.962430679005175e-05, + "loss": 3.1652, + "step": 4763 + }, + { + "epoch": 6.98, + "grad_norm": 1.7875831127166748, + "learning_rate": 3.958692348623634e-05, + "loss": 3.1795, + "step": 4764 + }, + { + "epoch": 6.98, + "grad_norm": 0.575053870677948, + "learning_rate": 3.954955347289289e-05, + "loss": 3.2002, + "step": 4765 + }, + { + "epoch": 6.98, + "grad_norm": 1.1401422023773193, + "learning_rate": 3.9512196758242556e-05, + "loss": 3.1345, + "step": 4766 + }, + { + "epoch": 6.99, + "grad_norm": 0.6520401835441589, + "learning_rate": 3.947485335050357e-05, + "loss": 3.2271, + "step": 4767 + }, + { + "epoch": 6.99, + "grad_norm": 0.7057487964630127, + "learning_rate": 3.9437523257891215e-05, + "loss": 3.2535, + "step": 4768 + }, + { + "epoch": 6.99, + "grad_norm": 1.187171220779419, + "learning_rate": 3.940020648861792e-05, + "loss": 3.153, + "step": 4769 + }, + { + "epoch": 6.99, + "grad_norm": 0.44475510716438293, + "learning_rate": 3.9362903050892996e-05, + "loss": 3.1748, + "step": 4770 + }, + { + "epoch": 6.99, + "grad_norm": 2.0689454078674316, + "learning_rate": 3.9325612952923065e-05, + "loss": 3.17, + "step": 4771 + }, + { + "epoch": 6.99, + "grad_norm": 1.3041399717330933, + "learning_rate": 3.92883362029117e-05, + "loss": 3.185, + "step": 4772 + }, + { + "epoch": 6.99, + "grad_norm": 0.5959802865982056, + "learning_rate": 3.925107280905943e-05, + "loss": 3.1533, + "step": 4773 + }, + { + "epoch": 7.0, + "grad_norm": 1.501555323600769, + "learning_rate": 3.9213822779564056e-05, + "loss": 3.1333, + "step": 4774 + }, + { + "epoch": 7.0, + "grad_norm": 0.4824700951576233, + "learning_rate": 3.917658612262033e-05, + "loss": 3.173, + "step": 4775 + }, + { + "epoch": 7.0, + "grad_norm": 0.9984436631202698, + "learning_rate": 3.913936284641997e-05, + "loss": 3.1509, + "step": 4776 + }, + { + "epoch": 7.0, + "grad_norm": 0.47394973039627075, + "learning_rate": 3.910215295915195e-05, + "loss": 3.1744, + "step": 4777 + }, + { + "epoch": 7.0, + "grad_norm": 1.570947527885437, + "learning_rate": 3.906495646900219e-05, + "loss": 3.1831, + "step": 4778 + }, + { + "epoch": 7.0, + "grad_norm": 0.9516702890396118, + "learning_rate": 3.902777338415358e-05, + "loss": 3.1569, + "step": 4779 + }, + { + "epoch": 7.01, + "grad_norm": 2.1789183616638184, + "learning_rate": 3.899060371278627e-05, + "loss": 3.1952, + "step": 4780 + }, + { + "epoch": 7.01, + "grad_norm": 3.014054536819458, + "learning_rate": 3.895344746307723e-05, + "loss": 3.2197, + "step": 4781 + }, + { + "epoch": 7.01, + "grad_norm": 1.0081833600997925, + "learning_rate": 3.8916304643200604e-05, + "loss": 3.1867, + "step": 4782 + }, + { + "epoch": 7.01, + "grad_norm": 1.2014774084091187, + "learning_rate": 3.887917526132767e-05, + "loss": 3.2003, + "step": 4783 + }, + { + "epoch": 7.01, + "grad_norm": 2.214982748031616, + "learning_rate": 3.8842059325626536e-05, + "loss": 3.1835, + "step": 4784 + }, + { + "epoch": 7.01, + "grad_norm": 1.1680030822753906, + "learning_rate": 3.880495684426243e-05, + "loss": 3.2276, + "step": 4785 + }, + { + "epoch": 7.01, + "grad_norm": 1.6248522996902466, + "learning_rate": 3.8767867825397796e-05, + "loss": 3.1552, + "step": 4786 + }, + { + "epoch": 7.0, + "grad_norm": 1.5934488773345947, + "learning_rate": 3.873079227719184e-05, + "loss": 3.1989, + "step": 4787 + }, + { + "epoch": 7.0, + "grad_norm": 1.3276259899139404, + "learning_rate": 3.869373020780094e-05, + "loss": 3.2109, + "step": 4788 + }, + { + "epoch": 7.0, + "grad_norm": 1.9052053689956665, + "learning_rate": 3.865668162537861e-05, + "loss": 3.1952, + "step": 4789 + }, + { + "epoch": 7.0, + "grad_norm": 1.7343652248382568, + "learning_rate": 3.8619646538075184e-05, + "loss": 3.2074, + "step": 4790 + }, + { + "epoch": 7.01, + "grad_norm": 1.1672378778457642, + "learning_rate": 3.858262495403816e-05, + "loss": 3.1704, + "step": 4791 + }, + { + "epoch": 7.01, + "grad_norm": 1.6275311708450317, + "learning_rate": 3.854561688141205e-05, + "loss": 3.1723, + "step": 4792 + }, + { + "epoch": 7.01, + "grad_norm": 2.027491569519043, + "learning_rate": 3.8508622328338376e-05, + "loss": 3.2521, + "step": 4793 + }, + { + "epoch": 7.01, + "grad_norm": 1.3763444423675537, + "learning_rate": 3.8471641302955684e-05, + "loss": 3.1991, + "step": 4794 + }, + { + "epoch": 7.01, + "grad_norm": 1.7834948301315308, + "learning_rate": 3.843467381339957e-05, + "loss": 3.1685, + "step": 4795 + }, + { + "epoch": 7.01, + "grad_norm": 1.548448085784912, + "learning_rate": 3.839771986780263e-05, + "loss": 3.1715, + "step": 4796 + }, + { + "epoch": 7.02, + "grad_norm": 2.0860743522644043, + "learning_rate": 3.836077947429446e-05, + "loss": 3.2339, + "step": 4797 + }, + { + "epoch": 7.02, + "grad_norm": 2.655271053314209, + "learning_rate": 3.8323852641001733e-05, + "loss": 3.1881, + "step": 4798 + }, + { + "epoch": 7.02, + "grad_norm": 1.268930435180664, + "learning_rate": 3.828693937604809e-05, + "loss": 3.1635, + "step": 4799 + }, + { + "epoch": 7.02, + "grad_norm": 1.7928105592727661, + "learning_rate": 3.8250039687554195e-05, + "loss": 3.1764, + "step": 4800 + }, + { + "epoch": 7.02, + "grad_norm": 2.381535768508911, + "learning_rate": 3.821315358363775e-05, + "loss": 3.1892, + "step": 4801 + }, + { + "epoch": 7.02, + "grad_norm": 1.9187071323394775, + "learning_rate": 3.817628107241343e-05, + "loss": 3.1783, + "step": 4802 + }, + { + "epoch": 7.02, + "grad_norm": 0.999586284160614, + "learning_rate": 3.813942216199299e-05, + "loss": 3.1682, + "step": 4803 + }, + { + "epoch": 7.03, + "grad_norm": 1.993181824684143, + "learning_rate": 3.810257686048503e-05, + "loss": 3.2129, + "step": 4804 + }, + { + "epoch": 7.03, + "grad_norm": 1.1408189535140991, + "learning_rate": 3.806574517599537e-05, + "loss": 3.2541, + "step": 4805 + }, + { + "epoch": 7.03, + "grad_norm": 1.8819730281829834, + "learning_rate": 3.802892711662676e-05, + "loss": 3.1628, + "step": 4806 + }, + { + "epoch": 7.03, + "grad_norm": 2.9541916847229004, + "learning_rate": 3.799212269047878e-05, + "loss": 3.224, + "step": 4807 + }, + { + "epoch": 7.03, + "grad_norm": 1.8541984558105469, + "learning_rate": 3.795533190564831e-05, + "loss": 3.1792, + "step": 4808 + }, + { + "epoch": 7.03, + "grad_norm": 0.8892889022827148, + "learning_rate": 3.791855477022903e-05, + "loss": 3.2253, + "step": 4809 + }, + { + "epoch": 7.03, + "grad_norm": 1.7917122840881348, + "learning_rate": 3.788179129231159e-05, + "loss": 3.1897, + "step": 4810 + }, + { + "epoch": 7.04, + "grad_norm": 1.6101915836334229, + "learning_rate": 3.78450414799838e-05, + "loss": 3.1593, + "step": 4811 + }, + { + "epoch": 7.04, + "grad_norm": 0.8278429508209229, + "learning_rate": 3.780830534133037e-05, + "loss": 3.1712, + "step": 4812 + }, + { + "epoch": 7.04, + "grad_norm": 1.5922514200210571, + "learning_rate": 3.777158288443291e-05, + "loss": 3.1608, + "step": 4813 + }, + { + "epoch": 7.04, + "grad_norm": 0.9403831362724304, + "learning_rate": 3.7734874117370244e-05, + "loss": 3.2142, + "step": 4814 + }, + { + "epoch": 7.04, + "grad_norm": 1.2470545768737793, + "learning_rate": 3.769817904821796e-05, + "loss": 3.2069, + "step": 4815 + }, + { + "epoch": 7.04, + "grad_norm": 0.7299978733062744, + "learning_rate": 3.766149768504876e-05, + "loss": 3.1938, + "step": 4816 + }, + { + "epoch": 7.04, + "grad_norm": 1.2859150171279907, + "learning_rate": 3.762483003593229e-05, + "loss": 3.1506, + "step": 4817 + }, + { + "epoch": 7.05, + "grad_norm": 0.8429161906242371, + "learning_rate": 3.7588176108935194e-05, + "loss": 3.1859, + "step": 4818 + }, + { + "epoch": 7.05, + "grad_norm": 1.0546910762786865, + "learning_rate": 3.75515359121211e-05, + "loss": 3.1802, + "step": 4819 + }, + { + "epoch": 7.05, + "grad_norm": 1.477935791015625, + "learning_rate": 3.7514909453550575e-05, + "loss": 3.19, + "step": 4820 + }, + { + "epoch": 7.05, + "grad_norm": 0.5975956916809082, + "learning_rate": 3.7478296741281215e-05, + "loss": 3.192, + "step": 4821 + }, + { + "epoch": 7.05, + "grad_norm": 0.9366810321807861, + "learning_rate": 3.744169778336758e-05, + "loss": 3.1711, + "step": 4822 + }, + { + "epoch": 7.05, + "grad_norm": 0.5407786965370178, + "learning_rate": 3.740511258786118e-05, + "loss": 3.2555, + "step": 4823 + }, + { + "epoch": 7.06, + "grad_norm": 0.5257308483123779, + "learning_rate": 3.736854116281052e-05, + "loss": 3.1868, + "step": 4824 + }, + { + "epoch": 7.06, + "grad_norm": 0.5943123698234558, + "learning_rate": 3.7331983516261073e-05, + "loss": 3.1969, + "step": 4825 + }, + { + "epoch": 7.06, + "grad_norm": 0.31985798478126526, + "learning_rate": 3.729543965625526e-05, + "loss": 3.2147, + "step": 4826 + }, + { + "epoch": 7.06, + "grad_norm": 0.932587206363678, + "learning_rate": 3.7258909590832506e-05, + "loss": 3.1519, + "step": 4827 + }, + { + "epoch": 7.06, + "grad_norm": 1.0996321439743042, + "learning_rate": 3.722239332802916e-05, + "loss": 3.2083, + "step": 4828 + }, + { + "epoch": 7.06, + "grad_norm": 0.7307466864585876, + "learning_rate": 3.718589087587857e-05, + "loss": 3.1716, + "step": 4829 + }, + { + "epoch": 7.06, + "grad_norm": 0.39040249586105347, + "learning_rate": 3.7149402242411023e-05, + "loss": 3.1738, + "step": 4830 + }, + { + "epoch": 7.07, + "grad_norm": 0.8237361907958984, + "learning_rate": 3.711292743565378e-05, + "loss": 3.186, + "step": 4831 + }, + { + "epoch": 7.07, + "grad_norm": 0.6313247084617615, + "learning_rate": 3.707646646363104e-05, + "loss": 3.186, + "step": 4832 + }, + { + "epoch": 7.07, + "grad_norm": 1.1303566694259644, + "learning_rate": 3.7040019334363996e-05, + "loss": 3.2012, + "step": 4833 + }, + { + "epoch": 7.07, + "grad_norm": 1.1949049234390259, + "learning_rate": 3.7003586055870755e-05, + "loss": 3.2074, + "step": 4834 + }, + { + "epoch": 7.07, + "grad_norm": 0.9406141638755798, + "learning_rate": 3.696716663616641e-05, + "loss": 3.1975, + "step": 4835 + }, + { + "epoch": 7.07, + "grad_norm": 0.7907782793045044, + "learning_rate": 3.6930761083262955e-05, + "loss": 3.1521, + "step": 4836 + }, + { + "epoch": 7.07, + "grad_norm": 0.45947736501693726, + "learning_rate": 3.689436940516944e-05, + "loss": 3.1649, + "step": 4837 + }, + { + "epoch": 7.08, + "grad_norm": 1.1769413948059082, + "learning_rate": 3.6857991609891675e-05, + "loss": 3.1585, + "step": 4838 + }, + { + "epoch": 7.08, + "grad_norm": 0.3579048812389374, + "learning_rate": 3.6821627705432614e-05, + "loss": 3.2008, + "step": 4839 + }, + { + "epoch": 7.08, + "grad_norm": 1.2127845287322998, + "learning_rate": 3.678527769979211e-05, + "loss": 3.194, + "step": 4840 + }, + { + "epoch": 7.08, + "grad_norm": 0.7427309155464172, + "learning_rate": 3.6748941600966815e-05, + "loss": 3.171, + "step": 4841 + }, + { + "epoch": 7.08, + "grad_norm": 1.3657749891281128, + "learning_rate": 3.671261941695045e-05, + "loss": 3.2067, + "step": 4842 + }, + { + "epoch": 7.08, + "grad_norm": 0.7372667193412781, + "learning_rate": 3.6676311155733746e-05, + "loss": 3.1599, + "step": 4843 + }, + { + "epoch": 7.09, + "grad_norm": 1.225062370300293, + "learning_rate": 3.664001682530418e-05, + "loss": 3.1744, + "step": 4844 + }, + { + "epoch": 7.09, + "grad_norm": 1.3079476356506348, + "learning_rate": 3.6603736433646264e-05, + "loss": 3.149, + "step": 4845 + }, + { + "epoch": 7.09, + "grad_norm": 0.3697514533996582, + "learning_rate": 3.656746998874153e-05, + "loss": 3.1942, + "step": 4846 + }, + { + "epoch": 7.09, + "grad_norm": 2.1522505283355713, + "learning_rate": 3.653121749856826e-05, + "loss": 3.1853, + "step": 4847 + }, + { + "epoch": 7.09, + "grad_norm": 0.901470959186554, + "learning_rate": 3.649497897110179e-05, + "loss": 3.1828, + "step": 4848 + }, + { + "epoch": 7.09, + "grad_norm": 0.8554129004478455, + "learning_rate": 3.645875441431435e-05, + "loss": 3.224, + "step": 4849 + }, + { + "epoch": 7.09, + "grad_norm": 1.8987047672271729, + "learning_rate": 3.642254383617511e-05, + "loss": 3.1925, + "step": 4850 + }, + { + "epoch": 7.1, + "grad_norm": 0.45335274934768677, + "learning_rate": 3.6386347244650155e-05, + "loss": 3.1787, + "step": 4851 + }, + { + "epoch": 7.1, + "grad_norm": 0.8638046979904175, + "learning_rate": 3.635016464770249e-05, + "loss": 3.1878, + "step": 4852 + }, + { + "epoch": 7.1, + "grad_norm": 0.6001507639884949, + "learning_rate": 3.6313996053292034e-05, + "loss": 3.1502, + "step": 4853 + }, + { + "epoch": 7.1, + "grad_norm": 0.9555931687355042, + "learning_rate": 3.627784146937563e-05, + "loss": 3.2091, + "step": 4854 + }, + { + "epoch": 7.1, + "grad_norm": 0.5545978546142578, + "learning_rate": 3.624170090390706e-05, + "loss": 3.2264, + "step": 4855 + }, + { + "epoch": 7.1, + "grad_norm": 0.2554129362106323, + "learning_rate": 3.620557436483701e-05, + "loss": 3.1756, + "step": 4856 + }, + { + "epoch": 7.1, + "grad_norm": 0.648738443851471, + "learning_rate": 3.6169461860113074e-05, + "loss": 3.1885, + "step": 4857 + }, + { + "epoch": 7.11, + "grad_norm": 0.39746615290641785, + "learning_rate": 3.613336339767974e-05, + "loss": 3.1645, + "step": 4858 + }, + { + "epoch": 7.11, + "grad_norm": 0.6752750277519226, + "learning_rate": 3.6097278985478454e-05, + "loss": 3.1669, + "step": 4859 + }, + { + "epoch": 7.11, + "grad_norm": 0.4994068443775177, + "learning_rate": 3.606120863144753e-05, + "loss": 3.2155, + "step": 4860 + }, + { + "epoch": 7.11, + "grad_norm": 0.9360056519508362, + "learning_rate": 3.6025152343522205e-05, + "loss": 3.198, + "step": 4861 + }, + { + "epoch": 7.11, + "grad_norm": 0.3693661093711853, + "learning_rate": 3.598911012963463e-05, + "loss": 3.1723, + "step": 4862 + }, + { + "epoch": 7.11, + "grad_norm": 0.7912516593933105, + "learning_rate": 3.5953081997713876e-05, + "loss": 3.1946, + "step": 4863 + }, + { + "epoch": 7.11, + "grad_norm": 0.5820804834365845, + "learning_rate": 3.591706795568579e-05, + "loss": 3.1551, + "step": 4864 + }, + { + "epoch": 7.12, + "grad_norm": 0.41594260931015015, + "learning_rate": 3.588106801147333e-05, + "loss": 3.2372, + "step": 4865 + }, + { + "epoch": 7.12, + "grad_norm": 1.0947422981262207, + "learning_rate": 3.5845082172996224e-05, + "loss": 3.1926, + "step": 4866 + }, + { + "epoch": 7.12, + "grad_norm": 0.4444226324558258, + "learning_rate": 3.580911044817104e-05, + "loss": 3.2154, + "step": 4867 + }, + { + "epoch": 7.12, + "grad_norm": 0.8094576001167297, + "learning_rate": 3.577315284491138e-05, + "loss": 3.1802, + "step": 4868 + }, + { + "epoch": 7.12, + "grad_norm": 0.6356644034385681, + "learning_rate": 3.573720937112772e-05, + "loss": 3.1484, + "step": 4869 + }, + { + "epoch": 7.12, + "grad_norm": 0.9300285577774048, + "learning_rate": 3.570128003472725e-05, + "loss": 3.1834, + "step": 4870 + }, + { + "epoch": 7.13, + "grad_norm": 0.419990211725235, + "learning_rate": 3.5665364843614336e-05, + "loss": 3.1857, + "step": 4871 + }, + { + "epoch": 7.13, + "grad_norm": 1.047782063484192, + "learning_rate": 3.562946380568997e-05, + "loss": 3.2032, + "step": 4872 + }, + { + "epoch": 7.13, + "grad_norm": 0.564288318157196, + "learning_rate": 3.5593576928852134e-05, + "loss": 3.1501, + "step": 4873 + }, + { + "epoch": 7.13, + "grad_norm": 0.6687262654304504, + "learning_rate": 3.555770422099581e-05, + "loss": 3.1657, + "step": 4874 + }, + { + "epoch": 7.13, + "grad_norm": 0.4760758876800537, + "learning_rate": 3.5521845690012655e-05, + "loss": 3.1708, + "step": 4875 + }, + { + "epoch": 7.13, + "grad_norm": 0.6651167273521423, + "learning_rate": 3.5486001343791295e-05, + "loss": 3.1546, + "step": 4876 + }, + { + "epoch": 7.13, + "grad_norm": 0.554828941822052, + "learning_rate": 3.5450171190217364e-05, + "loss": 3.1451, + "step": 4877 + }, + { + "epoch": 7.14, + "grad_norm": 1.176149606704712, + "learning_rate": 3.5414355237173136e-05, + "loss": 3.2025, + "step": 4878 + }, + { + "epoch": 7.14, + "grad_norm": 0.4831768572330475, + "learning_rate": 3.537855349253788e-05, + "loss": 3.2084, + "step": 4879 + }, + { + "epoch": 7.14, + "grad_norm": 0.946924090385437, + "learning_rate": 3.5342765964187864e-05, + "loss": 3.2116, + "step": 4880 + }, + { + "epoch": 7.14, + "grad_norm": 0.4359959661960602, + "learning_rate": 3.530699265999599e-05, + "loss": 3.1784, + "step": 4881 + }, + { + "epoch": 7.14, + "grad_norm": 0.9432086944580078, + "learning_rate": 3.5271233587832164e-05, + "loss": 3.2123, + "step": 4882 + }, + { + "epoch": 7.14, + "grad_norm": 0.4890124797821045, + "learning_rate": 3.523548875556316e-05, + "loss": 3.2041, + "step": 4883 + }, + { + "epoch": 7.14, + "grad_norm": 0.6941928267478943, + "learning_rate": 3.51997581710526e-05, + "loss": 3.2032, + "step": 4884 + }, + { + "epoch": 7.15, + "grad_norm": 0.5665786862373352, + "learning_rate": 3.5164041842160966e-05, + "loss": 3.193, + "step": 4885 + }, + { + "epoch": 7.15, + "grad_norm": 0.7066811919212341, + "learning_rate": 3.5128339776745625e-05, + "loss": 3.1766, + "step": 4886 + }, + { + "epoch": 7.15, + "grad_norm": 0.6680645942687988, + "learning_rate": 3.5092651982660774e-05, + "loss": 3.2185, + "step": 4887 + }, + { + "epoch": 7.15, + "grad_norm": 0.7763155698776245, + "learning_rate": 3.5056978467757506e-05, + "loss": 3.18, + "step": 4888 + }, + { + "epoch": 7.15, + "grad_norm": 0.8374306559562683, + "learning_rate": 3.502131923988373e-05, + "loss": 3.2049, + "step": 4889 + }, + { + "epoch": 7.15, + "grad_norm": 1.1140564680099487, + "learning_rate": 3.4985674306884274e-05, + "loss": 3.195, + "step": 4890 + }, + { + "epoch": 7.15, + "grad_norm": 0.9668914675712585, + "learning_rate": 3.495004367660075e-05, + "loss": 3.1662, + "step": 4891 + }, + { + "epoch": 7.16, + "grad_norm": 0.8925195336341858, + "learning_rate": 3.491442735687167e-05, + "loss": 3.198, + "step": 4892 + }, + { + "epoch": 7.16, + "grad_norm": 1.2694224119186401, + "learning_rate": 3.4878825355532394e-05, + "loss": 3.1972, + "step": 4893 + }, + { + "epoch": 7.16, + "grad_norm": 0.7892391085624695, + "learning_rate": 3.4843237680415156e-05, + "loss": 3.2424, + "step": 4894 + }, + { + "epoch": 7.16, + "grad_norm": 0.8622798323631287, + "learning_rate": 3.480766433934889e-05, + "loss": 3.2112, + "step": 4895 + }, + { + "epoch": 7.16, + "grad_norm": 0.8695278167724609, + "learning_rate": 3.477210534015961e-05, + "loss": 3.2158, + "step": 4896 + }, + { + "epoch": 7.16, + "grad_norm": 0.8544620871543884, + "learning_rate": 3.473656069067005e-05, + "loss": 3.2501, + "step": 4897 + }, + { + "epoch": 7.17, + "grad_norm": 0.6005237102508545, + "learning_rate": 3.470103039869968e-05, + "loss": 3.205, + "step": 4898 + }, + { + "epoch": 7.17, + "grad_norm": 0.796326220035553, + "learning_rate": 3.466551447206506e-05, + "loss": 3.1567, + "step": 4899 + }, + { + "epoch": 7.17, + "grad_norm": 0.40793824195861816, + "learning_rate": 3.463001291857943e-05, + "loss": 3.1772, + "step": 4900 + }, + { + "epoch": 7.17, + "grad_norm": 0.804324209690094, + "learning_rate": 3.459452574605281e-05, + "loss": 3.181, + "step": 4901 + }, + { + "epoch": 7.17, + "eval_loss": 3.0279037952423096, + "eval_runtime": 1163.8402, + "eval_samples_per_second": 14.729, + "eval_steps_per_second": 0.921, + "step": 4901 + }, + { + "epoch": 7.17, + "grad_norm": 0.47186848521232605, + "learning_rate": 3.455905296229222e-05, + "loss": 3.1301, + "step": 4902 + }, + { + "epoch": 7.17, + "grad_norm": 0.7227790951728821, + "learning_rate": 3.452359457510145e-05, + "loss": 3.1657, + "step": 4903 + }, + { + "epoch": 7.17, + "grad_norm": 0.9041692614555359, + "learning_rate": 3.4488150592281e-05, + "loss": 3.2102, + "step": 4904 + }, + { + "epoch": 7.18, + "grad_norm": 1.273749589920044, + "learning_rate": 3.4452721021628445e-05, + "loss": 3.2073, + "step": 4905 + }, + { + "epoch": 7.18, + "grad_norm": 1.3366678953170776, + "learning_rate": 3.441730587093795e-05, + "loss": 3.1913, + "step": 4906 + }, + { + "epoch": 7.18, + "grad_norm": 0.45730334520339966, + "learning_rate": 3.438190514800063e-05, + "loss": 3.2091, + "step": 4907 + }, + { + "epoch": 7.18, + "grad_norm": 1.0588980913162231, + "learning_rate": 3.4346518860604485e-05, + "loss": 3.2037, + "step": 4908 + }, + { + "epoch": 7.18, + "grad_norm": 0.6257373094558716, + "learning_rate": 3.4311147016534164e-05, + "loss": 3.1525, + "step": 4909 + }, + { + "epoch": 7.18, + "grad_norm": 0.7281100749969482, + "learning_rate": 3.4275789623571234e-05, + "loss": 3.2051, + "step": 4910 + }, + { + "epoch": 7.18, + "grad_norm": 0.8004705905914307, + "learning_rate": 3.42404466894942e-05, + "loss": 3.201, + "step": 4911 + }, + { + "epoch": 7.19, + "grad_norm": 0.34791865944862366, + "learning_rate": 3.4205118222078145e-05, + "loss": 3.1746, + "step": 4912 + }, + { + "epoch": 7.19, + "grad_norm": 1.0412063598632812, + "learning_rate": 3.416980422909516e-05, + "loss": 3.2195, + "step": 4913 + }, + { + "epoch": 7.19, + "grad_norm": 0.489105224609375, + "learning_rate": 3.413450471831407e-05, + "loss": 3.1446, + "step": 4914 + }, + { + "epoch": 7.19, + "grad_norm": 0.49872738122940063, + "learning_rate": 3.4099219697500514e-05, + "loss": 3.128, + "step": 4915 + }, + { + "epoch": 7.19, + "grad_norm": 0.7999401092529297, + "learning_rate": 3.4063949174416984e-05, + "loss": 3.1953, + "step": 4916 + }, + { + "epoch": 7.19, + "grad_norm": 0.4927428364753723, + "learning_rate": 3.402869315682275e-05, + "loss": 3.1533, + "step": 4917 + }, + { + "epoch": 7.19, + "grad_norm": 0.29562923312187195, + "learning_rate": 3.3993451652473916e-05, + "loss": 3.1752, + "step": 4918 + }, + { + "epoch": 7.2, + "grad_norm": 0.9475629329681396, + "learning_rate": 3.395822466912335e-05, + "loss": 3.1526, + "step": 4919 + }, + { + "epoch": 7.2, + "grad_norm": 0.9323421120643616, + "learning_rate": 3.3923012214520763e-05, + "loss": 3.1457, + "step": 4920 + }, + { + "epoch": 7.2, + "grad_norm": 0.30518239736557007, + "learning_rate": 3.388781429641267e-05, + "loss": 3.1892, + "step": 4921 + }, + { + "epoch": 7.2, + "grad_norm": 0.4029679000377655, + "learning_rate": 3.385263092254236e-05, + "loss": 3.1918, + "step": 4922 + }, + { + "epoch": 7.2, + "grad_norm": 0.7984765768051147, + "learning_rate": 3.381746210064995e-05, + "loss": 3.2006, + "step": 4923 + }, + { + "epoch": 7.2, + "grad_norm": 0.9310452342033386, + "learning_rate": 3.378230783847234e-05, + "loss": 3.1526, + "step": 4924 + }, + { + "epoch": 7.21, + "grad_norm": 0.5400900840759277, + "learning_rate": 3.374716814374323e-05, + "loss": 3.2063, + "step": 4925 + }, + { + "epoch": 7.21, + "grad_norm": 0.7315930128097534, + "learning_rate": 3.371204302419313e-05, + "loss": 3.1709, + "step": 4926 + }, + { + "epoch": 7.21, + "grad_norm": 0.8185734152793884, + "learning_rate": 3.367693248754933e-05, + "loss": 3.1659, + "step": 4927 + }, + { + "epoch": 7.21, + "grad_norm": 0.4014480710029602, + "learning_rate": 3.364183654153592e-05, + "loss": 3.1291, + "step": 4928 + }, + { + "epoch": 7.21, + "grad_norm": 0.45573729276657104, + "learning_rate": 3.360675519387371e-05, + "loss": 3.1648, + "step": 4929 + }, + { + "epoch": 7.21, + "grad_norm": 0.49166086316108704, + "learning_rate": 3.357168845228043e-05, + "loss": 3.168, + "step": 4930 + }, + { + "epoch": 7.21, + "grad_norm": 0.37833172082901, + "learning_rate": 3.353663632447054e-05, + "loss": 3.1675, + "step": 4931 + }, + { + "epoch": 7.22, + "grad_norm": 0.9355205297470093, + "learning_rate": 3.350159881815518e-05, + "loss": 3.1798, + "step": 4932 + }, + { + "epoch": 7.22, + "grad_norm": 0.888450562953949, + "learning_rate": 3.346657594104246e-05, + "loss": 3.1609, + "step": 4933 + }, + { + "epoch": 7.22, + "grad_norm": 0.39355748891830444, + "learning_rate": 3.343156770083717e-05, + "loss": 3.1912, + "step": 4934 + }, + { + "epoch": 7.22, + "grad_norm": 0.4100874960422516, + "learning_rate": 3.339657410524081e-05, + "loss": 3.1741, + "step": 4935 + }, + { + "epoch": 7.22, + "grad_norm": 0.6284832954406738, + "learning_rate": 3.336159516195182e-05, + "loss": 3.1456, + "step": 4936 + }, + { + "epoch": 7.22, + "grad_norm": 9.06452465057373, + "learning_rate": 3.3326630878665334e-05, + "loss": 3.1432, + "step": 4937 + }, + { + "epoch": 7.22, + "grad_norm": 1.4369940757751465, + "learning_rate": 3.329168126307319e-05, + "loss": 3.1878, + "step": 4938 + }, + { + "epoch": 7.23, + "grad_norm": 0.6629584431648254, + "learning_rate": 3.3256746322864127e-05, + "loss": 3.1782, + "step": 4939 + }, + { + "epoch": 7.23, + "grad_norm": 1.5383516550064087, + "learning_rate": 3.322182606572357e-05, + "loss": 3.1567, + "step": 4940 + }, + { + "epoch": 7.23, + "grad_norm": 1.498374581336975, + "learning_rate": 3.318692049933375e-05, + "loss": 3.2173, + "step": 4941 + }, + { + "epoch": 7.23, + "grad_norm": 0.5786330699920654, + "learning_rate": 3.315202963137366e-05, + "loss": 3.1327, + "step": 4942 + }, + { + "epoch": 7.23, + "grad_norm": 1.2023847103118896, + "learning_rate": 3.311715346951906e-05, + "loss": 3.2942, + "step": 4943 + }, + { + "epoch": 7.23, + "grad_norm": 0.6449021697044373, + "learning_rate": 3.3082292021442454e-05, + "loss": 3.1564, + "step": 4944 + }, + { + "epoch": 7.24, + "grad_norm": 0.36229845881462097, + "learning_rate": 3.304744529481315e-05, + "loss": 3.1787, + "step": 4945 + }, + { + "epoch": 7.24, + "grad_norm": 0.9449754953384399, + "learning_rate": 3.301261329729719e-05, + "loss": 3.1928, + "step": 4946 + }, + { + "epoch": 7.24, + "grad_norm": 0.9095726609230042, + "learning_rate": 3.297779603655736e-05, + "loss": 3.2106, + "step": 4947 + }, + { + "epoch": 7.24, + "grad_norm": 0.346010684967041, + "learning_rate": 3.2942993520253254e-05, + "loss": 3.1908, + "step": 4948 + }, + { + "epoch": 7.24, + "grad_norm": 0.7208975553512573, + "learning_rate": 3.2908205756041176e-05, + "loss": 3.1851, + "step": 4949 + }, + { + "epoch": 7.24, + "grad_norm": 0.5966201424598694, + "learning_rate": 3.2873432751574194e-05, + "loss": 3.2213, + "step": 4950 + }, + { + "epoch": 7.24, + "grad_norm": 0.3596455156803131, + "learning_rate": 3.2838674514502154e-05, + "loss": 3.2082, + "step": 4951 + }, + { + "epoch": 7.25, + "grad_norm": 0.8812998533248901, + "learning_rate": 3.2803931052471615e-05, + "loss": 3.1496, + "step": 4952 + }, + { + "epoch": 7.25, + "grad_norm": 0.7055163979530334, + "learning_rate": 3.276920237312593e-05, + "loss": 3.1859, + "step": 4953 + }, + { + "epoch": 7.25, + "grad_norm": 0.7300843596458435, + "learning_rate": 3.273448848410515e-05, + "loss": 3.1804, + "step": 4954 + }, + { + "epoch": 7.25, + "grad_norm": 1.0114784240722656, + "learning_rate": 3.269978939304612e-05, + "loss": 3.1612, + "step": 4955 + }, + { + "epoch": 7.25, + "grad_norm": 0.46303462982177734, + "learning_rate": 3.266510510758238e-05, + "loss": 3.1675, + "step": 4956 + }, + { + "epoch": 7.25, + "grad_norm": 0.722417950630188, + "learning_rate": 3.263043563534428e-05, + "loss": 3.1732, + "step": 4957 + }, + { + "epoch": 7.25, + "grad_norm": 0.6245961785316467, + "learning_rate": 3.2595780983958834e-05, + "loss": 3.1481, + "step": 4958 + }, + { + "epoch": 7.26, + "grad_norm": 0.31019535660743713, + "learning_rate": 3.256114116104986e-05, + "loss": 3.1808, + "step": 4959 + }, + { + "epoch": 7.26, + "grad_norm": 1.2723808288574219, + "learning_rate": 3.25265161742379e-05, + "loss": 3.1502, + "step": 4960 + }, + { + "epoch": 7.26, + "grad_norm": 0.6151775121688843, + "learning_rate": 3.249190603114013e-05, + "loss": 3.1393, + "step": 4961 + }, + { + "epoch": 7.26, + "grad_norm": 1.0856354236602783, + "learning_rate": 3.245731073937068e-05, + "loss": 3.1964, + "step": 4962 + }, + { + "epoch": 7.26, + "grad_norm": 0.665896475315094, + "learning_rate": 3.2422730306540186e-05, + "loss": 3.1715, + "step": 4963 + }, + { + "epoch": 7.26, + "grad_norm": 1.0534874200820923, + "learning_rate": 3.23881647402561e-05, + "loss": 3.1952, + "step": 4964 + }, + { + "epoch": 7.26, + "grad_norm": 1.2023156881332397, + "learning_rate": 3.235361404812274e-05, + "loss": 3.1622, + "step": 4965 + }, + { + "epoch": 7.27, + "grad_norm": 0.9104660749435425, + "learning_rate": 3.23190782377409e-05, + "loss": 3.1731, + "step": 4966 + }, + { + "epoch": 7.27, + "grad_norm": 1.6818856000900269, + "learning_rate": 3.228455731670823e-05, + "loss": 3.2543, + "step": 4967 + }, + { + "epoch": 7.27, + "grad_norm": 0.40944576263427734, + "learning_rate": 3.225005129261922e-05, + "loss": 3.1609, + "step": 4968 + }, + { + "epoch": 7.27, + "grad_norm": 2.257155656814575, + "learning_rate": 3.221556017306484e-05, + "loss": 3.2051, + "step": 4969 + }, + { + "epoch": 7.27, + "grad_norm": 1.4875580072402954, + "learning_rate": 3.218108396563293e-05, + "loss": 3.1654, + "step": 4970 + }, + { + "epoch": 7.27, + "grad_norm": 0.7136067152023315, + "learning_rate": 3.21466226779081e-05, + "loss": 3.1851, + "step": 4971 + }, + { + "epoch": 7.28, + "grad_norm": 1.1615363359451294, + "learning_rate": 3.2112176317471535e-05, + "loss": 3.1378, + "step": 4972 + }, + { + "epoch": 7.28, + "grad_norm": 1.23421311378479, + "learning_rate": 3.20777448919012e-05, + "loss": 3.1976, + "step": 4973 + }, + { + "epoch": 7.28, + "grad_norm": 0.3943505883216858, + "learning_rate": 3.204332840877179e-05, + "loss": 3.1514, + "step": 4974 + }, + { + "epoch": 7.28, + "grad_norm": 1.666935682296753, + "learning_rate": 3.200892687565471e-05, + "loss": 3.1276, + "step": 4975 + }, + { + "epoch": 7.28, + "grad_norm": 0.877071738243103, + "learning_rate": 3.1974540300118064e-05, + "loss": 3.1572, + "step": 4976 + }, + { + "epoch": 7.28, + "grad_norm": 1.395138144493103, + "learning_rate": 3.1940168689726666e-05, + "loss": 3.1789, + "step": 4977 + }, + { + "epoch": 7.28, + "grad_norm": 1.2696112394332886, + "learning_rate": 3.190581205204204e-05, + "loss": 3.1909, + "step": 4978 + }, + { + "epoch": 7.29, + "grad_norm": 0.9251769185066223, + "learning_rate": 3.1871470394622404e-05, + "loss": 3.1486, + "step": 4979 + }, + { + "epoch": 7.29, + "grad_norm": 0.9306465983390808, + "learning_rate": 3.183714372502272e-05, + "loss": 3.1897, + "step": 4980 + }, + { + "epoch": 7.29, + "grad_norm": 0.6166656613349915, + "learning_rate": 3.1802832050794595e-05, + "loss": 3.1902, + "step": 4981 + }, + { + "epoch": 7.29, + "grad_norm": 1.0859153270721436, + "learning_rate": 3.1768535379486395e-05, + "loss": 3.2179, + "step": 4982 + }, + { + "epoch": 7.29, + "grad_norm": 0.5373363494873047, + "learning_rate": 3.173425371864314e-05, + "loss": 3.1732, + "step": 4983 + }, + { + "epoch": 7.29, + "grad_norm": 0.6277896761894226, + "learning_rate": 3.169998707580657e-05, + "loss": 3.1735, + "step": 4984 + }, + { + "epoch": 7.29, + "grad_norm": 0.8959996104240417, + "learning_rate": 3.166573545851514e-05, + "loss": 3.1425, + "step": 4985 + }, + { + "epoch": 7.3, + "grad_norm": 0.5353897213935852, + "learning_rate": 3.16314988743039e-05, + "loss": 3.1447, + "step": 4986 + }, + { + "epoch": 7.3, + "grad_norm": 0.8923834562301636, + "learning_rate": 3.1597277330704756e-05, + "loss": 3.1684, + "step": 4987 + }, + { + "epoch": 7.3, + "grad_norm": 0.456402063369751, + "learning_rate": 3.1563070835246213e-05, + "loss": 3.1917, + "step": 4988 + }, + { + "epoch": 7.3, + "grad_norm": 0.744957685470581, + "learning_rate": 3.152887939545337e-05, + "loss": 3.205, + "step": 4989 + }, + { + "epoch": 7.3, + "grad_norm": 0.7469781041145325, + "learning_rate": 3.149470301884823e-05, + "loss": 3.1544, + "step": 4990 + }, + { + "epoch": 7.3, + "grad_norm": 0.48788437247276306, + "learning_rate": 3.146054171294935e-05, + "loss": 3.1809, + "step": 4991 + }, + { + "epoch": 7.3, + "grad_norm": 1.1828556060791016, + "learning_rate": 3.14263954852719e-05, + "loss": 3.1552, + "step": 4992 + }, + { + "epoch": 7.31, + "grad_norm": 0.35930806398391724, + "learning_rate": 3.1392264343327914e-05, + "loss": 3.1549, + "step": 4993 + }, + { + "epoch": 7.31, + "grad_norm": 1.010923981666565, + "learning_rate": 3.135814829462602e-05, + "loss": 3.181, + "step": 4994 + }, + { + "epoch": 7.31, + "grad_norm": 1.8422837257385254, + "learning_rate": 3.132404734667142e-05, + "loss": 3.1774, + "step": 4995 + }, + { + "epoch": 7.31, + "grad_norm": 0.7072595357894897, + "learning_rate": 3.1289961506966214e-05, + "loss": 3.1996, + "step": 4996 + }, + { + "epoch": 7.31, + "grad_norm": 0.8350168466567993, + "learning_rate": 3.125589078300898e-05, + "loss": 3.2288, + "step": 4997 + }, + { + "epoch": 7.31, + "grad_norm": 0.7736479043960571, + "learning_rate": 3.122183518229503e-05, + "loss": 3.1702, + "step": 4998 + }, + { + "epoch": 7.32, + "grad_norm": 0.6932786107063293, + "learning_rate": 3.1187794712316474e-05, + "loss": 3.1662, + "step": 4999 + }, + { + "epoch": 7.32, + "grad_norm": 0.5773895382881165, + "learning_rate": 3.11537693805619e-05, + "loss": 3.1859, + "step": 5000 + }, + { + "epoch": 7.32, + "grad_norm": 0.8364612460136414, + "learning_rate": 3.111975919451664e-05, + "loss": 3.1769, + "step": 5001 + }, + { + "epoch": 7.32, + "grad_norm": 0.3345867097377777, + "learning_rate": 3.10857641616628e-05, + "loss": 3.1939, + "step": 5002 + }, + { + "epoch": 7.32, + "grad_norm": 0.6873171925544739, + "learning_rate": 3.105178428947898e-05, + "loss": 3.2176, + "step": 5003 + }, + { + "epoch": 7.32, + "grad_norm": 0.5719138383865356, + "learning_rate": 3.10178195854405e-05, + "loss": 3.174, + "step": 5004 + }, + { + "epoch": 7.32, + "grad_norm": 0.9275087714195251, + "learning_rate": 3.0983870057019484e-05, + "loss": 3.2174, + "step": 5005 + }, + { + "epoch": 7.33, + "grad_norm": 1.1784770488739014, + "learning_rate": 3.09499357116845e-05, + "loss": 3.1778, + "step": 5006 + }, + { + "epoch": 7.33, + "grad_norm": 0.7836307287216187, + "learning_rate": 3.0916016556900904e-05, + "loss": 3.1627, + "step": 5007 + }, + { + "epoch": 7.33, + "grad_norm": 0.482113242149353, + "learning_rate": 3.088211260013068e-05, + "loss": 3.1841, + "step": 5008 + }, + { + "epoch": 7.33, + "grad_norm": 0.48355069756507874, + "learning_rate": 3.084822384883249e-05, + "loss": 3.1691, + "step": 5009 + }, + { + "epoch": 7.33, + "grad_norm": 0.4077546000480652, + "learning_rate": 3.0814350310461615e-05, + "loss": 3.1581, + "step": 5010 + }, + { + "epoch": 7.33, + "grad_norm": 0.7942917943000793, + "learning_rate": 3.078049199247e-05, + "loss": 3.1964, + "step": 5011 + }, + { + "epoch": 7.33, + "grad_norm": 0.38686656951904297, + "learning_rate": 3.0746648902306264e-05, + "loss": 3.1606, + "step": 5012 + }, + { + "epoch": 7.34, + "grad_norm": 1.3776711225509644, + "learning_rate": 3.071282104741565e-05, + "loss": 3.2248, + "step": 5013 + }, + { + "epoch": 7.34, + "grad_norm": 0.2463265061378479, + "learning_rate": 3.067900843524008e-05, + "loss": 3.1848, + "step": 5014 + }, + { + "epoch": 7.34, + "grad_norm": 0.5669382214546204, + "learning_rate": 3.064521107321806e-05, + "loss": 3.1501, + "step": 5015 + }, + { + "epoch": 7.34, + "grad_norm": 0.40979495644569397, + "learning_rate": 3.061142896878486e-05, + "loss": 3.1762, + "step": 5016 + }, + { + "epoch": 7.34, + "grad_norm": 0.5121455192565918, + "learning_rate": 3.05776621293722e-05, + "loss": 3.1893, + "step": 5017 + }, + { + "epoch": 7.34, + "grad_norm": 0.3571344017982483, + "learning_rate": 3.054391056240865e-05, + "loss": 3.1726, + "step": 5018 + }, + { + "epoch": 7.34, + "grad_norm": 0.3279051184654236, + "learning_rate": 3.0510174275319325e-05, + "loss": 3.1238, + "step": 5019 + }, + { + "epoch": 7.35, + "grad_norm": 0.5023763179779053, + "learning_rate": 3.0476453275525906e-05, + "loss": 3.1893, + "step": 5020 + }, + { + "epoch": 7.35, + "grad_norm": 0.45298734307289124, + "learning_rate": 3.0442747570446872e-05, + "loss": 3.192, + "step": 5021 + }, + { + "epoch": 7.35, + "grad_norm": 0.580197274684906, + "learning_rate": 3.0409057167497258e-05, + "loss": 3.1934, + "step": 5022 + }, + { + "epoch": 7.35, + "grad_norm": 0.4648624360561371, + "learning_rate": 3.037538207408862e-05, + "loss": 3.1733, + "step": 5023 + }, + { + "epoch": 7.35, + "grad_norm": 0.8324884176254272, + "learning_rate": 3.034172229762935e-05, + "loss": 3.1584, + "step": 5024 + }, + { + "epoch": 7.35, + "grad_norm": 0.6418886184692383, + "learning_rate": 3.0308077845524384e-05, + "loss": 3.1809, + "step": 5025 + }, + { + "epoch": 7.36, + "grad_norm": 0.7795079946517944, + "learning_rate": 3.0274448725175176e-05, + "loss": 3.1515, + "step": 5026 + }, + { + "epoch": 7.36, + "grad_norm": 0.5563576221466064, + "learning_rate": 3.024083494397999e-05, + "loss": 3.1772, + "step": 5027 + }, + { + "epoch": 7.36, + "grad_norm": 0.42387935519218445, + "learning_rate": 3.0207236509333647e-05, + "loss": 3.1866, + "step": 5028 + }, + { + "epoch": 7.36, + "grad_norm": 0.5205991864204407, + "learning_rate": 3.017365342862748e-05, + "loss": 3.1517, + "step": 5029 + }, + { + "epoch": 7.36, + "grad_norm": 0.5210472941398621, + "learning_rate": 3.0140085709249667e-05, + "loss": 3.172, + "step": 5030 + }, + { + "epoch": 7.36, + "grad_norm": 0.5895211100578308, + "learning_rate": 3.010653335858479e-05, + "loss": 3.239, + "step": 5031 + }, + { + "epoch": 7.36, + "grad_norm": 0.38003700971603394, + "learning_rate": 3.007299638401416e-05, + "loss": 3.1581, + "step": 5032 + }, + { + "epoch": 7.37, + "grad_norm": 0.3639942705631256, + "learning_rate": 3.0039474792915702e-05, + "loss": 3.2157, + "step": 5033 + }, + { + "epoch": 7.37, + "grad_norm": 1.5683830976486206, + "learning_rate": 3.0005968592663925e-05, + "loss": 3.1839, + "step": 5034 + }, + { + "epoch": 7.37, + "grad_norm": 0.5200375914573669, + "learning_rate": 2.9972477790629994e-05, + "loss": 3.1836, + "step": 5035 + }, + { + "epoch": 7.37, + "grad_norm": 0.7715541124343872, + "learning_rate": 2.9939002394181637e-05, + "loss": 3.1778, + "step": 5036 + }, + { + "epoch": 7.37, + "grad_norm": 1.0058480501174927, + "learning_rate": 2.9905542410683218e-05, + "loss": 3.1782, + "step": 5037 + }, + { + "epoch": 7.37, + "grad_norm": 0.6472846269607544, + "learning_rate": 2.987209784749573e-05, + "loss": 3.1847, + "step": 5038 + }, + { + "epoch": 7.37, + "grad_norm": 1.0543522834777832, + "learning_rate": 2.9838668711976724e-05, + "loss": 3.1228, + "step": 5039 + }, + { + "epoch": 7.38, + "grad_norm": 0.9698706865310669, + "learning_rate": 2.9805255011480415e-05, + "loss": 3.163, + "step": 5040 + }, + { + "epoch": 7.38, + "grad_norm": 0.7419912219047546, + "learning_rate": 2.9771856753357573e-05, + "loss": 3.1658, + "step": 5041 + }, + { + "epoch": 7.38, + "grad_norm": 0.8493951559066772, + "learning_rate": 2.9738473944955603e-05, + "loss": 3.1264, + "step": 5042 + }, + { + "epoch": 7.38, + "grad_norm": 0.9127371311187744, + "learning_rate": 2.970510659361848e-05, + "loss": 3.2144, + "step": 5043 + }, + { + "epoch": 7.38, + "grad_norm": 0.5002891421318054, + "learning_rate": 2.967175470668683e-05, + "loss": 3.1733, + "step": 5044 + }, + { + "epoch": 7.38, + "grad_norm": 0.7260929346084595, + "learning_rate": 2.9638418291497817e-05, + "loss": 3.1292, + "step": 5045 + }, + { + "epoch": 7.39, + "grad_norm": 0.8466237783432007, + "learning_rate": 2.9605097355385227e-05, + "loss": 3.194, + "step": 5046 + }, + { + "epoch": 7.39, + "grad_norm": 0.29703596234321594, + "learning_rate": 2.9571791905679468e-05, + "loss": 3.1485, + "step": 5047 + }, + { + "epoch": 7.39, + "grad_norm": 0.8655198216438293, + "learning_rate": 2.9538501949707487e-05, + "loss": 3.1638, + "step": 5048 + }, + { + "epoch": 7.39, + "grad_norm": 0.5336217880249023, + "learning_rate": 2.9505227494792874e-05, + "loss": 3.1671, + "step": 5049 + }, + { + "epoch": 7.39, + "grad_norm": 0.2816140949726105, + "learning_rate": 2.947196854825581e-05, + "loss": 3.1982, + "step": 5050 + }, + { + "epoch": 7.39, + "grad_norm": 0.6799694299697876, + "learning_rate": 2.9438725117412947e-05, + "loss": 3.1824, + "step": 5051 + }, + { + "epoch": 7.39, + "grad_norm": 0.7424922585487366, + "learning_rate": 2.9405497209577703e-05, + "loss": 3.1787, + "step": 5052 + }, + { + "epoch": 7.4, + "grad_norm": 0.5237893462181091, + "learning_rate": 2.9372284832059993e-05, + "loss": 3.1705, + "step": 5053 + }, + { + "epoch": 7.4, + "grad_norm": 0.7378460764884949, + "learning_rate": 2.933908799216625e-05, + "loss": 3.1769, + "step": 5054 + }, + { + "epoch": 7.4, + "grad_norm": 0.9311485886573792, + "learning_rate": 2.9305906697199615e-05, + "loss": 3.1574, + "step": 5055 + }, + { + "epoch": 7.4, + "grad_norm": 0.41424453258514404, + "learning_rate": 2.9272740954459776e-05, + "loss": 3.132, + "step": 5056 + }, + { + "epoch": 7.4, + "grad_norm": 0.4336400330066681, + "learning_rate": 2.92395907712429e-05, + "loss": 3.1724, + "step": 5057 + }, + { + "epoch": 7.4, + "grad_norm": 0.902377188205719, + "learning_rate": 2.920645615484181e-05, + "loss": 3.1471, + "step": 5058 + }, + { + "epoch": 7.4, + "grad_norm": 0.7323518991470337, + "learning_rate": 2.9173337112545985e-05, + "loss": 3.2012, + "step": 5059 + }, + { + "epoch": 7.41, + "grad_norm": 0.6311755776405334, + "learning_rate": 2.9140233651641314e-05, + "loss": 3.1655, + "step": 5060 + }, + { + "epoch": 7.41, + "grad_norm": 1.4856607913970947, + "learning_rate": 2.9107145779410317e-05, + "loss": 3.1768, + "step": 5061 + }, + { + "epoch": 7.41, + "grad_norm": 0.5299568176269531, + "learning_rate": 2.9074073503132217e-05, + "loss": 3.1368, + "step": 5062 + }, + { + "epoch": 7.41, + "grad_norm": 0.5073440074920654, + "learning_rate": 2.904101683008259e-05, + "loss": 3.1784, + "step": 5063 + }, + { + "epoch": 7.41, + "grad_norm": 0.7405803799629211, + "learning_rate": 2.9007975767533714e-05, + "loss": 3.145, + "step": 5064 + }, + { + "epoch": 7.41, + "grad_norm": 0.39113080501556396, + "learning_rate": 2.897495032275439e-05, + "loss": 3.148, + "step": 5065 + }, + { + "epoch": 7.41, + "grad_norm": 0.8356419801712036, + "learning_rate": 2.8941940503010022e-05, + "loss": 3.1611, + "step": 5066 + }, + { + "epoch": 7.42, + "grad_norm": 0.6192629337310791, + "learning_rate": 2.8908946315562524e-05, + "loss": 3.1742, + "step": 5067 + }, + { + "epoch": 7.42, + "grad_norm": 1.1198054552078247, + "learning_rate": 2.8875967767670408e-05, + "loss": 3.1789, + "step": 5068 + }, + { + "epoch": 7.42, + "grad_norm": 0.4452805519104004, + "learning_rate": 2.884300486658873e-05, + "loss": 3.1533, + "step": 5069 + }, + { + "epoch": 7.42, + "grad_norm": 0.8051960468292236, + "learning_rate": 2.881005761956911e-05, + "loss": 3.1688, + "step": 5070 + }, + { + "epoch": 7.42, + "eval_loss": 3.0202746391296387, + "eval_runtime": 1162.3223, + "eval_samples_per_second": 14.748, + "eval_steps_per_second": 0.922, + "step": 5070 + }, + { + "epoch": 7.42, + "grad_norm": 1.1692814826965332, + "learning_rate": 2.8777126033859726e-05, + "loss": 3.1559, + "step": 5071 + }, + { + "epoch": 7.42, + "grad_norm": 0.3186073303222656, + "learning_rate": 2.8744210116705296e-05, + "loss": 3.1507, + "step": 5072 + }, + { + "epoch": 7.43, + "grad_norm": 0.6718719005584717, + "learning_rate": 2.8711309875347102e-05, + "loss": 3.1667, + "step": 5073 + }, + { + "epoch": 7.43, + "grad_norm": 0.7967109680175781, + "learning_rate": 2.8678425317022973e-05, + "loss": 3.1415, + "step": 5074 + }, + { + "epoch": 7.43, + "grad_norm": 0.36231231689453125, + "learning_rate": 2.8645556448967315e-05, + "loss": 3.1264, + "step": 5075 + }, + { + "epoch": 7.43, + "grad_norm": 0.5855808258056641, + "learning_rate": 2.8612703278411025e-05, + "loss": 3.1217, + "step": 5076 + }, + { + "epoch": 7.43, + "grad_norm": 0.6386885046958923, + "learning_rate": 2.8579865812581608e-05, + "loss": 3.1476, + "step": 5077 + }, + { + "epoch": 7.43, + "grad_norm": 0.4413103759288788, + "learning_rate": 2.8547044058703076e-05, + "loss": 3.2046, + "step": 5078 + }, + { + "epoch": 7.43, + "grad_norm": 0.7146703004837036, + "learning_rate": 2.851423802399599e-05, + "loss": 3.2516, + "step": 5079 + }, + { + "epoch": 7.44, + "grad_norm": 0.7587980031967163, + "learning_rate": 2.8481447715677466e-05, + "loss": 3.2025, + "step": 5080 + }, + { + "epoch": 7.44, + "grad_norm": 1.2787212133407593, + "learning_rate": 2.8448673140961147e-05, + "loss": 3.1572, + "step": 5081 + }, + { + "epoch": 7.44, + "grad_norm": 0.5361132621765137, + "learning_rate": 2.8415914307057245e-05, + "loss": 3.1564, + "step": 5082 + }, + { + "epoch": 7.44, + "grad_norm": 1.6003608703613281, + "learning_rate": 2.8383171221172412e-05, + "loss": 3.1482, + "step": 5083 + }, + { + "epoch": 7.44, + "grad_norm": 0.9171953201293945, + "learning_rate": 2.8350443890510003e-05, + "loss": 3.1886, + "step": 5084 + }, + { + "epoch": 7.44, + "grad_norm": 0.39613091945648193, + "learning_rate": 2.8317732322269754e-05, + "loss": 3.1506, + "step": 5085 + }, + { + "epoch": 7.44, + "grad_norm": 1.17970609664917, + "learning_rate": 2.8285036523647944e-05, + "loss": 3.1526, + "step": 5086 + }, + { + "epoch": 7.45, + "grad_norm": 0.781102180480957, + "learning_rate": 2.8252356501837562e-05, + "loss": 3.1663, + "step": 5087 + }, + { + "epoch": 7.45, + "grad_norm": 0.49025315046310425, + "learning_rate": 2.8219692264027886e-05, + "loss": 3.1647, + "step": 5088 + }, + { + "epoch": 7.45, + "grad_norm": 2.0356221199035645, + "learning_rate": 2.8187043817404833e-05, + "loss": 3.2115, + "step": 5089 + }, + { + "epoch": 7.45, + "grad_norm": 1.2565196752548218, + "learning_rate": 2.815441116915093e-05, + "loss": 3.2334, + "step": 5090 + }, + { + "epoch": 7.45, + "grad_norm": 0.6307234168052673, + "learning_rate": 2.812179432644506e-05, + "loss": 3.1614, + "step": 5091 + }, + { + "epoch": 7.45, + "grad_norm": 0.8804566264152527, + "learning_rate": 2.808919329646269e-05, + "loss": 3.1703, + "step": 5092 + }, + { + "epoch": 7.45, + "grad_norm": 0.5766822695732117, + "learning_rate": 2.8056608086375934e-05, + "loss": 3.1264, + "step": 5093 + }, + { + "epoch": 7.46, + "grad_norm": 0.9480114579200745, + "learning_rate": 2.8024038703353227e-05, + "loss": 3.1382, + "step": 5094 + }, + { + "epoch": 7.46, + "grad_norm": 0.5312039852142334, + "learning_rate": 2.7991485154559617e-05, + "loss": 3.1555, + "step": 5095 + }, + { + "epoch": 7.46, + "grad_norm": 0.5020516514778137, + "learning_rate": 2.7958947447156746e-05, + "loss": 3.1825, + "step": 5096 + }, + { + "epoch": 7.46, + "grad_norm": 0.8643108010292053, + "learning_rate": 2.7926425588302606e-05, + "loss": 3.1903, + "step": 5097 + }, + { + "epoch": 7.46, + "grad_norm": 0.38518649339675903, + "learning_rate": 2.789391958515183e-05, + "loss": 3.1832, + "step": 5098 + }, + { + "epoch": 7.46, + "grad_norm": 0.9875998497009277, + "learning_rate": 2.7861429444855514e-05, + "loss": 3.1695, + "step": 5099 + }, + { + "epoch": 7.47, + "grad_norm": 0.5301055312156677, + "learning_rate": 2.7828955174561257e-05, + "loss": 3.1645, + "step": 5100 + }, + { + "epoch": 7.47, + "grad_norm": 0.7682579755783081, + "learning_rate": 2.7796496781413206e-05, + "loss": 3.1384, + "step": 5101 + }, + { + "epoch": 7.47, + "grad_norm": 1.0809067487716675, + "learning_rate": 2.7764054272551965e-05, + "loss": 3.1427, + "step": 5102 + }, + { + "epoch": 7.47, + "grad_norm": 0.6575438380241394, + "learning_rate": 2.773162765511469e-05, + "loss": 3.1146, + "step": 5103 + }, + { + "epoch": 7.47, + "grad_norm": 1.6334797143936157, + "learning_rate": 2.7699216936234994e-05, + "loss": 3.1582, + "step": 5104 + }, + { + "epoch": 7.47, + "grad_norm": 1.1224489212036133, + "learning_rate": 2.7666822123043046e-05, + "loss": 3.1531, + "step": 5105 + }, + { + "epoch": 7.47, + "grad_norm": 0.6989713311195374, + "learning_rate": 2.7634443222665475e-05, + "loss": 3.1587, + "step": 5106 + }, + { + "epoch": 7.48, + "grad_norm": 1.6318511962890625, + "learning_rate": 2.760208024222546e-05, + "loss": 3.1676, + "step": 5107 + }, + { + "epoch": 7.48, + "grad_norm": 1.0403834581375122, + "learning_rate": 2.7569733188842528e-05, + "loss": 3.2076, + "step": 5108 + }, + { + "epoch": 7.48, + "grad_norm": 0.9188293814659119, + "learning_rate": 2.7537402069632922e-05, + "loss": 3.1641, + "step": 5109 + }, + { + "epoch": 7.48, + "grad_norm": 1.1109488010406494, + "learning_rate": 2.7505086891709277e-05, + "loss": 3.1396, + "step": 5110 + }, + { + "epoch": 7.48, + "grad_norm": 0.40556856989860535, + "learning_rate": 2.747278766218061e-05, + "loss": 3.1475, + "step": 5111 + }, + { + "epoch": 7.48, + "grad_norm": 1.6644717454910278, + "learning_rate": 2.744050438815263e-05, + "loss": 3.1237, + "step": 5112 + }, + { + "epoch": 7.48, + "grad_norm": 1.2431093454360962, + "learning_rate": 2.740823707672744e-05, + "loss": 3.1342, + "step": 5113 + }, + { + "epoch": 7.49, + "grad_norm": 0.5421004295349121, + "learning_rate": 2.7375985735003552e-05, + "loss": 3.153, + "step": 5114 + }, + { + "epoch": 7.49, + "grad_norm": 1.7382439374923706, + "learning_rate": 2.7343750370076125e-05, + "loss": 3.1935, + "step": 5115 + }, + { + "epoch": 7.49, + "grad_norm": 0.757526695728302, + "learning_rate": 2.7311530989036726e-05, + "loss": 3.173, + "step": 5116 + }, + { + "epoch": 7.49, + "grad_norm": 1.094909906387329, + "learning_rate": 2.7279327598973303e-05, + "loss": 3.1654, + "step": 5117 + }, + { + "epoch": 7.49, + "grad_norm": 1.3784464597702026, + "learning_rate": 2.7247140206970534e-05, + "loss": 3.1394, + "step": 5118 + }, + { + "epoch": 7.49, + "grad_norm": 0.33335521817207336, + "learning_rate": 2.721496882010931e-05, + "loss": 3.1698, + "step": 5119 + }, + { + "epoch": 7.49, + "grad_norm": 0.5880882740020752, + "learning_rate": 2.7182813445467138e-05, + "loss": 3.1336, + "step": 5120 + }, + { + "epoch": 7.5, + "grad_norm": 1.3899345397949219, + "learning_rate": 2.7150674090118054e-05, + "loss": 3.2018, + "step": 5121 + }, + { + "epoch": 7.5, + "grad_norm": 0.4972773790359497, + "learning_rate": 2.7118550761132444e-05, + "loss": 3.174, + "step": 5122 + }, + { + "epoch": 7.5, + "grad_norm": 1.0646159648895264, + "learning_rate": 2.7086443465577195e-05, + "loss": 3.16, + "step": 5123 + }, + { + "epoch": 7.5, + "grad_norm": 0.5726083517074585, + "learning_rate": 2.705435221051581e-05, + "loss": 3.1459, + "step": 5124 + }, + { + "epoch": 7.5, + "grad_norm": 0.5148174166679382, + "learning_rate": 2.7022277003008035e-05, + "loss": 3.1702, + "step": 5125 + }, + { + "epoch": 7.5, + "grad_norm": 0.5842214226722717, + "learning_rate": 2.6990217850110222e-05, + "loss": 3.189, + "step": 5126 + }, + { + "epoch": 7.51, + "grad_norm": 0.450296550989151, + "learning_rate": 2.6958174758875253e-05, + "loss": 3.1581, + "step": 5127 + }, + { + "epoch": 7.51, + "grad_norm": 0.6981111168861389, + "learning_rate": 2.6926147736352302e-05, + "loss": 3.1709, + "step": 5128 + }, + { + "epoch": 7.51, + "grad_norm": 0.6360636949539185, + "learning_rate": 2.689413678958713e-05, + "loss": 3.163, + "step": 5129 + }, + { + "epoch": 7.51, + "grad_norm": 0.6991342306137085, + "learning_rate": 2.686214192562193e-05, + "loss": 3.1673, + "step": 5130 + }, + { + "epoch": 7.51, + "grad_norm": 0.4301992952823639, + "learning_rate": 2.6830163151495358e-05, + "loss": 3.15, + "step": 5131 + }, + { + "epoch": 7.51, + "grad_norm": 0.6416985988616943, + "learning_rate": 2.679820047424253e-05, + "loss": 3.1364, + "step": 5132 + }, + { + "epoch": 7.51, + "grad_norm": 0.6713948845863342, + "learning_rate": 2.6766253900895023e-05, + "loss": 3.1668, + "step": 5133 + }, + { + "epoch": 7.52, + "grad_norm": 0.4859183728694916, + "learning_rate": 2.6734323438480858e-05, + "loss": 3.153, + "step": 5134 + }, + { + "epoch": 7.52, + "grad_norm": 0.5950589179992676, + "learning_rate": 2.6702409094024538e-05, + "loss": 3.164, + "step": 5135 + }, + { + "epoch": 7.52, + "grad_norm": 0.4921453297138214, + "learning_rate": 2.6670510874547006e-05, + "loss": 3.1513, + "step": 5136 + }, + { + "epoch": 7.52, + "grad_norm": 0.3888862133026123, + "learning_rate": 2.6638628787065633e-05, + "loss": 3.1621, + "step": 5137 + }, + { + "epoch": 7.52, + "grad_norm": 0.43584099411964417, + "learning_rate": 2.660676283859429e-05, + "loss": 3.1327, + "step": 5138 + }, + { + "epoch": 7.52, + "grad_norm": 0.659436047077179, + "learning_rate": 2.6574913036143266e-05, + "loss": 3.1782, + "step": 5139 + }, + { + "epoch": 7.52, + "grad_norm": 0.33366990089416504, + "learning_rate": 2.6543079386719295e-05, + "loss": 3.1657, + "step": 5140 + }, + { + "epoch": 7.53, + "grad_norm": 0.6816606521606445, + "learning_rate": 2.651126189732561e-05, + "loss": 3.145, + "step": 5141 + }, + { + "epoch": 7.53, + "grad_norm": 0.5122085213661194, + "learning_rate": 2.6479460574961746e-05, + "loss": 3.1146, + "step": 5142 + }, + { + "epoch": 7.53, + "grad_norm": 0.7051358819007874, + "learning_rate": 2.6447675426623864e-05, + "loss": 3.1565, + "step": 5143 + }, + { + "epoch": 7.53, + "grad_norm": 0.3721584379673004, + "learning_rate": 2.6415906459304495e-05, + "loss": 3.165, + "step": 5144 + }, + { + "epoch": 7.53, + "grad_norm": 0.30766353011131287, + "learning_rate": 2.638415367999252e-05, + "loss": 3.1973, + "step": 5145 + }, + { + "epoch": 7.53, + "grad_norm": 0.41517648100852966, + "learning_rate": 2.63524170956734e-05, + "loss": 3.1766, + "step": 5146 + }, + { + "epoch": 7.53, + "grad_norm": 0.433137983083725, + "learning_rate": 2.6320696713328986e-05, + "loss": 3.2217, + "step": 5147 + }, + { + "epoch": 7.54, + "grad_norm": 0.3852587044239044, + "learning_rate": 2.6288992539937463e-05, + "loss": 3.1605, + "step": 5148 + }, + { + "epoch": 7.54, + "grad_norm": 0.5037500858306885, + "learning_rate": 2.625730458247362e-05, + "loss": 3.1981, + "step": 5149 + }, + { + "epoch": 7.54, + "grad_norm": 0.4805777966976166, + "learning_rate": 2.62256328479086e-05, + "loss": 3.1315, + "step": 5150 + }, + { + "epoch": 7.54, + "grad_norm": 0.7443678975105286, + "learning_rate": 2.6193977343209874e-05, + "loss": 3.173, + "step": 5151 + }, + { + "epoch": 7.54, + "grad_norm": 0.6087154746055603, + "learning_rate": 2.6162338075341564e-05, + "loss": 3.1921, + "step": 5152 + }, + { + "epoch": 7.54, + "grad_norm": 2.0148138999938965, + "learning_rate": 2.613071505126402e-05, + "loss": 3.1314, + "step": 5153 + }, + { + "epoch": 7.55, + "grad_norm": 0.5105882883071899, + "learning_rate": 2.6099108277934103e-05, + "loss": 3.1305, + "step": 5154 + }, + { + "epoch": 7.55, + "grad_norm": 0.4445195496082306, + "learning_rate": 2.6067517762305116e-05, + "loss": 3.1308, + "step": 5155 + }, + { + "epoch": 7.55, + "grad_norm": 0.3402569890022278, + "learning_rate": 2.6035943511326733e-05, + "loss": 3.1384, + "step": 5156 + }, + { + "epoch": 7.55, + "grad_norm": 0.42875614762306213, + "learning_rate": 2.6004385531945107e-05, + "loss": 3.1476, + "step": 5157 + }, + { + "epoch": 7.55, + "grad_norm": 0.46508556604385376, + "learning_rate": 2.5972843831102755e-05, + "loss": 3.1542, + "step": 5158 + }, + { + "epoch": 7.55, + "grad_norm": 0.7432370185852051, + "learning_rate": 2.594131841573866e-05, + "loss": 3.1507, + "step": 5159 + }, + { + "epoch": 7.55, + "grad_norm": 0.5330306887626648, + "learning_rate": 2.590980929278819e-05, + "loss": 3.1661, + "step": 5160 + }, + { + "epoch": 7.56, + "grad_norm": 0.37541455030441284, + "learning_rate": 2.587831646918315e-05, + "loss": 3.1454, + "step": 5161 + }, + { + "epoch": 7.56, + "grad_norm": 0.6398892402648926, + "learning_rate": 2.584683995185174e-05, + "loss": 3.1901, + "step": 5162 + }, + { + "epoch": 7.56, + "grad_norm": 0.9277903437614441, + "learning_rate": 2.5815379747718594e-05, + "loss": 3.1594, + "step": 5163 + }, + { + "epoch": 7.56, + "grad_norm": 0.8719270825386047, + "learning_rate": 2.5783935863704732e-05, + "loss": 3.1646, + "step": 5164 + }, + { + "epoch": 7.56, + "grad_norm": 0.811159074306488, + "learning_rate": 2.5752508306727608e-05, + "loss": 3.2009, + "step": 5165 + }, + { + "epoch": 7.56, + "grad_norm": 0.834298849105835, + "learning_rate": 2.5721097083701084e-05, + "loss": 3.1747, + "step": 5166 + }, + { + "epoch": 7.56, + "grad_norm": 1.0362366437911987, + "learning_rate": 2.5689702201535394e-05, + "loss": 3.1428, + "step": 5167 + }, + { + "epoch": 7.57, + "grad_norm": 0.8834418058395386, + "learning_rate": 2.5658323667137218e-05, + "loss": 3.1559, + "step": 5168 + }, + { + "epoch": 7.57, + "grad_norm": 0.4252467155456543, + "learning_rate": 2.5626961487409616e-05, + "loss": 3.1528, + "step": 5169 + }, + { + "epoch": 7.57, + "grad_norm": 1.2788466215133667, + "learning_rate": 2.559561566925206e-05, + "loss": 3.1789, + "step": 5170 + }, + { + "epoch": 7.57, + "grad_norm": 0.736139714717865, + "learning_rate": 2.556428621956043e-05, + "loss": 3.1242, + "step": 5171 + }, + { + "epoch": 7.57, + "grad_norm": 0.38194677233695984, + "learning_rate": 2.553297314522698e-05, + "loss": 3.1737, + "step": 5172 + }, + { + "epoch": 7.57, + "grad_norm": 0.5845624804496765, + "learning_rate": 2.5501676453140378e-05, + "loss": 3.1726, + "step": 5173 + }, + { + "epoch": 7.58, + "grad_norm": 0.7914636731147766, + "learning_rate": 2.5470396150185692e-05, + "loss": 3.1579, + "step": 5174 + }, + { + "epoch": 7.58, + "grad_norm": 0.39859166741371155, + "learning_rate": 2.5439132243244412e-05, + "loss": 3.1319, + "step": 5175 + }, + { + "epoch": 7.58, + "grad_norm": 1.092436671257019, + "learning_rate": 2.540788473919433e-05, + "loss": 3.1286, + "step": 5176 + }, + { + "epoch": 7.58, + "grad_norm": 0.6279719471931458, + "learning_rate": 2.5376653644909675e-05, + "loss": 3.1328, + "step": 5177 + }, + { + "epoch": 7.58, + "grad_norm": 1.2582124471664429, + "learning_rate": 2.534543896726117e-05, + "loss": 3.1486, + "step": 5178 + }, + { + "epoch": 7.58, + "grad_norm": 0.6606663465499878, + "learning_rate": 2.5314240713115756e-05, + "loss": 3.1386, + "step": 5179 + }, + { + "epoch": 7.58, + "grad_norm": 0.5657997727394104, + "learning_rate": 2.528305888933683e-05, + "loss": 3.1366, + "step": 5180 + }, + { + "epoch": 7.59, + "grad_norm": 0.7459825873374939, + "learning_rate": 2.5251893502784275e-05, + "loss": 3.1625, + "step": 5181 + }, + { + "epoch": 7.59, + "grad_norm": 0.6678188443183899, + "learning_rate": 2.522074456031418e-05, + "loss": 3.1981, + "step": 5182 + }, + { + "epoch": 7.59, + "grad_norm": 0.7135474681854248, + "learning_rate": 2.5189612068779112e-05, + "loss": 3.1958, + "step": 5183 + }, + { + "epoch": 7.59, + "grad_norm": 0.5656193494796753, + "learning_rate": 2.5158496035028077e-05, + "loss": 3.1218, + "step": 5184 + }, + { + "epoch": 7.59, + "grad_norm": 0.5109965205192566, + "learning_rate": 2.5127396465906327e-05, + "loss": 3.1465, + "step": 5185 + }, + { + "epoch": 7.59, + "grad_norm": 0.5121713876724243, + "learning_rate": 2.5096313368255585e-05, + "loss": 3.1401, + "step": 5186 + }, + { + "epoch": 7.59, + "grad_norm": 0.38621145486831665, + "learning_rate": 2.506524674891392e-05, + "loss": 3.1519, + "step": 5187 + }, + { + "epoch": 7.6, + "grad_norm": 0.557099461555481, + "learning_rate": 2.5034196614715778e-05, + "loss": 3.1632, + "step": 5188 + }, + { + "epoch": 7.6, + "grad_norm": 0.42920613288879395, + "learning_rate": 2.500316297249198e-05, + "loss": 3.1272, + "step": 5189 + }, + { + "epoch": 7.6, + "grad_norm": 0.765871524810791, + "learning_rate": 2.4972145829069714e-05, + "loss": 3.1568, + "step": 5190 + }, + { + "epoch": 7.6, + "grad_norm": 0.4879997670650482, + "learning_rate": 2.4941145191272563e-05, + "loss": 3.174, + "step": 5191 + }, + { + "epoch": 7.6, + "grad_norm": 0.5009108781814575, + "learning_rate": 2.4910161065920445e-05, + "loss": 3.1467, + "step": 5192 + }, + { + "epoch": 7.6, + "grad_norm": 0.6382935643196106, + "learning_rate": 2.4879193459829665e-05, + "loss": 3.1552, + "step": 5193 + }, + { + "epoch": 7.6, + "grad_norm": 0.57361900806427, + "learning_rate": 2.484824237981288e-05, + "loss": 3.1511, + "step": 5194 + }, + { + "epoch": 7.61, + "grad_norm": 0.2623114287853241, + "learning_rate": 2.4817307832679126e-05, + "loss": 3.1664, + "step": 5195 + }, + { + "epoch": 7.61, + "grad_norm": 0.5569726228713989, + "learning_rate": 2.4786389825233802e-05, + "loss": 3.1431, + "step": 5196 + }, + { + "epoch": 7.61, + "grad_norm": 0.5350569486618042, + "learning_rate": 2.4755488364278656e-05, + "loss": 3.173, + "step": 5197 + }, + { + "epoch": 7.61, + "grad_norm": 0.38636651635169983, + "learning_rate": 2.4724603456611805e-05, + "loss": 3.1996, + "step": 5198 + }, + { + "epoch": 7.61, + "grad_norm": 0.6508245468139648, + "learning_rate": 2.4693735109027716e-05, + "loss": 3.1693, + "step": 5199 + }, + { + "epoch": 7.61, + "grad_norm": 0.7632821798324585, + "learning_rate": 2.4662883328317222e-05, + "loss": 3.1948, + "step": 5200 + }, + { + "epoch": 7.62, + "grad_norm": 0.5081233978271484, + "learning_rate": 2.4632048121267547e-05, + "loss": 3.1938, + "step": 5201 + }, + { + "epoch": 7.62, + "grad_norm": 0.7682433128356934, + "learning_rate": 2.4601229494662125e-05, + "loss": 3.1553, + "step": 5202 + }, + { + "epoch": 7.62, + "grad_norm": 1.0782290697097778, + "learning_rate": 2.4570427455280943e-05, + "loss": 3.1649, + "step": 5203 + }, + { + "epoch": 7.62, + "grad_norm": 0.6353168487548828, + "learning_rate": 2.4539642009900253e-05, + "loss": 3.1517, + "step": 5204 + }, + { + "epoch": 7.62, + "grad_norm": 0.7350844144821167, + "learning_rate": 2.4508873165292535e-05, + "loss": 3.1375, + "step": 5205 + }, + { + "epoch": 7.62, + "grad_norm": 0.9207132458686829, + "learning_rate": 2.4478120928226846e-05, + "loss": 3.1478, + "step": 5206 + }, + { + "epoch": 7.62, + "grad_norm": 0.7946307063102722, + "learning_rate": 2.444738530546845e-05, + "loss": 3.1392, + "step": 5207 + }, + { + "epoch": 7.63, + "grad_norm": 0.9071816802024841, + "learning_rate": 2.441666630377889e-05, + "loss": 3.1275, + "step": 5208 + }, + { + "epoch": 7.63, + "grad_norm": 1.2288057804107666, + "learning_rate": 2.438596392991628e-05, + "loss": 3.1333, + "step": 5209 + }, + { + "epoch": 7.63, + "grad_norm": 1.0466946363449097, + "learning_rate": 2.435527819063482e-05, + "loss": 3.143, + "step": 5210 + }, + { + "epoch": 7.63, + "grad_norm": 0.5097655057907104, + "learning_rate": 2.432460909268518e-05, + "loss": 3.2069, + "step": 5211 + }, + { + "epoch": 7.63, + "grad_norm": 0.8036677241325378, + "learning_rate": 2.4293956642814453e-05, + "loss": 3.1592, + "step": 5212 + }, + { + "epoch": 7.63, + "grad_norm": 3.4330554008483887, + "learning_rate": 2.4263320847765868e-05, + "loss": 3.1551, + "step": 5213 + }, + { + "epoch": 7.63, + "grad_norm": 0.9436863660812378, + "learning_rate": 2.4232701714279115e-05, + "loss": 3.1353, + "step": 5214 + }, + { + "epoch": 7.64, + "grad_norm": 0.30926942825317383, + "learning_rate": 2.420209924909026e-05, + "loss": 3.1577, + "step": 5215 + }, + { + "epoch": 7.64, + "grad_norm": 0.7689909338951111, + "learning_rate": 2.417151345893157e-05, + "loss": 3.1326, + "step": 5216 + }, + { + "epoch": 7.64, + "grad_norm": 0.8262006640434265, + "learning_rate": 2.4140944350531712e-05, + "loss": 3.1372, + "step": 5217 + }, + { + "epoch": 7.64, + "grad_norm": 0.35750913619995117, + "learning_rate": 2.411039193061577e-05, + "loss": 3.1526, + "step": 5218 + }, + { + "epoch": 7.64, + "grad_norm": 0.7210152745246887, + "learning_rate": 2.407985620590498e-05, + "loss": 3.1702, + "step": 5219 + }, + { + "epoch": 7.64, + "grad_norm": 0.37686386704444885, + "learning_rate": 2.404933718311704e-05, + "loss": 3.1609, + "step": 5220 + }, + { + "epoch": 7.64, + "grad_norm": 0.38531693816185, + "learning_rate": 2.401883486896591e-05, + "loss": 3.16, + "step": 5221 + }, + { + "epoch": 7.65, + "grad_norm": 3.7467172145843506, + "learning_rate": 2.3988349270161913e-05, + "loss": 3.1974, + "step": 5222 + }, + { + "epoch": 7.65, + "grad_norm": 0.4558284878730774, + "learning_rate": 2.3957880393411667e-05, + "loss": 3.1724, + "step": 5223 + }, + { + "epoch": 7.65, + "grad_norm": 0.5120177268981934, + "learning_rate": 2.3927428245418127e-05, + "loss": 3.1466, + "step": 5224 + }, + { + "epoch": 7.65, + "grad_norm": 0.47643178701400757, + "learning_rate": 2.3896992832880548e-05, + "loss": 3.1508, + "step": 5225 + }, + { + "epoch": 7.65, + "grad_norm": 0.333384245634079, + "learning_rate": 2.3866574162494538e-05, + "loss": 3.1346, + "step": 5226 + }, + { + "epoch": 7.65, + "grad_norm": 0.5989437699317932, + "learning_rate": 2.383617224095198e-05, + "loss": 3.1354, + "step": 5227 + }, + { + "epoch": 7.66, + "grad_norm": 0.35983359813690186, + "learning_rate": 2.38057870749411e-05, + "loss": 3.1599, + "step": 5228 + }, + { + "epoch": 7.66, + "grad_norm": 0.3712514042854309, + "learning_rate": 2.3775418671146443e-05, + "loss": 3.1584, + "step": 5229 + }, + { + "epoch": 7.66, + "grad_norm": 0.3716100752353668, + "learning_rate": 2.3745067036248836e-05, + "loss": 3.1766, + "step": 5230 + }, + { + "epoch": 7.66, + "grad_norm": 0.27831578254699707, + "learning_rate": 2.3714732176925446e-05, + "loss": 3.1285, + "step": 5231 + }, + { + "epoch": 7.66, + "grad_norm": 0.47703278064727783, + "learning_rate": 2.368441409984976e-05, + "loss": 3.1467, + "step": 5232 + }, + { + "epoch": 7.66, + "grad_norm": 0.577670156955719, + "learning_rate": 2.3654112811691477e-05, + "loss": 3.1222, + "step": 5233 + }, + { + "epoch": 7.66, + "grad_norm": 0.44678911566734314, + "learning_rate": 2.3623828319116748e-05, + "loss": 3.1783, + "step": 5234 + }, + { + "epoch": 7.67, + "grad_norm": 0.9070417881011963, + "learning_rate": 2.359356062878797e-05, + "loss": 3.1324, + "step": 5235 + }, + { + "epoch": 7.67, + "grad_norm": 0.5070209503173828, + "learning_rate": 2.356330974736375e-05, + "loss": 3.129, + "step": 5236 + }, + { + "epoch": 7.67, + "grad_norm": 0.6462606191635132, + "learning_rate": 2.353307568149915e-05, + "loss": 3.1474, + "step": 5237 + }, + { + "epoch": 7.67, + "grad_norm": 0.34412530064582825, + "learning_rate": 2.3502858437845476e-05, + "loss": 3.1597, + "step": 5238 + }, + { + "epoch": 7.67, + "grad_norm": 0.4765468239784241, + "learning_rate": 2.347265802305022e-05, + "loss": 3.1474, + "step": 5239 + }, + { + "epoch": 7.67, + "eval_loss": 3.013108015060425, + "eval_runtime": 1170.7492, + "eval_samples_per_second": 14.642, + "eval_steps_per_second": 0.916, + "step": 5239 + }, + { + "epoch": 7.67, + "grad_norm": 0.5959687232971191, + "learning_rate": 2.3442474443757366e-05, + "loss": 3.1172, + "step": 5240 + }, + { + "epoch": 7.67, + "grad_norm": 0.27279970049858093, + "learning_rate": 2.3412307706607108e-05, + "loss": 3.1737, + "step": 5241 + }, + { + "epoch": 7.68, + "grad_norm": 0.5665635466575623, + "learning_rate": 2.3382157818235816e-05, + "loss": 3.1308, + "step": 5242 + }, + { + "epoch": 7.68, + "grad_norm": 0.3132018446922302, + "learning_rate": 2.3352024785276404e-05, + "loss": 3.1657, + "step": 5243 + }, + { + "epoch": 7.68, + "grad_norm": 0.622239351272583, + "learning_rate": 2.3321908614357824e-05, + "loss": 3.1689, + "step": 5244 + }, + { + "epoch": 7.68, + "grad_norm": 0.5310043692588806, + "learning_rate": 2.3291809312105462e-05, + "loss": 3.2016, + "step": 5245 + }, + { + "epoch": 7.68, + "grad_norm": 0.3280445337295532, + "learning_rate": 2.3261726885141023e-05, + "loss": 3.1654, + "step": 5246 + }, + { + "epoch": 7.68, + "grad_norm": 0.4684401750564575, + "learning_rate": 2.3231661340082366e-05, + "loss": 3.1932, + "step": 5247 + }, + { + "epoch": 7.68, + "grad_norm": 0.5372167825698853, + "learning_rate": 2.3201612683543704e-05, + "loss": 3.2027, + "step": 5248 + }, + { + "epoch": 7.69, + "grad_norm": 0.865881621837616, + "learning_rate": 2.3171580922135627e-05, + "loss": 3.1653, + "step": 5249 + }, + { + "epoch": 7.69, + "grad_norm": 0.650731086730957, + "learning_rate": 2.3141566062464847e-05, + "loss": 3.1495, + "step": 5250 + }, + { + "epoch": 7.69, + "grad_norm": 1.0147230625152588, + "learning_rate": 2.311156811113444e-05, + "loss": 3.1422, + "step": 5251 + }, + { + "epoch": 7.69, + "grad_norm": 0.34079644083976746, + "learning_rate": 2.308158707474377e-05, + "loss": 3.125, + "step": 5252 + }, + { + "epoch": 7.69, + "grad_norm": 0.8186140656471252, + "learning_rate": 2.3051622959888464e-05, + "loss": 3.1991, + "step": 5253 + }, + { + "epoch": 7.69, + "grad_norm": 0.8478550314903259, + "learning_rate": 2.3021675773160424e-05, + "loss": 3.1535, + "step": 5254 + }, + { + "epoch": 7.7, + "grad_norm": 0.6345941424369812, + "learning_rate": 2.299174552114781e-05, + "loss": 3.1704, + "step": 5255 + }, + { + "epoch": 7.7, + "grad_norm": 0.9661188721656799, + "learning_rate": 2.296183221043511e-05, + "loss": 3.1492, + "step": 5256 + }, + { + "epoch": 7.7, + "grad_norm": 0.4356244206428528, + "learning_rate": 2.2931935847603037e-05, + "loss": 3.159, + "step": 5257 + }, + { + "epoch": 7.7, + "grad_norm": 0.6363608837127686, + "learning_rate": 2.290205643922859e-05, + "loss": 3.1579, + "step": 5258 + }, + { + "epoch": 7.7, + "grad_norm": 1.1210684776306152, + "learning_rate": 2.2872193991885026e-05, + "loss": 3.1412, + "step": 5259 + }, + { + "epoch": 7.7, + "grad_norm": 0.38606759905815125, + "learning_rate": 2.2842348512141908e-05, + "loss": 3.1999, + "step": 5260 + }, + { + "epoch": 7.7, + "grad_norm": 1.1856557130813599, + "learning_rate": 2.281252000656503e-05, + "loss": 3.1595, + "step": 5261 + }, + { + "epoch": 7.71, + "grad_norm": 1.063137412071228, + "learning_rate": 2.2782708481716463e-05, + "loss": 3.1713, + "step": 5262 + }, + { + "epoch": 7.71, + "grad_norm": 0.6546114087104797, + "learning_rate": 2.275291394415454e-05, + "loss": 3.1581, + "step": 5263 + }, + { + "epoch": 7.71, + "grad_norm": 1.1830710172653198, + "learning_rate": 2.2723136400433865e-05, + "loss": 3.1447, + "step": 5264 + }, + { + "epoch": 7.71, + "grad_norm": 0.6315742135047913, + "learning_rate": 2.269337585710529e-05, + "loss": 3.1487, + "step": 5265 + }, + { + "epoch": 7.71, + "grad_norm": 0.8535947799682617, + "learning_rate": 2.2663632320715976e-05, + "loss": 3.1631, + "step": 5266 + }, + { + "epoch": 7.71, + "grad_norm": 1.0756810903549194, + "learning_rate": 2.2633905797809197e-05, + "loss": 3.1389, + "step": 5267 + }, + { + "epoch": 7.71, + "grad_norm": 0.7916353344917297, + "learning_rate": 2.2604196294924694e-05, + "loss": 3.13, + "step": 5268 + }, + { + "epoch": 7.72, + "grad_norm": 0.46646764874458313, + "learning_rate": 2.257450381859836e-05, + "loss": 3.1633, + "step": 5269 + }, + { + "epoch": 7.72, + "grad_norm": 0.6788039803504944, + "learning_rate": 2.254482837536224e-05, + "loss": 3.1178, + "step": 5270 + }, + { + "epoch": 7.72, + "grad_norm": 0.5274389982223511, + "learning_rate": 2.251516997174482e-05, + "loss": 3.1824, + "step": 5271 + }, + { + "epoch": 7.72, + "grad_norm": 0.5712849497795105, + "learning_rate": 2.2485528614270757e-05, + "loss": 3.1324, + "step": 5272 + }, + { + "epoch": 7.72, + "grad_norm": 0.39241471886634827, + "learning_rate": 2.2455904309460886e-05, + "loss": 3.1527, + "step": 5273 + }, + { + "epoch": 7.72, + "grad_norm": 0.3917776346206665, + "learning_rate": 2.242629706383237e-05, + "loss": 3.1554, + "step": 5274 + }, + { + "epoch": 7.73, + "grad_norm": 0.44407832622528076, + "learning_rate": 2.2396706883898676e-05, + "loss": 3.1729, + "step": 5275 + }, + { + "epoch": 7.73, + "grad_norm": 0.6070168614387512, + "learning_rate": 2.236713377616936e-05, + "loss": 3.1218, + "step": 5276 + }, + { + "epoch": 7.73, + "grad_norm": 0.43469035625457764, + "learning_rate": 2.2337577747150353e-05, + "loss": 3.1484, + "step": 5277 + }, + { + "epoch": 7.73, + "grad_norm": 0.3631550967693329, + "learning_rate": 2.2308038803343755e-05, + "loss": 3.1192, + "step": 5278 + }, + { + "epoch": 7.73, + "grad_norm": 0.45660415291786194, + "learning_rate": 2.227851695124796e-05, + "loss": 3.1704, + "step": 5279 + }, + { + "epoch": 7.73, + "grad_norm": 0.3175928592681885, + "learning_rate": 2.224901219735758e-05, + "loss": 3.1534, + "step": 5280 + }, + { + "epoch": 7.73, + "grad_norm": 0.6207453608512878, + "learning_rate": 2.221952454816344e-05, + "loss": 3.191, + "step": 5281 + }, + { + "epoch": 7.74, + "grad_norm": 0.5500271320343018, + "learning_rate": 2.219005401015264e-05, + "loss": 3.1639, + "step": 5282 + }, + { + "epoch": 7.74, + "grad_norm": 0.5747382044792175, + "learning_rate": 2.21606005898085e-05, + "loss": 3.132, + "step": 5283 + }, + { + "epoch": 7.74, + "grad_norm": 0.26599442958831787, + "learning_rate": 2.2131164293610572e-05, + "loss": 3.1491, + "step": 5284 + }, + { + "epoch": 7.74, + "grad_norm": 0.46995365619659424, + "learning_rate": 2.2101745128034656e-05, + "loss": 3.1383, + "step": 5285 + }, + { + "epoch": 7.74, + "grad_norm": 0.23462985455989838, + "learning_rate": 2.207234309955275e-05, + "loss": 3.1604, + "step": 5286 + }, + { + "epoch": 7.74, + "grad_norm": 0.45813092589378357, + "learning_rate": 2.2042958214633115e-05, + "loss": 3.1589, + "step": 5287 + }, + { + "epoch": 7.74, + "grad_norm": 0.48195019364356995, + "learning_rate": 2.2013590479740232e-05, + "loss": 3.1074, + "step": 5288 + }, + { + "epoch": 7.75, + "grad_norm": 0.9091559052467346, + "learning_rate": 2.1984239901334802e-05, + "loss": 3.1518, + "step": 5289 + }, + { + "epoch": 7.75, + "grad_norm": 0.9648109078407288, + "learning_rate": 2.195490648587375e-05, + "loss": 3.2077, + "step": 5290 + }, + { + "epoch": 7.75, + "grad_norm": 0.6132333874702454, + "learning_rate": 2.1925590239810235e-05, + "loss": 3.1481, + "step": 5291 + }, + { + "epoch": 7.75, + "grad_norm": 0.4865446984767914, + "learning_rate": 2.1896291169593643e-05, + "loss": 3.1402, + "step": 5292 + }, + { + "epoch": 7.75, + "grad_norm": 0.561164140701294, + "learning_rate": 2.186700928166956e-05, + "loss": 3.1597, + "step": 5293 + }, + { + "epoch": 7.75, + "grad_norm": 0.46440842747688293, + "learning_rate": 2.1837744582479814e-05, + "loss": 3.1493, + "step": 5294 + }, + { + "epoch": 7.75, + "grad_norm": 0.5474674701690674, + "learning_rate": 2.1808497078462442e-05, + "loss": 3.1622, + "step": 5295 + }, + { + "epoch": 7.76, + "grad_norm": 0.304200679063797, + "learning_rate": 2.1779266776051688e-05, + "loss": 3.1756, + "step": 5296 + }, + { + "epoch": 7.76, + "grad_norm": 0.534535825252533, + "learning_rate": 2.175005368167804e-05, + "loss": 3.2078, + "step": 5297 + }, + { + "epoch": 7.76, + "grad_norm": 0.7717536687850952, + "learning_rate": 2.17208578017682e-05, + "loss": 3.1603, + "step": 5298 + }, + { + "epoch": 7.76, + "grad_norm": 0.8810925483703613, + "learning_rate": 2.169167914274498e-05, + "loss": 3.1335, + "step": 5299 + }, + { + "epoch": 7.76, + "grad_norm": 0.9111099243164062, + "learning_rate": 2.166251771102761e-05, + "loss": 3.1247, + "step": 5300 + }, + { + "epoch": 7.76, + "grad_norm": 0.5138559937477112, + "learning_rate": 2.163337351303132e-05, + "loss": 3.183, + "step": 5301 + }, + { + "epoch": 7.77, + "grad_norm": 0.5875454545021057, + "learning_rate": 2.1604246555167638e-05, + "loss": 3.1424, + "step": 5302 + }, + { + "epoch": 7.77, + "grad_norm": 0.35727164149284363, + "learning_rate": 2.157513684384439e-05, + "loss": 3.1641, + "step": 5303 + }, + { + "epoch": 7.77, + "grad_norm": 0.720236599445343, + "learning_rate": 2.1546044385465426e-05, + "loss": 3.1459, + "step": 5304 + }, + { + "epoch": 7.77, + "grad_norm": 0.49461638927459717, + "learning_rate": 2.151696918643089e-05, + "loss": 3.1613, + "step": 5305 + }, + { + "epoch": 7.77, + "grad_norm": 0.5594753623008728, + "learning_rate": 2.1487911253137217e-05, + "loss": 3.1251, + "step": 5306 + }, + { + "epoch": 7.77, + "grad_norm": 1.0345263481140137, + "learning_rate": 2.1458870591976877e-05, + "loss": 3.1507, + "step": 5307 + }, + { + "epoch": 7.77, + "grad_norm": 0.31530436873435974, + "learning_rate": 2.1429847209338628e-05, + "loss": 3.1615, + "step": 5308 + }, + { + "epoch": 7.78, + "grad_norm": 0.3019442558288574, + "learning_rate": 2.1400841111607483e-05, + "loss": 3.1626, + "step": 5309 + }, + { + "epoch": 7.78, + "grad_norm": 0.40825822949409485, + "learning_rate": 2.137185230516453e-05, + "loss": 3.1289, + "step": 5310 + }, + { + "epoch": 7.78, + "grad_norm": 0.4319787323474884, + "learning_rate": 2.134288079638712e-05, + "loss": 3.1428, + "step": 5311 + }, + { + "epoch": 7.78, + "grad_norm": 0.5881616473197937, + "learning_rate": 2.131392659164879e-05, + "loss": 3.1675, + "step": 5312 + }, + { + "epoch": 7.78, + "grad_norm": 0.4166703224182129, + "learning_rate": 2.128498969731929e-05, + "loss": 3.1611, + "step": 5313 + }, + { + "epoch": 7.78, + "grad_norm": 0.7921746373176575, + "learning_rate": 2.125607011976454e-05, + "loss": 3.1281, + "step": 5314 + }, + { + "epoch": 7.78, + "grad_norm": 0.454723060131073, + "learning_rate": 2.1227167865346632e-05, + "loss": 3.1568, + "step": 5315 + }, + { + "epoch": 7.79, + "grad_norm": 0.49115943908691406, + "learning_rate": 2.1198282940423887e-05, + "loss": 3.1556, + "step": 5316 + }, + { + "epoch": 7.79, + "grad_norm": 0.4243347644805908, + "learning_rate": 2.1169415351350786e-05, + "loss": 3.1701, + "step": 5317 + }, + { + "epoch": 7.79, + "grad_norm": 0.9867798089981079, + "learning_rate": 2.1140565104478004e-05, + "loss": 3.1729, + "step": 5318 + }, + { + "epoch": 7.79, + "grad_norm": 1.0699418783187866, + "learning_rate": 2.1111732206152424e-05, + "loss": 3.1711, + "step": 5319 + }, + { + "epoch": 7.79, + "grad_norm": 0.7602143287658691, + "learning_rate": 2.1082916662717055e-05, + "loss": 3.1824, + "step": 5320 + }, + { + "epoch": 7.79, + "grad_norm": 0.7598912715911865, + "learning_rate": 2.1054118480511154e-05, + "loss": 3.1232, + "step": 5321 + }, + { + "epoch": 7.79, + "grad_norm": 0.7849280834197998, + "learning_rate": 2.1025337665870104e-05, + "loss": 3.1169, + "step": 5322 + }, + { + "epoch": 7.8, + "grad_norm": 0.5915687084197998, + "learning_rate": 2.0996574225125532e-05, + "loss": 3.1269, + "step": 5323 + }, + { + "epoch": 7.8, + "grad_norm": 0.9598039984703064, + "learning_rate": 2.096782816460513e-05, + "loss": 3.1504, + "step": 5324 + }, + { + "epoch": 7.8, + "grad_norm": 0.6661960482597351, + "learning_rate": 2.0939099490632896e-05, + "loss": 3.1284, + "step": 5325 + }, + { + "epoch": 7.8, + "grad_norm": 0.4720488488674164, + "learning_rate": 2.0910388209528964e-05, + "loss": 3.1584, + "step": 5326 + }, + { + "epoch": 7.8, + "grad_norm": 0.6807810664176941, + "learning_rate": 2.0881694327609536e-05, + "loss": 3.1586, + "step": 5327 + }, + { + "epoch": 7.8, + "grad_norm": 0.7132989764213562, + "learning_rate": 2.085301785118716e-05, + "loss": 3.1526, + "step": 5328 + }, + { + "epoch": 7.81, + "grad_norm": 0.4347662031650543, + "learning_rate": 2.0824358786570453e-05, + "loss": 3.1645, + "step": 5329 + }, + { + "epoch": 7.81, + "grad_norm": 1.1308764219284058, + "learning_rate": 2.0795717140064163e-05, + "loss": 3.162, + "step": 5330 + }, + { + "epoch": 7.81, + "grad_norm": 0.759826123714447, + "learning_rate": 2.076709291796931e-05, + "loss": 3.184, + "step": 5331 + }, + { + "epoch": 7.81, + "grad_norm": 0.6993570923805237, + "learning_rate": 2.0738486126583056e-05, + "loss": 3.1636, + "step": 5332 + }, + { + "epoch": 7.81, + "grad_norm": 0.5368728637695312, + "learning_rate": 2.0709896772198612e-05, + "loss": 3.1551, + "step": 5333 + }, + { + "epoch": 7.81, + "grad_norm": 0.5664252042770386, + "learning_rate": 2.0681324861105556e-05, + "loss": 3.1349, + "step": 5334 + }, + { + "epoch": 7.81, + "grad_norm": 0.6612221598625183, + "learning_rate": 2.0652770399589427e-05, + "loss": 3.1495, + "step": 5335 + }, + { + "epoch": 7.82, + "grad_norm": 0.48035261034965515, + "learning_rate": 2.0624233393932024e-05, + "loss": 3.1624, + "step": 5336 + }, + { + "epoch": 7.82, + "grad_norm": 0.36596471071243286, + "learning_rate": 2.059571385041138e-05, + "loss": 3.1674, + "step": 5337 + }, + { + "epoch": 7.82, + "grad_norm": 0.6980510950088501, + "learning_rate": 2.056721177530151e-05, + "loss": 3.1126, + "step": 5338 + }, + { + "epoch": 7.82, + "grad_norm": 0.567153811454773, + "learning_rate": 2.0538727174872684e-05, + "loss": 3.1449, + "step": 5339 + }, + { + "epoch": 7.82, + "grad_norm": 0.6353252530097961, + "learning_rate": 2.0510260055391418e-05, + "loss": 3.1646, + "step": 5340 + }, + { + "epoch": 7.82, + "grad_norm": 0.6695141792297363, + "learning_rate": 2.0481810423120196e-05, + "loss": 3.1426, + "step": 5341 + }, + { + "epoch": 7.82, + "grad_norm": 1.0574151277542114, + "learning_rate": 2.045337828431775e-05, + "loss": 3.1435, + "step": 5342 + }, + { + "epoch": 7.83, + "grad_norm": 0.3314790725708008, + "learning_rate": 2.0424963645239027e-05, + "loss": 3.1263, + "step": 5343 + }, + { + "epoch": 7.83, + "grad_norm": 0.708838939666748, + "learning_rate": 2.0396566512134996e-05, + "loss": 3.178, + "step": 5344 + }, + { + "epoch": 7.83, + "grad_norm": 0.670356810092926, + "learning_rate": 2.036818689125284e-05, + "loss": 3.1153, + "step": 5345 + }, + { + "epoch": 7.83, + "grad_norm": 0.6451178193092346, + "learning_rate": 2.0339824788835916e-05, + "loss": 3.1816, + "step": 5346 + }, + { + "epoch": 7.83, + "grad_norm": 0.5722768306732178, + "learning_rate": 2.031148021112367e-05, + "loss": 3.1584, + "step": 5347 + }, + { + "epoch": 7.83, + "grad_norm": 0.8647962808609009, + "learning_rate": 2.028315316435173e-05, + "loss": 3.118, + "step": 5348 + }, + { + "epoch": 7.83, + "grad_norm": 0.5145653486251831, + "learning_rate": 2.0254843654751852e-05, + "loss": 3.1143, + "step": 5349 + }, + { + "epoch": 7.84, + "grad_norm": 0.7415019273757935, + "learning_rate": 2.0226551688551954e-05, + "loss": 3.1961, + "step": 5350 + }, + { + "epoch": 7.84, + "grad_norm": 0.6742720603942871, + "learning_rate": 2.0198277271976052e-05, + "loss": 3.107, + "step": 5351 + }, + { + "epoch": 7.84, + "grad_norm": 0.5243716835975647, + "learning_rate": 2.017002041124435e-05, + "loss": 3.1568, + "step": 5352 + }, + { + "epoch": 7.84, + "grad_norm": 0.6386037468910217, + "learning_rate": 2.0141781112573154e-05, + "loss": 3.1501, + "step": 5353 + }, + { + "epoch": 7.84, + "grad_norm": 0.9167563915252686, + "learning_rate": 2.011355938217496e-05, + "loss": 3.1591, + "step": 5354 + }, + { + "epoch": 7.84, + "grad_norm": 0.3091890811920166, + "learning_rate": 2.0085355226258285e-05, + "loss": 3.1714, + "step": 5355 + }, + { + "epoch": 7.85, + "grad_norm": 0.6145365834236145, + "learning_rate": 2.00571686510279e-05, + "loss": 3.1247, + "step": 5356 + }, + { + "epoch": 7.85, + "grad_norm": 0.3485010862350464, + "learning_rate": 2.00289996626847e-05, + "loss": 3.1696, + "step": 5357 + }, + { + "epoch": 7.85, + "grad_norm": 0.5586562156677246, + "learning_rate": 2.0000848267425575e-05, + "loss": 3.1249, + "step": 5358 + }, + { + "epoch": 7.85, + "grad_norm": 0.45884886384010315, + "learning_rate": 1.9972714471443733e-05, + "loss": 3.1264, + "step": 5359 + }, + { + "epoch": 7.85, + "grad_norm": 0.3801732361316681, + "learning_rate": 1.9944598280928408e-05, + "loss": 3.1699, + "step": 5360 + }, + { + "epoch": 7.85, + "grad_norm": 0.5741209983825684, + "learning_rate": 1.9916499702064894e-05, + "loss": 3.1625, + "step": 5361 + }, + { + "epoch": 7.85, + "grad_norm": 0.23692893981933594, + "learning_rate": 1.9888418741034787e-05, + "loss": 3.1282, + "step": 5362 + }, + { + "epoch": 7.86, + "grad_norm": 0.23494169116020203, + "learning_rate": 1.9860355404015696e-05, + "loss": 3.1757, + "step": 5363 + }, + { + "epoch": 7.86, + "grad_norm": 0.5047478675842285, + "learning_rate": 1.9832309697181294e-05, + "loss": 3.1371, + "step": 5364 + }, + { + "epoch": 7.86, + "grad_norm": 0.39867815375328064, + "learning_rate": 1.9804281626701525e-05, + "loss": 3.1522, + "step": 5365 + }, + { + "epoch": 7.86, + "grad_norm": 0.4054205119609833, + "learning_rate": 1.977627119874237e-05, + "loss": 3.1453, + "step": 5366 + }, + { + "epoch": 7.86, + "grad_norm": 0.37046530842781067, + "learning_rate": 1.9748278419465872e-05, + "loss": 3.1295, + "step": 5367 + }, + { + "epoch": 7.86, + "grad_norm": 0.4440865218639374, + "learning_rate": 1.972030329503035e-05, + "loss": 3.1436, + "step": 5368 + }, + { + "epoch": 7.86, + "grad_norm": 0.25685054063796997, + "learning_rate": 1.9692345831590064e-05, + "loss": 3.147, + "step": 5369 + }, + { + "epoch": 7.87, + "grad_norm": 0.423784077167511, + "learning_rate": 1.966440603529549e-05, + "loss": 3.161, + "step": 5370 + }, + { + "epoch": 7.87, + "grad_norm": 0.407570481300354, + "learning_rate": 1.963648391229321e-05, + "loss": 3.1711, + "step": 5371 + }, + { + "epoch": 7.87, + "grad_norm": 1.7502979040145874, + "learning_rate": 1.9608579468725895e-05, + "loss": 3.1717, + "step": 5372 + }, + { + "epoch": 7.87, + "grad_norm": 0.37750712037086487, + "learning_rate": 1.958069271073234e-05, + "loss": 3.1518, + "step": 5373 + }, + { + "epoch": 7.87, + "grad_norm": 0.44295188784599304, + "learning_rate": 1.9552823644447437e-05, + "loss": 3.1532, + "step": 5374 + }, + { + "epoch": 7.87, + "grad_norm": 0.6612734794616699, + "learning_rate": 1.9524972276002207e-05, + "loss": 3.1507, + "step": 5375 + }, + { + "epoch": 7.88, + "grad_norm": 0.5701128244400024, + "learning_rate": 1.9497138611523748e-05, + "loss": 3.1726, + "step": 5376 + }, + { + "epoch": 7.88, + "grad_norm": 0.48842257261276245, + "learning_rate": 1.9469322657135293e-05, + "loss": 3.1591, + "step": 5377 + }, + { + "epoch": 7.88, + "grad_norm": 0.40046364068984985, + "learning_rate": 1.944152441895616e-05, + "loss": 3.1994, + "step": 5378 + }, + { + "epoch": 7.88, + "grad_norm": 0.62181556224823, + "learning_rate": 1.9413743903101776e-05, + "loss": 3.1446, + "step": 5379 + }, + { + "epoch": 7.88, + "grad_norm": 0.7460499405860901, + "learning_rate": 1.9385981115683672e-05, + "loss": 3.1688, + "step": 5380 + }, + { + "epoch": 7.88, + "grad_norm": 0.2939889132976532, + "learning_rate": 1.9358236062809464e-05, + "loss": 3.1436, + "step": 5381 + }, + { + "epoch": 7.88, + "grad_norm": 0.6719468832015991, + "learning_rate": 1.9330508750582897e-05, + "loss": 3.1482, + "step": 5382 + }, + { + "epoch": 7.89, + "grad_norm": 0.41709572076797485, + "learning_rate": 1.930279918510377e-05, + "loss": 3.1564, + "step": 5383 + }, + { + "epoch": 7.89, + "grad_norm": 0.5399413704872131, + "learning_rate": 1.9275107372468014e-05, + "loss": 3.1684, + "step": 5384 + }, + { + "epoch": 7.89, + "grad_norm": 0.6927580237388611, + "learning_rate": 1.9247433318767648e-05, + "loss": 3.1473, + "step": 5385 + }, + { + "epoch": 7.89, + "grad_norm": 0.5529474020004272, + "learning_rate": 1.921977703009077e-05, + "loss": 3.1235, + "step": 5386 + }, + { + "epoch": 7.89, + "grad_norm": 1.1675437688827515, + "learning_rate": 1.9192138512521584e-05, + "loss": 3.1587, + "step": 5387 + }, + { + "epoch": 7.89, + "grad_norm": 0.3985271751880646, + "learning_rate": 1.916451777214039e-05, + "loss": 3.1479, + "step": 5388 + }, + { + "epoch": 7.89, + "grad_norm": 0.8106251358985901, + "learning_rate": 1.913691481502351e-05, + "loss": 3.1356, + "step": 5389 + }, + { + "epoch": 7.9, + "grad_norm": 0.767872154712677, + "learning_rate": 1.910932964724347e-05, + "loss": 3.1462, + "step": 5390 + }, + { + "epoch": 7.9, + "grad_norm": 0.41784894466400146, + "learning_rate": 1.9081762274868832e-05, + "loss": 3.1637, + "step": 5391 + }, + { + "epoch": 7.9, + "grad_norm": 0.8937169313430786, + "learning_rate": 1.9054212703964147e-05, + "loss": 3.1743, + "step": 5392 + }, + { + "epoch": 7.9, + "grad_norm": 3.9727253913879395, + "learning_rate": 1.9026680940590224e-05, + "loss": 3.1128, + "step": 5393 + }, + { + "epoch": 7.9, + "grad_norm": 0.4264329969882965, + "learning_rate": 1.8999166990803864e-05, + "loss": 3.1802, + "step": 5394 + }, + { + "epoch": 7.9, + "grad_norm": 0.4454759657382965, + "learning_rate": 1.8971670860657887e-05, + "loss": 3.1421, + "step": 5395 + }, + { + "epoch": 7.9, + "grad_norm": 0.8566815257072449, + "learning_rate": 1.8944192556201266e-05, + "loss": 3.1259, + "step": 5396 + }, + { + "epoch": 7.91, + "grad_norm": 0.42065227031707764, + "learning_rate": 1.8916732083479138e-05, + "loss": 3.1286, + "step": 5397 + }, + { + "epoch": 7.91, + "grad_norm": 1.0363951921463013, + "learning_rate": 1.888928944853251e-05, + "loss": 3.1475, + "step": 5398 + }, + { + "epoch": 7.91, + "grad_norm": 0.8358655571937561, + "learning_rate": 1.886186465739861e-05, + "loss": 3.1292, + "step": 5399 + }, + { + "epoch": 7.91, + "grad_norm": 0.5414959788322449, + "learning_rate": 1.8834457716110778e-05, + "loss": 3.1235, + "step": 5400 + }, + { + "epoch": 7.91, + "grad_norm": 0.6389888525009155, + "learning_rate": 1.8807068630698267e-05, + "loss": 3.1118, + "step": 5401 + }, + { + "epoch": 7.91, + "grad_norm": 0.5987576246261597, + "learning_rate": 1.8779697407186526e-05, + "loss": 3.1534, + "step": 5402 + }, + { + "epoch": 7.92, + "grad_norm": 0.9241575598716736, + "learning_rate": 1.8752344051597036e-05, + "loss": 3.207, + "step": 5403 + }, + { + "epoch": 7.92, + "grad_norm": 0.5730569958686829, + "learning_rate": 1.8725008569947365e-05, + "loss": 3.1383, + "step": 5404 + }, + { + "epoch": 7.92, + "grad_norm": 0.43397852778434753, + "learning_rate": 1.8697690968251126e-05, + "loss": 3.1723, + "step": 5405 + }, + { + "epoch": 7.92, + "grad_norm": 0.4378036558628082, + "learning_rate": 1.8670391252518017e-05, + "loss": 3.1829, + "step": 5406 + }, + { + "epoch": 7.92, + "grad_norm": 0.8141224980354309, + "learning_rate": 1.864310942875377e-05, + "loss": 3.1639, + "step": 5407 + }, + { + "epoch": 7.92, + "grad_norm": 0.25414714217185974, + "learning_rate": 1.8615845502960228e-05, + "loss": 3.1672, + "step": 5408 + }, + { + "epoch": 7.92, + "eval_loss": 3.008017063140869, + "eval_runtime": 1168.0951, + "eval_samples_per_second": 14.675, + "eval_steps_per_second": 0.918, + "step": 5408 + }, + { + "epoch": 7.92, + "grad_norm": 0.5402032136917114, + "learning_rate": 1.858859948113526e-05, + "loss": 3.1674, + "step": 5409 + }, + { + "epoch": 7.93, + "grad_norm": 0.5880429148674011, + "learning_rate": 1.8561371369272818e-05, + "loss": 3.1517, + "step": 5410 + }, + { + "epoch": 7.93, + "grad_norm": 0.44701412320137024, + "learning_rate": 1.8534161173362897e-05, + "loss": 3.1729, + "step": 5411 + }, + { + "epoch": 7.93, + "grad_norm": 0.36186301708221436, + "learning_rate": 1.8506968899391554e-05, + "loss": 3.1521, + "step": 5412 + }, + { + "epoch": 7.93, + "grad_norm": 0.35009482502937317, + "learning_rate": 1.847979455334091e-05, + "loss": 3.1199, + "step": 5413 + }, + { + "epoch": 7.93, + "grad_norm": 0.34212368726730347, + "learning_rate": 1.8452638141189138e-05, + "loss": 3.1637, + "step": 5414 + }, + { + "epoch": 7.93, + "grad_norm": 0.35525619983673096, + "learning_rate": 1.8425499668910472e-05, + "loss": 3.1613, + "step": 5415 + }, + { + "epoch": 7.93, + "grad_norm": 0.9575212597846985, + "learning_rate": 1.8398379142475196e-05, + "loss": 3.1801, + "step": 5416 + }, + { + "epoch": 7.94, + "grad_norm": 0.2845044434070587, + "learning_rate": 1.837127656784966e-05, + "loss": 3.1833, + "step": 5417 + }, + { + "epoch": 7.94, + "grad_norm": 0.29446759819984436, + "learning_rate": 1.834419195099617e-05, + "loss": 3.1385, + "step": 5418 + }, + { + "epoch": 7.94, + "grad_norm": 0.3464059829711914, + "learning_rate": 1.831712529787325e-05, + "loss": 3.1476, + "step": 5419 + }, + { + "epoch": 7.94, + "grad_norm": 0.24105912446975708, + "learning_rate": 1.829007661443538e-05, + "loss": 3.1684, + "step": 5420 + }, + { + "epoch": 7.94, + "grad_norm": 0.4570894241333008, + "learning_rate": 1.8263045906633002e-05, + "loss": 3.1824, + "step": 5421 + }, + { + "epoch": 7.94, + "grad_norm": 0.5457885265350342, + "learning_rate": 1.8236033180412793e-05, + "loss": 3.1701, + "step": 5422 + }, + { + "epoch": 7.94, + "grad_norm": 0.29805809259414673, + "learning_rate": 1.8209038441717306e-05, + "loss": 3.1578, + "step": 5423 + }, + { + "epoch": 7.95, + "grad_norm": 0.47300076484680176, + "learning_rate": 1.8182061696485207e-05, + "loss": 3.1372, + "step": 5424 + }, + { + "epoch": 7.95, + "grad_norm": 0.7307349443435669, + "learning_rate": 1.815510295065126e-05, + "loss": 3.1615, + "step": 5425 + }, + { + "epoch": 7.95, + "grad_norm": 0.28895077109336853, + "learning_rate": 1.8128162210146136e-05, + "loss": 3.1576, + "step": 5426 + }, + { + "epoch": 7.95, + "grad_norm": 0.4990285634994507, + "learning_rate": 1.8101239480896626e-05, + "loss": 3.1346, + "step": 5427 + }, + { + "epoch": 7.95, + "grad_norm": 0.42847058176994324, + "learning_rate": 1.8074334768825606e-05, + "loss": 3.1791, + "step": 5428 + }, + { + "epoch": 7.95, + "grad_norm": 0.3061131536960602, + "learning_rate": 1.804744807985188e-05, + "loss": 3.1364, + "step": 5429 + }, + { + "epoch": 7.96, + "grad_norm": 0.4757770001888275, + "learning_rate": 1.8020579419890326e-05, + "loss": 3.1907, + "step": 5430 + }, + { + "epoch": 7.96, + "grad_norm": 0.4148907959461212, + "learning_rate": 1.799372879485196e-05, + "loss": 3.2014, + "step": 5431 + }, + { + "epoch": 7.96, + "grad_norm": 0.30733799934387207, + "learning_rate": 1.7966896210643634e-05, + "loss": 3.1103, + "step": 5432 + }, + { + "epoch": 7.96, + "grad_norm": 0.40864160656929016, + "learning_rate": 1.794008167316836e-05, + "loss": 3.1728, + "step": 5433 + }, + { + "epoch": 7.96, + "grad_norm": 0.5808351039886475, + "learning_rate": 1.7913285188325235e-05, + "loss": 3.1226, + "step": 5434 + }, + { + "epoch": 7.96, + "grad_norm": 0.5464041829109192, + "learning_rate": 1.7886506762009225e-05, + "loss": 3.1737, + "step": 5435 + }, + { + "epoch": 7.96, + "grad_norm": 0.587399959564209, + "learning_rate": 1.7859746400111422e-05, + "loss": 3.1581, + "step": 5436 + }, + { + "epoch": 7.97, + "grad_norm": 0.36822310090065, + "learning_rate": 1.7833004108518925e-05, + "loss": 3.1762, + "step": 5437 + }, + { + "epoch": 7.97, + "grad_norm": 2.8975210189819336, + "learning_rate": 1.7806279893114875e-05, + "loss": 3.1847, + "step": 5438 + }, + { + "epoch": 7.97, + "grad_norm": 0.3770262897014618, + "learning_rate": 1.77795737597784e-05, + "loss": 3.1737, + "step": 5439 + }, + { + "epoch": 7.97, + "grad_norm": 0.4018249213695526, + "learning_rate": 1.7752885714384694e-05, + "loss": 3.1248, + "step": 5440 + }, + { + "epoch": 7.97, + "grad_norm": 0.26656123995780945, + "learning_rate": 1.772621576280492e-05, + "loss": 3.1343, + "step": 5441 + }, + { + "epoch": 7.97, + "grad_norm": 0.29161426424980164, + "learning_rate": 1.7699563910906314e-05, + "loss": 3.1363, + "step": 5442 + }, + { + "epoch": 7.97, + "grad_norm": 0.4624899625778198, + "learning_rate": 1.767293016455209e-05, + "loss": 3.1913, + "step": 5443 + }, + { + "epoch": 7.98, + "grad_norm": 0.529693067073822, + "learning_rate": 1.7646314529601505e-05, + "loss": 3.1301, + "step": 5444 + }, + { + "epoch": 7.98, + "grad_norm": 0.25641757249832153, + "learning_rate": 1.7619717011909842e-05, + "loss": 3.1114, + "step": 5445 + }, + { + "epoch": 7.98, + "grad_norm": 0.26415735483169556, + "learning_rate": 1.7593137617328282e-05, + "loss": 3.1425, + "step": 5446 + }, + { + "epoch": 7.98, + "grad_norm": 0.6229835748672485, + "learning_rate": 1.7566576351704224e-05, + "loss": 3.1519, + "step": 5447 + }, + { + "epoch": 7.98, + "grad_norm": 0.6144125461578369, + "learning_rate": 1.7540033220880936e-05, + "loss": 3.1652, + "step": 5448 + }, + { + "epoch": 7.98, + "grad_norm": 0.30449098348617554, + "learning_rate": 1.7513508230697673e-05, + "loss": 3.1811, + "step": 5449 + }, + { + "epoch": 7.98, + "grad_norm": 0.4084562361240387, + "learning_rate": 1.748700138698982e-05, + "loss": 3.185, + "step": 5450 + }, + { + "epoch": 7.99, + "grad_norm": 0.6092720627784729, + "learning_rate": 1.7460512695588704e-05, + "loss": 3.1706, + "step": 5451 + }, + { + "epoch": 7.99, + "grad_norm": 0.5944820046424866, + "learning_rate": 1.7434042162321597e-05, + "loss": 3.165, + "step": 5452 + }, + { + "epoch": 7.99, + "grad_norm": 0.3383338749408722, + "learning_rate": 1.7407589793011912e-05, + "loss": 3.171, + "step": 5453 + }, + { + "epoch": 7.99, + "grad_norm": 0.7481070160865784, + "learning_rate": 1.738115559347898e-05, + "loss": 3.1244, + "step": 5454 + }, + { + "epoch": 7.99, + "grad_norm": 0.6213645339012146, + "learning_rate": 1.735473956953808e-05, + "loss": 3.198, + "step": 5455 + }, + { + "epoch": 7.99, + "grad_norm": 0.5867860913276672, + "learning_rate": 1.7328341727000664e-05, + "loss": 3.1474, + "step": 5456 + }, + { + "epoch": 8.0, + "grad_norm": 0.4407004714012146, + "learning_rate": 1.7301962071674004e-05, + "loss": 3.1451, + "step": 5457 + }, + { + "epoch": 8.0, + "grad_norm": 0.6691040992736816, + "learning_rate": 1.7275600609361443e-05, + "loss": 3.1377, + "step": 5458 + }, + { + "epoch": 8.0, + "grad_norm": 0.63309645652771, + "learning_rate": 1.72492573458624e-05, + "loss": 3.1297, + "step": 5459 + }, + { + "epoch": 8.0, + "grad_norm": 0.481118768453598, + "learning_rate": 1.722293228697216e-05, + "loss": 3.2125, + "step": 5460 + }, + { + "epoch": 8.0, + "grad_norm": 0.5907300710678101, + "learning_rate": 1.7196625438482027e-05, + "loss": 3.1457, + "step": 5461 + }, + { + "epoch": 8.0, + "grad_norm": 0.7083212733268738, + "learning_rate": 1.7170336806179433e-05, + "loss": 3.1235, + "step": 5462 + }, + { + "epoch": 8.0, + "grad_norm": 0.9391943216323853, + "learning_rate": 1.7144066395847615e-05, + "loss": 3.1399, + "step": 5463 + }, + { + "epoch": 8.01, + "grad_norm": 0.5589471459388733, + "learning_rate": 1.7117814213265893e-05, + "loss": 3.129, + "step": 5464 + }, + { + "epoch": 8.01, + "grad_norm": 0.7485381960868835, + "learning_rate": 1.7091580264209627e-05, + "loss": 3.1755, + "step": 5465 + }, + { + "epoch": 8.01, + "grad_norm": 0.8377997279167175, + "learning_rate": 1.706536455445007e-05, + "loss": 3.159, + "step": 5466 + }, + { + "epoch": 8.01, + "grad_norm": 0.469113290309906, + "learning_rate": 1.7039167089754492e-05, + "loss": 3.1715, + "step": 5467 + }, + { + "epoch": 8.01, + "grad_norm": 0.610246479511261, + "learning_rate": 1.701298787588619e-05, + "loss": 3.1326, + "step": 5468 + }, + { + "epoch": 8.01, + "grad_norm": 0.643669068813324, + "learning_rate": 1.698682691860439e-05, + "loss": 3.1593, + "step": 5469 + }, + { + "epoch": 8.01, + "grad_norm": 0.5371613502502441, + "learning_rate": 1.6960684223664337e-05, + "loss": 3.1264, + "step": 5470 + }, + { + "epoch": 8.0, + "grad_norm": 0.4291907250881195, + "learning_rate": 1.6934559796817263e-05, + "loss": 3.1593, + "step": 5471 + }, + { + "epoch": 8.0, + "grad_norm": 0.6830047965049744, + "learning_rate": 1.690845364381034e-05, + "loss": 3.1404, + "step": 5472 + }, + { + "epoch": 8.0, + "grad_norm": 0.419023722410202, + "learning_rate": 1.6882365770386766e-05, + "loss": 3.1497, + "step": 5473 + }, + { + "epoch": 8.01, + "grad_norm": 0.6376973986625671, + "learning_rate": 1.685629618228569e-05, + "loss": 3.1803, + "step": 5474 + }, + { + "epoch": 8.01, + "grad_norm": 0.814529299736023, + "learning_rate": 1.6830244885242252e-05, + "loss": 3.1458, + "step": 5475 + }, + { + "epoch": 8.01, + "grad_norm": 0.5563649535179138, + "learning_rate": 1.6804211884987563e-05, + "loss": 3.1326, + "step": 5476 + }, + { + "epoch": 8.01, + "grad_norm": 0.8076274394989014, + "learning_rate": 1.677819718724871e-05, + "loss": 3.1377, + "step": 5477 + }, + { + "epoch": 8.01, + "grad_norm": 0.39461883902549744, + "learning_rate": 1.675220079774874e-05, + "loss": 3.1151, + "step": 5478 + }, + { + "epoch": 8.01, + "grad_norm": 0.4176673889160156, + "learning_rate": 1.6726222722206718e-05, + "loss": 3.1486, + "step": 5479 + }, + { + "epoch": 8.01, + "grad_norm": 0.4115258455276489, + "learning_rate": 1.6700262966337578e-05, + "loss": 3.1341, + "step": 5480 + }, + { + "epoch": 8.02, + "grad_norm": 0.7755107879638672, + "learning_rate": 1.6674321535852343e-05, + "loss": 3.1913, + "step": 5481 + }, + { + "epoch": 8.02, + "grad_norm": 0.29228803515434265, + "learning_rate": 1.664839843645799e-05, + "loss": 3.1166, + "step": 5482 + }, + { + "epoch": 8.02, + "grad_norm": 0.7680675983428955, + "learning_rate": 1.662249367385732e-05, + "loss": 3.1529, + "step": 5483 + }, + { + "epoch": 8.02, + "grad_norm": 1.2723925113677979, + "learning_rate": 1.6596607253749308e-05, + "loss": 3.148, + "step": 5484 + }, + { + "epoch": 8.02, + "grad_norm": 0.5549447536468506, + "learning_rate": 1.6570739181828775e-05, + "loss": 3.1678, + "step": 5485 + }, + { + "epoch": 8.02, + "grad_norm": 0.6857723593711853, + "learning_rate": 1.654488946378645e-05, + "loss": 3.2432, + "step": 5486 + }, + { + "epoch": 8.02, + "grad_norm": 0.9772923588752747, + "learning_rate": 1.651905810530918e-05, + "loss": 3.1304, + "step": 5487 + }, + { + "epoch": 8.03, + "grad_norm": 0.33374878764152527, + "learning_rate": 1.649324511207969e-05, + "loss": 3.1321, + "step": 5488 + }, + { + "epoch": 8.03, + "grad_norm": 0.5361008644104004, + "learning_rate": 1.6467450489776582e-05, + "loss": 3.1434, + "step": 5489 + }, + { + "epoch": 8.03, + "grad_norm": 0.5644283890724182, + "learning_rate": 1.644167424407461e-05, + "loss": 3.1415, + "step": 5490 + }, + { + "epoch": 8.03, + "grad_norm": 0.3192185163497925, + "learning_rate": 1.641591638064429e-05, + "loss": 3.1669, + "step": 5491 + }, + { + "epoch": 8.03, + "grad_norm": 2.5619218349456787, + "learning_rate": 1.639017690515222e-05, + "loss": 3.1329, + "step": 5492 + }, + { + "epoch": 8.03, + "grad_norm": 0.5575476288795471, + "learning_rate": 1.6364455823260906e-05, + "loss": 3.1527, + "step": 5493 + }, + { + "epoch": 8.03, + "grad_norm": 0.6022101044654846, + "learning_rate": 1.6338753140628792e-05, + "loss": 3.1372, + "step": 5494 + }, + { + "epoch": 8.04, + "grad_norm": 0.5677317380905151, + "learning_rate": 1.631306886291033e-05, + "loss": 3.1356, + "step": 5495 + }, + { + "epoch": 8.04, + "grad_norm": 0.7927194833755493, + "learning_rate": 1.6287402995755864e-05, + "loss": 3.1394, + "step": 5496 + }, + { + "epoch": 8.04, + "grad_norm": 0.3976461887359619, + "learning_rate": 1.6261755544811728e-05, + "loss": 3.129, + "step": 5497 + }, + { + "epoch": 8.04, + "grad_norm": 0.6121003031730652, + "learning_rate": 1.6236126515720184e-05, + "loss": 3.1544, + "step": 5498 + }, + { + "epoch": 8.04, + "grad_norm": 0.4400195777416229, + "learning_rate": 1.621051591411944e-05, + "loss": 3.1182, + "step": 5499 + }, + { + "epoch": 8.04, + "grad_norm": 0.4340709149837494, + "learning_rate": 1.6184923745643667e-05, + "loss": 3.1473, + "step": 5500 + }, + { + "epoch": 8.05, + "grad_norm": 0.8094406723976135, + "learning_rate": 1.615935001592297e-05, + "loss": 3.1462, + "step": 5501 + }, + { + "epoch": 8.05, + "grad_norm": 0.40194961428642273, + "learning_rate": 1.6133794730583384e-05, + "loss": 3.1298, + "step": 5502 + }, + { + "epoch": 8.05, + "grad_norm": 0.44503289461135864, + "learning_rate": 1.6108257895246925e-05, + "loss": 3.1624, + "step": 5503 + }, + { + "epoch": 8.05, + "grad_norm": 0.23914110660552979, + "learning_rate": 1.6082739515531497e-05, + "loss": 3.1345, + "step": 5504 + }, + { + "epoch": 8.05, + "grad_norm": 0.37540099024772644, + "learning_rate": 1.6057239597051e-05, + "loss": 3.1487, + "step": 5505 + }, + { + "epoch": 8.05, + "grad_norm": 0.31591150164604187, + "learning_rate": 1.603175814541522e-05, + "loss": 3.1514, + "step": 5506 + }, + { + "epoch": 8.05, + "grad_norm": 0.5472672581672668, + "learning_rate": 1.6006295166229924e-05, + "loss": 3.1564, + "step": 5507 + }, + { + "epoch": 8.06, + "grad_norm": 0.42764073610305786, + "learning_rate": 1.598085066509679e-05, + "loss": 3.1504, + "step": 5508 + }, + { + "epoch": 8.06, + "grad_norm": 0.45837774872779846, + "learning_rate": 1.5955424647613438e-05, + "loss": 3.1613, + "step": 5509 + }, + { + "epoch": 8.06, + "grad_norm": 0.3727029263973236, + "learning_rate": 1.593001711937343e-05, + "loss": 3.1321, + "step": 5510 + }, + { + "epoch": 8.06, + "grad_norm": 0.38336122035980225, + "learning_rate": 1.5904628085966234e-05, + "loss": 3.1526, + "step": 5511 + }, + { + "epoch": 8.06, + "grad_norm": 0.33697742223739624, + "learning_rate": 1.587925755297728e-05, + "loss": 3.1447, + "step": 5512 + }, + { + "epoch": 8.06, + "grad_norm": 0.42843082547187805, + "learning_rate": 1.5853905525987932e-05, + "loss": 3.1343, + "step": 5513 + }, + { + "epoch": 8.06, + "grad_norm": 0.4407825469970703, + "learning_rate": 1.5828572010575426e-05, + "loss": 3.1385, + "step": 5514 + }, + { + "epoch": 8.07, + "grad_norm": 0.3060319125652313, + "learning_rate": 1.5803257012312956e-05, + "loss": 3.1189, + "step": 5515 + }, + { + "epoch": 8.07, + "grad_norm": 0.3679053783416748, + "learning_rate": 1.5777960536769743e-05, + "loss": 3.1458, + "step": 5516 + }, + { + "epoch": 8.07, + "grad_norm": 0.5479602813720703, + "learning_rate": 1.5752682589510747e-05, + "loss": 3.1543, + "step": 5517 + }, + { + "epoch": 8.07, + "grad_norm": 0.6712786555290222, + "learning_rate": 1.5727423176096955e-05, + "loss": 3.1429, + "step": 5518 + }, + { + "epoch": 8.07, + "grad_norm": 0.24910399317741394, + "learning_rate": 1.5702182302085343e-05, + "loss": 3.2076, + "step": 5519 + }, + { + "epoch": 8.07, + "grad_norm": 0.42461472749710083, + "learning_rate": 1.567695997302867e-05, + "loss": 3.1566, + "step": 5520 + }, + { + "epoch": 8.07, + "grad_norm": 0.2487383484840393, + "learning_rate": 1.5651756194475666e-05, + "loss": 3.1567, + "step": 5521 + }, + { + "epoch": 8.08, + "grad_norm": 0.602077841758728, + "learning_rate": 1.5626570971971068e-05, + "loss": 3.1662, + "step": 5522 + }, + { + "epoch": 8.08, + "grad_norm": 0.5867159962654114, + "learning_rate": 1.560140431105539e-05, + "loss": 3.1886, + "step": 5523 + }, + { + "epoch": 8.08, + "grad_norm": 0.2727794945240021, + "learning_rate": 1.5576256217265152e-05, + "loss": 3.127, + "step": 5524 + }, + { + "epoch": 8.08, + "grad_norm": 0.33053722977638245, + "learning_rate": 1.5551126696132756e-05, + "loss": 3.1243, + "step": 5525 + }, + { + "epoch": 8.08, + "grad_norm": 0.5441333055496216, + "learning_rate": 1.552601575318654e-05, + "loss": 3.1573, + "step": 5526 + }, + { + "epoch": 8.08, + "grad_norm": 0.4156208634376526, + "learning_rate": 1.550092339395074e-05, + "loss": 3.1993, + "step": 5527 + }, + { + "epoch": 8.09, + "grad_norm": 0.5446397066116333, + "learning_rate": 1.5475849623945505e-05, + "loss": 3.1158, + "step": 5528 + }, + { + "epoch": 8.09, + "grad_norm": 0.4556122124195099, + "learning_rate": 1.545079444868689e-05, + "loss": 3.1386, + "step": 5529 + }, + { + "epoch": 8.09, + "grad_norm": 0.6932422518730164, + "learning_rate": 1.5425757873686875e-05, + "loss": 3.1687, + "step": 5530 + }, + { + "epoch": 8.09, + "grad_norm": 0.28542694449424744, + "learning_rate": 1.5400739904453344e-05, + "loss": 3.1192, + "step": 5531 + }, + { + "epoch": 8.09, + "grad_norm": 0.3106916844844818, + "learning_rate": 1.537574054649007e-05, + "loss": 3.1941, + "step": 5532 + }, + { + "epoch": 8.09, + "grad_norm": 0.5748286247253418, + "learning_rate": 1.5350759805296744e-05, + "loss": 3.1547, + "step": 5533 + }, + { + "epoch": 8.09, + "grad_norm": 0.2282506823539734, + "learning_rate": 1.532579768636898e-05, + "loss": 3.1223, + "step": 5534 + }, + { + "epoch": 8.1, + "grad_norm": 0.36305752396583557, + "learning_rate": 1.5300854195198257e-05, + "loss": 3.1469, + "step": 5535 + }, + { + "epoch": 8.1, + "grad_norm": 0.3789515197277069, + "learning_rate": 1.527592933727198e-05, + "loss": 3.1726, + "step": 5536 + }, + { + "epoch": 8.1, + "grad_norm": 0.32144132256507874, + "learning_rate": 1.5251023118073448e-05, + "loss": 3.1429, + "step": 5537 + }, + { + "epoch": 8.1, + "grad_norm": 0.47844165563583374, + "learning_rate": 1.5226135543081876e-05, + "loss": 3.1361, + "step": 5538 + }, + { + "epoch": 8.1, + "grad_norm": 0.3552798628807068, + "learning_rate": 1.520126661777238e-05, + "loss": 3.1202, + "step": 5539 + }, + { + "epoch": 8.1, + "grad_norm": 0.24844308197498322, + "learning_rate": 1.5176416347615885e-05, + "loss": 3.1319, + "step": 5540 + }, + { + "epoch": 8.1, + "grad_norm": 0.7230424880981445, + "learning_rate": 1.515158473807935e-05, + "loss": 3.1792, + "step": 5541 + }, + { + "epoch": 8.11, + "grad_norm": 0.3808114528656006, + "learning_rate": 1.5126771794625572e-05, + "loss": 3.1422, + "step": 5542 + }, + { + "epoch": 8.11, + "grad_norm": 0.6194649934768677, + "learning_rate": 1.5101977522713152e-05, + "loss": 3.1325, + "step": 5543 + }, + { + "epoch": 8.11, + "grad_norm": 0.6643770933151245, + "learning_rate": 1.5077201927796747e-05, + "loss": 3.1341, + "step": 5544 + }, + { + "epoch": 8.11, + "grad_norm": 0.20721003413200378, + "learning_rate": 1.50524450153268e-05, + "loss": 3.1474, + "step": 5545 + }, + { + "epoch": 8.11, + "grad_norm": 0.3616236746311188, + "learning_rate": 1.5027706790749619e-05, + "loss": 3.1442, + "step": 5546 + }, + { + "epoch": 8.11, + "grad_norm": 1.006027340888977, + "learning_rate": 1.5002987259507539e-05, + "loss": 3.2174, + "step": 5547 + }, + { + "epoch": 8.12, + "grad_norm": 0.40877261757850647, + "learning_rate": 1.4978286427038601e-05, + "loss": 3.203, + "step": 5548 + }, + { + "epoch": 8.12, + "grad_norm": 0.6168388724327087, + "learning_rate": 1.4953604298776847e-05, + "loss": 3.1284, + "step": 5549 + }, + { + "epoch": 8.12, + "grad_norm": 0.846990704536438, + "learning_rate": 1.4928940880152232e-05, + "loss": 3.1267, + "step": 5550 + }, + { + "epoch": 8.12, + "grad_norm": 0.47257474064826965, + "learning_rate": 1.490429617659047e-05, + "loss": 3.1828, + "step": 5551 + }, + { + "epoch": 8.12, + "grad_norm": 0.5591023564338684, + "learning_rate": 1.4879670193513251e-05, + "loss": 3.1506, + "step": 5552 + }, + { + "epoch": 8.12, + "grad_norm": 1.040032982826233, + "learning_rate": 1.4855062936338183e-05, + "loss": 3.2159, + "step": 5553 + }, + { + "epoch": 8.12, + "grad_norm": 0.566142737865448, + "learning_rate": 1.4830474410478623e-05, + "loss": 3.1234, + "step": 5554 + }, + { + "epoch": 8.13, + "grad_norm": 0.6576405763626099, + "learning_rate": 1.4805904621343903e-05, + "loss": 3.1391, + "step": 5555 + }, + { + "epoch": 8.13, + "grad_norm": 0.622337818145752, + "learning_rate": 1.4781353574339252e-05, + "loss": 3.1823, + "step": 5556 + }, + { + "epoch": 8.13, + "grad_norm": 0.45685091614723206, + "learning_rate": 1.4756821274865695e-05, + "loss": 3.1574, + "step": 5557 + }, + { + "epoch": 8.13, + "grad_norm": 0.45965492725372314, + "learning_rate": 1.473230772832015e-05, + "loss": 3.1754, + "step": 5558 + }, + { + "epoch": 8.13, + "grad_norm": 0.5647199153900146, + "learning_rate": 1.4707812940095511e-05, + "loss": 3.1166, + "step": 5559 + }, + { + "epoch": 8.13, + "grad_norm": 0.6305474638938904, + "learning_rate": 1.4683336915580414e-05, + "loss": 3.134, + "step": 5560 + }, + { + "epoch": 8.13, + "grad_norm": 0.23273073136806488, + "learning_rate": 1.4658879660159419e-05, + "loss": 3.1207, + "step": 5561 + }, + { + "epoch": 8.14, + "grad_norm": 0.5071410536766052, + "learning_rate": 1.4634441179212965e-05, + "loss": 3.1623, + "step": 5562 + }, + { + "epoch": 8.14, + "grad_norm": 0.34845244884490967, + "learning_rate": 1.4610021478117352e-05, + "loss": 3.1172, + "step": 5563 + }, + { + "epoch": 8.14, + "grad_norm": 0.43220025300979614, + "learning_rate": 1.4585620562244773e-05, + "loss": 3.1265, + "step": 5564 + }, + { + "epoch": 8.14, + "grad_norm": 0.3767712712287903, + "learning_rate": 1.4561238436963231e-05, + "loss": 3.1457, + "step": 5565 + }, + { + "epoch": 8.14, + "grad_norm": 0.4660220742225647, + "learning_rate": 1.453687510763666e-05, + "loss": 3.1654, + "step": 5566 + }, + { + "epoch": 8.14, + "grad_norm": 0.3686300814151764, + "learning_rate": 1.4512530579624806e-05, + "loss": 3.1813, + "step": 5567 + }, + { + "epoch": 8.14, + "grad_norm": 0.40277543663978577, + "learning_rate": 1.4488204858283328e-05, + "loss": 3.1764, + "step": 5568 + }, + { + "epoch": 8.15, + "grad_norm": 0.4759199917316437, + "learning_rate": 1.446389794896369e-05, + "loss": 3.2285, + "step": 5569 + }, + { + "epoch": 8.15, + "grad_norm": 0.6583293676376343, + "learning_rate": 1.4439609857013304e-05, + "loss": 3.1806, + "step": 5570 + }, + { + "epoch": 8.15, + "grad_norm": 2.4492862224578857, + "learning_rate": 1.4415340587775295e-05, + "loss": 3.1365, + "step": 5571 + }, + { + "epoch": 8.15, + "grad_norm": 0.5203627943992615, + "learning_rate": 1.4391090146588816e-05, + "loss": 3.1313, + "step": 5572 + }, + { + "epoch": 8.15, + "grad_norm": 0.3894268870353699, + "learning_rate": 1.4366858538788796e-05, + "loss": 3.1248, + "step": 5573 + }, + { + "epoch": 8.15, + "grad_norm": 0.5935128331184387, + "learning_rate": 1.4342645769705977e-05, + "loss": 3.1253, + "step": 5574 + }, + { + "epoch": 8.16, + "grad_norm": 0.44079282879829407, + "learning_rate": 1.4318451844667058e-05, + "loss": 3.1611, + "step": 5575 + }, + { + "epoch": 8.16, + "grad_norm": 7.371267318725586, + "learning_rate": 1.429427676899454e-05, + "loss": 3.1284, + "step": 5576 + }, + { + "epoch": 8.16, + "grad_norm": 0.3392414152622223, + "learning_rate": 1.4270120548006716e-05, + "loss": 3.1202, + "step": 5577 + }, + { + "epoch": 8.16, + "eval_loss": 3.003582000732422, + "eval_runtime": 1164.277, + "eval_samples_per_second": 14.723, + "eval_steps_per_second": 0.921, + "step": 5577 + }, + { + "epoch": 8.16, + "grad_norm": 0.4595149755477905, + "learning_rate": 1.4245983187017854e-05, + "loss": 3.1606, + "step": 5578 + }, + { + "epoch": 8.16, + "grad_norm": 0.20213989913463593, + "learning_rate": 1.4221864691338015e-05, + "loss": 3.1206, + "step": 5579 + }, + { + "epoch": 8.16, + "grad_norm": 0.389874666929245, + "learning_rate": 1.419776506627305e-05, + "loss": 3.1593, + "step": 5580 + }, + { + "epoch": 8.16, + "grad_norm": 0.7267783284187317, + "learning_rate": 1.417368431712478e-05, + "loss": 3.1118, + "step": 5581 + }, + { + "epoch": 8.17, + "grad_norm": 0.2689005136489868, + "learning_rate": 1.4149622449190768e-05, + "loss": 3.1423, + "step": 5582 + }, + { + "epoch": 8.17, + "grad_norm": 0.2806105613708496, + "learning_rate": 1.4125579467764438e-05, + "loss": 3.1678, + "step": 5583 + }, + { + "epoch": 8.17, + "grad_norm": 0.22201931476593018, + "learning_rate": 1.4101555378135178e-05, + "loss": 3.1606, + "step": 5584 + }, + { + "epoch": 8.17, + "grad_norm": 0.5508745908737183, + "learning_rate": 1.407755018558804e-05, + "loss": 3.1305, + "step": 5585 + }, + { + "epoch": 8.17, + "grad_norm": 0.5338426232337952, + "learning_rate": 1.4053563895404032e-05, + "loss": 3.1826, + "step": 5586 + }, + { + "epoch": 8.17, + "grad_norm": 0.45947423577308655, + "learning_rate": 1.4029596512859989e-05, + "loss": 3.1468, + "step": 5587 + }, + { + "epoch": 8.17, + "grad_norm": 0.2771443724632263, + "learning_rate": 1.4005648043228558e-05, + "loss": 3.1542, + "step": 5588 + }, + { + "epoch": 8.18, + "grad_norm": 0.372646689414978, + "learning_rate": 1.398171849177825e-05, + "loss": 3.1321, + "step": 5589 + }, + { + "epoch": 8.18, + "grad_norm": 0.682151734828949, + "learning_rate": 1.3957807863773398e-05, + "loss": 3.1698, + "step": 5590 + }, + { + "epoch": 8.18, + "grad_norm": 0.39036300778388977, + "learning_rate": 1.3933916164474192e-05, + "loss": 3.129, + "step": 5591 + }, + { + "epoch": 8.18, + "grad_norm": 0.4261132478713989, + "learning_rate": 1.3910043399136651e-05, + "loss": 3.1635, + "step": 5592 + }, + { + "epoch": 8.18, + "grad_norm": 0.4003067910671234, + "learning_rate": 1.3886189573012597e-05, + "loss": 3.1571, + "step": 5593 + }, + { + "epoch": 8.18, + "grad_norm": 1.0211633443832397, + "learning_rate": 1.3862354691349732e-05, + "loss": 3.1219, + "step": 5594 + }, + { + "epoch": 8.18, + "grad_norm": 0.3579561114311218, + "learning_rate": 1.3838538759391573e-05, + "loss": 3.1424, + "step": 5595 + }, + { + "epoch": 8.19, + "grad_norm": 0.5200756788253784, + "learning_rate": 1.3814741782377461e-05, + "loss": 3.1432, + "step": 5596 + }, + { + "epoch": 8.19, + "grad_norm": 0.9823718667030334, + "learning_rate": 1.3790963765542553e-05, + "loss": 3.1739, + "step": 5597 + }, + { + "epoch": 8.19, + "grad_norm": 0.4610528349876404, + "learning_rate": 1.3767204714117876e-05, + "loss": 3.1228, + "step": 5598 + }, + { + "epoch": 8.19, + "grad_norm": 0.46500396728515625, + "learning_rate": 1.3743464633330261e-05, + "loss": 3.1569, + "step": 5599 + }, + { + "epoch": 8.19, + "grad_norm": 0.48705846071243286, + "learning_rate": 1.3719743528402362e-05, + "loss": 3.1635, + "step": 5600 + }, + { + "epoch": 8.19, + "grad_norm": 0.4838919937610626, + "learning_rate": 1.369604140455265e-05, + "loss": 3.1584, + "step": 5601 + }, + { + "epoch": 8.2, + "grad_norm": 0.46300509572029114, + "learning_rate": 1.3672358266995456e-05, + "loss": 3.1597, + "step": 5602 + }, + { + "epoch": 8.2, + "grad_norm": 0.32473286986351013, + "learning_rate": 1.3648694120940896e-05, + "loss": 3.1018, + "step": 5603 + }, + { + "epoch": 8.2, + "grad_norm": 0.7043519020080566, + "learning_rate": 1.3625048971594957e-05, + "loss": 3.1544, + "step": 5604 + }, + { + "epoch": 8.2, + "grad_norm": 0.36091065406799316, + "learning_rate": 1.3601422824159338e-05, + "loss": 3.124, + "step": 5605 + }, + { + "epoch": 8.2, + "grad_norm": 0.2936200499534607, + "learning_rate": 1.3577815683831708e-05, + "loss": 3.1584, + "step": 5606 + }, + { + "epoch": 8.2, + "grad_norm": 0.5868943333625793, + "learning_rate": 1.3554227555805465e-05, + "loss": 3.1279, + "step": 5607 + }, + { + "epoch": 8.2, + "grad_norm": 0.4498915672302246, + "learning_rate": 1.3530658445269783e-05, + "loss": 3.1589, + "step": 5608 + }, + { + "epoch": 8.21, + "grad_norm": 0.7671617269515991, + "learning_rate": 1.350710835740977e-05, + "loss": 3.1504, + "step": 5609 + }, + { + "epoch": 8.21, + "grad_norm": 3.664649486541748, + "learning_rate": 1.3483577297406303e-05, + "loss": 3.1071, + "step": 5610 + }, + { + "epoch": 8.21, + "grad_norm": 0.42762038111686707, + "learning_rate": 1.3460065270435984e-05, + "loss": 3.133, + "step": 5611 + }, + { + "epoch": 8.21, + "grad_norm": 0.3746001720428467, + "learning_rate": 1.3436572281671333e-05, + "loss": 3.1491, + "step": 5612 + }, + { + "epoch": 8.21, + "grad_norm": 0.5500866770744324, + "learning_rate": 1.3413098336280694e-05, + "loss": 3.1355, + "step": 5613 + }, + { + "epoch": 8.21, + "grad_norm": 0.5390032529830933, + "learning_rate": 1.3389643439428124e-05, + "loss": 3.1345, + "step": 5614 + }, + { + "epoch": 8.21, + "grad_norm": 0.30622246861457825, + "learning_rate": 1.3366207596273561e-05, + "loss": 3.1552, + "step": 5615 + }, + { + "epoch": 8.22, + "grad_norm": 0.3916447162628174, + "learning_rate": 1.3342790811972727e-05, + "loss": 3.1234, + "step": 5616 + }, + { + "epoch": 8.22, + "grad_norm": 0.5330905914306641, + "learning_rate": 1.3319393091677168e-05, + "loss": 3.1187, + "step": 5617 + }, + { + "epoch": 8.22, + "grad_norm": 0.700232744216919, + "learning_rate": 1.3296014440534222e-05, + "loss": 3.1759, + "step": 5618 + }, + { + "epoch": 8.22, + "grad_norm": 0.5803514122962952, + "learning_rate": 1.3272654863687028e-05, + "loss": 3.1565, + "step": 5619 + }, + { + "epoch": 8.22, + "grad_norm": 0.35048308968544006, + "learning_rate": 1.3249314366274546e-05, + "loss": 3.1275, + "step": 5620 + }, + { + "epoch": 8.22, + "grad_norm": 0.6838834285736084, + "learning_rate": 1.3225992953431532e-05, + "loss": 3.1853, + "step": 5621 + }, + { + "epoch": 8.22, + "grad_norm": 0.37435978651046753, + "learning_rate": 1.3202690630288528e-05, + "loss": 3.1035, + "step": 5622 + }, + { + "epoch": 8.23, + "grad_norm": 0.3659183979034424, + "learning_rate": 1.31794074019719e-05, + "loss": 3.1258, + "step": 5623 + }, + { + "epoch": 8.23, + "grad_norm": 0.9304698705673218, + "learning_rate": 1.3156143273603794e-05, + "loss": 3.1519, + "step": 5624 + }, + { + "epoch": 8.23, + "grad_norm": 0.29267510771751404, + "learning_rate": 1.3132898250302172e-05, + "loss": 3.1569, + "step": 5625 + }, + { + "epoch": 8.23, + "grad_norm": 0.7396655678749084, + "learning_rate": 1.3109672337180768e-05, + "loss": 3.1207, + "step": 5626 + }, + { + "epoch": 8.23, + "grad_norm": 0.7278848886489868, + "learning_rate": 1.3086465539349146e-05, + "loss": 3.1382, + "step": 5627 + }, + { + "epoch": 8.23, + "grad_norm": 0.5418538451194763, + "learning_rate": 1.3063277861912637e-05, + "loss": 3.1373, + "step": 5628 + }, + { + "epoch": 8.24, + "grad_norm": 0.7232877016067505, + "learning_rate": 1.304010930997237e-05, + "loss": 3.1655, + "step": 5629 + }, + { + "epoch": 8.24, + "grad_norm": 0.8563187122344971, + "learning_rate": 1.3016959888625269e-05, + "loss": 3.1261, + "step": 5630 + }, + { + "epoch": 8.24, + "grad_norm": 0.671583354473114, + "learning_rate": 1.299382960296407e-05, + "loss": 3.1386, + "step": 5631 + }, + { + "epoch": 8.24, + "grad_norm": 0.49029281735420227, + "learning_rate": 1.2970718458077258e-05, + "loss": 3.1581, + "step": 5632 + }, + { + "epoch": 8.24, + "grad_norm": 0.7014363408088684, + "learning_rate": 1.2947626459049146e-05, + "loss": 3.1533, + "step": 5633 + }, + { + "epoch": 8.24, + "grad_norm": 0.8424006104469299, + "learning_rate": 1.2924553610959821e-05, + "loss": 3.1507, + "step": 5634 + }, + { + "epoch": 8.24, + "grad_norm": 0.45622050762176514, + "learning_rate": 1.290149991888514e-05, + "loss": 3.155, + "step": 5635 + }, + { + "epoch": 8.25, + "grad_norm": 0.5782467126846313, + "learning_rate": 1.2878465387896788e-05, + "loss": 3.1422, + "step": 5636 + }, + { + "epoch": 8.25, + "grad_norm": 0.8892111778259277, + "learning_rate": 1.2855450023062155e-05, + "loss": 3.1708, + "step": 5637 + }, + { + "epoch": 8.25, + "grad_norm": 0.40920886397361755, + "learning_rate": 1.2832453829444534e-05, + "loss": 3.1274, + "step": 5638 + }, + { + "epoch": 8.25, + "grad_norm": 0.720196008682251, + "learning_rate": 1.2809476812102883e-05, + "loss": 3.2203, + "step": 5639 + }, + { + "epoch": 8.25, + "grad_norm": 0.7434779405593872, + "learning_rate": 1.2786518976091977e-05, + "loss": 3.1798, + "step": 5640 + }, + { + "epoch": 8.25, + "grad_norm": 0.5027379393577576, + "learning_rate": 1.2763580326462476e-05, + "loss": 3.1386, + "step": 5641 + }, + { + "epoch": 8.25, + "grad_norm": 0.6186532974243164, + "learning_rate": 1.2740660868260633e-05, + "loss": 3.1423, + "step": 5642 + }, + { + "epoch": 8.26, + "grad_norm": 0.6929596662521362, + "learning_rate": 1.2717760606528605e-05, + "loss": 3.1433, + "step": 5643 + }, + { + "epoch": 8.26, + "grad_norm": 0.3948609530925751, + "learning_rate": 1.2694879546304338e-05, + "loss": 3.1478, + "step": 5644 + }, + { + "epoch": 8.26, + "grad_norm": 0.4978497624397278, + "learning_rate": 1.2672017692621463e-05, + "loss": 3.1263, + "step": 5645 + }, + { + "epoch": 8.26, + "grad_norm": 0.6163767576217651, + "learning_rate": 1.264917505050942e-05, + "loss": 3.1549, + "step": 5646 + }, + { + "epoch": 8.26, + "grad_norm": 0.49492985010147095, + "learning_rate": 1.2626351624993515e-05, + "loss": 3.1269, + "step": 5647 + }, + { + "epoch": 8.26, + "grad_norm": 0.3817135691642761, + "learning_rate": 1.2603547421094675e-05, + "loss": 3.1334, + "step": 5648 + }, + { + "epoch": 8.27, + "grad_norm": 0.5009942650794983, + "learning_rate": 1.2580762443829707e-05, + "loss": 3.175, + "step": 5649 + }, + { + "epoch": 8.27, + "grad_norm": 0.5749830007553101, + "learning_rate": 1.2557996698211139e-05, + "loss": 3.143, + "step": 5650 + }, + { + "epoch": 8.27, + "grad_norm": 0.44454360008239746, + "learning_rate": 1.2535250189247282e-05, + "loss": 3.1268, + "step": 5651 + }, + { + "epoch": 8.27, + "grad_norm": 1.2505532503128052, + "learning_rate": 1.2512522921942226e-05, + "loss": 3.1423, + "step": 5652 + }, + { + "epoch": 8.27, + "grad_norm": 0.49076762795448303, + "learning_rate": 1.2489814901295804e-05, + "loss": 3.1443, + "step": 5653 + }, + { + "epoch": 8.27, + "grad_norm": 0.4465857148170471, + "learning_rate": 1.2467126132303642e-05, + "loss": 3.16, + "step": 5654 + }, + { + "epoch": 8.27, + "grad_norm": 0.5789979696273804, + "learning_rate": 1.2444456619957113e-05, + "loss": 3.124, + "step": 5655 + }, + { + "epoch": 8.28, + "grad_norm": 0.3660390079021454, + "learning_rate": 1.2421806369243361e-05, + "loss": 3.104, + "step": 5656 + }, + { + "epoch": 8.28, + "grad_norm": 0.7501944899559021, + "learning_rate": 1.2399175385145278e-05, + "loss": 3.1535, + "step": 5657 + }, + { + "epoch": 8.28, + "grad_norm": 0.7539718151092529, + "learning_rate": 1.2376563672641562e-05, + "loss": 3.1868, + "step": 5658 + }, + { + "epoch": 8.28, + "grad_norm": 0.3036991059780121, + "learning_rate": 1.2353971236706564e-05, + "loss": 3.1484, + "step": 5659 + }, + { + "epoch": 8.28, + "grad_norm": 0.8954304456710815, + "learning_rate": 1.233139808231053e-05, + "loss": 3.1336, + "step": 5660 + }, + { + "epoch": 8.28, + "grad_norm": 0.4805630147457123, + "learning_rate": 1.2308844214419423e-05, + "loss": 3.1618, + "step": 5661 + }, + { + "epoch": 8.28, + "grad_norm": 0.39005419611930847, + "learning_rate": 1.2286309637994864e-05, + "loss": 3.1422, + "step": 5662 + }, + { + "epoch": 8.29, + "grad_norm": 0.5603565573692322, + "learning_rate": 1.2263794357994373e-05, + "loss": 3.1942, + "step": 5663 + }, + { + "epoch": 8.29, + "grad_norm": 0.558560311794281, + "learning_rate": 1.2241298379371168e-05, + "loss": 3.1461, + "step": 5664 + }, + { + "epoch": 8.29, + "grad_norm": 0.408338725566864, + "learning_rate": 1.2218821707074136e-05, + "loss": 3.1577, + "step": 5665 + }, + { + "epoch": 8.29, + "grad_norm": 0.6621556878089905, + "learning_rate": 1.2196364346048072e-05, + "loss": 3.1458, + "step": 5666 + }, + { + "epoch": 8.29, + "grad_norm": 0.5381448268890381, + "learning_rate": 1.2173926301233429e-05, + "loss": 3.1414, + "step": 5667 + }, + { + "epoch": 8.29, + "grad_norm": 0.33626240491867065, + "learning_rate": 1.215150757756639e-05, + "loss": 3.162, + "step": 5668 + }, + { + "epoch": 8.29, + "grad_norm": 0.31144070625305176, + "learning_rate": 1.2129108179978954e-05, + "loss": 3.1176, + "step": 5669 + }, + { + "epoch": 8.3, + "grad_norm": 0.28460976481437683, + "learning_rate": 1.2106728113398869e-05, + "loss": 3.1216, + "step": 5670 + }, + { + "epoch": 8.3, + "grad_norm": 0.5316192507743835, + "learning_rate": 1.2084367382749517e-05, + "loss": 3.1335, + "step": 5671 + }, + { + "epoch": 8.3, + "grad_norm": 0.4095101058483124, + "learning_rate": 1.2062025992950187e-05, + "loss": 3.1225, + "step": 5672 + }, + { + "epoch": 8.3, + "grad_norm": 0.45470380783081055, + "learning_rate": 1.2039703948915804e-05, + "loss": 3.1579, + "step": 5673 + }, + { + "epoch": 8.3, + "grad_norm": 0.30135294795036316, + "learning_rate": 1.2017401255557037e-05, + "loss": 3.1336, + "step": 5674 + }, + { + "epoch": 8.3, + "grad_norm": 0.7919064164161682, + "learning_rate": 1.1995117917780396e-05, + "loss": 3.1425, + "step": 5675 + }, + { + "epoch": 8.31, + "grad_norm": 0.42938122153282166, + "learning_rate": 1.1972853940488015e-05, + "loss": 3.1453, + "step": 5676 + }, + { + "epoch": 8.31, + "grad_norm": 0.3541744351387024, + "learning_rate": 1.1950609328577823e-05, + "loss": 3.1628, + "step": 5677 + }, + { + "epoch": 8.31, + "grad_norm": 0.2010623663663864, + "learning_rate": 1.1928384086943533e-05, + "loss": 3.1708, + "step": 5678 + }, + { + "epoch": 8.31, + "grad_norm": 0.6417263746261597, + "learning_rate": 1.1906178220474485e-05, + "loss": 3.1305, + "step": 5679 + }, + { + "epoch": 8.31, + "grad_norm": 0.3581140637397766, + "learning_rate": 1.1883991734055844e-05, + "loss": 3.1437, + "step": 5680 + }, + { + "epoch": 8.31, + "grad_norm": 0.40174880623817444, + "learning_rate": 1.186182463256853e-05, + "loss": 3.1109, + "step": 5681 + }, + { + "epoch": 8.31, + "grad_norm": 0.6692104935646057, + "learning_rate": 1.1839676920889098e-05, + "loss": 3.1739, + "step": 5682 + }, + { + "epoch": 8.32, + "grad_norm": 0.2795259952545166, + "learning_rate": 1.1817548603889917e-05, + "loss": 3.1526, + "step": 5683 + }, + { + "epoch": 8.32, + "grad_norm": 0.21500855684280396, + "learning_rate": 1.1795439686439069e-05, + "loss": 3.1436, + "step": 5684 + }, + { + "epoch": 8.32, + "grad_norm": 0.7537773251533508, + "learning_rate": 1.1773350173400377e-05, + "loss": 3.1438, + "step": 5685 + }, + { + "epoch": 8.32, + "grad_norm": 0.520550549030304, + "learning_rate": 1.1751280069633375e-05, + "loss": 3.1659, + "step": 5686 + }, + { + "epoch": 8.32, + "grad_norm": 0.27544260025024414, + "learning_rate": 1.1729229379993345e-05, + "loss": 3.1702, + "step": 5687 + }, + { + "epoch": 8.32, + "grad_norm": 0.36824995279312134, + "learning_rate": 1.1707198109331274e-05, + "loss": 3.128, + "step": 5688 + }, + { + "epoch": 8.32, + "grad_norm": 0.268276184797287, + "learning_rate": 1.168518626249392e-05, + "loss": 3.1329, + "step": 5689 + }, + { + "epoch": 8.33, + "grad_norm": 0.43285050988197327, + "learning_rate": 1.1663193844323727e-05, + "loss": 3.1342, + "step": 5690 + }, + { + "epoch": 8.33, + "grad_norm": 0.28300830721855164, + "learning_rate": 1.1641220859658874e-05, + "loss": 3.1185, + "step": 5691 + }, + { + "epoch": 8.33, + "grad_norm": 0.33784517645835876, + "learning_rate": 1.1619267313333281e-05, + "loss": 3.1615, + "step": 5692 + }, + { + "epoch": 8.33, + "grad_norm": 0.3320251405239105, + "learning_rate": 1.1597333210176586e-05, + "loss": 3.1881, + "step": 5693 + }, + { + "epoch": 8.33, + "grad_norm": 0.3278450667858124, + "learning_rate": 1.1575418555014127e-05, + "loss": 3.1494, + "step": 5694 + }, + { + "epoch": 8.33, + "grad_norm": 0.30050769448280334, + "learning_rate": 1.1553523352667018e-05, + "loss": 3.1541, + "step": 5695 + }, + { + "epoch": 8.33, + "grad_norm": 0.26941055059432983, + "learning_rate": 1.1531647607951979e-05, + "loss": 3.1437, + "step": 5696 + }, + { + "epoch": 8.34, + "grad_norm": 0.5336388349533081, + "learning_rate": 1.1509791325681607e-05, + "loss": 3.1115, + "step": 5697 + }, + { + "epoch": 8.34, + "grad_norm": 0.30893999338150024, + "learning_rate": 1.1487954510664122e-05, + "loss": 3.1613, + "step": 5698 + }, + { + "epoch": 8.34, + "grad_norm": 0.48479652404785156, + "learning_rate": 1.1466137167703428e-05, + "loss": 3.183, + "step": 5699 + }, + { + "epoch": 8.34, + "grad_norm": 0.33634239435195923, + "learning_rate": 1.1444339301599239e-05, + "loss": 3.1524, + "step": 5700 + }, + { + "epoch": 8.34, + "grad_norm": 0.45775729417800903, + "learning_rate": 1.1422560917146962e-05, + "loss": 3.1708, + "step": 5701 + }, + { + "epoch": 8.34, + "grad_norm": 0.2623753845691681, + "learning_rate": 1.1400802019137613e-05, + "loss": 3.1317, + "step": 5702 + }, + { + "epoch": 8.35, + "grad_norm": 0.4868522584438324, + "learning_rate": 1.1379062612358082e-05, + "loss": 3.1486, + "step": 5703 + }, + { + "epoch": 8.35, + "grad_norm": 0.32729241251945496, + "learning_rate": 1.1357342701590889e-05, + "loss": 3.1077, + "step": 5704 + }, + { + "epoch": 8.35, + "grad_norm": 0.4166317582130432, + "learning_rate": 1.1335642291614201e-05, + "loss": 3.1305, + "step": 5705 + }, + { + "epoch": 8.35, + "grad_norm": 0.3061630129814148, + "learning_rate": 1.1313961387202044e-05, + "loss": 3.1469, + "step": 5706 + }, + { + "epoch": 8.35, + "grad_norm": 0.32585567235946655, + "learning_rate": 1.1292299993124012e-05, + "loss": 3.1448, + "step": 5707 + }, + { + "epoch": 8.35, + "grad_norm": 0.5888083577156067, + "learning_rate": 1.1270658114145493e-05, + "loss": 3.1815, + "step": 5708 + }, + { + "epoch": 8.35, + "grad_norm": 0.28794151544570923, + "learning_rate": 1.1249035755027559e-05, + "loss": 3.1343, + "step": 5709 + }, + { + "epoch": 8.36, + "grad_norm": 0.6074895262718201, + "learning_rate": 1.122743292052697e-05, + "loss": 3.1481, + "step": 5710 + }, + { + "epoch": 8.36, + "grad_norm": 0.6733052730560303, + "learning_rate": 1.1205849615396202e-05, + "loss": 3.1235, + "step": 5711 + }, + { + "epoch": 8.36, + "grad_norm": 0.2199082374572754, + "learning_rate": 1.1184285844383458e-05, + "loss": 3.1172, + "step": 5712 + }, + { + "epoch": 8.36, + "grad_norm": 0.32932931184768677, + "learning_rate": 1.1162741612232608e-05, + "loss": 3.1226, + "step": 5713 + }, + { + "epoch": 8.36, + "grad_norm": 0.7292448878288269, + "learning_rate": 1.114121692368324e-05, + "loss": 3.1721, + "step": 5714 + }, + { + "epoch": 8.36, + "grad_norm": 0.34164637327194214, + "learning_rate": 1.1119711783470643e-05, + "loss": 3.1505, + "step": 5715 + }, + { + "epoch": 8.36, + "grad_norm": 0.2828170657157898, + "learning_rate": 1.1098226196325811e-05, + "loss": 3.1372, + "step": 5716 + }, + { + "epoch": 8.37, + "grad_norm": 0.3506002724170685, + "learning_rate": 1.107676016697542e-05, + "loss": 3.1324, + "step": 5717 + }, + { + "epoch": 8.37, + "grad_norm": 0.5819188952445984, + "learning_rate": 1.1055313700141857e-05, + "loss": 3.1621, + "step": 5718 + }, + { + "epoch": 8.37, + "grad_norm": 0.2485204041004181, + "learning_rate": 1.1033886800543191e-05, + "loss": 3.17, + "step": 5719 + }, + { + "epoch": 8.37, + "grad_norm": 0.3673901855945587, + "learning_rate": 1.1012479472893212e-05, + "loss": 3.1383, + "step": 5720 + }, + { + "epoch": 8.37, + "grad_norm": 0.3317386507987976, + "learning_rate": 1.099109172190138e-05, + "loss": 3.1596, + "step": 5721 + }, + { + "epoch": 8.37, + "grad_norm": 0.5169317722320557, + "learning_rate": 1.096972355227286e-05, + "loss": 3.1451, + "step": 5722 + }, + { + "epoch": 8.37, + "grad_norm": 0.4397293031215668, + "learning_rate": 1.0948374968708497e-05, + "loss": 3.1589, + "step": 5723 + }, + { + "epoch": 8.38, + "grad_norm": 0.23546314239501953, + "learning_rate": 1.0927045975904848e-05, + "loss": 3.1252, + "step": 5724 + }, + { + "epoch": 8.38, + "grad_norm": 0.2676627039909363, + "learning_rate": 1.0905736578554138e-05, + "loss": 3.1651, + "step": 5725 + }, + { + "epoch": 8.38, + "grad_norm": 0.9150519371032715, + "learning_rate": 1.0884446781344303e-05, + "loss": 3.1197, + "step": 5726 + }, + { + "epoch": 8.38, + "grad_norm": 0.5215425491333008, + "learning_rate": 1.0863176588958956e-05, + "loss": 3.1501, + "step": 5727 + }, + { + "epoch": 8.38, + "grad_norm": 0.48437726497650146, + "learning_rate": 1.0841926006077385e-05, + "loss": 3.1656, + "step": 5728 + }, + { + "epoch": 8.38, + "grad_norm": 0.7072523832321167, + "learning_rate": 1.0820695037374595e-05, + "loss": 3.1917, + "step": 5729 + }, + { + "epoch": 8.39, + "grad_norm": 0.4144083857536316, + "learning_rate": 1.0799483687521216e-05, + "loss": 3.1268, + "step": 5730 + }, + { + "epoch": 8.39, + "grad_norm": 0.2927367389202118, + "learning_rate": 1.0778291961183639e-05, + "loss": 3.1329, + "step": 5731 + }, + { + "epoch": 8.39, + "grad_norm": 0.3951784372329712, + "learning_rate": 1.0757119863023924e-05, + "loss": 3.1293, + "step": 5732 + }, + { + "epoch": 8.39, + "grad_norm": 0.6036468744277954, + "learning_rate": 1.0735967397699731e-05, + "loss": 3.1348, + "step": 5733 + }, + { + "epoch": 8.39, + "grad_norm": 0.4498322606086731, + "learning_rate": 1.0714834569864474e-05, + "loss": 3.1327, + "step": 5734 + }, + { + "epoch": 8.39, + "grad_norm": 0.4379383325576782, + "learning_rate": 1.0693721384167288e-05, + "loss": 3.1598, + "step": 5735 + }, + { + "epoch": 8.39, + "grad_norm": 0.3004567325115204, + "learning_rate": 1.0672627845252881e-05, + "loss": 3.1152, + "step": 5736 + }, + { + "epoch": 8.4, + "grad_norm": 0.7020764946937561, + "learning_rate": 1.0651553957761673e-05, + "loss": 3.1358, + "step": 5737 + }, + { + "epoch": 8.4, + "grad_norm": 0.3879508078098297, + "learning_rate": 1.0630499726329846e-05, + "loss": 3.148, + "step": 5738 + }, + { + "epoch": 8.4, + "grad_norm": 0.34494709968566895, + "learning_rate": 1.0609465155589137e-05, + "loss": 3.1244, + "step": 5739 + }, + { + "epoch": 8.4, + "grad_norm": 0.45144951343536377, + "learning_rate": 1.0588450250167026e-05, + "loss": 3.139, + "step": 5740 + }, + { + "epoch": 8.4, + "grad_norm": 0.4698631167411804, + "learning_rate": 1.0567455014686645e-05, + "loss": 3.1501, + "step": 5741 + }, + { + "epoch": 8.4, + "grad_norm": 0.3781033158302307, + "learning_rate": 1.0546479453766811e-05, + "loss": 3.1047, + "step": 5742 + }, + { + "epoch": 8.4, + "grad_norm": 0.9458195567131042, + "learning_rate": 1.052552357202201e-05, + "loss": 3.1385, + "step": 5743 + }, + { + "epoch": 8.41, + "grad_norm": 0.1932523101568222, + "learning_rate": 1.0504587374062391e-05, + "loss": 3.1302, + "step": 5744 + }, + { + "epoch": 8.41, + "grad_norm": 0.541278064250946, + "learning_rate": 1.0483670864493778e-05, + "loss": 3.1655, + "step": 5745 + }, + { + "epoch": 8.41, + "grad_norm": 1.0147125720977783, + "learning_rate": 1.0462774047917656e-05, + "loss": 3.1368, + "step": 5746 + }, + { + "epoch": 8.41, + "eval_loss": 2.9998810291290283, + "eval_runtime": 1167.1096, + "eval_samples_per_second": 14.688, + "eval_steps_per_second": 0.919, + "step": 5746 + }, + { + "epoch": 8.41, + "grad_norm": 0.746507465839386, + "learning_rate": 1.0441896928931206e-05, + "loss": 3.1443, + "step": 5747 + }, + { + "epoch": 8.41, + "grad_norm": 0.41886982321739197, + "learning_rate": 1.0421039512127229e-05, + "loss": 3.1343, + "step": 5748 + }, + { + "epoch": 8.41, + "grad_norm": 1.3022066354751587, + "learning_rate": 1.0400201802094223e-05, + "loss": 3.1096, + "step": 5749 + }, + { + "epoch": 8.42, + "grad_norm": 0.7943986058235168, + "learning_rate": 1.0379383803416364e-05, + "loss": 3.1392, + "step": 5750 + }, + { + "epoch": 8.42, + "grad_norm": 0.40819865465164185, + "learning_rate": 1.0358585520673446e-05, + "loss": 3.1234, + "step": 5751 + }, + { + "epoch": 8.42, + "grad_norm": 0.6574892401695251, + "learning_rate": 1.0337806958440966e-05, + "loss": 3.1238, + "step": 5752 + }, + { + "epoch": 8.42, + "grad_norm": 0.598135232925415, + "learning_rate": 1.0317048121290074e-05, + "loss": 3.1246, + "step": 5753 + }, + { + "epoch": 8.42, + "grad_norm": 0.543494701385498, + "learning_rate": 1.0296309013787553e-05, + "loss": 3.125, + "step": 5754 + }, + { + "epoch": 8.42, + "grad_norm": 0.754518985748291, + "learning_rate": 1.0275589640495908e-05, + "loss": 3.1477, + "step": 5755 + }, + { + "epoch": 8.42, + "grad_norm": 0.27329519391059875, + "learning_rate": 1.0254890005973205e-05, + "loss": 3.1207, + "step": 5756 + }, + { + "epoch": 8.43, + "grad_norm": 0.7898842096328735, + "learning_rate": 1.0234210114773257e-05, + "loss": 3.1593, + "step": 5757 + }, + { + "epoch": 8.43, + "grad_norm": 0.5713753700256348, + "learning_rate": 1.0213549971445535e-05, + "loss": 3.1235, + "step": 5758 + }, + { + "epoch": 8.43, + "grad_norm": 0.3577013909816742, + "learning_rate": 1.0192909580535049e-05, + "loss": 3.1596, + "step": 5759 + }, + { + "epoch": 8.43, + "grad_norm": 0.6142879128456116, + "learning_rate": 1.017228894658262e-05, + "loss": 3.148, + "step": 5760 + }, + { + "epoch": 8.43, + "grad_norm": 0.71822589635849, + "learning_rate": 1.0151688074124643e-05, + "loss": 3.1361, + "step": 5761 + }, + { + "epoch": 8.43, + "grad_norm": 0.7075265049934387, + "learning_rate": 1.0131106967693116e-05, + "loss": 3.1091, + "step": 5762 + }, + { + "epoch": 8.43, + "grad_norm": 0.4524644911289215, + "learning_rate": 1.0110545631815815e-05, + "loss": 3.1273, + "step": 5763 + }, + { + "epoch": 8.44, + "grad_norm": 0.451507568359375, + "learning_rate": 1.0090004071016046e-05, + "loss": 3.141, + "step": 5764 + }, + { + "epoch": 8.44, + "grad_norm": 0.5503039956092834, + "learning_rate": 1.0069482289812816e-05, + "loss": 3.1444, + "step": 5765 + }, + { + "epoch": 8.44, + "grad_norm": 0.2659579813480377, + "learning_rate": 1.0048980292720822e-05, + "loss": 3.1474, + "step": 5766 + }, + { + "epoch": 8.44, + "grad_norm": 0.25024622678756714, + "learning_rate": 1.0028498084250325e-05, + "loss": 3.1663, + "step": 5767 + }, + { + "epoch": 8.44, + "grad_norm": 0.6892327070236206, + "learning_rate": 1.0008035668907279e-05, + "loss": 3.1352, + "step": 5768 + }, + { + "epoch": 8.44, + "grad_norm": 0.32359933853149414, + "learning_rate": 9.987593051193311e-06, + "loss": 3.1438, + "step": 5769 + }, + { + "epoch": 8.44, + "grad_norm": 0.36346298456192017, + "learning_rate": 9.967170235605616e-06, + "loss": 3.1841, + "step": 5770 + }, + { + "epoch": 8.45, + "grad_norm": 0.42581936717033386, + "learning_rate": 9.946767226637088e-06, + "loss": 3.1161, + "step": 5771 + }, + { + "epoch": 8.45, + "grad_norm": 0.5385969281196594, + "learning_rate": 9.926384028776293e-06, + "loss": 3.1526, + "step": 5772 + }, + { + "epoch": 8.45, + "grad_norm": 0.4430517256259918, + "learning_rate": 9.90602064650734e-06, + "loss": 3.179, + "step": 5773 + }, + { + "epoch": 8.45, + "grad_norm": 0.7233391404151917, + "learning_rate": 9.885677084310063e-06, + "loss": 3.1161, + "step": 5774 + }, + { + "epoch": 8.45, + "grad_norm": 0.36595597863197327, + "learning_rate": 9.865353346659911e-06, + "loss": 3.1277, + "step": 5775 + }, + { + "epoch": 8.45, + "grad_norm": 0.7899670600891113, + "learning_rate": 9.845049438027976e-06, + "loss": 3.159, + "step": 5776 + }, + { + "epoch": 8.46, + "grad_norm": 0.3608589172363281, + "learning_rate": 9.824765362880962e-06, + "loss": 3.1284, + "step": 5777 + }, + { + "epoch": 8.46, + "grad_norm": 0.4470582902431488, + "learning_rate": 9.804501125681243e-06, + "loss": 3.1371, + "step": 5778 + }, + { + "epoch": 8.46, + "grad_norm": 0.45787444710731506, + "learning_rate": 9.784256730886809e-06, + "loss": 3.1443, + "step": 5779 + }, + { + "epoch": 8.46, + "grad_norm": 0.28761786222457886, + "learning_rate": 9.7640321829513e-06, + "loss": 3.1522, + "step": 5780 + }, + { + "epoch": 8.46, + "grad_norm": 0.8798906207084656, + "learning_rate": 9.743827486323975e-06, + "loss": 3.1395, + "step": 5781 + }, + { + "epoch": 8.46, + "grad_norm": 0.5044177174568176, + "learning_rate": 9.723642645449737e-06, + "loss": 3.1485, + "step": 5782 + }, + { + "epoch": 8.46, + "grad_norm": 0.43212029337882996, + "learning_rate": 9.703477664769123e-06, + "loss": 3.1306, + "step": 5783 + }, + { + "epoch": 8.47, + "grad_norm": 0.38974976539611816, + "learning_rate": 9.683332548718248e-06, + "loss": 3.1621, + "step": 5784 + }, + { + "epoch": 8.47, + "grad_norm": 0.425968736410141, + "learning_rate": 9.66320730172896e-06, + "loss": 3.1229, + "step": 5785 + }, + { + "epoch": 8.47, + "grad_norm": 2.754948377609253, + "learning_rate": 9.64310192822867e-06, + "loss": 3.1593, + "step": 5786 + }, + { + "epoch": 8.47, + "grad_norm": 0.7210294008255005, + "learning_rate": 9.623016432640375e-06, + "loss": 3.1309, + "step": 5787 + }, + { + "epoch": 8.47, + "grad_norm": 0.17482349276542664, + "learning_rate": 9.602950819382806e-06, + "loss": 3.1226, + "step": 5788 + }, + { + "epoch": 8.47, + "grad_norm": 0.35730379819869995, + "learning_rate": 9.582905092870254e-06, + "loss": 3.155, + "step": 5789 + }, + { + "epoch": 8.47, + "grad_norm": 0.3369613587856293, + "learning_rate": 9.562879257512602e-06, + "loss": 3.135, + "step": 5790 + }, + { + "epoch": 8.48, + "grad_norm": 0.9082013368606567, + "learning_rate": 9.542873317715439e-06, + "loss": 3.1319, + "step": 5791 + }, + { + "epoch": 8.48, + "grad_norm": 0.5040355920791626, + "learning_rate": 9.522887277879956e-06, + "loss": 3.1234, + "step": 5792 + }, + { + "epoch": 8.48, + "grad_norm": 0.47945570945739746, + "learning_rate": 9.502921142402887e-06, + "loss": 3.1499, + "step": 5793 + }, + { + "epoch": 8.48, + "grad_norm": 0.9931325912475586, + "learning_rate": 9.482974915676702e-06, + "loss": 3.158, + "step": 5794 + }, + { + "epoch": 8.48, + "grad_norm": 0.5050073266029358, + "learning_rate": 9.46304860208943e-06, + "loss": 3.1161, + "step": 5795 + }, + { + "epoch": 8.48, + "grad_norm": 0.21985039114952087, + "learning_rate": 9.443142206024668e-06, + "loss": 3.1283, + "step": 5796 + }, + { + "epoch": 8.48, + "grad_norm": 0.38089579343795776, + "learning_rate": 9.423255731861791e-06, + "loss": 3.1516, + "step": 5797 + }, + { + "epoch": 8.49, + "grad_norm": 0.8046565651893616, + "learning_rate": 9.403389183975609e-06, + "loss": 3.1305, + "step": 5798 + }, + { + "epoch": 8.49, + "grad_norm": 0.7955847382545471, + "learning_rate": 9.383542566736635e-06, + "loss": 3.1691, + "step": 5799 + }, + { + "epoch": 8.49, + "grad_norm": 0.2818087339401245, + "learning_rate": 9.363715884511049e-06, + "loss": 3.1686, + "step": 5800 + }, + { + "epoch": 8.49, + "grad_norm": 0.37908995151519775, + "learning_rate": 9.343909141660535e-06, + "loss": 3.1416, + "step": 5801 + }, + { + "epoch": 8.49, + "grad_norm": 0.9841012358665466, + "learning_rate": 9.324122342542429e-06, + "loss": 3.1679, + "step": 5802 + }, + { + "epoch": 8.49, + "grad_norm": 0.7049186825752258, + "learning_rate": 9.304355491509776e-06, + "loss": 3.1367, + "step": 5803 + }, + { + "epoch": 8.5, + "grad_norm": 0.4633147120475769, + "learning_rate": 9.284608592911082e-06, + "loss": 3.1225, + "step": 5804 + }, + { + "epoch": 8.5, + "grad_norm": 0.8353789448738098, + "learning_rate": 9.26488165109054e-06, + "loss": 3.1142, + "step": 5805 + }, + { + "epoch": 8.5, + "grad_norm": 0.740578830242157, + "learning_rate": 9.245174670387958e-06, + "loss": 3.1183, + "step": 5806 + }, + { + "epoch": 8.5, + "grad_norm": 0.40835264325141907, + "learning_rate": 9.225487655138742e-06, + "loss": 3.1315, + "step": 5807 + }, + { + "epoch": 8.5, + "grad_norm": 0.6107511520385742, + "learning_rate": 9.205820609673887e-06, + "loss": 3.1433, + "step": 5808 + }, + { + "epoch": 8.5, + "grad_norm": 0.8635002970695496, + "learning_rate": 9.18617353832003e-06, + "loss": 3.1264, + "step": 5809 + }, + { + "epoch": 8.5, + "grad_norm": 0.37021952867507935, + "learning_rate": 9.166546445399394e-06, + "loss": 3.1448, + "step": 5810 + }, + { + "epoch": 8.51, + "grad_norm": 0.7059236168861389, + "learning_rate": 9.146939335229788e-06, + "loss": 3.1168, + "step": 5811 + }, + { + "epoch": 8.51, + "grad_norm": 0.4614633619785309, + "learning_rate": 9.127352212124662e-06, + "loss": 3.1413, + "step": 5812 + }, + { + "epoch": 8.51, + "grad_norm": 0.4562102258205414, + "learning_rate": 9.107785080393061e-06, + "loss": 3.1227, + "step": 5813 + }, + { + "epoch": 8.51, + "grad_norm": 0.5066782832145691, + "learning_rate": 9.088237944339594e-06, + "loss": 3.1707, + "step": 5814 + }, + { + "epoch": 8.51, + "grad_norm": 0.47414615750312805, + "learning_rate": 9.068710808264524e-06, + "loss": 3.1667, + "step": 5815 + }, + { + "epoch": 8.51, + "grad_norm": 0.6087870001792908, + "learning_rate": 9.04920367646368e-06, + "loss": 3.1496, + "step": 5816 + }, + { + "epoch": 8.51, + "grad_norm": 0.5062857270240784, + "learning_rate": 9.029716553228517e-06, + "loss": 3.1633, + "step": 5817 + }, + { + "epoch": 8.52, + "grad_norm": 0.28521206974983215, + "learning_rate": 9.01024944284602e-06, + "loss": 3.1468, + "step": 5818 + }, + { + "epoch": 8.52, + "grad_norm": 0.46825653314590454, + "learning_rate": 8.99080234959888e-06, + "loss": 3.1665, + "step": 5819 + }, + { + "epoch": 8.52, + "grad_norm": 0.682184636592865, + "learning_rate": 8.971375277765326e-06, + "loss": 3.1446, + "step": 5820 + }, + { + "epoch": 8.52, + "grad_norm": 0.2446981519460678, + "learning_rate": 8.951968231619124e-06, + "loss": 3.189, + "step": 5821 + }, + { + "epoch": 8.52, + "grad_norm": 0.23873744904994965, + "learning_rate": 8.932581215429748e-06, + "loss": 3.1293, + "step": 5822 + }, + { + "epoch": 8.52, + "grad_norm": 0.33204108476638794, + "learning_rate": 8.913214233462231e-06, + "loss": 3.13, + "step": 5823 + }, + { + "epoch": 8.52, + "grad_norm": 0.32620128989219666, + "learning_rate": 8.893867289977097e-06, + "loss": 3.1604, + "step": 5824 + }, + { + "epoch": 8.53, + "grad_norm": 0.6817585825920105, + "learning_rate": 8.874540389230612e-06, + "loss": 3.1272, + "step": 5825 + }, + { + "epoch": 8.53, + "grad_norm": 0.2696726322174072, + "learning_rate": 8.855233535474571e-06, + "loss": 3.139, + "step": 5826 + }, + { + "epoch": 8.53, + "grad_norm": 0.46153542399406433, + "learning_rate": 8.835946732956302e-06, + "loss": 3.136, + "step": 5827 + }, + { + "epoch": 8.53, + "grad_norm": 0.5768734216690063, + "learning_rate": 8.816679985918785e-06, + "loss": 3.1362, + "step": 5828 + }, + { + "epoch": 8.53, + "grad_norm": 0.5621840953826904, + "learning_rate": 8.797433298600622e-06, + "loss": 3.1742, + "step": 5829 + }, + { + "epoch": 8.53, + "grad_norm": 1.3828614950180054, + "learning_rate": 8.7782066752359e-06, + "loss": 3.1435, + "step": 5830 + }, + { + "epoch": 8.54, + "grad_norm": 0.3548658788204193, + "learning_rate": 8.75900012005436e-06, + "loss": 3.2216, + "step": 5831 + }, + { + "epoch": 8.54, + "grad_norm": 0.25544172525405884, + "learning_rate": 8.739813637281336e-06, + "loss": 3.1576, + "step": 5832 + }, + { + "epoch": 8.54, + "grad_norm": 0.6139233112335205, + "learning_rate": 8.720647231137713e-06, + "loss": 3.162, + "step": 5833 + }, + { + "epoch": 8.54, + "grad_norm": 0.3499756157398224, + "learning_rate": 8.701500905839966e-06, + "loss": 3.1198, + "step": 5834 + }, + { + "epoch": 8.54, + "grad_norm": 0.5180540084838867, + "learning_rate": 8.682374665600168e-06, + "loss": 3.1238, + "step": 5835 + }, + { + "epoch": 8.54, + "grad_norm": 0.29475992918014526, + "learning_rate": 8.663268514625966e-06, + "loss": 3.1271, + "step": 5836 + }, + { + "epoch": 8.54, + "grad_norm": 0.6238717436790466, + "learning_rate": 8.644182457120575e-06, + "loss": 3.1083, + "step": 5837 + }, + { + "epoch": 8.55, + "grad_norm": 0.24445649981498718, + "learning_rate": 8.625116497282803e-06, + "loss": 3.1107, + "step": 5838 + }, + { + "epoch": 8.55, + "grad_norm": 0.6731110215187073, + "learning_rate": 8.60607063930703e-06, + "loss": 3.1931, + "step": 5839 + }, + { + "epoch": 8.55, + "grad_norm": 0.3173396587371826, + "learning_rate": 8.587044887383221e-06, + "loss": 3.1099, + "step": 5840 + }, + { + "epoch": 8.55, + "grad_norm": 0.27860239148139954, + "learning_rate": 8.568039245696924e-06, + "loss": 3.143, + "step": 5841 + }, + { + "epoch": 8.55, + "grad_norm": 0.25917676091194153, + "learning_rate": 8.549053718429235e-06, + "loss": 3.1283, + "step": 5842 + }, + { + "epoch": 8.55, + "grad_norm": 0.3333747386932373, + "learning_rate": 8.530088309756856e-06, + "loss": 3.1291, + "step": 5843 + }, + { + "epoch": 8.55, + "grad_norm": 0.22568973898887634, + "learning_rate": 8.511143023852041e-06, + "loss": 3.1724, + "step": 5844 + }, + { + "epoch": 8.56, + "grad_norm": 0.3453158140182495, + "learning_rate": 8.492217864882623e-06, + "loss": 3.1325, + "step": 5845 + }, + { + "epoch": 8.56, + "grad_norm": 0.4224380850791931, + "learning_rate": 8.473312837012026e-06, + "loss": 3.1217, + "step": 5846 + }, + { + "epoch": 8.56, + "grad_norm": 0.3284725844860077, + "learning_rate": 8.454427944399224e-06, + "loss": 3.1243, + "step": 5847 + }, + { + "epoch": 8.56, + "grad_norm": 0.47590208053588867, + "learning_rate": 8.435563191198758e-06, + "loss": 3.1304, + "step": 5848 + }, + { + "epoch": 8.56, + "grad_norm": 0.3408375680446625, + "learning_rate": 8.416718581560746e-06, + "loss": 3.1498, + "step": 5849 + }, + { + "epoch": 8.56, + "grad_norm": 0.28439176082611084, + "learning_rate": 8.39789411963089e-06, + "loss": 3.1231, + "step": 5850 + }, + { + "epoch": 8.57, + "grad_norm": 0.36393728852272034, + "learning_rate": 8.379089809550456e-06, + "loss": 3.1541, + "step": 5851 + }, + { + "epoch": 8.57, + "grad_norm": 0.7330798506736755, + "learning_rate": 8.360305655456225e-06, + "loss": 3.112, + "step": 5852 + }, + { + "epoch": 8.57, + "grad_norm": 1.5145879983901978, + "learning_rate": 8.34154166148059e-06, + "loss": 3.1475, + "step": 5853 + }, + { + "epoch": 8.57, + "grad_norm": 0.6329509615898132, + "learning_rate": 8.322797831751561e-06, + "loss": 3.1115, + "step": 5854 + }, + { + "epoch": 8.57, + "grad_norm": 0.5651994347572327, + "learning_rate": 8.304074170392595e-06, + "loss": 3.1165, + "step": 5855 + }, + { + "epoch": 8.57, + "grad_norm": 0.5233526825904846, + "learning_rate": 8.285370681522775e-06, + "loss": 3.1514, + "step": 5856 + }, + { + "epoch": 8.57, + "grad_norm": 0.27295824885368347, + "learning_rate": 8.266687369256798e-06, + "loss": 3.0979, + "step": 5857 + }, + { + "epoch": 8.58, + "grad_norm": 0.3821253478527069, + "learning_rate": 8.248024237704821e-06, + "loss": 3.1523, + "step": 5858 + }, + { + "epoch": 8.58, + "grad_norm": 0.3980197310447693, + "learning_rate": 8.229381290972583e-06, + "loss": 3.1384, + "step": 5859 + }, + { + "epoch": 8.58, + "grad_norm": 0.3785715103149414, + "learning_rate": 8.210758533161488e-06, + "loss": 3.1334, + "step": 5860 + }, + { + "epoch": 8.58, + "grad_norm": 0.43636593222618103, + "learning_rate": 8.192155968368353e-06, + "loss": 3.174, + "step": 5861 + }, + { + "epoch": 8.58, + "grad_norm": 0.35782870650291443, + "learning_rate": 8.173573600685614e-06, + "loss": 3.1439, + "step": 5862 + }, + { + "epoch": 8.58, + "grad_norm": 0.28232669830322266, + "learning_rate": 8.155011434201333e-06, + "loss": 3.1474, + "step": 5863 + }, + { + "epoch": 8.58, + "grad_norm": 0.4299192428588867, + "learning_rate": 8.136469472998986e-06, + "loss": 3.151, + "step": 5864 + }, + { + "epoch": 8.59, + "grad_norm": 0.24973128736019135, + "learning_rate": 8.117947721157726e-06, + "loss": 3.1622, + "step": 5865 + }, + { + "epoch": 8.59, + "grad_norm": 0.3395419716835022, + "learning_rate": 8.099446182752202e-06, + "loss": 3.1173, + "step": 5866 + }, + { + "epoch": 8.59, + "grad_norm": 0.328425794839859, + "learning_rate": 8.080964861852625e-06, + "loss": 3.177, + "step": 5867 + }, + { + "epoch": 8.59, + "grad_norm": 0.509938657283783, + "learning_rate": 8.062503762524754e-06, + "loss": 3.174, + "step": 5868 + }, + { + "epoch": 8.59, + "grad_norm": 0.16413269937038422, + "learning_rate": 8.044062888829928e-06, + "loss": 3.1419, + "step": 5869 + }, + { + "epoch": 8.59, + "grad_norm": 0.6065314412117004, + "learning_rate": 8.025642244825005e-06, + "loss": 3.1232, + "step": 5870 + }, + { + "epoch": 8.59, + "grad_norm": 0.5938077569007874, + "learning_rate": 8.007241834562395e-06, + "loss": 3.1267, + "step": 5871 + }, + { + "epoch": 8.6, + "grad_norm": 0.2933906316757202, + "learning_rate": 7.988861662090085e-06, + "loss": 3.1554, + "step": 5872 + }, + { + "epoch": 8.6, + "grad_norm": 0.36898481845855713, + "learning_rate": 7.970501731451573e-06, + "loss": 3.1333, + "step": 5873 + }, + { + "epoch": 8.6, + "grad_norm": 0.5361416935920715, + "learning_rate": 7.95216204668593e-06, + "loss": 3.1383, + "step": 5874 + }, + { + "epoch": 8.6, + "grad_norm": 0.8210395574569702, + "learning_rate": 7.933842611827746e-06, + "loss": 3.1449, + "step": 5875 + }, + { + "epoch": 8.6, + "grad_norm": 0.7221469879150391, + "learning_rate": 7.915543430907202e-06, + "loss": 3.15, + "step": 5876 + }, + { + "epoch": 8.6, + "grad_norm": 0.25091636180877686, + "learning_rate": 7.897264507949986e-06, + "loss": 3.1368, + "step": 5877 + }, + { + "epoch": 8.61, + "grad_norm": 0.9012389183044434, + "learning_rate": 7.879005846977305e-06, + "loss": 3.1382, + "step": 5878 + }, + { + "epoch": 8.61, + "grad_norm": 0.8301836848258972, + "learning_rate": 7.86076745200599e-06, + "loss": 3.1148, + "step": 5879 + }, + { + "epoch": 8.61, + "grad_norm": 0.5587486624717712, + "learning_rate": 7.842549327048365e-06, + "loss": 3.1546, + "step": 5880 + }, + { + "epoch": 8.61, + "grad_norm": 0.4740186333656311, + "learning_rate": 7.824351476112234e-06, + "loss": 3.1153, + "step": 5881 + }, + { + "epoch": 8.61, + "grad_norm": 0.7473950982093811, + "learning_rate": 7.806173903201063e-06, + "loss": 3.1274, + "step": 5882 + }, + { + "epoch": 8.61, + "grad_norm": 1.0179914236068726, + "learning_rate": 7.788016612313798e-06, + "loss": 3.1127, + "step": 5883 + }, + { + "epoch": 8.61, + "grad_norm": 0.5002656579017639, + "learning_rate": 7.769879607444863e-06, + "loss": 3.1289, + "step": 5884 + }, + { + "epoch": 8.62, + "grad_norm": 0.2755470871925354, + "learning_rate": 7.751762892584358e-06, + "loss": 3.1618, + "step": 5885 + }, + { + "epoch": 8.62, + "grad_norm": 0.5702336430549622, + "learning_rate": 7.733666471717772e-06, + "loss": 3.1114, + "step": 5886 + }, + { + "epoch": 8.62, + "grad_norm": 1.010840892791748, + "learning_rate": 7.715590348826196e-06, + "loss": 3.1286, + "step": 5887 + }, + { + "epoch": 8.62, + "grad_norm": 0.9508697986602783, + "learning_rate": 7.69753452788633e-06, + "loss": 3.0807, + "step": 5888 + }, + { + "epoch": 8.62, + "grad_norm": 0.23731788992881775, + "learning_rate": 7.679499012870261e-06, + "loss": 3.1327, + "step": 5889 + }, + { + "epoch": 8.62, + "grad_norm": 0.806941032409668, + "learning_rate": 7.661483807745683e-06, + "loss": 3.1694, + "step": 5890 + }, + { + "epoch": 8.62, + "grad_norm": 0.7129907011985779, + "learning_rate": 7.643488916475882e-06, + "loss": 3.1413, + "step": 5891 + }, + { + "epoch": 8.63, + "grad_norm": 0.8230921030044556, + "learning_rate": 7.625514343019557e-06, + "loss": 3.1352, + "step": 5892 + }, + { + "epoch": 8.63, + "grad_norm": 0.6420795917510986, + "learning_rate": 7.60756009133099e-06, + "loss": 3.1604, + "step": 5893 + }, + { + "epoch": 8.63, + "grad_norm": 0.5177065134048462, + "learning_rate": 7.5896261653600535e-06, + "loss": 3.1341, + "step": 5894 + }, + { + "epoch": 8.63, + "grad_norm": 1.5158321857452393, + "learning_rate": 7.571712569052037e-06, + "loss": 3.1549, + "step": 5895 + }, + { + "epoch": 8.63, + "grad_norm": 0.8775774836540222, + "learning_rate": 7.5538193063478095e-06, + "loss": 3.1505, + "step": 5896 + }, + { + "epoch": 8.63, + "grad_norm": 0.23683175444602966, + "learning_rate": 7.535946381183812e-06, + "loss": 3.1582, + "step": 5897 + }, + { + "epoch": 8.63, + "grad_norm": 0.3066456615924835, + "learning_rate": 7.518093797491943e-06, + "loss": 3.1531, + "step": 5898 + }, + { + "epoch": 8.64, + "grad_norm": 0.45848727226257324, + "learning_rate": 7.50026155919964e-06, + "loss": 3.1185, + "step": 5899 + }, + { + "epoch": 8.64, + "grad_norm": 0.4867948591709137, + "learning_rate": 7.482449670229896e-06, + "loss": 3.1427, + "step": 5900 + }, + { + "epoch": 8.64, + "grad_norm": 0.34094056487083435, + "learning_rate": 7.464658134501201e-06, + "loss": 3.1407, + "step": 5901 + }, + { + "epoch": 8.64, + "grad_norm": 0.3147621750831604, + "learning_rate": 7.446886955927568e-06, + "loss": 3.1378, + "step": 5902 + }, + { + "epoch": 8.64, + "grad_norm": 0.40778762102127075, + "learning_rate": 7.429136138418535e-06, + "loss": 3.1767, + "step": 5903 + }, + { + "epoch": 8.64, + "grad_norm": 0.40993595123291016, + "learning_rate": 7.4114056858791805e-06, + "loss": 3.1326, + "step": 5904 + }, + { + "epoch": 8.65, + "grad_norm": 0.25892430543899536, + "learning_rate": 7.393695602210071e-06, + "loss": 3.1982, + "step": 5905 + }, + { + "epoch": 8.65, + "grad_norm": 0.4945090711116791, + "learning_rate": 7.376005891307302e-06, + "loss": 3.1257, + "step": 5906 + }, + { + "epoch": 8.65, + "grad_norm": 0.39548638463020325, + "learning_rate": 7.358336557062507e-06, + "loss": 3.1185, + "step": 5907 + }, + { + "epoch": 8.65, + "grad_norm": 0.5258082151412964, + "learning_rate": 7.340687603362828e-06, + "loss": 3.1174, + "step": 5908 + }, + { + "epoch": 8.65, + "grad_norm": 0.34763848781585693, + "learning_rate": 7.3230590340908626e-06, + "loss": 3.1277, + "step": 5909 + }, + { + "epoch": 8.65, + "grad_norm": 0.2732061743736267, + "learning_rate": 7.30545085312484e-06, + "loss": 3.1047, + "step": 5910 + }, + { + "epoch": 8.65, + "grad_norm": 0.4295293986797333, + "learning_rate": 7.2878630643384295e-06, + "loss": 3.128, + "step": 5911 + }, + { + "epoch": 8.66, + "grad_norm": 0.41177213191986084, + "learning_rate": 7.270295671600791e-06, + "loss": 3.1348, + "step": 5912 + }, + { + "epoch": 8.66, + "grad_norm": 0.24607589840888977, + "learning_rate": 7.252748678776666e-06, + "loss": 3.1461, + "step": 5913 + }, + { + "epoch": 8.66, + "grad_norm": 0.3988043963909149, + "learning_rate": 7.235222089726279e-06, + "loss": 3.1677, + "step": 5914 + }, + { + "epoch": 8.66, + "grad_norm": 0.48878470063209534, + "learning_rate": 7.217715908305333e-06, + "loss": 3.1104, + "step": 5915 + }, + { + "epoch": 8.66, + "eval_loss": 2.9968373775482178, + "eval_runtime": 1167.9535, + "eval_samples_per_second": 14.677, + "eval_steps_per_second": 0.918, + "step": 5915 + }, + { + "epoch": 8.66, + "grad_norm": 0.22077223658561707, + "learning_rate": 7.200230138365094e-06, + "loss": 3.1239, + "step": 5916 + }, + { + "epoch": 8.66, + "grad_norm": 0.3054116368293762, + "learning_rate": 7.182764783752327e-06, + "loss": 3.1718, + "step": 5917 + }, + { + "epoch": 8.66, + "grad_norm": 0.3858598470687866, + "learning_rate": 7.165319848309238e-06, + "loss": 3.1447, + "step": 5918 + }, + { + "epoch": 8.67, + "grad_norm": 0.6377255916595459, + "learning_rate": 7.1478953358736665e-06, + "loss": 3.1457, + "step": 5919 + }, + { + "epoch": 8.67, + "grad_norm": 0.4466152787208557, + "learning_rate": 7.1304912502788365e-06, + "loss": 3.1208, + "step": 5920 + }, + { + "epoch": 8.67, + "grad_norm": 0.528590738773346, + "learning_rate": 7.11310759535353e-06, + "loss": 3.1644, + "step": 5921 + }, + { + "epoch": 8.67, + "grad_norm": 0.32386285066604614, + "learning_rate": 7.095744374922087e-06, + "loss": 3.1177, + "step": 5922 + }, + { + "epoch": 8.67, + "grad_norm": 0.3173229694366455, + "learning_rate": 7.0784015928042516e-06, + "loss": 3.1442, + "step": 5923 + }, + { + "epoch": 8.67, + "grad_norm": 0.32550540566444397, + "learning_rate": 7.061079252815328e-06, + "loss": 3.1537, + "step": 5924 + }, + { + "epoch": 8.67, + "grad_norm": 0.37319254875183105, + "learning_rate": 7.043777358766124e-06, + "loss": 3.1576, + "step": 5925 + }, + { + "epoch": 8.68, + "grad_norm": 0.30680203437805176, + "learning_rate": 7.026495914462938e-06, + "loss": 3.1132, + "step": 5926 + }, + { + "epoch": 8.68, + "grad_norm": 0.37015992403030396, + "learning_rate": 7.009234923707564e-06, + "loss": 3.1623, + "step": 5927 + }, + { + "epoch": 8.68, + "grad_norm": 0.4574234187602997, + "learning_rate": 6.991994390297307e-06, + "loss": 3.0993, + "step": 5928 + }, + { + "epoch": 8.68, + "grad_norm": 0.39552009105682373, + "learning_rate": 6.974774318024979e-06, + "loss": 3.1413, + "step": 5929 + }, + { + "epoch": 8.68, + "grad_norm": 0.6035201549530029, + "learning_rate": 6.9575747106788715e-06, + "loss": 3.1387, + "step": 5930 + }, + { + "epoch": 8.68, + "grad_norm": 0.3027237355709076, + "learning_rate": 6.9403955720427905e-06, + "loss": 3.127, + "step": 5931 + }, + { + "epoch": 8.69, + "grad_norm": 0.5305870175361633, + "learning_rate": 6.923236905896024e-06, + "loss": 3.1486, + "step": 5932 + }, + { + "epoch": 8.69, + "grad_norm": 0.3524973690509796, + "learning_rate": 6.906098716013354e-06, + "loss": 3.0994, + "step": 5933 + }, + { + "epoch": 8.69, + "grad_norm": 0.2866142690181732, + "learning_rate": 6.888981006165096e-06, + "loss": 3.1436, + "step": 5934 + }, + { + "epoch": 8.69, + "grad_norm": 0.3267587721347809, + "learning_rate": 6.871883780117005e-06, + "loss": 3.1174, + "step": 5935 + }, + { + "epoch": 8.69, + "grad_norm": 0.573542594909668, + "learning_rate": 6.854807041630362e-06, + "loss": 3.144, + "step": 5936 + }, + { + "epoch": 8.69, + "grad_norm": 0.23285147547721863, + "learning_rate": 6.837750794461939e-06, + "loss": 3.1232, + "step": 5937 + }, + { + "epoch": 8.69, + "grad_norm": 0.5822926163673401, + "learning_rate": 6.820715042364001e-06, + "loss": 3.1213, + "step": 5938 + }, + { + "epoch": 8.7, + "grad_norm": 0.42947494983673096, + "learning_rate": 6.803699789084295e-06, + "loss": 3.1078, + "step": 5939 + }, + { + "epoch": 8.7, + "grad_norm": 0.4609920382499695, + "learning_rate": 6.786705038366059e-06, + "loss": 3.1454, + "step": 5940 + }, + { + "epoch": 8.7, + "grad_norm": 0.46844562888145447, + "learning_rate": 6.7697307939480255e-06, + "loss": 3.1492, + "step": 5941 + }, + { + "epoch": 8.7, + "grad_norm": 0.4904913902282715, + "learning_rate": 6.75277705956443e-06, + "loss": 3.1276, + "step": 5942 + }, + { + "epoch": 8.7, + "grad_norm": 0.4832492470741272, + "learning_rate": 6.735843838944944e-06, + "loss": 3.128, + "step": 5943 + }, + { + "epoch": 8.7, + "grad_norm": 0.898986279964447, + "learning_rate": 6.718931135814787e-06, + "loss": 3.1299, + "step": 5944 + }, + { + "epoch": 8.7, + "grad_norm": 0.9148275256156921, + "learning_rate": 6.702038953894663e-06, + "loss": 3.1266, + "step": 5945 + }, + { + "epoch": 8.71, + "grad_norm": 0.2876090407371521, + "learning_rate": 6.685167296900685e-06, + "loss": 3.1499, + "step": 5946 + }, + { + "epoch": 8.71, + "grad_norm": 0.33294424414634705, + "learning_rate": 6.668316168544542e-06, + "loss": 3.1116, + "step": 5947 + }, + { + "epoch": 8.71, + "grad_norm": 0.3966740071773529, + "learning_rate": 6.651485572533378e-06, + "loss": 3.1724, + "step": 5948 + }, + { + "epoch": 8.71, + "grad_norm": 0.2978176176548004, + "learning_rate": 6.6346755125697765e-06, + "loss": 3.1483, + "step": 5949 + }, + { + "epoch": 8.71, + "grad_norm": 0.258612722158432, + "learning_rate": 6.617885992351846e-06, + "loss": 3.1448, + "step": 5950 + }, + { + "epoch": 8.71, + "grad_norm": 0.33197665214538574, + "learning_rate": 6.6011170155732e-06, + "loss": 3.1419, + "step": 5951 + }, + { + "epoch": 8.72, + "grad_norm": 0.4256979823112488, + "learning_rate": 6.584368585922874e-06, + "loss": 3.1285, + "step": 5952 + }, + { + "epoch": 8.72, + "grad_norm": 0.26601269841194153, + "learning_rate": 6.56764070708541e-06, + "loss": 3.1451, + "step": 5953 + }, + { + "epoch": 8.72, + "grad_norm": 0.37005001306533813, + "learning_rate": 6.5509333827408315e-06, + "loss": 3.1264, + "step": 5954 + }, + { + "epoch": 8.72, + "grad_norm": 0.4265497028827667, + "learning_rate": 6.534246616564654e-06, + "loss": 3.1624, + "step": 5955 + }, + { + "epoch": 8.72, + "grad_norm": 0.283138632774353, + "learning_rate": 6.5175804122278304e-06, + "loss": 3.1231, + "step": 5956 + }, + { + "epoch": 8.72, + "grad_norm": 0.4558214843273163, + "learning_rate": 6.500934773396827e-06, + "loss": 3.14, + "step": 5957 + }, + { + "epoch": 8.72, + "grad_norm": 0.23857197165489197, + "learning_rate": 6.48430970373356e-06, + "loss": 3.1686, + "step": 5958 + }, + { + "epoch": 8.73, + "grad_norm": 0.40693220496177673, + "learning_rate": 6.467705206895458e-06, + "loss": 3.1246, + "step": 5959 + }, + { + "epoch": 8.73, + "grad_norm": 0.3224029839038849, + "learning_rate": 6.451121286535378e-06, + "loss": 3.1159, + "step": 5960 + }, + { + "epoch": 8.73, + "grad_norm": 0.36411646008491516, + "learning_rate": 6.434557946301678e-06, + "loss": 3.1206, + "step": 5961 + }, + { + "epoch": 8.73, + "grad_norm": 0.4473271369934082, + "learning_rate": 6.418015189838189e-06, + "loss": 3.1924, + "step": 5962 + }, + { + "epoch": 8.73, + "grad_norm": 0.4669060707092285, + "learning_rate": 6.401493020784199e-06, + "loss": 3.1359, + "step": 5963 + }, + { + "epoch": 8.73, + "grad_norm": 0.4600774347782135, + "learning_rate": 6.384991442774468e-06, + "loss": 3.101, + "step": 5964 + }, + { + "epoch": 8.73, + "grad_norm": 0.3088182210922241, + "learning_rate": 6.368510459439248e-06, + "loss": 3.093, + "step": 5965 + }, + { + "epoch": 8.74, + "grad_norm": 0.5628402829170227, + "learning_rate": 6.352050074404226e-06, + "loss": 3.1187, + "step": 5966 + }, + { + "epoch": 8.74, + "grad_norm": 0.2512408196926117, + "learning_rate": 6.335610291290594e-06, + "loss": 3.1644, + "step": 5967 + }, + { + "epoch": 8.74, + "grad_norm": 0.3120940327644348, + "learning_rate": 6.319191113714984e-06, + "loss": 3.1145, + "step": 5968 + }, + { + "epoch": 8.74, + "grad_norm": 0.3800273835659027, + "learning_rate": 6.3027925452895154e-06, + "loss": 3.1278, + "step": 5969 + }, + { + "epoch": 8.74, + "grad_norm": 0.28590822219848633, + "learning_rate": 6.2864145896217475e-06, + "loss": 3.1181, + "step": 5970 + }, + { + "epoch": 8.74, + "grad_norm": 0.7252241373062134, + "learning_rate": 6.270057250314731e-06, + "loss": 3.1463, + "step": 5971 + }, + { + "epoch": 8.74, + "grad_norm": 0.3182050287723541, + "learning_rate": 6.2537205309669665e-06, + "loss": 3.1443, + "step": 5972 + }, + { + "epoch": 8.75, + "grad_norm": 0.2928391098976135, + "learning_rate": 6.237404435172434e-06, + "loss": 3.1282, + "step": 5973 + }, + { + "epoch": 8.75, + "grad_norm": 0.3506748676300049, + "learning_rate": 6.221108966520561e-06, + "loss": 3.156, + "step": 5974 + }, + { + "epoch": 8.75, + "grad_norm": 0.42503002285957336, + "learning_rate": 6.204834128596215e-06, + "loss": 3.088, + "step": 5975 + }, + { + "epoch": 8.75, + "grad_norm": 0.7118878960609436, + "learning_rate": 6.1885799249797985e-06, + "loss": 3.1189, + "step": 5976 + }, + { + "epoch": 8.75, + "grad_norm": 0.6801698207855225, + "learning_rate": 6.172346359247094e-06, + "loss": 3.1899, + "step": 5977 + }, + { + "epoch": 8.75, + "grad_norm": 0.5313051342964172, + "learning_rate": 6.156133434969369e-06, + "loss": 3.1453, + "step": 5978 + }, + { + "epoch": 8.76, + "grad_norm": 0.5182479023933411, + "learning_rate": 6.139941155713402e-06, + "loss": 3.1668, + "step": 5979 + }, + { + "epoch": 8.76, + "grad_norm": 0.40206000208854675, + "learning_rate": 6.123769525041345e-06, + "loss": 3.1273, + "step": 5980 + }, + { + "epoch": 8.76, + "grad_norm": 0.7351994514465332, + "learning_rate": 6.10761854651084e-06, + "loss": 3.1369, + "step": 5981 + }, + { + "epoch": 8.76, + "grad_norm": 0.2961914837360382, + "learning_rate": 6.091488223675057e-06, + "loss": 3.1798, + "step": 5982 + }, + { + "epoch": 8.76, + "grad_norm": 0.45421791076660156, + "learning_rate": 6.075378560082501e-06, + "loss": 3.1241, + "step": 5983 + }, + { + "epoch": 8.76, + "grad_norm": 0.4831741750240326, + "learning_rate": 6.059289559277181e-06, + "loss": 3.1234, + "step": 5984 + }, + { + "epoch": 8.76, + "grad_norm": 0.36750924587249756, + "learning_rate": 6.043221224798634e-06, + "loss": 3.1557, + "step": 5985 + }, + { + "epoch": 8.77, + "grad_norm": 0.3191670775413513, + "learning_rate": 6.027173560181731e-06, + "loss": 3.1495, + "step": 5986 + }, + { + "epoch": 8.77, + "grad_norm": 0.47878777980804443, + "learning_rate": 6.011146568956871e-06, + "loss": 3.1158, + "step": 5987 + }, + { + "epoch": 8.77, + "grad_norm": 0.5439360737800598, + "learning_rate": 5.995140254649878e-06, + "loss": 3.128, + "step": 5988 + }, + { + "epoch": 8.77, + "grad_norm": 0.37722069025039673, + "learning_rate": 5.979154620782035e-06, + "loss": 3.1135, + "step": 5989 + }, + { + "epoch": 8.77, + "grad_norm": 0.27893126010894775, + "learning_rate": 5.963189670870073e-06, + "loss": 3.1115, + "step": 5990 + }, + { + "epoch": 8.77, + "grad_norm": 0.3667724132537842, + "learning_rate": 5.947245408426172e-06, + "loss": 3.1513, + "step": 5991 + }, + { + "epoch": 8.77, + "grad_norm": 0.34727010130882263, + "learning_rate": 5.931321836957982e-06, + "loss": 3.1473, + "step": 5992 + }, + { + "epoch": 8.78, + "grad_norm": 0.5472817420959473, + "learning_rate": 5.915418959968555e-06, + "loss": 3.1246, + "step": 5993 + }, + { + "epoch": 8.78, + "grad_norm": 0.46271124482154846, + "learning_rate": 5.899536780956438e-06, + "loss": 3.1131, + "step": 5994 + }, + { + "epoch": 8.78, + "grad_norm": 0.39179834723472595, + "learning_rate": 5.8836753034155925e-06, + "loss": 3.1154, + "step": 5995 + }, + { + "epoch": 8.78, + "grad_norm": 0.5964497923851013, + "learning_rate": 5.867834530835437e-06, + "loss": 3.1636, + "step": 5996 + }, + { + "epoch": 8.78, + "grad_norm": 0.3770958483219147, + "learning_rate": 5.85201446670085e-06, + "loss": 3.1155, + "step": 5997 + }, + { + "epoch": 8.78, + "grad_norm": 0.6412743330001831, + "learning_rate": 5.8362151144921165e-06, + "loss": 3.1764, + "step": 5998 + }, + { + "epoch": 8.78, + "grad_norm": 0.39634811878204346, + "learning_rate": 5.82043647768502e-06, + "loss": 3.134, + "step": 5999 + }, + { + "epoch": 8.79, + "grad_norm": 0.2597801089286804, + "learning_rate": 5.8046785597507096e-06, + "loss": 3.1321, + "step": 6000 + }, + { + "epoch": 8.79, + "grad_norm": 0.4642222821712494, + "learning_rate": 5.7889413641558535e-06, + "loss": 3.1118, + "step": 6001 + }, + { + "epoch": 8.79, + "grad_norm": 0.8907680511474609, + "learning_rate": 5.77322489436255e-06, + "loss": 3.1598, + "step": 6002 + }, + { + "epoch": 8.79, + "grad_norm": 0.6435474753379822, + "learning_rate": 5.757529153828245e-06, + "loss": 3.1429, + "step": 6003 + }, + { + "epoch": 8.79, + "grad_norm": 6.3963494300842285, + "learning_rate": 5.741854146005965e-06, + "loss": 3.2169, + "step": 6004 + }, + { + "epoch": 8.79, + "grad_norm": 0.403425931930542, + "learning_rate": 5.726199874344096e-06, + "loss": 3.134, + "step": 6005 + }, + { + "epoch": 8.8, + "grad_norm": 0.4647369980812073, + "learning_rate": 5.710566342286428e-06, + "loss": 3.1384, + "step": 6006 + }, + { + "epoch": 8.8, + "grad_norm": 0.27036553621292114, + "learning_rate": 5.6949535532722864e-06, + "loss": 3.1547, + "step": 6007 + }, + { + "epoch": 8.8, + "grad_norm": 0.27433207631111145, + "learning_rate": 5.6793615107363695e-06, + "loss": 3.1312, + "step": 6008 + }, + { + "epoch": 8.8, + "grad_norm": 0.4025347828865051, + "learning_rate": 5.6637902181087774e-06, + "loss": 3.1355, + "step": 6009 + }, + { + "epoch": 8.8, + "grad_norm": 0.4300908148288727, + "learning_rate": 5.64823967881516e-06, + "loss": 3.1247, + "step": 6010 + }, + { + "epoch": 8.8, + "grad_norm": 0.42426836490631104, + "learning_rate": 5.632709896276478e-06, + "loss": 3.1543, + "step": 6011 + }, + { + "epoch": 8.8, + "grad_norm": 0.43459025025367737, + "learning_rate": 5.617200873909178e-06, + "loss": 3.109, + "step": 6012 + }, + { + "epoch": 8.81, + "grad_norm": 0.42334458231925964, + "learning_rate": 5.601712615125199e-06, + "loss": 3.1084, + "step": 6013 + }, + { + "epoch": 8.81, + "grad_norm": 0.33407434821128845, + "learning_rate": 5.58624512333179e-06, + "loss": 3.1399, + "step": 6014 + }, + { + "epoch": 8.81, + "grad_norm": 0.29522469639778137, + "learning_rate": 5.570798401931721e-06, + "loss": 3.126, + "step": 6015 + }, + { + "epoch": 8.81, + "grad_norm": 0.3701851963996887, + "learning_rate": 5.5553724543231825e-06, + "loss": 3.1287, + "step": 6016 + }, + { + "epoch": 8.81, + "grad_norm": 0.4674670994281769, + "learning_rate": 5.539967283899749e-06, + "loss": 3.1133, + "step": 6017 + }, + { + "epoch": 8.81, + "grad_norm": 0.8055880665779114, + "learning_rate": 5.5245828940504675e-06, + "loss": 3.1813, + "step": 6018 + }, + { + "epoch": 8.81, + "grad_norm": 0.18001839518547058, + "learning_rate": 5.5092192881598286e-06, + "loss": 3.158, + "step": 6019 + }, + { + "epoch": 8.82, + "grad_norm": 0.4075320065021515, + "learning_rate": 5.4938764696076726e-06, + "loss": 3.1312, + "step": 6020 + }, + { + "epoch": 8.82, + "grad_norm": 0.7519275546073914, + "learning_rate": 5.478554441769357e-06, + "loss": 3.1851, + "step": 6021 + }, + { + "epoch": 8.82, + "grad_norm": 0.7654166221618652, + "learning_rate": 5.463253208015595e-06, + "loss": 3.101, + "step": 6022 + }, + { + "epoch": 8.82, + "grad_norm": 0.5612555742263794, + "learning_rate": 5.447972771712584e-06, + "loss": 3.1467, + "step": 6023 + }, + { + "epoch": 8.82, + "grad_norm": 0.3048601448535919, + "learning_rate": 5.4327131362218895e-06, + "loss": 3.1308, + "step": 6024 + }, + { + "epoch": 8.82, + "grad_norm": 0.3827259838581085, + "learning_rate": 5.4174743049005495e-06, + "loss": 3.1142, + "step": 6025 + }, + { + "epoch": 8.82, + "grad_norm": 0.5008516311645508, + "learning_rate": 5.402256281101004e-06, + "loss": 3.1305, + "step": 6026 + }, + { + "epoch": 8.83, + "grad_norm": 0.5051065683364868, + "learning_rate": 5.387059068171107e-06, + "loss": 3.1348, + "step": 6027 + }, + { + "epoch": 8.83, + "grad_norm": 0.6256129145622253, + "learning_rate": 5.3718826694541425e-06, + "loss": 3.1257, + "step": 6028 + }, + { + "epoch": 8.83, + "grad_norm": 0.838629961013794, + "learning_rate": 5.356727088288826e-06, + "loss": 3.1441, + "step": 6029 + }, + { + "epoch": 8.83, + "grad_norm": 0.42572328448295593, + "learning_rate": 5.341592328009282e-06, + "loss": 3.1358, + "step": 6030 + }, + { + "epoch": 8.83, + "grad_norm": 0.5678095817565918, + "learning_rate": 5.326478391945045e-06, + "loss": 3.1257, + "step": 6031 + }, + { + "epoch": 8.83, + "grad_norm": 0.608961284160614, + "learning_rate": 5.311385283421089e-06, + "loss": 3.1949, + "step": 6032 + }, + { + "epoch": 8.84, + "grad_norm": 0.48100271821022034, + "learning_rate": 5.296313005757813e-06, + "loss": 3.1098, + "step": 6033 + }, + { + "epoch": 8.84, + "grad_norm": 0.370830774307251, + "learning_rate": 5.281261562270978e-06, + "loss": 3.1347, + "step": 6034 + }, + { + "epoch": 8.84, + "grad_norm": 0.3011663258075714, + "learning_rate": 5.266230956271834e-06, + "loss": 3.158, + "step": 6035 + }, + { + "epoch": 8.84, + "grad_norm": 0.573705792427063, + "learning_rate": 5.251221191067013e-06, + "loss": 3.165, + "step": 6036 + }, + { + "epoch": 8.84, + "grad_norm": 0.9574636816978455, + "learning_rate": 5.236232269958519e-06, + "loss": 3.1486, + "step": 6037 + }, + { + "epoch": 8.84, + "grad_norm": 0.2764081656932831, + "learning_rate": 5.221264196243869e-06, + "loss": 3.1264, + "step": 6038 + }, + { + "epoch": 8.84, + "grad_norm": 0.3352329432964325, + "learning_rate": 5.206316973215919e-06, + "loss": 3.105, + "step": 6039 + }, + { + "epoch": 8.85, + "grad_norm": 0.5301186442375183, + "learning_rate": 5.191390604162938e-06, + "loss": 3.1347, + "step": 6040 + }, + { + "epoch": 8.85, + "grad_norm": 0.44588613510131836, + "learning_rate": 5.176485092368655e-06, + "loss": 3.143, + "step": 6041 + }, + { + "epoch": 8.85, + "grad_norm": 0.5561344027519226, + "learning_rate": 5.161600441112191e-06, + "loss": 3.1433, + "step": 6042 + }, + { + "epoch": 8.85, + "grad_norm": 0.3855009973049164, + "learning_rate": 5.146736653668016e-06, + "loss": 3.1325, + "step": 6043 + }, + { + "epoch": 8.85, + "grad_norm": 0.4017171263694763, + "learning_rate": 5.131893733306137e-06, + "loss": 3.1241, + "step": 6044 + }, + { + "epoch": 8.85, + "grad_norm": 0.6462271213531494, + "learning_rate": 5.1170716832918405e-06, + "loss": 3.1079, + "step": 6045 + }, + { + "epoch": 8.85, + "grad_norm": 0.6944392323493958, + "learning_rate": 5.102270506885898e-06, + "loss": 3.1106, + "step": 6046 + }, + { + "epoch": 8.86, + "grad_norm": 0.43211668729782104, + "learning_rate": 5.087490207344458e-06, + "loss": 3.1481, + "step": 6047 + }, + { + "epoch": 8.86, + "grad_norm": 0.48566940426826477, + "learning_rate": 5.072730787919111e-06, + "loss": 3.1552, + "step": 6048 + }, + { + "epoch": 8.86, + "grad_norm": 0.3225969672203064, + "learning_rate": 5.057992251856802e-06, + "loss": 3.1173, + "step": 6049 + }, + { + "epoch": 8.86, + "grad_norm": 0.6626749634742737, + "learning_rate": 5.043274602399939e-06, + "loss": 3.1274, + "step": 6050 + }, + { + "epoch": 8.86, + "grad_norm": 0.5008309483528137, + "learning_rate": 5.028577842786287e-06, + "loss": 3.1181, + "step": 6051 + }, + { + "epoch": 8.86, + "grad_norm": 0.726362943649292, + "learning_rate": 5.013901976249036e-06, + "loss": 3.1433, + "step": 6052 + }, + { + "epoch": 8.87, + "grad_norm": 0.2972591817378998, + "learning_rate": 4.999247006016783e-06, + "loss": 3.138, + "step": 6053 + }, + { + "epoch": 8.87, + "grad_norm": 0.3870035409927368, + "learning_rate": 4.984612935313515e-06, + "loss": 3.2039, + "step": 6054 + }, + { + "epoch": 8.87, + "grad_norm": 0.6251978278160095, + "learning_rate": 4.969999767358646e-06, + "loss": 3.1232, + "step": 6055 + }, + { + "epoch": 8.87, + "grad_norm": 0.4746638238430023, + "learning_rate": 4.9554075053669605e-06, + "loss": 3.1353, + "step": 6056 + }, + { + "epoch": 8.87, + "grad_norm": 0.26788651943206787, + "learning_rate": 4.9408361525486665e-06, + "loss": 3.1221, + "step": 6057 + }, + { + "epoch": 8.87, + "grad_norm": 0.3060562014579773, + "learning_rate": 4.9262857121093575e-06, + "loss": 3.1414, + "step": 6058 + }, + { + "epoch": 8.87, + "grad_norm": 0.36984533071517944, + "learning_rate": 4.911756187250039e-06, + "loss": 3.1289, + "step": 6059 + }, + { + "epoch": 8.88, + "grad_norm": 0.28999122977256775, + "learning_rate": 4.897247581167108e-06, + "loss": 3.1094, + "step": 6060 + }, + { + "epoch": 8.88, + "grad_norm": 0.3002908229827881, + "learning_rate": 4.882759897052358e-06, + "loss": 3.1673, + "step": 6061 + }, + { + "epoch": 8.88, + "grad_norm": 0.3484486937522888, + "learning_rate": 4.868293138092972e-06, + "loss": 3.1189, + "step": 6062 + }, + { + "epoch": 8.88, + "grad_norm": 0.25520235300064087, + "learning_rate": 4.85384730747156e-06, + "loss": 3.1485, + "step": 6063 + }, + { + "epoch": 8.88, + "grad_norm": 0.5077893733978271, + "learning_rate": 4.839422408366101e-06, + "loss": 3.1668, + "step": 6064 + }, + { + "epoch": 8.88, + "grad_norm": 0.28061893582344055, + "learning_rate": 4.82501844394998e-06, + "loss": 3.1337, + "step": 6065 + }, + { + "epoch": 8.88, + "grad_norm": 0.3427242636680603, + "learning_rate": 4.810635417391951e-06, + "loss": 3.1098, + "step": 6066 + }, + { + "epoch": 8.89, + "grad_norm": 0.23328736424446106, + "learning_rate": 4.796273331856227e-06, + "loss": 3.1272, + "step": 6067 + }, + { + "epoch": 8.89, + "grad_norm": 0.3861193358898163, + "learning_rate": 4.781932190502325e-06, + "loss": 3.1417, + "step": 6068 + }, + { + "epoch": 8.89, + "grad_norm": 0.5966961979866028, + "learning_rate": 4.7676119964852e-06, + "loss": 3.139, + "step": 6069 + }, + { + "epoch": 8.89, + "grad_norm": 0.24302925169467926, + "learning_rate": 4.753312752955252e-06, + "loss": 3.1153, + "step": 6070 + }, + { + "epoch": 8.89, + "grad_norm": 0.39507362246513367, + "learning_rate": 4.7390344630581565e-06, + "loss": 3.1404, + "step": 6071 + }, + { + "epoch": 8.89, + "grad_norm": 0.3054639995098114, + "learning_rate": 4.7247771299350655e-06, + "loss": 3.1653, + "step": 6072 + }, + { + "epoch": 8.89, + "grad_norm": 0.23265469074249268, + "learning_rate": 4.710540756722526e-06, + "loss": 3.129, + "step": 6073 + }, + { + "epoch": 8.9, + "grad_norm": 0.27230361104011536, + "learning_rate": 4.696325346552399e-06, + "loss": 3.1594, + "step": 6074 + }, + { + "epoch": 8.9, + "grad_norm": 0.317284494638443, + "learning_rate": 4.682130902551995e-06, + "loss": 3.1249, + "step": 6075 + }, + { + "epoch": 8.9, + "grad_norm": 0.33463653922080994, + "learning_rate": 4.6679574278440275e-06, + "loss": 3.1616, + "step": 6076 + }, + { + "epoch": 8.9, + "grad_norm": 0.25718221068382263, + "learning_rate": 4.653804925546512e-06, + "loss": 3.1111, + "step": 6077 + }, + { + "epoch": 8.9, + "grad_norm": 0.263304740190506, + "learning_rate": 4.639673398772937e-06, + "loss": 3.1289, + "step": 6078 + }, + { + "epoch": 8.9, + "grad_norm": 0.5322175025939941, + "learning_rate": 4.625562850632148e-06, + "loss": 3.1378, + "step": 6079 + }, + { + "epoch": 8.91, + "grad_norm": 1.2477030754089355, + "learning_rate": 4.6114732842283515e-06, + "loss": 3.1447, + "step": 6080 + }, + { + "epoch": 8.91, + "grad_norm": 0.2595040202140808, + "learning_rate": 4.597404702661167e-06, + "loss": 3.1427, + "step": 6081 + }, + { + "epoch": 8.91, + "grad_norm": 0.3155920207500458, + "learning_rate": 4.583357109025588e-06, + "loss": 3.1459, + "step": 6082 + }, + { + "epoch": 8.91, + "grad_norm": 0.3636661469936371, + "learning_rate": 4.569330506411995e-06, + "loss": 3.1673, + "step": 6083 + }, + { + "epoch": 8.91, + "grad_norm": 0.31277310848236084, + "learning_rate": 4.555324897906132e-06, + "loss": 3.1236, + "step": 6084 + }, + { + "epoch": 8.91, + "eval_loss": 2.993858814239502, + "eval_runtime": 1169.2074, + "eval_samples_per_second": 14.661, + "eval_steps_per_second": 0.917, + "step": 6084 + }, + { + "epoch": 8.91, + "grad_norm": 0.44307875633239746, + "learning_rate": 4.541340286589169e-06, + "loss": 3.1468, + "step": 6085 + }, + { + "epoch": 8.91, + "grad_norm": 0.37170204520225525, + "learning_rate": 4.527376675537587e-06, + "loss": 3.1259, + "step": 6086 + }, + { + "epoch": 8.92, + "grad_norm": 0.40819883346557617, + "learning_rate": 4.513434067823319e-06, + "loss": 3.1084, + "step": 6087 + }, + { + "epoch": 8.92, + "grad_norm": 0.2787991166114807, + "learning_rate": 4.499512466513633e-06, + "loss": 3.1221, + "step": 6088 + }, + { + "epoch": 8.92, + "grad_norm": 0.3242098391056061, + "learning_rate": 4.485611874671181e-06, + "loss": 3.1169, + "step": 6089 + }, + { + "epoch": 8.92, + "grad_norm": 0.5514737963676453, + "learning_rate": 4.471732295354014e-06, + "loss": 3.1218, + "step": 6090 + }, + { + "epoch": 8.92, + "grad_norm": 1.0488824844360352, + "learning_rate": 4.457873731615536e-06, + "loss": 3.1422, + "step": 6091 + }, + { + "epoch": 8.92, + "grad_norm": 0.3109232485294342, + "learning_rate": 4.444036186504552e-06, + "loss": 3.1272, + "step": 6092 + }, + { + "epoch": 8.92, + "grad_norm": 0.3036029636859894, + "learning_rate": 4.430219663065227e-06, + "loss": 3.149, + "step": 6093 + }, + { + "epoch": 8.93, + "grad_norm": 0.2809702754020691, + "learning_rate": 4.416424164337063e-06, + "loss": 3.14, + "step": 6094 + }, + { + "epoch": 8.93, + "grad_norm": 0.35515812039375305, + "learning_rate": 4.402649693355032e-06, + "loss": 3.1336, + "step": 6095 + }, + { + "epoch": 8.93, + "grad_norm": 0.2919917106628418, + "learning_rate": 4.38889625314941e-06, + "loss": 3.1479, + "step": 6096 + }, + { + "epoch": 8.93, + "grad_norm": 0.6312490105628967, + "learning_rate": 4.375163846745833e-06, + "loss": 3.149, + "step": 6097 + }, + { + "epoch": 8.93, + "grad_norm": 0.4424973130226135, + "learning_rate": 4.361452477165373e-06, + "loss": 3.113, + "step": 6098 + }, + { + "epoch": 8.93, + "grad_norm": 0.2590726613998413, + "learning_rate": 4.347762147424439e-06, + "loss": 3.1016, + "step": 6099 + }, + { + "epoch": 8.93, + "grad_norm": 0.3446723520755768, + "learning_rate": 4.334092860534777e-06, + "loss": 3.1077, + "step": 6100 + }, + { + "epoch": 8.94, + "grad_norm": 0.8220780491828918, + "learning_rate": 4.320444619503583e-06, + "loss": 3.1439, + "step": 6101 + }, + { + "epoch": 8.94, + "grad_norm": 0.4808022081851959, + "learning_rate": 4.306817427333343e-06, + "loss": 3.1463, + "step": 6102 + }, + { + "epoch": 8.94, + "grad_norm": 0.2719718813896179, + "learning_rate": 4.293211287021947e-06, + "loss": 3.1431, + "step": 6103 + }, + { + "epoch": 8.94, + "grad_norm": 0.2795056402683258, + "learning_rate": 4.279626201562703e-06, + "loss": 3.167, + "step": 6104 + }, + { + "epoch": 8.94, + "grad_norm": 0.3881921172142029, + "learning_rate": 4.2660621739441855e-06, + "loss": 3.1496, + "step": 6105 + }, + { + "epoch": 8.94, + "grad_norm": 0.8131992816925049, + "learning_rate": 4.252519207150407e-06, + "loss": 3.1027, + "step": 6106 + }, + { + "epoch": 8.95, + "grad_norm": 0.6725051999092102, + "learning_rate": 4.238997304160741e-06, + "loss": 3.1003, + "step": 6107 + }, + { + "epoch": 8.95, + "grad_norm": 0.38035905361175537, + "learning_rate": 4.225496467949907e-06, + "loss": 3.1203, + "step": 6108 + }, + { + "epoch": 8.95, + "grad_norm": 0.28128498792648315, + "learning_rate": 4.212016701487975e-06, + "loss": 3.1223, + "step": 6109 + }, + { + "epoch": 8.95, + "grad_norm": 1.1468039751052856, + "learning_rate": 4.198558007740461e-06, + "loss": 3.1659, + "step": 6110 + }, + { + "epoch": 8.95, + "grad_norm": 0.5970154404640198, + "learning_rate": 4.18512038966814e-06, + "loss": 3.1411, + "step": 6111 + }, + { + "epoch": 8.95, + "grad_norm": 0.38064298033714294, + "learning_rate": 4.171703850227204e-06, + "loss": 3.1204, + "step": 6112 + }, + { + "epoch": 8.95, + "grad_norm": 0.6189457178115845, + "learning_rate": 4.1583083923692145e-06, + "loss": 3.1542, + "step": 6113 + }, + { + "epoch": 8.96, + "grad_norm": 0.6810131072998047, + "learning_rate": 4.144934019041069e-06, + "loss": 3.1261, + "step": 6114 + }, + { + "epoch": 8.96, + "grad_norm": 0.2683347463607788, + "learning_rate": 4.131580733185048e-06, + "loss": 3.1208, + "step": 6115 + }, + { + "epoch": 8.96, + "grad_norm": 0.42770662903785706, + "learning_rate": 4.118248537738789e-06, + "loss": 3.1367, + "step": 6116 + }, + { + "epoch": 8.96, + "grad_norm": 0.4592244029045105, + "learning_rate": 4.104937435635281e-06, + "loss": 3.1958, + "step": 6117 + }, + { + "epoch": 8.96, + "grad_norm": 0.4542753994464874, + "learning_rate": 4.091647429802869e-06, + "loss": 3.1095, + "step": 6118 + }, + { + "epoch": 8.96, + "grad_norm": 0.5182382464408875, + "learning_rate": 4.078378523165282e-06, + "loss": 3.1165, + "step": 6119 + }, + { + "epoch": 8.96, + "grad_norm": 0.5061773061752319, + "learning_rate": 4.065130718641574e-06, + "loss": 3.1431, + "step": 6120 + }, + { + "epoch": 8.97, + "grad_norm": 0.8390511870384216, + "learning_rate": 4.051904019146202e-06, + "loss": 3.1532, + "step": 6121 + }, + { + "epoch": 8.97, + "grad_norm": 0.7076882123947144, + "learning_rate": 4.038698427588894e-06, + "loss": 3.1292, + "step": 6122 + }, + { + "epoch": 8.97, + "grad_norm": 0.31397783756256104, + "learning_rate": 4.025513946874848e-06, + "loss": 3.1056, + "step": 6123 + }, + { + "epoch": 8.97, + "grad_norm": 0.2746732831001282, + "learning_rate": 4.012350579904556e-06, + "loss": 3.1059, + "step": 6124 + }, + { + "epoch": 8.97, + "grad_norm": 0.3005313277244568, + "learning_rate": 3.999208329573822e-06, + "loss": 3.1013, + "step": 6125 + }, + { + "epoch": 8.97, + "grad_norm": 0.32004761695861816, + "learning_rate": 3.986087198773914e-06, + "loss": 3.1482, + "step": 6126 + }, + { + "epoch": 8.97, + "grad_norm": 0.36849233508110046, + "learning_rate": 3.972987190391364e-06, + "loss": 3.1764, + "step": 6127 + }, + { + "epoch": 8.98, + "grad_norm": 0.4946931004524231, + "learning_rate": 3.959908307308069e-06, + "loss": 3.1072, + "step": 6128 + }, + { + "epoch": 8.98, + "grad_norm": 0.5236996412277222, + "learning_rate": 3.946850552401326e-06, + "loss": 3.1182, + "step": 6129 + }, + { + "epoch": 8.98, + "grad_norm": 0.392853319644928, + "learning_rate": 3.93381392854375e-06, + "loss": 3.1449, + "step": 6130 + }, + { + "epoch": 8.98, + "grad_norm": 0.4347004294395447, + "learning_rate": 3.920798438603279e-06, + "loss": 3.1185, + "step": 6131 + }, + { + "epoch": 8.98, + "grad_norm": 0.37783846259117126, + "learning_rate": 3.907804085443278e-06, + "loss": 3.1043, + "step": 6132 + }, + { + "epoch": 8.98, + "grad_norm": 0.5223934650421143, + "learning_rate": 3.8948308719224056e-06, + "loss": 3.1673, + "step": 6133 + }, + { + "epoch": 8.99, + "grad_norm": 0.32108426094055176, + "learning_rate": 3.881878800894645e-06, + "loss": 3.1235, + "step": 6134 + }, + { + "epoch": 8.99, + "grad_norm": 0.3647514879703522, + "learning_rate": 3.868947875209417e-06, + "loss": 3.1346, + "step": 6135 + }, + { + "epoch": 8.99, + "grad_norm": 0.2049417793750763, + "learning_rate": 3.8560380977114006e-06, + "loss": 3.1303, + "step": 6136 + }, + { + "epoch": 8.99, + "grad_norm": 0.2983146607875824, + "learning_rate": 3.84314947124067e-06, + "loss": 3.1216, + "step": 6137 + }, + { + "epoch": 8.99, + "grad_norm": 0.3992176949977875, + "learning_rate": 3.830281998632656e-06, + "loss": 3.1652, + "step": 6138 + }, + { + "epoch": 8.99, + "grad_norm": 0.48177677392959595, + "learning_rate": 3.817435682718096e-06, + "loss": 3.1406, + "step": 6139 + }, + { + "epoch": 8.99, + "grad_norm": 0.4557303190231323, + "learning_rate": 3.8046105263231048e-06, + "loss": 3.119, + "step": 6140 + }, + { + "epoch": 9.0, + "grad_norm": 0.6173531413078308, + "learning_rate": 3.7918065322691153e-06, + "loss": 3.1731, + "step": 6141 + }, + { + "epoch": 9.0, + "grad_norm": 0.3435233533382416, + "learning_rate": 3.7790237033729302e-06, + "loss": 3.1451, + "step": 6142 + }, + { + "epoch": 9.0, + "grad_norm": 9.693846702575684, + "learning_rate": 3.7662620424466888e-06, + "loss": 3.1799, + "step": 6143 + }, + { + "epoch": 9.0, + "grad_norm": 0.8565120697021484, + "learning_rate": 3.753521552297867e-06, + "loss": 3.1593, + "step": 6144 + }, + { + "epoch": 9.0, + "grad_norm": 0.5679046511650085, + "learning_rate": 3.7408022357292903e-06, + "loss": 3.107, + "step": 6145 + }, + { + "epoch": 9.0, + "grad_norm": 0.5068061351776123, + "learning_rate": 3.72810409553912e-06, + "loss": 3.1332, + "step": 6146 + }, + { + "epoch": 9.0, + "grad_norm": 0.3536016345024109, + "learning_rate": 3.715427134520866e-06, + "loss": 3.1241, + "step": 6147 + }, + { + "epoch": 9.01, + "grad_norm": 0.2531319558620453, + "learning_rate": 3.7027713554633638e-06, + "loss": 3.1666, + "step": 6148 + }, + { + "epoch": 9.01, + "grad_norm": 0.25378838181495667, + "learning_rate": 3.6901367611508196e-06, + "loss": 3.1469, + "step": 6149 + }, + { + "epoch": 9.01, + "grad_norm": 0.4261099696159363, + "learning_rate": 3.677523354362733e-06, + "loss": 3.111, + "step": 6150 + }, + { + "epoch": 9.01, + "grad_norm": 0.34264785051345825, + "learning_rate": 3.664931137873995e-06, + "loss": 3.133, + "step": 6151 + }, + { + "epoch": 9.01, + "grad_norm": 0.3564015328884125, + "learning_rate": 3.6523601144548003e-06, + "loss": 3.1489, + "step": 6152 + }, + { + "epoch": 9.01, + "grad_norm": 0.28124377131462097, + "learning_rate": 3.639810286870682e-06, + "loss": 3.117, + "step": 6153 + }, + { + "epoch": 9.02, + "grad_norm": 3.4681382179260254, + "learning_rate": 3.6272816578825196e-06, + "loss": 3.163, + "step": 6154 + }, + { + "epoch": 9.0, + "grad_norm": 0.20317670702934265, + "learning_rate": 3.6147742302465425e-06, + "loss": 3.1435, + "step": 6155 + }, + { + "epoch": 9.0, + "grad_norm": 0.466596394777298, + "learning_rate": 3.6022880067142717e-06, + "loss": 3.1636, + "step": 6156 + }, + { + "epoch": 9.0, + "grad_norm": 0.2808545231819153, + "learning_rate": 3.5898229900326206e-06, + "loss": 3.1395, + "step": 6157 + }, + { + "epoch": 9.0, + "grad_norm": 0.3181181848049164, + "learning_rate": 3.5773791829438187e-06, + "loss": 3.1, + "step": 6158 + }, + { + "epoch": 9.01, + "grad_norm": 0.38059529662132263, + "learning_rate": 3.564956588185364e-06, + "loss": 3.1424, + "step": 6159 + }, + { + "epoch": 9.01, + "grad_norm": 0.34742945432662964, + "learning_rate": 3.552555208490205e-06, + "loss": 3.1805, + "step": 6160 + }, + { + "epoch": 9.01, + "grad_norm": 0.5437316298484802, + "learning_rate": 3.540175046586536e-06, + "loss": 3.128, + "step": 6161 + }, + { + "epoch": 9.01, + "grad_norm": 0.5122477412223816, + "learning_rate": 3.527816105197901e-06, + "loss": 3.1659, + "step": 6162 + }, + { + "epoch": 9.01, + "grad_norm": 0.47210362553596497, + "learning_rate": 3.515478387043203e-06, + "loss": 3.1248, + "step": 6163 + }, + { + "epoch": 9.01, + "grad_norm": 0.27898654341697693, + "learning_rate": 3.5031618948366596e-06, + "loss": 3.1208, + "step": 6164 + }, + { + "epoch": 9.02, + "grad_norm": 0.3407975435256958, + "learning_rate": 3.4908666312877927e-06, + "loss": 3.0918, + "step": 6165 + }, + { + "epoch": 9.02, + "grad_norm": 0.38441261649131775, + "learning_rate": 3.478592599101482e-06, + "loss": 3.1519, + "step": 6166 + }, + { + "epoch": 9.02, + "grad_norm": 0.7627063393592834, + "learning_rate": 3.466339800977958e-06, + "loss": 3.1297, + "step": 6167 + }, + { + "epoch": 9.02, + "grad_norm": 0.4208107590675354, + "learning_rate": 3.454108239612741e-06, + "loss": 3.1548, + "step": 6168 + }, + { + "epoch": 9.02, + "grad_norm": 0.26444071531295776, + "learning_rate": 3.441897917696679e-06, + "loss": 3.1222, + "step": 6169 + }, + { + "epoch": 9.02, + "grad_norm": 0.23935873806476593, + "learning_rate": 3.429708837915968e-06, + "loss": 3.1342, + "step": 6170 + }, + { + "epoch": 9.02, + "grad_norm": 0.30464550852775574, + "learning_rate": 3.4175410029521406e-06, + "loss": 3.1283, + "step": 6171 + }, + { + "epoch": 9.03, + "grad_norm": 0.3734738230705261, + "learning_rate": 3.4053944154820107e-06, + "loss": 3.1016, + "step": 6172 + }, + { + "epoch": 9.03, + "grad_norm": 0.5887541174888611, + "learning_rate": 3.3932690781777745e-06, + "loss": 3.1465, + "step": 6173 + }, + { + "epoch": 9.03, + "grad_norm": 0.283188134431839, + "learning_rate": 3.381164993706898e-06, + "loss": 3.1611, + "step": 6174 + }, + { + "epoch": 9.03, + "grad_norm": 0.41735076904296875, + "learning_rate": 3.369082164732218e-06, + "loss": 3.1355, + "step": 6175 + }, + { + "epoch": 9.03, + "grad_norm": 0.24843092262744904, + "learning_rate": 3.357020593911875e-06, + "loss": 3.0913, + "step": 6176 + }, + { + "epoch": 9.03, + "grad_norm": 0.4017045497894287, + "learning_rate": 3.344980283899324e-06, + "loss": 3.1586, + "step": 6177 + }, + { + "epoch": 9.03, + "grad_norm": 0.3119039535522461, + "learning_rate": 3.332961237343357e-06, + "loss": 3.1401, + "step": 6178 + }, + { + "epoch": 9.04, + "grad_norm": 0.4604090452194214, + "learning_rate": 3.3209634568880933e-06, + "loss": 3.1625, + "step": 6179 + }, + { + "epoch": 9.04, + "grad_norm": 0.37948447465896606, + "learning_rate": 3.308986945172943e-06, + "loss": 3.1111, + "step": 6180 + }, + { + "epoch": 9.04, + "grad_norm": 0.5480170845985413, + "learning_rate": 3.297031704832676e-06, + "loss": 3.1893, + "step": 6181 + }, + { + "epoch": 9.04, + "grad_norm": 0.4930878281593323, + "learning_rate": 3.285097738497356e-06, + "loss": 3.1135, + "step": 6182 + }, + { + "epoch": 9.04, + "grad_norm": 0.6153199076652527, + "learning_rate": 3.273185048792382e-06, + "loss": 3.0977, + "step": 6183 + }, + { + "epoch": 9.04, + "grad_norm": 0.21819010376930237, + "learning_rate": 3.2612936383384583e-06, + "loss": 3.1449, + "step": 6184 + }, + { + "epoch": 9.04, + "grad_norm": 0.302505224943161, + "learning_rate": 3.2494235097516255e-06, + "loss": 3.1441, + "step": 6185 + }, + { + "epoch": 9.05, + "grad_norm": 0.36193329095840454, + "learning_rate": 3.2375746656432284e-06, + "loss": 3.1302, + "step": 6186 + }, + { + "epoch": 9.05, + "grad_norm": 0.5750117897987366, + "learning_rate": 3.2257471086199253e-06, + "loss": 3.1713, + "step": 6187 + }, + { + "epoch": 9.05, + "grad_norm": 0.5752658843994141, + "learning_rate": 3.2139408412837133e-06, + "loss": 3.1361, + "step": 6188 + }, + { + "epoch": 9.05, + "grad_norm": 0.2937156558036804, + "learning_rate": 3.202155866231893e-06, + "loss": 3.1271, + "step": 6189 + }, + { + "epoch": 9.05, + "grad_norm": 0.25414547324180603, + "learning_rate": 3.1903921860570564e-06, + "loss": 3.171, + "step": 6190 + }, + { + "epoch": 9.05, + "grad_norm": 0.35953837633132935, + "learning_rate": 3.1786498033471445e-06, + "loss": 3.1337, + "step": 6191 + }, + { + "epoch": 9.06, + "grad_norm": 0.3548029959201813, + "learning_rate": 3.1669287206854247e-06, + "loss": 3.1119, + "step": 6192 + }, + { + "epoch": 9.06, + "grad_norm": 0.30916643142700195, + "learning_rate": 3.1552289406504343e-06, + "loss": 3.1406, + "step": 6193 + }, + { + "epoch": 9.06, + "grad_norm": 0.264125794172287, + "learning_rate": 3.1435504658160363e-06, + "loss": 3.1219, + "step": 6194 + }, + { + "epoch": 9.06, + "grad_norm": 0.3955405056476593, + "learning_rate": 3.131893298751465e-06, + "loss": 3.1724, + "step": 6195 + }, + { + "epoch": 9.06, + "grad_norm": 0.4040085971355438, + "learning_rate": 3.120257442021168e-06, + "loss": 3.1436, + "step": 6196 + }, + { + "epoch": 9.06, + "grad_norm": 0.3696833848953247, + "learning_rate": 3.108642898184966e-06, + "loss": 3.1164, + "step": 6197 + }, + { + "epoch": 9.06, + "grad_norm": 0.630946695804596, + "learning_rate": 3.0970496697980023e-06, + "loss": 3.0944, + "step": 6198 + }, + { + "epoch": 9.07, + "grad_norm": 0.33301976323127747, + "learning_rate": 3.085477759410693e-06, + "loss": 3.1384, + "step": 6199 + }, + { + "epoch": 9.07, + "grad_norm": 0.2601894438266754, + "learning_rate": 3.0739271695687688e-06, + "loss": 3.0923, + "step": 6200 + }, + { + "epoch": 9.07, + "grad_norm": 0.31439009308815, + "learning_rate": 3.06239790281333e-06, + "loss": 3.1017, + "step": 6201 + }, + { + "epoch": 9.07, + "grad_norm": 0.30821776390075684, + "learning_rate": 3.050889961680681e-06, + "loss": 3.1428, + "step": 6202 + }, + { + "epoch": 9.07, + "grad_norm": 0.30211207270622253, + "learning_rate": 3.03940334870253e-06, + "loss": 3.1257, + "step": 6203 + }, + { + "epoch": 9.07, + "grad_norm": 0.20325028896331787, + "learning_rate": 3.027938066405833e-06, + "loss": 3.1026, + "step": 6204 + }, + { + "epoch": 9.07, + "grad_norm": 0.4516938030719757, + "learning_rate": 3.0164941173128846e-06, + "loss": 3.1275, + "step": 6205 + }, + { + "epoch": 9.08, + "grad_norm": 0.3202283978462219, + "learning_rate": 3.0050715039412703e-06, + "loss": 3.1101, + "step": 6206 + }, + { + "epoch": 9.08, + "grad_norm": 0.17919191718101501, + "learning_rate": 2.993670228803902e-06, + "loss": 3.1383, + "step": 6207 + }, + { + "epoch": 9.08, + "grad_norm": 0.3291981518268585, + "learning_rate": 2.9822902944089737e-06, + "loss": 3.1203, + "step": 6208 + }, + { + "epoch": 9.08, + "grad_norm": 0.3368469476699829, + "learning_rate": 2.9709317032599935e-06, + "loss": 3.1112, + "step": 6209 + }, + { + "epoch": 9.08, + "grad_norm": 0.2890414297580719, + "learning_rate": 2.9595944578557855e-06, + "loss": 3.1559, + "step": 6210 + }, + { + "epoch": 9.08, + "grad_norm": 0.32051295042037964, + "learning_rate": 2.9482785606904538e-06, + "loss": 3.1038, + "step": 6211 + }, + { + "epoch": 9.08, + "grad_norm": 0.38786590099334717, + "learning_rate": 2.936984014253441e-06, + "loss": 3.1456, + "step": 6212 + }, + { + "epoch": 9.09, + "grad_norm": 0.5022422671318054, + "learning_rate": 2.925710821029437e-06, + "loss": 3.117, + "step": 6213 + }, + { + "epoch": 9.09, + "grad_norm": 0.21428152918815613, + "learning_rate": 2.9144589834985024e-06, + "loss": 3.1454, + "step": 6214 + }, + { + "epoch": 9.09, + "grad_norm": 0.5008912086486816, + "learning_rate": 2.9032285041359573e-06, + "loss": 3.1365, + "step": 6215 + }, + { + "epoch": 9.09, + "grad_norm": 0.5330930352210999, + "learning_rate": 2.8920193854124143e-06, + "loss": 3.1251, + "step": 6216 + }, + { + "epoch": 9.09, + "grad_norm": 0.32792308926582336, + "learning_rate": 2.880831629793823e-06, + "loss": 3.1732, + "step": 6217 + }, + { + "epoch": 9.09, + "grad_norm": 0.3495442569255829, + "learning_rate": 2.8696652397414147e-06, + "loss": 3.1015, + "step": 6218 + }, + { + "epoch": 9.1, + "grad_norm": 0.4573792815208435, + "learning_rate": 2.8585202177117023e-06, + "loss": 3.1154, + "step": 6219 + }, + { + "epoch": 9.1, + "grad_norm": 0.24023696780204773, + "learning_rate": 2.8473965661565347e-06, + "loss": 3.1305, + "step": 6220 + }, + { + "epoch": 9.1, + "grad_norm": 0.4277721047401428, + "learning_rate": 2.8362942875230445e-06, + "loss": 3.1318, + "step": 6221 + }, + { + "epoch": 9.1, + "grad_norm": 0.9642704129219055, + "learning_rate": 2.8252133842536333e-06, + "loss": 3.1498, + "step": 6222 + }, + { + "epoch": 9.1, + "grad_norm": 0.4266915023326874, + "learning_rate": 2.8141538587860616e-06, + "loss": 3.1429, + "step": 6223 + }, + { + "epoch": 9.1, + "grad_norm": 0.29061654210090637, + "learning_rate": 2.803115713553317e-06, + "loss": 3.1669, + "step": 6224 + }, + { + "epoch": 9.1, + "grad_norm": 0.553413987159729, + "learning_rate": 2.7920989509837234e-06, + "loss": 3.1616, + "step": 6225 + }, + { + "epoch": 9.11, + "grad_norm": 0.9510021805763245, + "learning_rate": 2.7811035735009205e-06, + "loss": 3.1329, + "step": 6226 + }, + { + "epoch": 9.11, + "grad_norm": 0.29453250765800476, + "learning_rate": 2.770129583523784e-06, + "loss": 3.1351, + "step": 6227 + }, + { + "epoch": 9.11, + "grad_norm": 0.4295230805873871, + "learning_rate": 2.759176983466527e-06, + "loss": 3.1458, + "step": 6228 + }, + { + "epoch": 9.11, + "grad_norm": 0.5276914238929749, + "learning_rate": 2.7482457757386783e-06, + "loss": 3.173, + "step": 6229 + }, + { + "epoch": 9.11, + "grad_norm": 0.38097521662712097, + "learning_rate": 2.737335962744991e-06, + "loss": 3.1138, + "step": 6230 + }, + { + "epoch": 9.11, + "grad_norm": 0.2351088970899582, + "learning_rate": 2.726447546885569e-06, + "loss": 3.1499, + "step": 6231 + }, + { + "epoch": 9.11, + "grad_norm": 0.4661239981651306, + "learning_rate": 2.715580530555795e-06, + "loss": 3.1113, + "step": 6232 + }, + { + "epoch": 9.12, + "grad_norm": 0.31686079502105713, + "learning_rate": 2.704734916146323e-06, + "loss": 3.1467, + "step": 6233 + }, + { + "epoch": 9.12, + "grad_norm": 0.2289198338985443, + "learning_rate": 2.6939107060431235e-06, + "loss": 3.1672, + "step": 6234 + }, + { + "epoch": 9.12, + "grad_norm": 0.3613332211971283, + "learning_rate": 2.6831079026274687e-06, + "loss": 3.1464, + "step": 6235 + }, + { + "epoch": 9.12, + "grad_norm": 2.2319555282592773, + "learning_rate": 2.6723265082758687e-06, + "loss": 3.132, + "step": 6236 + }, + { + "epoch": 9.12, + "grad_norm": 0.38924992084503174, + "learning_rate": 2.661566525360193e-06, + "loss": 3.1881, + "step": 6237 + }, + { + "epoch": 9.12, + "grad_norm": 0.42754054069519043, + "learning_rate": 2.6508279562475368e-06, + "loss": 3.159, + "step": 6238 + }, + { + "epoch": 9.12, + "grad_norm": 0.3254989981651306, + "learning_rate": 2.6401108033003217e-06, + "loss": 3.1244, + "step": 6239 + }, + { + "epoch": 9.13, + "grad_norm": 0.18462346494197845, + "learning_rate": 2.629415068876262e-06, + "loss": 3.1495, + "step": 6240 + }, + { + "epoch": 9.13, + "grad_norm": 0.2884494662284851, + "learning_rate": 2.6187407553283305e-06, + "loss": 3.1256, + "step": 6241 + }, + { + "epoch": 9.13, + "grad_norm": 0.331559419631958, + "learning_rate": 2.608087865004816e-06, + "loss": 3.2043, + "step": 6242 + }, + { + "epoch": 9.13, + "grad_norm": 0.5967766046524048, + "learning_rate": 2.5974564002492762e-06, + "loss": 3.1137, + "step": 6243 + }, + { + "epoch": 9.13, + "grad_norm": 0.561845064163208, + "learning_rate": 2.5868463634005747e-06, + "loss": 3.0911, + "step": 6244 + }, + { + "epoch": 9.13, + "grad_norm": 0.3471025228500366, + "learning_rate": 2.5762577567928324e-06, + "loss": 3.1196, + "step": 6245 + }, + { + "epoch": 9.14, + "grad_norm": 0.36495816707611084, + "learning_rate": 2.5656905827554867e-06, + "loss": 3.1474, + "step": 6246 + }, + { + "epoch": 9.14, + "grad_norm": 0.3217298984527588, + "learning_rate": 2.555144843613222e-06, + "loss": 3.1222, + "step": 6247 + }, + { + "epoch": 9.14, + "grad_norm": 0.22464439272880554, + "learning_rate": 2.5446205416860603e-06, + "loss": 3.1343, + "step": 6248 + }, + { + "epoch": 9.14, + "grad_norm": 0.267767995595932, + "learning_rate": 2.5341176792892608e-06, + "loss": 3.1311, + "step": 6249 + }, + { + "epoch": 9.14, + "grad_norm": 0.469012051820755, + "learning_rate": 2.523636258733375e-06, + "loss": 3.1447, + "step": 6250 + }, + { + "epoch": 9.14, + "grad_norm": 0.2566981613636017, + "learning_rate": 2.5131762823242587e-06, + "loss": 3.1439, + "step": 6251 + }, + { + "epoch": 9.14, + "grad_norm": 0.40152233839035034, + "learning_rate": 2.502737752363038e-06, + "loss": 3.1142, + "step": 6252 + }, + { + "epoch": 9.15, + "grad_norm": 0.344911128282547, + "learning_rate": 2.492320671146098e-06, + "loss": 3.1055, + "step": 6253 + }, + { + "epoch": 9.15, + "eval_loss": 2.9924094676971436, + "eval_runtime": 1166.3468, + "eval_samples_per_second": 14.697, + "eval_steps_per_second": 0.919, + "step": 6253 + }, + { + "epoch": 9.15, + "grad_norm": 0.9697498679161072, + "learning_rate": 2.4819250409651607e-06, + "loss": 3.122, + "step": 6254 + }, + { + "epoch": 9.15, + "grad_norm": 0.27645161747932434, + "learning_rate": 2.471550864107175e-06, + "loss": 3.1275, + "step": 6255 + }, + { + "epoch": 9.15, + "grad_norm": 0.20699070394039154, + "learning_rate": 2.461198142854382e-06, + "loss": 3.1085, + "step": 6256 + }, + { + "epoch": 9.15, + "grad_norm": 0.3643249273300171, + "learning_rate": 2.450866879484337e-06, + "loss": 3.1565, + "step": 6257 + }, + { + "epoch": 9.15, + "grad_norm": 0.3047739863395691, + "learning_rate": 2.4405570762698225e-06, + "loss": 3.1603, + "step": 6258 + }, + { + "epoch": 9.15, + "grad_norm": 0.36578232049942017, + "learning_rate": 2.4302687354789243e-06, + "loss": 3.1402, + "step": 6259 + }, + { + "epoch": 9.16, + "grad_norm": 0.42692938446998596, + "learning_rate": 2.420001859375054e-06, + "loss": 3.1303, + "step": 6260 + }, + { + "epoch": 9.16, + "grad_norm": 0.21095794439315796, + "learning_rate": 2.4097564502167934e-06, + "loss": 3.1089, + "step": 6261 + }, + { + "epoch": 9.16, + "grad_norm": 0.2429499626159668, + "learning_rate": 2.3995325102581067e-06, + "loss": 3.1317, + "step": 6262 + }, + { + "epoch": 9.16, + "grad_norm": 0.3850218951702118, + "learning_rate": 2.3893300417481725e-06, + "loss": 3.1063, + "step": 6263 + }, + { + "epoch": 9.16, + "grad_norm": 0.25018832087516785, + "learning_rate": 2.3791490469314724e-06, + "loss": 3.1371, + "step": 6264 + }, + { + "epoch": 9.16, + "grad_norm": 0.27892038226127625, + "learning_rate": 2.36898952804776e-06, + "loss": 3.1259, + "step": 6265 + }, + { + "epoch": 9.17, + "grad_norm": 0.2211170643568039, + "learning_rate": 2.358851487332059e-06, + "loss": 3.1246, + "step": 6266 + }, + { + "epoch": 9.17, + "grad_norm": 0.880876898765564, + "learning_rate": 2.348734927014662e-06, + "loss": 3.1408, + "step": 6267 + }, + { + "epoch": 9.17, + "grad_norm": 0.510045051574707, + "learning_rate": 2.338639849321156e-06, + "loss": 3.1284, + "step": 6268 + }, + { + "epoch": 9.17, + "grad_norm": 0.3161579966545105, + "learning_rate": 2.3285662564723865e-06, + "loss": 3.1282, + "step": 6269 + }, + { + "epoch": 9.17, + "grad_norm": 0.4412688910961151, + "learning_rate": 2.3185141506844697e-06, + "loss": 3.161, + "step": 6270 + }, + { + "epoch": 9.17, + "grad_norm": 0.3650240898132324, + "learning_rate": 2.3084835341688147e-06, + "loss": 3.168, + "step": 6271 + }, + { + "epoch": 9.17, + "grad_norm": 0.23962388932704926, + "learning_rate": 2.2984744091320786e-06, + "loss": 3.1433, + "step": 6272 + }, + { + "epoch": 9.18, + "grad_norm": 0.3383813500404358, + "learning_rate": 2.2884867777762e-06, + "loss": 3.1386, + "step": 6273 + }, + { + "epoch": 9.18, + "grad_norm": 0.25145041942596436, + "learning_rate": 2.2785206422984097e-06, + "loss": 3.1351, + "step": 6274 + }, + { + "epoch": 9.18, + "grad_norm": 0.25064894556999207, + "learning_rate": 2.2685760048911654e-06, + "loss": 3.1289, + "step": 6275 + }, + { + "epoch": 9.18, + "grad_norm": 0.4283466041088104, + "learning_rate": 2.258652867742239e-06, + "loss": 3.1328, + "step": 6276 + }, + { + "epoch": 9.18, + "grad_norm": 0.3982808589935303, + "learning_rate": 2.2487512330346517e-06, + "loss": 3.1139, + "step": 6277 + }, + { + "epoch": 9.18, + "grad_norm": 0.5738875269889832, + "learning_rate": 2.2388711029466824e-06, + "loss": 3.133, + "step": 6278 + }, + { + "epoch": 9.18, + "grad_norm": 0.383221834897995, + "learning_rate": 2.2290124796519153e-06, + "loss": 3.147, + "step": 6279 + }, + { + "epoch": 9.19, + "grad_norm": 0.36027202010154724, + "learning_rate": 2.2191753653191817e-06, + "loss": 3.1099, + "step": 6280 + }, + { + "epoch": 9.19, + "grad_norm": 0.27547046542167664, + "learning_rate": 2.2093597621125397e-06, + "loss": 3.1073, + "step": 6281 + }, + { + "epoch": 9.19, + "grad_norm": 0.3234958350658417, + "learning_rate": 2.199565672191406e-06, + "loss": 3.1286, + "step": 6282 + }, + { + "epoch": 9.19, + "grad_norm": 0.3891391158103943, + "learning_rate": 2.189793097710402e-06, + "loss": 3.1205, + "step": 6283 + }, + { + "epoch": 9.19, + "grad_norm": 0.2948206663131714, + "learning_rate": 2.1800420408193967e-06, + "loss": 3.1266, + "step": 6284 + }, + { + "epoch": 9.19, + "grad_norm": 0.2731497287750244, + "learning_rate": 2.1703125036635965e-06, + "loss": 3.1349, + "step": 6285 + }, + { + "epoch": 9.19, + "grad_norm": 0.4376061260700226, + "learning_rate": 2.160604488383422e-06, + "loss": 3.2027, + "step": 6286 + }, + { + "epoch": 9.2, + "grad_norm": 0.23648232221603394, + "learning_rate": 2.1509179971145764e-06, + "loss": 3.157, + "step": 6287 + }, + { + "epoch": 9.2, + "grad_norm": 0.2671348452568054, + "learning_rate": 2.1412530319879887e-06, + "loss": 3.1551, + "step": 6288 + }, + { + "epoch": 9.2, + "grad_norm": 0.3133906424045563, + "learning_rate": 2.1316095951299465e-06, + "loss": 3.1321, + "step": 6289 + }, + { + "epoch": 9.2, + "grad_norm": 0.27077266573905945, + "learning_rate": 2.121987688661886e-06, + "loss": 3.1356, + "step": 6290 + }, + { + "epoch": 9.2, + "grad_norm": 0.3760845959186554, + "learning_rate": 2.112387314700592e-06, + "loss": 3.1598, + "step": 6291 + }, + { + "epoch": 9.2, + "grad_norm": 0.6859793066978455, + "learning_rate": 2.1028084753580755e-06, + "loss": 3.1154, + "step": 6292 + }, + { + "epoch": 9.21, + "grad_norm": 0.3447398245334625, + "learning_rate": 2.093251172741617e-06, + "loss": 3.1377, + "step": 6293 + }, + { + "epoch": 9.21, + "grad_norm": 0.28955718874931335, + "learning_rate": 2.083715408953757e-06, + "loss": 3.1093, + "step": 6294 + }, + { + "epoch": 9.21, + "grad_norm": 0.44994664192199707, + "learning_rate": 2.074201186092306e-06, + "loss": 3.1114, + "step": 6295 + }, + { + "epoch": 9.21, + "grad_norm": 0.46158942580223083, + "learning_rate": 2.0647085062503237e-06, + "loss": 3.1042, + "step": 6296 + }, + { + "epoch": 9.21, + "grad_norm": 0.3395864963531494, + "learning_rate": 2.0552373715161276e-06, + "loss": 3.1546, + "step": 6297 + }, + { + "epoch": 9.21, + "grad_norm": 0.48832792043685913, + "learning_rate": 2.0457877839733186e-06, + "loss": 3.1421, + "step": 6298 + }, + { + "epoch": 9.21, + "grad_norm": 0.4269855320453644, + "learning_rate": 2.0363597457007445e-06, + "loss": 3.1524, + "step": 6299 + }, + { + "epoch": 9.22, + "grad_norm": 0.33162572979927063, + "learning_rate": 2.0269532587724905e-06, + "loss": 3.084, + "step": 6300 + }, + { + "epoch": 9.22, + "grad_norm": 0.5742428302764893, + "learning_rate": 2.017568325257946e-06, + "loss": 3.1574, + "step": 6301 + }, + { + "epoch": 9.22, + "grad_norm": 0.2641449570655823, + "learning_rate": 2.0082049472217146e-06, + "loss": 3.1393, + "step": 6302 + }, + { + "epoch": 9.22, + "grad_norm": 0.3563360273838043, + "learning_rate": 1.9988631267236824e-06, + "loss": 3.1342, + "step": 6303 + }, + { + "epoch": 9.22, + "grad_norm": 0.25904589891433716, + "learning_rate": 1.989542865818983e-06, + "loss": 3.1186, + "step": 6304 + }, + { + "epoch": 9.22, + "grad_norm": 0.28990599513053894, + "learning_rate": 1.9802441665580207e-06, + "loss": 3.1156, + "step": 6305 + }, + { + "epoch": 9.22, + "grad_norm": 0.47935912013053894, + "learning_rate": 1.970967030986448e-06, + "loss": 3.1126, + "step": 6306 + }, + { + "epoch": 9.23, + "grad_norm": 0.41986241936683655, + "learning_rate": 1.961711461145155e-06, + "loss": 3.0965, + "step": 6307 + }, + { + "epoch": 9.23, + "grad_norm": 0.16427509486675262, + "learning_rate": 1.9524774590703232e-06, + "loss": 3.1306, + "step": 6308 + }, + { + "epoch": 9.23, + "grad_norm": 0.2631678581237793, + "learning_rate": 1.9432650267933836e-06, + "loss": 3.1338, + "step": 6309 + }, + { + "epoch": 9.23, + "grad_norm": 0.44342029094696045, + "learning_rate": 1.93407416634096e-06, + "loss": 3.1189, + "step": 6310 + }, + { + "epoch": 9.23, + "grad_norm": 0.1969728022813797, + "learning_rate": 1.9249048797350234e-06, + "loss": 3.1197, + "step": 6311 + }, + { + "epoch": 9.23, + "grad_norm": 0.3396991789340973, + "learning_rate": 1.915757168992749e-06, + "loss": 3.1112, + "step": 6312 + }, + { + "epoch": 9.23, + "grad_norm": 0.5259889960289001, + "learning_rate": 1.9066310361265605e-06, + "loss": 3.121, + "step": 6313 + }, + { + "epoch": 9.24, + "grad_norm": 0.298654705286026, + "learning_rate": 1.897526483144163e-06, + "loss": 3.183, + "step": 6314 + }, + { + "epoch": 9.24, + "grad_norm": 0.7138015031814575, + "learning_rate": 1.8884435120484656e-06, + "loss": 3.1578, + "step": 6315 + }, + { + "epoch": 9.24, + "grad_norm": 0.20870919525623322, + "learning_rate": 1.8793821248376808e-06, + "loss": 3.1192, + "step": 6316 + }, + { + "epoch": 9.24, + "grad_norm": 0.425702840089798, + "learning_rate": 1.8703423235052808e-06, + "loss": 3.2129, + "step": 6317 + }, + { + "epoch": 9.24, + "grad_norm": 0.6385764479637146, + "learning_rate": 1.861324110039908e-06, + "loss": 3.1574, + "step": 6318 + }, + { + "epoch": 9.24, + "grad_norm": 0.3600045442581177, + "learning_rate": 1.8523274864255424e-06, + "loss": 3.1124, + "step": 6319 + }, + { + "epoch": 9.25, + "grad_norm": 0.5064054131507874, + "learning_rate": 1.8433524546413783e-06, + "loss": 3.1285, + "step": 6320 + }, + { + "epoch": 9.25, + "grad_norm": 0.41301196813583374, + "learning_rate": 1.8343990166618474e-06, + "loss": 3.1213, + "step": 6321 + }, + { + "epoch": 9.25, + "grad_norm": 0.9420896768569946, + "learning_rate": 1.825467174456652e-06, + "loss": 3.151, + "step": 6322 + }, + { + "epoch": 9.25, + "grad_norm": 0.25728273391723633, + "learning_rate": 1.8165569299907647e-06, + "loss": 3.1376, + "step": 6323 + }, + { + "epoch": 9.25, + "grad_norm": 2.1151819229125977, + "learning_rate": 1.8076682852243398e-06, + "loss": 3.1215, + "step": 6324 + }, + { + "epoch": 9.25, + "grad_norm": 0.3092057704925537, + "learning_rate": 1.7988012421128465e-06, + "loss": 3.1563, + "step": 6325 + }, + { + "epoch": 9.25, + "grad_norm": 1.1969783306121826, + "learning_rate": 1.7899558026069573e-06, + "loss": 3.163, + "step": 6326 + }, + { + "epoch": 9.26, + "grad_norm": 0.20899808406829834, + "learning_rate": 1.7811319686526161e-06, + "loss": 3.1611, + "step": 6327 + }, + { + "epoch": 9.26, + "grad_norm": 0.3563082218170166, + "learning_rate": 1.7723297421910145e-06, + "loss": 3.14, + "step": 6328 + }, + { + "epoch": 9.26, + "grad_norm": 0.22852715849876404, + "learning_rate": 1.76354912515857e-06, + "loss": 3.1375, + "step": 6329 + }, + { + "epoch": 9.26, + "grad_norm": 0.34858134388923645, + "learning_rate": 1.7547901194869708e-06, + "loss": 3.1257, + "step": 6330 + }, + { + "epoch": 9.26, + "grad_norm": 0.31731095910072327, + "learning_rate": 1.7460527271031314e-06, + "loss": 3.1926, + "step": 6331 + }, + { + "epoch": 9.26, + "grad_norm": 0.334920197725296, + "learning_rate": 1.7373369499292248e-06, + "loss": 3.1301, + "step": 6332 + }, + { + "epoch": 9.26, + "grad_norm": 0.196299210190773, + "learning_rate": 1.7286427898826619e-06, + "loss": 3.168, + "step": 6333 + }, + { + "epoch": 9.27, + "grad_norm": 0.35018235445022583, + "learning_rate": 1.7199702488761015e-06, + "loss": 3.1454, + "step": 6334 + }, + { + "epoch": 9.27, + "grad_norm": 0.5034624934196472, + "learning_rate": 1.71131932881744e-06, + "loss": 3.1558, + "step": 6335 + }, + { + "epoch": 9.27, + "grad_norm": 0.3856576979160309, + "learning_rate": 1.7026900316098215e-06, + "loss": 3.1224, + "step": 6336 + }, + { + "epoch": 9.27, + "grad_norm": 0.2683032751083374, + "learning_rate": 1.6940823591516497e-06, + "loss": 3.1571, + "step": 6337 + }, + { + "epoch": 9.27, + "grad_norm": 0.29963627457618713, + "learning_rate": 1.685496313336532e-06, + "loss": 3.1182, + "step": 6338 + }, + { + "epoch": 9.27, + "grad_norm": 0.2658440172672272, + "learning_rate": 1.6769318960533464e-06, + "loss": 3.1226, + "step": 6339 + }, + { + "epoch": 9.27, + "grad_norm": 0.48988819122314453, + "learning_rate": 1.6683891091862303e-06, + "loss": 3.0946, + "step": 6340 + }, + { + "epoch": 9.28, + "grad_norm": 2.378725528717041, + "learning_rate": 1.6598679546144913e-06, + "loss": 3.1256, + "step": 6341 + }, + { + "epoch": 9.28, + "grad_norm": 0.3069950044155121, + "learning_rate": 1.6513684342127745e-06, + "loss": 3.1259, + "step": 6342 + }, + { + "epoch": 9.28, + "grad_norm": 0.5627717971801758, + "learning_rate": 1.6428905498509061e-06, + "loss": 3.1109, + "step": 6343 + }, + { + "epoch": 9.28, + "grad_norm": 0.31544429063796997, + "learning_rate": 1.6344343033939391e-06, + "loss": 3.162, + "step": 6344 + }, + { + "epoch": 9.28, + "grad_norm": 0.28528231382369995, + "learning_rate": 1.6259996967022184e-06, + "loss": 3.1213, + "step": 6345 + }, + { + "epoch": 9.28, + "grad_norm": 0.2968144118785858, + "learning_rate": 1.6175867316313043e-06, + "loss": 3.1765, + "step": 6346 + }, + { + "epoch": 9.29, + "grad_norm": 0.6702325940132141, + "learning_rate": 1.6091954100319495e-06, + "loss": 3.1315, + "step": 6347 + }, + { + "epoch": 9.29, + "grad_norm": 0.39867013692855835, + "learning_rate": 1.6008257337502442e-06, + "loss": 3.2009, + "step": 6348 + }, + { + "epoch": 9.29, + "grad_norm": 0.23865939676761627, + "learning_rate": 1.5924777046274263e-06, + "loss": 3.1321, + "step": 6349 + }, + { + "epoch": 9.29, + "grad_norm": 0.303800493478775, + "learning_rate": 1.584151324500005e-06, + "loss": 3.1584, + "step": 6350 + }, + { + "epoch": 9.29, + "grad_norm": 0.341605544090271, + "learning_rate": 1.5758465951997591e-06, + "loss": 3.1407, + "step": 6351 + }, + { + "epoch": 9.29, + "grad_norm": 0.24277356266975403, + "learning_rate": 1.5675635185536386e-06, + "loss": 3.1352, + "step": 6352 + }, + { + "epoch": 9.29, + "grad_norm": 0.37486472725868225, + "learning_rate": 1.5593020963838855e-06, + "loss": 3.0938, + "step": 6353 + }, + { + "epoch": 9.3, + "grad_norm": 0.23244240880012512, + "learning_rate": 1.5510623305079464e-06, + "loss": 3.1551, + "step": 6354 + }, + { + "epoch": 9.3, + "grad_norm": 0.3731347322463989, + "learning_rate": 1.5428442227385265e-06, + "loss": 3.1361, + "step": 6355 + }, + { + "epoch": 9.3, + "grad_norm": 0.4529644846916199, + "learning_rate": 1.5346477748835354e-06, + "loss": 3.1445, + "step": 6356 + }, + { + "epoch": 9.3, + "grad_norm": 0.3394308388233185, + "learning_rate": 1.5264729887461638e-06, + "loss": 3.0985, + "step": 6357 + }, + { + "epoch": 9.3, + "grad_norm": 0.3395843505859375, + "learning_rate": 1.5183198661247845e-06, + "loss": 3.1638, + "step": 6358 + }, + { + "epoch": 9.3, + "grad_norm": 0.442415326833725, + "learning_rate": 1.5101884088130403e-06, + "loss": 3.1293, + "step": 6359 + }, + { + "epoch": 9.3, + "grad_norm": 0.39445722103118896, + "learning_rate": 1.5020786185997892e-06, + "loss": 3.0915, + "step": 6360 + }, + { + "epoch": 9.31, + "grad_norm": 0.36471104621887207, + "learning_rate": 1.4939904972691487e-06, + "loss": 3.1528, + "step": 6361 + }, + { + "epoch": 9.31, + "grad_norm": 0.1835634410381317, + "learning_rate": 1.4859240466004399e-06, + "loss": 3.1295, + "step": 6362 + }, + { + "epoch": 9.31, + "grad_norm": 0.3105097711086273, + "learning_rate": 1.4778792683682207e-06, + "loss": 3.118, + "step": 6363 + }, + { + "epoch": 9.31, + "grad_norm": 0.3407469391822815, + "learning_rate": 1.4698561643423092e-06, + "loss": 3.1322, + "step": 6364 + }, + { + "epoch": 9.31, + "grad_norm": 0.26898640394210815, + "learning_rate": 1.4618547362877156e-06, + "loss": 3.1145, + "step": 6365 + }, + { + "epoch": 9.31, + "grad_norm": 0.45513013005256653, + "learning_rate": 1.4538749859646983e-06, + "loss": 3.1567, + "step": 6366 + }, + { + "epoch": 9.32, + "grad_norm": 0.37628865242004395, + "learning_rate": 1.4459169151287643e-06, + "loss": 3.126, + "step": 6367 + }, + { + "epoch": 9.32, + "grad_norm": 0.306885302066803, + "learning_rate": 1.4379805255306357e-06, + "loss": 3.1249, + "step": 6368 + }, + { + "epoch": 9.32, + "grad_norm": 0.3588414490222931, + "learning_rate": 1.4300658189162486e-06, + "loss": 3.1584, + "step": 6369 + }, + { + "epoch": 9.32, + "grad_norm": 0.2908707857131958, + "learning_rate": 1.4221727970267995e-06, + "loss": 3.123, + "step": 6370 + }, + { + "epoch": 9.32, + "grad_norm": 0.20172357559204102, + "learning_rate": 1.4143014615986993e-06, + "loss": 3.1318, + "step": 6371 + }, + { + "epoch": 9.32, + "grad_norm": 0.24213376641273499, + "learning_rate": 1.4064518143635742e-06, + "loss": 3.1375, + "step": 6372 + }, + { + "epoch": 9.32, + "grad_norm": 0.3989996314048767, + "learning_rate": 1.3986238570483201e-06, + "loss": 3.1308, + "step": 6373 + }, + { + "epoch": 9.33, + "grad_norm": 0.4063306450843811, + "learning_rate": 1.3908175913750154e-06, + "loss": 3.139, + "step": 6374 + }, + { + "epoch": 9.33, + "grad_norm": 0.40335598587989807, + "learning_rate": 1.3830330190609863e-06, + "loss": 3.157, + "step": 6375 + }, + { + "epoch": 9.33, + "grad_norm": 0.3092147409915924, + "learning_rate": 1.3752701418187852e-06, + "loss": 3.1316, + "step": 6376 + }, + { + "epoch": 9.33, + "grad_norm": 0.34485480189323425, + "learning_rate": 1.3675289613562237e-06, + "loss": 3.1208, + "step": 6377 + }, + { + "epoch": 9.33, + "grad_norm": 0.18705491721630096, + "learning_rate": 1.3598094793762506e-06, + "loss": 3.1279, + "step": 6378 + }, + { + "epoch": 9.33, + "grad_norm": 0.37322962284088135, + "learning_rate": 1.352111697577163e-06, + "loss": 3.1436, + "step": 6379 + }, + { + "epoch": 9.33, + "grad_norm": 0.2731330692768097, + "learning_rate": 1.3444356176523843e-06, + "loss": 3.1352, + "step": 6380 + }, + { + "epoch": 9.34, + "grad_norm": 0.27044248580932617, + "learning_rate": 1.3367812412906078e-06, + "loss": 3.1731, + "step": 6381 + }, + { + "epoch": 9.34, + "grad_norm": 0.45415210723876953, + "learning_rate": 1.3291485701757533e-06, + "loss": 3.0973, + "step": 6382 + }, + { + "epoch": 9.34, + "grad_norm": 0.38562557101249695, + "learning_rate": 1.3215376059869445e-06, + "loss": 3.0987, + "step": 6383 + }, + { + "epoch": 9.34, + "grad_norm": 0.4931897521018982, + "learning_rate": 1.3139483503985418e-06, + "loss": 3.1229, + "step": 6384 + }, + { + "epoch": 9.34, + "grad_norm": 0.27991271018981934, + "learning_rate": 1.306380805080143e-06, + "loss": 3.1336, + "step": 6385 + }, + { + "epoch": 9.34, + "grad_norm": 0.3207654058933258, + "learning_rate": 1.2988349716965608e-06, + "loss": 3.1297, + "step": 6386 + }, + { + "epoch": 9.34, + "grad_norm": 0.2610193192958832, + "learning_rate": 1.2913108519078121e-06, + "loss": 3.111, + "step": 6387 + }, + { + "epoch": 9.35, + "grad_norm": 0.29374003410339355, + "learning_rate": 1.2838084473691615e-06, + "loss": 3.1361, + "step": 6388 + }, + { + "epoch": 9.35, + "grad_norm": 0.3695549964904785, + "learning_rate": 1.2763277597310775e-06, + "loss": 3.1543, + "step": 6389 + }, + { + "epoch": 9.35, + "grad_norm": 0.43370383977890015, + "learning_rate": 1.268868790639277e-06, + "loss": 3.1427, + "step": 6390 + }, + { + "epoch": 9.35, + "grad_norm": 0.30709391832351685, + "learning_rate": 1.26143154173467e-06, + "loss": 3.1415, + "step": 6391 + }, + { + "epoch": 9.35, + "grad_norm": 0.26942434906959534, + "learning_rate": 1.2540160146534142e-06, + "loss": 3.1113, + "step": 6392 + }, + { + "epoch": 9.35, + "grad_norm": 0.2789570093154907, + "learning_rate": 1.2466222110268711e-06, + "loss": 3.1754, + "step": 6393 + }, + { + "epoch": 9.36, + "grad_norm": 0.31712105870246887, + "learning_rate": 1.2392501324816175e-06, + "loss": 3.1267, + "step": 6394 + }, + { + "epoch": 9.36, + "grad_norm": 0.28174713253974915, + "learning_rate": 1.2318997806394784e-06, + "loss": 3.1297, + "step": 6395 + }, + { + "epoch": 9.36, + "grad_norm": 0.39602330327033997, + "learning_rate": 1.2245711571174602e-06, + "loss": 3.1361, + "step": 6396 + }, + { + "epoch": 9.36, + "grad_norm": 0.36304354667663574, + "learning_rate": 1.2172642635278396e-06, + "loss": 3.1454, + "step": 6397 + }, + { + "epoch": 9.36, + "grad_norm": 0.6136435270309448, + "learning_rate": 1.2099791014780537e-06, + "loss": 3.1827, + "step": 6398 + }, + { + "epoch": 9.36, + "grad_norm": 0.2997877597808838, + "learning_rate": 1.20271567257082e-06, + "loss": 3.1271, + "step": 6399 + }, + { + "epoch": 9.36, + "grad_norm": 0.2577243447303772, + "learning_rate": 1.195473978404016e-06, + "loss": 3.1513, + "step": 6400 + }, + { + "epoch": 9.37, + "grad_norm": 0.29769569635391235, + "learning_rate": 1.1882540205707782e-06, + "loss": 3.1101, + "step": 6401 + }, + { + "epoch": 9.37, + "grad_norm": 0.21552865207195282, + "learning_rate": 1.1810558006594585e-06, + "loss": 3.1019, + "step": 6402 + }, + { + "epoch": 9.37, + "grad_norm": 0.2695240080356598, + "learning_rate": 1.1738793202536013e-06, + "loss": 3.164, + "step": 6403 + }, + { + "epoch": 9.37, + "grad_norm": 0.49010011553764343, + "learning_rate": 1.1667245809319882e-06, + "loss": 3.1666, + "step": 6404 + }, + { + "epoch": 9.37, + "grad_norm": 0.2555443346500397, + "learning_rate": 1.1595915842686266e-06, + "loss": 3.1167, + "step": 6405 + }, + { + "epoch": 9.37, + "grad_norm": 0.4333082437515259, + "learning_rate": 1.152480331832706e-06, + "loss": 3.1168, + "step": 6406 + }, + { + "epoch": 9.37, + "grad_norm": 0.36112117767333984, + "learning_rate": 1.1453908251886636e-06, + "loss": 3.1347, + "step": 6407 + }, + { + "epoch": 9.38, + "grad_norm": 0.34627094864845276, + "learning_rate": 1.138323065896163e-06, + "loss": 3.1279, + "step": 6408 + }, + { + "epoch": 9.38, + "grad_norm": 0.24105395376682281, + "learning_rate": 1.131277055510027e-06, + "loss": 3.1302, + "step": 6409 + }, + { + "epoch": 9.38, + "grad_norm": 0.4641140401363373, + "learning_rate": 1.1242527955803494e-06, + "loss": 3.1347, + "step": 6410 + }, + { + "epoch": 9.38, + "grad_norm": 0.5431744456291199, + "learning_rate": 1.1172502876524382e-06, + "loss": 3.1303, + "step": 6411 + }, + { + "epoch": 9.38, + "grad_norm": 0.4685412645339966, + "learning_rate": 1.1102695332667723e-06, + "loss": 3.1154, + "step": 6412 + }, + { + "epoch": 9.38, + "grad_norm": 0.6576271653175354, + "learning_rate": 1.1033105339590677e-06, + "loss": 3.0996, + "step": 6413 + }, + { + "epoch": 9.38, + "grad_norm": 0.45234495401382446, + "learning_rate": 1.0963732912602886e-06, + "loss": 3.1551, + "step": 6414 + }, + { + "epoch": 9.39, + "grad_norm": 1.590352177619934, + "learning_rate": 1.0894578066965588e-06, + "loss": 3.1459, + "step": 6415 + }, + { + "epoch": 9.39, + "grad_norm": 0.31640344858169556, + "learning_rate": 1.082564081789239e-06, + "loss": 3.1576, + "step": 6416 + }, + { + "epoch": 9.39, + "grad_norm": 0.35781222581863403, + "learning_rate": 1.075692118054905e-06, + "loss": 3.1269, + "step": 6417 + }, + { + "epoch": 9.39, + "grad_norm": 0.32014670968055725, + "learning_rate": 1.0688419170053477e-06, + "loss": 3.115, + "step": 6418 + }, + { + "epoch": 9.39, + "grad_norm": 0.2519568204879761, + "learning_rate": 1.0620134801475501e-06, + "loss": 3.1214, + "step": 6419 + }, + { + "epoch": 9.39, + "grad_norm": 0.31856784224510193, + "learning_rate": 1.055206808983744e-06, + "loss": 3.1433, + "step": 6420 + }, + { + "epoch": 9.4, + "grad_norm": 0.2100788950920105, + "learning_rate": 1.048421905011343e-06, + "loss": 3.1443, + "step": 6421 + }, + { + "epoch": 9.4, + "grad_norm": 0.20311523973941803, + "learning_rate": 1.0416587697229752e-06, + "loss": 3.1563, + "step": 6422 + }, + { + "epoch": 9.4, + "eval_loss": 2.9917690753936768, + "eval_runtime": 1170.1133, + "eval_samples_per_second": 14.65, + "eval_steps_per_second": 0.916, + "step": 6422 + }, + { + "epoch": 9.4, + "grad_norm": 0.18928997218608856, + "learning_rate": 1.0349174046064835e-06, + "loss": 3.1323, + "step": 6423 + }, + { + "epoch": 9.4, + "grad_norm": 0.21475400030612946, + "learning_rate": 1.0281978111449375e-06, + "loss": 3.1322, + "step": 6424 + }, + { + "epoch": 9.4, + "grad_norm": 0.5063421130180359, + "learning_rate": 1.0214999908165879e-06, + "loss": 3.1366, + "step": 6425 + }, + { + "epoch": 9.4, + "grad_norm": 0.3121015131473541, + "learning_rate": 1.0148239450949115e-06, + "loss": 3.1141, + "step": 6426 + }, + { + "epoch": 9.4, + "grad_norm": 0.33075249195098877, + "learning_rate": 1.0081696754486115e-06, + "loss": 3.099, + "step": 6427 + }, + { + "epoch": 9.41, + "grad_norm": 0.25428274273872375, + "learning_rate": 1.001537183341561e-06, + "loss": 3.1343, + "step": 6428 + }, + { + "epoch": 9.41, + "grad_norm": 0.4780999720096588, + "learning_rate": 9.949264702328708e-07, + "loss": 3.1992, + "step": 6429 + }, + { + "epoch": 9.41, + "grad_norm": 0.23802733421325684, + "learning_rate": 9.883375375768555e-07, + "loss": 3.1453, + "step": 6430 + }, + { + "epoch": 9.41, + "grad_norm": 0.33159157633781433, + "learning_rate": 9.817703868230443e-07, + "loss": 3.1608, + "step": 6431 + }, + { + "epoch": 9.41, + "grad_norm": 1.387945532798767, + "learning_rate": 9.752250194161371e-07, + "loss": 3.1124, + "step": 6432 + }, + { + "epoch": 9.41, + "grad_norm": 0.4307352304458618, + "learning_rate": 9.687014367961045e-07, + "loss": 3.1762, + "step": 6433 + }, + { + "epoch": 9.41, + "grad_norm": 0.30907031893730164, + "learning_rate": 9.621996403980759e-07, + "loss": 3.1014, + "step": 6434 + }, + { + "epoch": 9.42, + "grad_norm": 0.2936875820159912, + "learning_rate": 9.557196316524076e-07, + "loss": 3.1718, + "step": 6435 + }, + { + "epoch": 9.42, + "grad_norm": 0.2288825958967209, + "learning_rate": 9.492614119846477e-07, + "loss": 3.1688, + "step": 6436 + }, + { + "epoch": 9.42, + "grad_norm": 0.272483229637146, + "learning_rate": 9.428249828155822e-07, + "loss": 3.1595, + "step": 6437 + }, + { + "epoch": 9.42, + "grad_norm": 0.33346354961395264, + "learning_rate": 9.36410345561145e-07, + "loss": 3.1253, + "step": 6438 + }, + { + "epoch": 9.42, + "grad_norm": 0.23747757077217102, + "learning_rate": 9.30017501632563e-07, + "loss": 3.0796, + "step": 6439 + }, + { + "epoch": 9.42, + "grad_norm": 0.32719528675079346, + "learning_rate": 9.236464524361777e-07, + "loss": 3.1361, + "step": 6440 + }, + { + "epoch": 9.42, + "grad_norm": 0.2457912415266037, + "learning_rate": 9.172971993735901e-07, + "loss": 3.1148, + "step": 6441 + }, + { + "epoch": 9.43, + "grad_norm": 0.3018089234828949, + "learning_rate": 9.109697438416054e-07, + "loss": 3.1641, + "step": 6442 + }, + { + "epoch": 9.43, + "grad_norm": 0.4333385229110718, + "learning_rate": 9.046640872322099e-07, + "loss": 3.1158, + "step": 6443 + }, + { + "epoch": 9.43, + "grad_norm": 0.41932156682014465, + "learning_rate": 8.98380230932605e-07, + "loss": 3.1115, + "step": 6444 + }, + { + "epoch": 9.43, + "grad_norm": 0.4317435026168823, + "learning_rate": 8.921181763252074e-07, + "loss": 3.1436, + "step": 6445 + }, + { + "epoch": 9.43, + "grad_norm": 0.275638610124588, + "learning_rate": 8.858779247876037e-07, + "loss": 3.1135, + "step": 6446 + }, + { + "epoch": 9.43, + "grad_norm": 0.3342507481575012, + "learning_rate": 8.796594776926292e-07, + "loss": 3.1085, + "step": 6447 + }, + { + "epoch": 9.44, + "grad_norm": 0.6510316729545593, + "learning_rate": 8.734628364083009e-07, + "loss": 3.0985, + "step": 6448 + }, + { + "epoch": 9.44, + "grad_norm": 0.1988769769668579, + "learning_rate": 8.672880022978392e-07, + "loss": 3.1544, + "step": 6449 + }, + { + "epoch": 9.44, + "grad_norm": 0.3602251410484314, + "learning_rate": 8.611349767196464e-07, + "loss": 3.1167, + "step": 6450 + }, + { + "epoch": 9.44, + "grad_norm": 0.3572871685028076, + "learning_rate": 8.55003761027362e-07, + "loss": 3.1644, + "step": 6451 + }, + { + "epoch": 9.44, + "grad_norm": 0.31289011240005493, + "learning_rate": 8.488943565698071e-07, + "loss": 3.1489, + "step": 6452 + }, + { + "epoch": 9.44, + "grad_norm": 0.46560922265052795, + "learning_rate": 8.428067646910176e-07, + "loss": 3.1438, + "step": 6453 + }, + { + "epoch": 9.44, + "grad_norm": 0.35999464988708496, + "learning_rate": 8.367409867302112e-07, + "loss": 3.1807, + "step": 6454 + }, + { + "epoch": 9.45, + "grad_norm": 0.28948530554771423, + "learning_rate": 8.306970240218315e-07, + "loss": 3.1289, + "step": 6455 + }, + { + "epoch": 9.45, + "grad_norm": 0.34700512886047363, + "learning_rate": 8.246748778955038e-07, + "loss": 3.124, + "step": 6456 + }, + { + "epoch": 9.45, + "grad_norm": 0.3402508795261383, + "learning_rate": 8.186745496760573e-07, + "loss": 3.1207, + "step": 6457 + }, + { + "epoch": 9.45, + "grad_norm": 0.22101683914661407, + "learning_rate": 8.126960406835249e-07, + "loss": 3.1162, + "step": 6458 + }, + { + "epoch": 9.45, + "grad_norm": 0.22484362125396729, + "learning_rate": 8.067393522331435e-07, + "loss": 3.1116, + "step": 6459 + }, + { + "epoch": 9.45, + "grad_norm": 0.2751406729221344, + "learning_rate": 8.008044856353313e-07, + "loss": 3.142, + "step": 6460 + }, + { + "epoch": 9.45, + "grad_norm": 0.3033272922039032, + "learning_rate": 7.948914421957332e-07, + "loss": 3.1218, + "step": 6461 + }, + { + "epoch": 9.46, + "grad_norm": 0.3599221706390381, + "learning_rate": 7.890002232151861e-07, + "loss": 3.1326, + "step": 6462 + }, + { + "epoch": 9.46, + "grad_norm": 2.09816312789917, + "learning_rate": 7.831308299896866e-07, + "loss": 3.1378, + "step": 6463 + }, + { + "epoch": 9.46, + "grad_norm": 0.33876097202301025, + "learning_rate": 7.772832638105021e-07, + "loss": 3.1112, + "step": 6464 + }, + { + "epoch": 9.46, + "grad_norm": 0.37547340989112854, + "learning_rate": 7.714575259640366e-07, + "loss": 3.1411, + "step": 6465 + }, + { + "epoch": 9.46, + "grad_norm": 0.31911343336105347, + "learning_rate": 7.656536177318984e-07, + "loss": 3.1397, + "step": 6466 + }, + { + "epoch": 9.46, + "grad_norm": 0.21420590579509735, + "learning_rate": 7.598715403909329e-07, + "loss": 3.1372, + "step": 6467 + }, + { + "epoch": 9.47, + "grad_norm": 0.3588124215602875, + "learning_rate": 7.541112952131669e-07, + "loss": 3.1421, + "step": 6468 + }, + { + "epoch": 9.47, + "grad_norm": 1.5312747955322266, + "learning_rate": 7.483728834657755e-07, + "loss": 3.1242, + "step": 6469 + }, + { + "epoch": 9.47, + "grad_norm": 0.19234134256839752, + "learning_rate": 7.426563064112046e-07, + "loss": 3.1337, + "step": 6470 + }, + { + "epoch": 9.47, + "grad_norm": 0.32724443078041077, + "learning_rate": 7.369615653070594e-07, + "loss": 3.1235, + "step": 6471 + }, + { + "epoch": 9.47, + "grad_norm": 0.22673092782497406, + "learning_rate": 7.312886614061265e-07, + "loss": 3.1116, + "step": 6472 + }, + { + "epoch": 9.47, + "grad_norm": 0.2321528047323227, + "learning_rate": 7.256375959564299e-07, + "loss": 3.1596, + "step": 6473 + }, + { + "epoch": 9.47, + "grad_norm": 0.2677914798259735, + "learning_rate": 7.200083702011528e-07, + "loss": 3.1541, + "step": 6474 + }, + { + "epoch": 9.48, + "grad_norm": 0.698582649230957, + "learning_rate": 7.144009853786826e-07, + "loss": 3.2143, + "step": 6475 + }, + { + "epoch": 9.48, + "grad_norm": 0.29191771149635315, + "learning_rate": 7.088154427226213e-07, + "loss": 3.1183, + "step": 6476 + }, + { + "epoch": 9.48, + "grad_norm": 0.23747608065605164, + "learning_rate": 7.032517434617413e-07, + "loss": 3.1147, + "step": 6477 + }, + { + "epoch": 9.48, + "grad_norm": 0.333160400390625, + "learning_rate": 6.977098888200196e-07, + "loss": 3.0996, + "step": 6478 + }, + { + "epoch": 9.48, + "grad_norm": 0.4941348731517792, + "learning_rate": 6.921898800166249e-07, + "loss": 3.1212, + "step": 6479 + }, + { + "epoch": 9.48, + "grad_norm": 0.35406991839408875, + "learning_rate": 6.866917182659194e-07, + "loss": 3.1587, + "step": 6480 + }, + { + "epoch": 9.48, + "grad_norm": 0.2900647521018982, + "learning_rate": 6.812154047774799e-07, + "loss": 3.1404, + "step": 6481 + }, + { + "epoch": 9.49, + "grad_norm": 0.2967754602432251, + "learning_rate": 6.757609407560428e-07, + "loss": 3.1544, + "step": 6482 + }, + { + "epoch": 9.49, + "grad_norm": 0.19114892184734344, + "learning_rate": 6.703283274015593e-07, + "loss": 3.0933, + "step": 6483 + }, + { + "epoch": 9.49, + "grad_norm": 0.27145031094551086, + "learning_rate": 6.649175659091622e-07, + "loss": 3.1036, + "step": 6484 + }, + { + "epoch": 9.49, + "grad_norm": 0.38012662529945374, + "learning_rate": 6.595286574691772e-07, + "loss": 3.1216, + "step": 6485 + }, + { + "epoch": 9.49, + "grad_norm": 0.23291641473770142, + "learning_rate": 6.54161603267145e-07, + "loss": 3.1082, + "step": 6486 + }, + { + "epoch": 9.49, + "grad_norm": 0.37107568979263306, + "learning_rate": 6.488164044837763e-07, + "loss": 3.1471, + "step": 6487 + }, + { + "epoch": 9.49, + "grad_norm": 0.23927997052669525, + "learning_rate": 6.434930622949753e-07, + "loss": 3.1683, + "step": 6488 + }, + { + "epoch": 9.5, + "grad_norm": 0.1931021809577942, + "learning_rate": 6.381915778718495e-07, + "loss": 3.1302, + "step": 6489 + }, + { + "epoch": 9.5, + "grad_norm": 0.3359212577342987, + "learning_rate": 6.32911952380677e-07, + "loss": 3.0989, + "step": 6490 + }, + { + "epoch": 9.5, + "grad_norm": 0.3253166377544403, + "learning_rate": 6.2765418698294e-07, + "loss": 3.1024, + "step": 6491 + }, + { + "epoch": 9.5, + "grad_norm": 0.2889295518398285, + "learning_rate": 6.224182828353242e-07, + "loss": 3.1311, + "step": 6492 + }, + { + "epoch": 9.5, + "grad_norm": 0.3093731999397278, + "learning_rate": 6.172042410896972e-07, + "loss": 3.1478, + "step": 6493 + }, + { + "epoch": 9.5, + "grad_norm": 0.218590646982193, + "learning_rate": 6.120120628930859e-07, + "loss": 3.1206, + "step": 6494 + }, + { + "epoch": 9.51, + "grad_norm": 0.31948429346084595, + "learning_rate": 6.068417493877654e-07, + "loss": 3.1603, + "step": 6495 + }, + { + "epoch": 9.51, + "grad_norm": 0.1962941735982895, + "learning_rate": 6.016933017111482e-07, + "loss": 3.1132, + "step": 6496 + }, + { + "epoch": 9.51, + "grad_norm": 0.20247966051101685, + "learning_rate": 5.965667209958614e-07, + "loss": 3.1348, + "step": 6497 + }, + { + "epoch": 9.51, + "grad_norm": 0.4274284839630127, + "learning_rate": 5.914620083697365e-07, + "loss": 3.1147, + "step": 6498 + }, + { + "epoch": 9.51, + "grad_norm": 0.34700536727905273, + "learning_rate": 5.863791649557637e-07, + "loss": 3.1406, + "step": 6499 + }, + { + "epoch": 9.51, + "grad_norm": 0.20800916850566864, + "learning_rate": 5.813181918721267e-07, + "loss": 3.1337, + "step": 6500 + }, + { + "epoch": 9.51, + "grad_norm": 0.26811301708221436, + "learning_rate": 5.762790902322123e-07, + "loss": 3.1002, + "step": 6501 + }, + { + "epoch": 9.52, + "grad_norm": 0.22953198850154877, + "learning_rate": 5.712618611446008e-07, + "loss": 3.1508, + "step": 6502 + }, + { + "epoch": 9.52, + "grad_norm": 0.3208388090133667, + "learning_rate": 5.662665057130201e-07, + "loss": 3.1635, + "step": 6503 + }, + { + "epoch": 9.52, + "grad_norm": 0.36171430349349976, + "learning_rate": 5.612930250364468e-07, + "loss": 3.1083, + "step": 6504 + }, + { + "epoch": 9.52, + "grad_norm": 0.43349719047546387, + "learning_rate": 5.563414202089945e-07, + "loss": 3.131, + "step": 6505 + }, + { + "epoch": 9.52, + "grad_norm": 0.7703474760055542, + "learning_rate": 5.514116923199919e-07, + "loss": 3.1445, + "step": 6506 + }, + { + "epoch": 9.52, + "grad_norm": 0.41442474722862244, + "learning_rate": 5.465038424539381e-07, + "loss": 3.1357, + "step": 6507 + }, + { + "epoch": 9.52, + "grad_norm": 0.29321762919425964, + "learning_rate": 5.416178716905251e-07, + "loss": 3.1088, + "step": 6508 + }, + { + "epoch": 9.53, + "grad_norm": 0.84801185131073, + "learning_rate": 5.367537811046485e-07, + "loss": 3.1096, + "step": 6509 + }, + { + "epoch": 9.53, + "grad_norm": 0.2752794325351715, + "learning_rate": 5.319115717663636e-07, + "loss": 3.1595, + "step": 6510 + }, + { + "epoch": 9.53, + "grad_norm": 0.27682313323020935, + "learning_rate": 5.270912447409293e-07, + "loss": 3.127, + "step": 6511 + }, + { + "epoch": 9.53, + "grad_norm": 0.2727954387664795, + "learning_rate": 5.222928010887862e-07, + "loss": 3.1745, + "step": 6512 + }, + { + "epoch": 9.53, + "grad_norm": 0.2682643234729767, + "learning_rate": 5.175162418655566e-07, + "loss": 3.1418, + "step": 6513 + }, + { + "epoch": 9.53, + "grad_norm": 0.24266032874584198, + "learning_rate": 5.127615681220443e-07, + "loss": 3.1784, + "step": 6514 + }, + { + "epoch": 9.53, + "grad_norm": 0.5018312335014343, + "learning_rate": 5.080287809042572e-07, + "loss": 3.122, + "step": 6515 + }, + { + "epoch": 9.54, + "grad_norm": 0.23751623928546906, + "learning_rate": 5.033178812533735e-07, + "loss": 3.1308, + "step": 6516 + }, + { + "epoch": 9.54, + "grad_norm": 0.3432961404323578, + "learning_rate": 4.98628870205764e-07, + "loss": 3.1387, + "step": 6517 + }, + { + "epoch": 9.54, + "grad_norm": 1.0361169576644897, + "learning_rate": 4.939617487929593e-07, + "loss": 3.1305, + "step": 6518 + }, + { + "epoch": 9.54, + "grad_norm": 0.3029511868953705, + "learning_rate": 4.89316518041727e-07, + "loss": 3.1152, + "step": 6519 + }, + { + "epoch": 9.54, + "grad_norm": 0.2776906490325928, + "learning_rate": 4.846931789739606e-07, + "loss": 3.1205, + "step": 6520 + }, + { + "epoch": 9.54, + "grad_norm": 0.9315544962882996, + "learning_rate": 4.800917326067689e-07, + "loss": 3.1491, + "step": 6521 + }, + { + "epoch": 9.55, + "grad_norm": 0.2468407303094864, + "learning_rate": 4.755121799524531e-07, + "loss": 3.1326, + "step": 6522 + }, + { + "epoch": 9.55, + "grad_norm": 0.3743303418159485, + "learning_rate": 4.709545220184741e-07, + "loss": 3.1277, + "step": 6523 + }, + { + "epoch": 9.55, + "grad_norm": 0.30229365825653076, + "learning_rate": 4.664187598074743e-07, + "loss": 3.2156, + "step": 6524 + }, + { + "epoch": 9.55, + "grad_norm": 0.2736436426639557, + "learning_rate": 4.61904894317311e-07, + "loss": 3.1003, + "step": 6525 + }, + { + "epoch": 9.55, + "grad_norm": 0.3961028456687927, + "learning_rate": 4.57412926541001e-07, + "loss": 3.1433, + "step": 6526 + }, + { + "epoch": 9.55, + "grad_norm": 0.2155308723449707, + "learning_rate": 4.529428574667427e-07, + "loss": 3.1291, + "step": 6527 + }, + { + "epoch": 9.55, + "grad_norm": 0.34587162733078003, + "learning_rate": 4.4849468807791615e-07, + "loss": 3.1351, + "step": 6528 + }, + { + "epoch": 9.56, + "grad_norm": 0.38426241278648376, + "learning_rate": 4.4406841935308305e-07, + "loss": 3.1023, + "step": 6529 + }, + { + "epoch": 9.56, + "grad_norm": 0.3618767559528351, + "learning_rate": 4.3966405226602006e-07, + "loss": 3.1678, + "step": 6530 + }, + { + "epoch": 9.56, + "grad_norm": 0.6507797241210938, + "learning_rate": 4.352815877856298e-07, + "loss": 3.1222, + "step": 6531 + }, + { + "epoch": 9.56, + "grad_norm": 0.38927918672561646, + "learning_rate": 4.309210268760411e-07, + "loss": 3.1351, + "step": 6532 + }, + { + "epoch": 9.56, + "grad_norm": 0.22871467471122742, + "learning_rate": 4.2658237049655323e-07, + "loss": 3.1175, + "step": 6533 + }, + { + "epoch": 9.56, + "grad_norm": 0.21361161768436432, + "learning_rate": 4.2226561960162503e-07, + "loss": 3.1446, + "step": 6534 + }, + { + "epoch": 9.56, + "grad_norm": 0.4756571650505066, + "learning_rate": 4.179707751409079e-07, + "loss": 3.1917, + "step": 6535 + }, + { + "epoch": 9.57, + "grad_norm": 0.34980106353759766, + "learning_rate": 4.136978380592682e-07, + "loss": 3.1541, + "step": 6536 + }, + { + "epoch": 9.57, + "grad_norm": 0.3700217306613922, + "learning_rate": 4.0944680929670963e-07, + "loss": 3.1239, + "step": 6537 + }, + { + "epoch": 9.57, + "grad_norm": 0.3132055997848511, + "learning_rate": 4.0521768978840636e-07, + "loss": 3.1432, + "step": 6538 + }, + { + "epoch": 9.57, + "grad_norm": 0.4539642035961151, + "learning_rate": 4.0101048046478075e-07, + "loss": 3.1253, + "step": 6539 + }, + { + "epoch": 9.57, + "grad_norm": 0.2445564568042755, + "learning_rate": 3.96825182251348e-07, + "loss": 3.1127, + "step": 6540 + }, + { + "epoch": 9.57, + "grad_norm": 0.19848205149173737, + "learning_rate": 3.9266179606887164e-07, + "loss": 3.11, + "step": 6541 + }, + { + "epoch": 9.57, + "grad_norm": 0.3294031322002411, + "learning_rate": 3.8852032283326346e-07, + "loss": 3.1112, + "step": 6542 + }, + { + "epoch": 9.58, + "grad_norm": 0.28210368752479553, + "learning_rate": 3.8440076345561684e-07, + "loss": 3.1453, + "step": 6543 + }, + { + "epoch": 9.58, + "grad_norm": 0.6630240678787231, + "learning_rate": 3.803031188421957e-07, + "loss": 3.1793, + "step": 6544 + }, + { + "epoch": 9.58, + "grad_norm": 0.2525712549686432, + "learning_rate": 3.7622738989447905e-07, + "loss": 3.1022, + "step": 6545 + }, + { + "epoch": 9.58, + "grad_norm": 0.2561974823474884, + "learning_rate": 3.72173577509094e-07, + "loss": 3.1025, + "step": 6546 + }, + { + "epoch": 9.58, + "grad_norm": 0.3544372618198395, + "learning_rate": 3.6814168257783833e-07, + "loss": 3.1123, + "step": 6547 + }, + { + "epoch": 9.58, + "grad_norm": 0.23596583306789398, + "learning_rate": 3.641317059877025e-07, + "loss": 3.1274, + "step": 6548 + }, + { + "epoch": 9.59, + "grad_norm": 0.6944418549537659, + "learning_rate": 3.601436486208698e-07, + "loss": 3.1257, + "step": 6549 + }, + { + "epoch": 9.59, + "grad_norm": 0.5012179613113403, + "learning_rate": 3.561775113546828e-07, + "loss": 3.1273, + "step": 6550 + }, + { + "epoch": 9.59, + "grad_norm": 0.2207842916250229, + "learning_rate": 3.522332950616436e-07, + "loss": 3.127, + "step": 6551 + }, + { + "epoch": 9.59, + "grad_norm": 0.4304347634315491, + "learning_rate": 3.483110006094803e-07, + "loss": 3.1117, + "step": 6552 + }, + { + "epoch": 9.59, + "grad_norm": 0.2506290674209595, + "learning_rate": 3.444106288610582e-07, + "loss": 3.1624, + "step": 6553 + }, + { + "epoch": 9.59, + "grad_norm": 0.24125221371650696, + "learning_rate": 3.405321806744355e-07, + "loss": 3.1096, + "step": 6554 + }, + { + "epoch": 9.59, + "grad_norm": 0.26603150367736816, + "learning_rate": 3.3667565690285177e-07, + "loss": 3.1007, + "step": 6555 + }, + { + "epoch": 9.6, + "grad_norm": 2.8406460285186768, + "learning_rate": 3.3284105839470616e-07, + "loss": 3.1371, + "step": 6556 + }, + { + "epoch": 9.6, + "grad_norm": 0.3415467441082001, + "learning_rate": 3.2902838599357943e-07, + "loss": 3.1232, + "step": 6557 + }, + { + "epoch": 9.6, + "grad_norm": 2.049724817276001, + "learning_rate": 3.252376405382562e-07, + "loss": 3.1637, + "step": 6558 + }, + { + "epoch": 9.6, + "grad_norm": 0.36443763971328735, + "learning_rate": 3.214688228626583e-07, + "loss": 3.1664, + "step": 6559 + }, + { + "epoch": 9.6, + "grad_norm": 0.3832044303417206, + "learning_rate": 3.177219337958892e-07, + "loss": 3.1173, + "step": 6560 + }, + { + "epoch": 9.6, + "grad_norm": 0.5116267204284668, + "learning_rate": 3.139969741622673e-07, + "loss": 3.1362, + "step": 6561 + }, + { + "epoch": 9.6, + "grad_norm": 0.25494176149368286, + "learning_rate": 3.1029394478124805e-07, + "loss": 3.1506, + "step": 6562 + }, + { + "epoch": 9.61, + "grad_norm": 0.5267819762229919, + "learning_rate": 3.066128464674578e-07, + "loss": 3.1126, + "step": 6563 + }, + { + "epoch": 9.61, + "grad_norm": 0.5442273616790771, + "learning_rate": 3.0295368003073756e-07, + "loss": 3.1846, + "step": 6564 + }, + { + "epoch": 9.61, + "grad_norm": 0.5021152496337891, + "learning_rate": 2.993164462760656e-07, + "loss": 3.1337, + "step": 6565 + }, + { + "epoch": 9.61, + "grad_norm": 0.35817328095436096, + "learning_rate": 2.9570114600360186e-07, + "loss": 3.144, + "step": 6566 + }, + { + "epoch": 9.61, + "grad_norm": 0.6157897114753723, + "learning_rate": 2.921077800087102e-07, + "loss": 3.1411, + "step": 6567 + }, + { + "epoch": 9.61, + "grad_norm": 0.3213179111480713, + "learning_rate": 2.885363490818804e-07, + "loss": 3.1111, + "step": 6568 + }, + { + "epoch": 9.61, + "grad_norm": 0.24565556645393372, + "learning_rate": 2.849868540088285e-07, + "loss": 3.1359, + "step": 6569 + }, + { + "epoch": 9.62, + "grad_norm": 0.7529188990592957, + "learning_rate": 2.814592955704076e-07, + "loss": 3.1213, + "step": 6570 + }, + { + "epoch": 9.62, + "grad_norm": 1.2467443943023682, + "learning_rate": 2.779536745426525e-07, + "loss": 3.1394, + "step": 6571 + }, + { + "epoch": 9.62, + "grad_norm": 0.3905908465385437, + "learning_rate": 2.744699916967797e-07, + "loss": 3.1464, + "step": 6572 + }, + { + "epoch": 9.62, + "grad_norm": 0.5479477047920227, + "learning_rate": 2.710082477991871e-07, + "loss": 3.1293, + "step": 6573 + }, + { + "epoch": 9.62, + "grad_norm": 0.3964822590351105, + "learning_rate": 2.675684436114212e-07, + "loss": 3.0963, + "step": 6574 + }, + { + "epoch": 9.62, + "grad_norm": 0.19991499185562134, + "learning_rate": 2.6415057989022107e-07, + "loss": 3.1233, + "step": 6575 + }, + { + "epoch": 9.63, + "grad_norm": 0.268430233001709, + "learning_rate": 2.607546573874853e-07, + "loss": 3.1624, + "step": 6576 + }, + { + "epoch": 9.63, + "grad_norm": 0.4299977123737335, + "learning_rate": 2.573806768503162e-07, + "loss": 3.1199, + "step": 6577 + }, + { + "epoch": 9.63, + "grad_norm": 0.5895475149154663, + "learning_rate": 2.540286390209423e-07, + "loss": 3.1271, + "step": 6578 + }, + { + "epoch": 9.63, + "grad_norm": 0.3114526867866516, + "learning_rate": 2.506985446368071e-07, + "loss": 3.1572, + "step": 6579 + }, + { + "epoch": 9.63, + "grad_norm": 0.28470930457115173, + "learning_rate": 2.4739039443049115e-07, + "loss": 3.1482, + "step": 6580 + }, + { + "epoch": 9.63, + "grad_norm": 0.2567174434661865, + "learning_rate": 2.4410418912979014e-07, + "loss": 3.1423, + "step": 6581 + }, + { + "epoch": 9.63, + "grad_norm": 0.4415115714073181, + "learning_rate": 2.408399294576258e-07, + "loss": 3.1218, + "step": 6582 + }, + { + "epoch": 9.64, + "grad_norm": 0.2566371560096741, + "learning_rate": 2.375976161321236e-07, + "loss": 3.1582, + "step": 6583 + }, + { + "epoch": 9.64, + "grad_norm": 0.1854650229215622, + "learning_rate": 2.343772498665686e-07, + "loss": 3.1261, + "step": 6584 + }, + { + "epoch": 9.64, + "grad_norm": 0.5996760725975037, + "learning_rate": 2.311788313694052e-07, + "loss": 3.1236, + "step": 6585 + }, + { + "epoch": 9.64, + "grad_norm": 0.21690422296524048, + "learning_rate": 2.2800236134429276e-07, + "loss": 3.155, + "step": 6586 + }, + { + "epoch": 9.64, + "grad_norm": 0.753720223903656, + "learning_rate": 2.2484784049001672e-07, + "loss": 3.1147, + "step": 6587 + }, + { + "epoch": 9.64, + "grad_norm": 0.3540690839290619, + "learning_rate": 2.2171526950054421e-07, + "loss": 3.1183, + "step": 6588 + }, + { + "epoch": 9.64, + "grad_norm": 0.19896644353866577, + "learning_rate": 2.1860464906503507e-07, + "loss": 3.1367, + "step": 6589 + }, + { + "epoch": 9.65, + "grad_norm": 0.31346434354782104, + "learning_rate": 2.1551597986780859e-07, + "loss": 3.1322, + "step": 6590 + }, + { + "epoch": 9.65, + "grad_norm": 0.19907262921333313, + "learning_rate": 2.124492625883212e-07, + "loss": 3.1373, + "step": 6591 + }, + { + "epoch": 9.65, + "eval_loss": 2.9914393424987793, + "eval_runtime": 1163.7003, + "eval_samples_per_second": 14.731, + "eval_steps_per_second": 0.921, + "step": 6591 + }, + { + "epoch": 9.65, + "grad_norm": 0.3037753999233246, + "learning_rate": 2.0940449790125549e-07, + "loss": 3.1501, + "step": 6592 + }, + { + "epoch": 9.65, + "grad_norm": 0.2875763773918152, + "learning_rate": 2.063816864764423e-07, + "loss": 3.1402, + "step": 6593 + }, + { + "epoch": 9.65, + "grad_norm": 0.5775416493415833, + "learning_rate": 2.0338082897886079e-07, + "loss": 3.1158, + "step": 6594 + }, + { + "epoch": 9.65, + "grad_norm": 0.2470092624425888, + "learning_rate": 2.0040192606869402e-07, + "loss": 3.1199, + "step": 6595 + }, + { + "epoch": 9.66, + "grad_norm": 0.2634389102458954, + "learning_rate": 1.974449784012733e-07, + "loss": 3.1039, + "step": 6596 + }, + { + "epoch": 9.66, + "grad_norm": 0.24670274555683136, + "learning_rate": 1.9450998662711163e-07, + "loss": 3.1321, + "step": 6597 + }, + { + "epoch": 9.66, + "grad_norm": 0.22088152170181274, + "learning_rate": 1.915969513918925e-07, + "loss": 3.1406, + "step": 6598 + }, + { + "epoch": 9.66, + "grad_norm": 0.30397117137908936, + "learning_rate": 1.8870587333644775e-07, + "loss": 3.1404, + "step": 6599 + }, + { + "epoch": 9.66, + "grad_norm": 0.23878346383571625, + "learning_rate": 1.8583675309681304e-07, + "loss": 3.1015, + "step": 6600 + }, + { + "epoch": 9.66, + "grad_norm": 0.3267252445220947, + "learning_rate": 1.8298959130416127e-07, + "loss": 3.155, + "step": 6601 + }, + { + "epoch": 9.66, + "grad_norm": 0.23852427303791046, + "learning_rate": 1.8016438858485807e-07, + "loss": 3.1309, + "step": 6602 + }, + { + "epoch": 9.67, + "grad_norm": 0.2040443867444992, + "learning_rate": 1.773611455604285e-07, + "loss": 3.1556, + "step": 6603 + }, + { + "epoch": 9.67, + "grad_norm": 0.21314983069896698, + "learning_rate": 1.745798628475681e-07, + "loss": 3.1152, + "step": 6604 + }, + { + "epoch": 9.67, + "grad_norm": 0.15410488843917847, + "learning_rate": 1.7182054105813194e-07, + "loss": 3.1187, + "step": 6605 + }, + { + "epoch": 9.67, + "grad_norm": 0.2223520427942276, + "learning_rate": 1.6908318079915663e-07, + "loss": 3.1321, + "step": 6606 + }, + { + "epoch": 9.67, + "grad_norm": 0.2847364544868469, + "learning_rate": 1.6636778267284936e-07, + "loss": 3.1052, + "step": 6607 + }, + { + "epoch": 9.67, + "grad_norm": 0.4363754093647003, + "learning_rate": 1.636743472765656e-07, + "loss": 3.1249, + "step": 6608 + }, + { + "epoch": 9.67, + "grad_norm": 0.26813626289367676, + "learning_rate": 1.6100287520286473e-07, + "loss": 3.1313, + "step": 6609 + }, + { + "epoch": 9.68, + "grad_norm": 0.21203118562698364, + "learning_rate": 1.5835336703943216e-07, + "loss": 3.1671, + "step": 6610 + }, + { + "epoch": 9.68, + "grad_norm": 0.29765477776527405, + "learning_rate": 1.557258233691572e-07, + "loss": 3.1452, + "step": 6611 + }, + { + "epoch": 9.68, + "grad_norm": 0.21558693051338196, + "learning_rate": 1.5312024477006637e-07, + "loss": 3.1426, + "step": 6612 + }, + { + "epoch": 9.68, + "grad_norm": 0.3551577031612396, + "learning_rate": 1.505366318153789e-07, + "loss": 3.1334, + "step": 6613 + }, + { + "epoch": 9.68, + "grad_norm": 0.21328043937683105, + "learning_rate": 1.4797498507347352e-07, + "loss": 3.1, + "step": 6614 + }, + { + "epoch": 9.68, + "grad_norm": 0.49471306800842285, + "learning_rate": 1.4543530510788828e-07, + "loss": 3.1552, + "step": 6615 + }, + { + "epoch": 9.68, + "grad_norm": 0.43984168767929077, + "learning_rate": 1.4291759247734294e-07, + "loss": 3.1227, + "step": 6616 + }, + { + "epoch": 9.69, + "grad_norm": 0.34304744005203247, + "learning_rate": 1.4042184773571666e-07, + "loss": 3.1755, + "step": 6617 + }, + { + "epoch": 9.69, + "grad_norm": 0.16853567957878113, + "learning_rate": 1.3794807143205912e-07, + "loss": 3.1358, + "step": 6618 + }, + { + "epoch": 9.69, + "grad_norm": 0.26293420791625977, + "learning_rate": 1.3549626411056836e-07, + "loss": 3.1494, + "step": 6619 + }, + { + "epoch": 9.69, + "grad_norm": 0.3390566408634186, + "learning_rate": 1.3306642631064625e-07, + "loss": 3.122, + "step": 6620 + }, + { + "epoch": 9.69, + "grad_norm": 0.36053574085235596, + "learning_rate": 1.3065855856683184e-07, + "loss": 3.1467, + "step": 6621 + }, + { + "epoch": 9.69, + "grad_norm": 0.2748585045337677, + "learning_rate": 1.2827266140883476e-07, + "loss": 3.0955, + "step": 6622 + }, + { + "epoch": 9.7, + "grad_norm": 0.29848429560661316, + "learning_rate": 1.2590873536154623e-07, + "loss": 3.1919, + "step": 6623 + }, + { + "epoch": 9.7, + "grad_norm": 0.36377960443496704, + "learning_rate": 1.2356678094500584e-07, + "loss": 3.1346, + "step": 6624 + }, + { + "epoch": 9.7, + "grad_norm": 0.22711066901683807, + "learning_rate": 1.2124679867443477e-07, + "loss": 3.1583, + "step": 6625 + }, + { + "epoch": 9.7, + "grad_norm": 0.2531357407569885, + "learning_rate": 1.1894878906020257e-07, + "loss": 3.1648, + "step": 6626 + }, + { + "epoch": 9.7, + "grad_norm": 0.37433087825775146, + "learning_rate": 1.1667275260787148e-07, + "loss": 3.1278, + "step": 6627 + }, + { + "epoch": 9.7, + "grad_norm": 0.31404930353164673, + "learning_rate": 1.1441868981815207e-07, + "loss": 3.1007, + "step": 6628 + }, + { + "epoch": 9.7, + "grad_norm": 0.3023819625377655, + "learning_rate": 1.1218660118690328e-07, + "loss": 3.1456, + "step": 6629 + }, + { + "epoch": 9.71, + "grad_norm": 0.24183262884616852, + "learning_rate": 1.0997648720519893e-07, + "loss": 3.1549, + "step": 6630 + }, + { + "epoch": 9.71, + "grad_norm": 0.1486586481332779, + "learning_rate": 1.077883483592168e-07, + "loss": 3.1399, + "step": 6631 + }, + { + "epoch": 9.71, + "grad_norm": 0.3949415981769562, + "learning_rate": 1.056221851303607e-07, + "loss": 3.1365, + "step": 6632 + }, + { + "epoch": 9.71, + "grad_norm": 0.6169975996017456, + "learning_rate": 1.0347799799516056e-07, + "loss": 3.1777, + "step": 6633 + }, + { + "epoch": 9.71, + "grad_norm": 0.24591876566410065, + "learning_rate": 1.0135578742532793e-07, + "loss": 3.0985, + "step": 6634 + }, + { + "epoch": 9.71, + "grad_norm": 0.3505055606365204, + "learning_rate": 9.925555388772267e-08, + "loss": 3.1615, + "step": 6635 + }, + { + "epoch": 9.71, + "grad_norm": 0.36763930320739746, + "learning_rate": 9.717729784439745e-08, + "loss": 3.1427, + "step": 6636 + }, + { + "epoch": 9.72, + "grad_norm": 0.36300286650657654, + "learning_rate": 9.512101975254207e-08, + "loss": 3.1379, + "step": 6637 + }, + { + "epoch": 9.72, + "grad_norm": 0.2554195523262024, + "learning_rate": 9.308672006453911e-08, + "loss": 3.0936, + "step": 6638 + }, + { + "epoch": 9.72, + "grad_norm": 0.1804320514202118, + "learning_rate": 9.107439922789729e-08, + "loss": 3.1193, + "step": 6639 + }, + { + "epoch": 9.72, + "grad_norm": 0.44588401913642883, + "learning_rate": 8.908405768534022e-08, + "loss": 3.1212, + "step": 6640 + }, + { + "epoch": 9.72, + "grad_norm": 0.19104427099227905, + "learning_rate": 8.711569587470659e-08, + "loss": 3.11, + "step": 6641 + }, + { + "epoch": 9.72, + "grad_norm": 0.4022594690322876, + "learning_rate": 8.516931422903885e-08, + "loss": 3.2002, + "step": 6642 + }, + { + "epoch": 9.72, + "grad_norm": 0.283624529838562, + "learning_rate": 8.324491317652782e-08, + "loss": 3.1126, + "step": 6643 + }, + { + "epoch": 9.73, + "grad_norm": 0.26203152537345886, + "learning_rate": 8.134249314051267e-08, + "loss": 3.1584, + "step": 6644 + }, + { + "epoch": 9.73, + "grad_norm": 0.21490979194641113, + "learning_rate": 7.946205453953637e-08, + "loss": 3.1602, + "step": 6645 + }, + { + "epoch": 9.73, + "grad_norm": 0.27834105491638184, + "learning_rate": 7.76035977872569e-08, + "loss": 3.099, + "step": 6646 + }, + { + "epoch": 9.73, + "grad_norm": 0.31908273696899414, + "learning_rate": 7.576712329254721e-08, + "loss": 3.1468, + "step": 6647 + }, + { + "epoch": 9.73, + "grad_norm": 0.1860104501247406, + "learning_rate": 7.395263145939524e-08, + "loss": 3.1205, + "step": 6648 + }, + { + "epoch": 9.73, + "grad_norm": 0.4036659002304077, + "learning_rate": 7.21601226870039e-08, + "loss": 3.1568, + "step": 6649 + }, + { + "epoch": 9.74, + "grad_norm": 0.3992268741130829, + "learning_rate": 7.038959736969108e-08, + "loss": 3.1151, + "step": 6650 + }, + { + "epoch": 9.74, + "grad_norm": 0.31392067670822144, + "learning_rate": 6.864105589696746e-08, + "loss": 3.1275, + "step": 6651 + }, + { + "epoch": 9.74, + "grad_norm": 0.267260879278183, + "learning_rate": 6.691449865350308e-08, + "loss": 3.1309, + "step": 6652 + }, + { + "epoch": 9.74, + "grad_norm": 0.26456329226493835, + "learning_rate": 6.520992601912746e-08, + "loss": 3.1379, + "step": 6653 + }, + { + "epoch": 9.74, + "grad_norm": 0.4101209044456482, + "learning_rate": 6.352733836884062e-08, + "loss": 3.1462, + "step": 6654 + }, + { + "epoch": 9.74, + "grad_norm": 0.23847658932209015, + "learning_rate": 6.186673607279092e-08, + "loss": 3.1299, + "step": 6655 + }, + { + "epoch": 9.74, + "grad_norm": 0.24062010645866394, + "learning_rate": 6.022811949629725e-08, + "loss": 3.122, + "step": 6656 + }, + { + "epoch": 9.75, + "grad_norm": 3.0939149856567383, + "learning_rate": 5.861148899984903e-08, + "loss": 3.1619, + "step": 6657 + }, + { + "epoch": 9.75, + "grad_norm": 0.21735475957393646, + "learning_rate": 5.701684493910619e-08, + "loss": 3.113, + "step": 6658 + }, + { + "epoch": 9.75, + "grad_norm": 0.2022494077682495, + "learning_rate": 5.544418766486592e-08, + "loss": 3.1144, + "step": 6659 + }, + { + "epoch": 9.75, + "grad_norm": 0.27019014954566956, + "learning_rate": 5.389351752310701e-08, + "loss": 3.1156, + "step": 6660 + }, + { + "epoch": 9.75, + "grad_norm": 0.23590201139450073, + "learning_rate": 5.236483485496768e-08, + "loss": 3.1197, + "step": 6661 + }, + { + "epoch": 9.75, + "grad_norm": 0.3665682077407837, + "learning_rate": 5.08581399967345e-08, + "loss": 3.1041, + "step": 6662 + }, + { + "epoch": 9.75, + "grad_norm": 0.23733316361904144, + "learning_rate": 4.937343327988675e-08, + "loss": 3.135, + "step": 6663 + }, + { + "epoch": 9.76, + "grad_norm": 0.23805661499500275, + "learning_rate": 4.7910715031052046e-08, + "loss": 3.1082, + "step": 6664 + }, + { + "epoch": 9.76, + "grad_norm": 0.2705228924751282, + "learning_rate": 4.6469985571995225e-08, + "loss": 3.1506, + "step": 6665 + }, + { + "epoch": 9.76, + "grad_norm": 0.1711297333240509, + "learning_rate": 4.505124521969606e-08, + "loss": 3.1259, + "step": 6666 + }, + { + "epoch": 9.76, + "grad_norm": 0.3597375154495239, + "learning_rate": 4.3654494286249345e-08, + "loss": 3.1425, + "step": 6667 + }, + { + "epoch": 9.76, + "grad_norm": 0.9291727542877197, + "learning_rate": 4.2279733078931516e-08, + "loss": 3.1775, + "step": 6668 + }, + { + "epoch": 9.76, + "grad_norm": 0.35449373722076416, + "learning_rate": 4.092696190018952e-08, + "loss": 3.1286, + "step": 6669 + }, + { + "epoch": 9.76, + "grad_norm": 0.3513713479042053, + "learning_rate": 3.959618104762974e-08, + "loss": 3.1597, + "step": 6670 + }, + { + "epoch": 9.77, + "grad_norm": 0.21470598876476288, + "learning_rate": 3.828739081399579e-08, + "loss": 3.153, + "step": 6671 + }, + { + "epoch": 9.77, + "grad_norm": 0.40011513233184814, + "learning_rate": 3.700059148722401e-08, + "loss": 3.1228, + "step": 6672 + }, + { + "epoch": 9.77, + "grad_norm": 0.1759519726037979, + "learning_rate": 3.5735783350399064e-08, + "loss": 3.1599, + "step": 6673 + }, + { + "epoch": 9.77, + "grad_norm": 0.16630885004997253, + "learning_rate": 3.4492966681765046e-08, + "loss": 3.1679, + "step": 6674 + }, + { + "epoch": 9.77, + "grad_norm": 0.5002431273460388, + "learning_rate": 3.3272141754747686e-08, + "loss": 3.1186, + "step": 6675 + }, + { + "epoch": 9.77, + "grad_norm": 0.24642068147659302, + "learning_rate": 3.207330883790993e-08, + "loss": 3.1312, + "step": 6676 + }, + { + "epoch": 9.78, + "grad_norm": 0.24573257565498352, + "learning_rate": 3.089646819498526e-08, + "loss": 3.1215, + "step": 6677 + }, + { + "epoch": 9.78, + "grad_norm": 0.3076600134372711, + "learning_rate": 2.9741620084877685e-08, + "loss": 3.1408, + "step": 6678 + }, + { + "epoch": 9.78, + "grad_norm": 0.32427266240119934, + "learning_rate": 2.8608764761639538e-08, + "loss": 3.1366, + "step": 6679 + }, + { + "epoch": 9.78, + "grad_norm": 0.2206949144601822, + "learning_rate": 2.7497902474504788e-08, + "loss": 3.1552, + "step": 6680 + }, + { + "epoch": 9.78, + "grad_norm": 0.2506512403488159, + "learning_rate": 2.640903346783352e-08, + "loss": 3.1244, + "step": 6681 + }, + { + "epoch": 9.78, + "grad_norm": 0.279695600271225, + "learning_rate": 2.5342157981189662e-08, + "loss": 3.1487, + "step": 6682 + }, + { + "epoch": 9.78, + "grad_norm": 0.21855193376541138, + "learning_rate": 2.4297276249263255e-08, + "loss": 3.1121, + "step": 6683 + }, + { + "epoch": 9.79, + "grad_norm": 0.2374710589647293, + "learning_rate": 2.3274388501937082e-08, + "loss": 3.1674, + "step": 6684 + }, + { + "epoch": 9.79, + "grad_norm": 0.32136070728302, + "learning_rate": 2.2273494964220044e-08, + "loss": 3.1362, + "step": 6685 + }, + { + "epoch": 9.79, + "grad_norm": 0.4361661374568939, + "learning_rate": 2.1294595856324872e-08, + "loss": 3.1293, + "step": 6686 + }, + { + "epoch": 9.79, + "grad_norm": 0.2575805187225342, + "learning_rate": 2.033769139359043e-08, + "loss": 3.1001, + "step": 6687 + }, + { + "epoch": 9.79, + "grad_norm": 0.4384787380695343, + "learning_rate": 1.9402781786515002e-08, + "loss": 3.0979, + "step": 6688 + }, + { + "epoch": 9.79, + "grad_norm": 0.39492806792259216, + "learning_rate": 1.8489867240800706e-08, + "loss": 3.1892, + "step": 6689 + }, + { + "epoch": 9.79, + "grad_norm": 0.2683747112751007, + "learning_rate": 1.7598947957264687e-08, + "loss": 3.095, + "step": 6690 + }, + { + "epoch": 9.8, + "grad_norm": 0.3403540551662445, + "learning_rate": 1.673002413191682e-08, + "loss": 3.1194, + "step": 6691 + }, + { + "epoch": 9.8, + "grad_norm": 0.3888680040836334, + "learning_rate": 1.58830959558931e-08, + "loss": 3.1255, + "step": 6692 + }, + { + "epoch": 9.8, + "grad_norm": 0.3829781413078308, + "learning_rate": 1.505816361553336e-08, + "loss": 3.1308, + "step": 6693 + }, + { + "epoch": 9.8, + "grad_norm": 0.248688206076622, + "learning_rate": 1.4255227292303553e-08, + "loss": 3.146, + "step": 6694 + }, + { + "epoch": 9.8, + "grad_norm": 2.7760934829711914, + "learning_rate": 1.3474287162851262e-08, + "loss": 3.1347, + "step": 6695 + }, + { + "epoch": 9.8, + "grad_norm": 0.2652311623096466, + "learning_rate": 1.2715343398972402e-08, + "loss": 3.1425, + "step": 6696 + }, + { + "epoch": 9.81, + "grad_norm": 0.20968441665172577, + "learning_rate": 1.197839616763341e-08, + "loss": 3.1408, + "step": 6697 + }, + { + "epoch": 9.81, + "grad_norm": 0.25183215737342834, + "learning_rate": 1.1263445630960157e-08, + "loss": 3.1272, + "step": 6698 + }, + { + "epoch": 9.81, + "grad_norm": 0.21223939955234528, + "learning_rate": 1.0570491946237936e-08, + "loss": 3.1286, + "step": 6699 + }, + { + "epoch": 9.81, + "grad_norm": 0.22320213913917542, + "learning_rate": 9.899535265900373e-09, + "loss": 3.121, + "step": 6700 + }, + { + "epoch": 9.81, + "grad_norm": 0.3158072829246521, + "learning_rate": 9.250575737562717e-09, + "loss": 3.1429, + "step": 6701 + }, + { + "epoch": 9.81, + "grad_norm": 0.23429007828235626, + "learning_rate": 8.623613503988548e-09, + "loss": 3.1318, + "step": 6702 + }, + { + "epoch": 9.81, + "grad_norm": 0.3295222818851471, + "learning_rate": 8.018648703111975e-09, + "loss": 3.0962, + "step": 6703 + }, + { + "epoch": 9.82, + "grad_norm": 0.2013850212097168, + "learning_rate": 7.435681468015432e-09, + "loss": 3.0993, + "step": 6704 + }, + { + "epoch": 9.82, + "grad_norm": 0.3494795262813568, + "learning_rate": 6.87471192694078e-09, + "loss": 3.1368, + "step": 6705 + }, + { + "epoch": 9.82, + "grad_norm": 0.3240787386894226, + "learning_rate": 6.335740203311513e-09, + "loss": 3.1125, + "step": 6706 + }, + { + "epoch": 9.82, + "grad_norm": 0.2250877022743225, + "learning_rate": 5.81876641569945e-09, + "loss": 3.1517, + "step": 6707 + }, + { + "epoch": 9.82, + "grad_norm": 0.25935402512550354, + "learning_rate": 5.3237906778136334e-09, + "loss": 3.1106, + "step": 6708 + }, + { + "epoch": 9.82, + "grad_norm": 0.3084031641483307, + "learning_rate": 4.850813098566942e-09, + "loss": 3.1059, + "step": 6709 + }, + { + "epoch": 9.82, + "grad_norm": 0.22343696653842926, + "learning_rate": 4.399833781998375e-09, + "loss": 3.115, + "step": 6710 + }, + { + "epoch": 9.83, + "grad_norm": 0.6314164996147156, + "learning_rate": 3.970852827339666e-09, + "loss": 3.1202, + "step": 6711 + }, + { + "epoch": 9.83, + "grad_norm": 0.31552714109420776, + "learning_rate": 3.5638703289375684e-09, + "loss": 3.1364, + "step": 6712 + }, + { + "epoch": 9.83, + "grad_norm": 0.2638113498687744, + "learning_rate": 3.178886376331569e-09, + "loss": 3.1112, + "step": 6713 + }, + { + "epoch": 9.83, + "grad_norm": 0.24450135231018066, + "learning_rate": 2.8159010542316845e-09, + "loss": 3.1224, + "step": 6714 + }, + { + "epoch": 9.83, + "grad_norm": 0.25562459230422974, + "learning_rate": 2.4749144424851546e-09, + "loss": 3.1271, + "step": 6715 + }, + { + "epoch": 9.83, + "grad_norm": 0.3367336094379425, + "learning_rate": 2.1559266160986467e-09, + "loss": 3.1251, + "step": 6716 + }, + { + "epoch": 9.83, + "grad_norm": 0.23420773446559906, + "learning_rate": 1.858937645249359e-09, + "loss": 3.1274, + "step": 6717 + }, + { + "epoch": 9.84, + "grad_norm": 0.20982877910137177, + "learning_rate": 1.5839475952739157e-09, + "loss": 3.1819, + "step": 6718 + }, + { + "epoch": 9.84, + "grad_norm": 0.2672583758831024, + "learning_rate": 1.3309565266794722e-09, + "loss": 3.1235, + "step": 6719 + }, + { + "epoch": 9.84, + "grad_norm": 22.543325424194336, + "learning_rate": 1.099964495110406e-09, + "loss": 3.1494, + "step": 6720 + }, + { + "epoch": 9.84, + "grad_norm": 0.3094947338104248, + "learning_rate": 8.909715513816253e-10, + "loss": 3.0934, + "step": 6721 + }, + { + "epoch": 9.84, + "grad_norm": 0.23040957748889923, + "learning_rate": 7.039777414785675e-10, + "loss": 3.1071, + "step": 6722 + }, + { + "epoch": 9.84, + "grad_norm": 0.25470995903015137, + "learning_rate": 5.38983106534996e-10, + "loss": 3.1336, + "step": 6723 + }, + { + "epoch": 9.85, + "grad_norm": 0.27829986810684204, + "learning_rate": 3.9598768285520336e-10, + "loss": 3.1287, + "step": 6724 + }, + { + "epoch": 9.85, + "grad_norm": 0.5278626084327698, + "learning_rate": 2.7499150188070587e-10, + "loss": 3.1516, + "step": 6725 + }, + { + "epoch": 9.85, + "grad_norm": 0.23420289158821106, + "learning_rate": 1.7599459024575381e-10, + "loss": 3.1289, + "step": 6726 + }, + { + "epoch": 9.85, + "grad_norm": 0.21472777426242828, + "learning_rate": 9.899696971071848e-11, + "loss": 3.1088, + "step": 6727 + }, + { + "epoch": 9.85, + "grad_norm": 0.2794855237007141, + "learning_rate": 4.399865723980767e-11, + "loss": 3.165, + "step": 6728 + }, + { + "epoch": 9.85, + "grad_norm": 1.3788336515426636, + "learning_rate": 1.099966491224791e-11, + "loss": 3.1195, + "step": 6729 + }, + { + "epoch": 9.85, + "grad_norm": 0.28580406308174133, + "learning_rate": 0.0, + "loss": 3.0934, + "step": 6730 + } + ], + "logging_steps": 1, + "max_steps": 6730, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "save_steps": 673, + "total_flos": 1.662169839530777e+20, + "train_batch_size": 2, + "trial_name": null, + "trial_params": null +} diff --git a/checkpoint-6730/training_args.bin b/checkpoint-6730/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..c9318904b092ceb1505e2e8f2b5300dd991baa7f --- /dev/null +++ b/checkpoint-6730/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a4ab9da40c5b7630b88beccc23069550f831ccb9b69b82772516407dd51150c +size 5688 diff --git a/config.json b/config.json new file mode 100644 index 0000000000000000000000000000000000000000..6b4dd143e04f5808fbc1160726a8e245b4531254 --- /dev/null +++ b/config.json @@ -0,0 +1,44 @@ +{ + "_name_or_path": "unsloth/gemma-7b", + "architectures": [ + "GemmaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 2, + "eos_token_id": 1, + "head_dim": 256, + "hidden_act": "gelu", + "hidden_activation": null, + "hidden_size": 3072, + "initializer_range": 0.02, + "intermediate_size": 24576, + "max_position_embeddings": 8192, + "model_type": "gemma", + "num_attention_heads": 16, + "num_hidden_layers": 28, + "num_key_value_heads": 16, + "pad_token_id": 0, + "quantization_config": { + "_load_in_4bit": true, + "_load_in_8bit": false, + "bnb_4bit_compute_dtype": "bfloat16", + "bnb_4bit_quant_storage": "bfloat16", + "bnb_4bit_quant_type": "nf4", + "bnb_4bit_use_double_quant": true, + "llm_int8_enable_fp32_cpu_offload": false, + "llm_int8_has_fp16_weight": false, + "llm_int8_skip_modules": null, + "llm_int8_threshold": 6.0, + "load_in_4bit": true, + "load_in_8bit": false, + "quant_method": "bitsandbytes" + }, + "rms_norm_eps": 1e-06, + "rope_scaling": null, + "rope_theta": 10000.0, + "torch_dtype": "bfloat16", + "transformers_version": "4.40.0.dev0", + "use_cache": false, + "vocab_size": 256000 +} diff --git a/special_tokens_map.json b/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80 --- /dev/null +++ b/special_tokens_map.json @@ -0,0 +1,34 @@ +{ + "additional_special_tokens": [ + "", + "" + ], + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/tokenizer.json b/tokenizer.json new file mode 100644 index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044 --- /dev/null +++ b/tokenizer.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee +size 17477929 diff --git a/tokenizer.model b/tokenizer.model new file mode 100644 index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8 --- /dev/null +++ b/tokenizer.model @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2 +size 4241003 diff --git a/tokenizer_config.json b/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..1451ca400a880c5056ab40cb960930cf827de2de --- /dev/null +++ b/tokenizer_config.json @@ -0,0 +1,70 @@ +{ + "add_bos_token": true, + "add_eos_token": false, + "added_tokens_decoder": { + "0": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "1": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "2": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "3": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "106": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "107": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "additional_special_tokens": [ + "", + "" + ], + "bos_token": "", + "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '' + role + '\n' + message['content'] | trim + '\n' }}{% endfor %}{% if add_generation_prompt %}{{'model\n'}}{% endif %}", + "clean_up_tokenization_spaces": false, + "eos_token": "", + "legacy": null, + "model_max_length": 1000000000000000019884624838656, + "pad_token": "", + "sp_model_kwargs": {}, + "spaces_between_special_tokens": false, + "tokenizer_class": "GemmaTokenizer", + "unk_token": "", + "use_default_system_prompt": false +} diff --git a/training_args.bin b/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..c9318904b092ceb1505e2e8f2b5300dd991baa7f --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a4ab9da40c5b7630b88beccc23069550f831ccb9b69b82772516407dd51150c +size 5688