End of training

README.md

@@ -21,17 +21,13 @@ axolotl version: `0.4.1`
 adapter: lora
 base_model: unsloth/Meta-Llama-3.1-8B
 bf16: auto
-bnb_config_kwargs:
-  bnb_4bit_quant_type: nf4
-  bnb_4bit_use_double_quant: true
 chat_template: llama3
-cosine_min_lr_ratio: 0.1
-data_processes: 16
 dataset_prepared_path: null
 datasets:
 - data_files:
   - 296bd32b0a7d0eae_train_data.json
   ds_type: json
+  format: custom
   path: /workspace/input_data/296bd32b0a7d0eae_train_data.json
   type:
     field_input: level
@@ -41,79 +37,60 @@ datasets:
     system_prompt: ''
 debug: null
 deepspeed: null
-device_map: '{'''':torch.cuda.current_device()}'
-do_eval: true
-early_stopping_patience: 1
-eval_batch_size: 1
-eval_sample_packing: false
-eval_steps: 25
-evaluation_strategy: steps
-flash_attention: false
+early_stopping_patience: null
+eval_max_new_tokens: 128
+eval_table_size: null
+evals_per_epoch: 4
+flash_attention: true
 fp16: null
 fsdp: null
 fsdp_config: null
-gradient_accumulation_steps: 32
-gradient_checkpointing: true
-group_by_length: true
+gradient_accumulation_steps: 4
+gradient_checkpointing: false
+group_by_length: false
 hub_model_id: cwaud/0f6de495-2b3e-4109-828d-b91842a9e39d
 hub_repo: cwaud
 hub_strategy: checkpoint
 hub_token: null
-learning_rate: 0.0001
+learning_rate: 0.002
 load_in_4bit: false
-load_in_8bit: false
+load_in_8bit: true
 local_rank: null
 logging_steps: 1
-lora_alpha: 64
+lora_alpha: 16
 lora_dropout: 0.05
 lora_fan_in_fan_out: null
 lora_model_dir: null
-lora_r: 32
+lora_r: 8
 lora_target_linear: true
-lora_target_modules:
-- q_proj
-- v_proj
 lr_scheduler: cosine
-max_grad_norm: 1.0
-max_memory:
-  0: 70GiB
-  1: 70GiB
-  2: 70GiB
-  3: 70GiB
-max_steps: 83
+max_steps: 100
 micro_batch_size: 1
 mlflow_experiment_name: /tmp/296bd32b0a7d0eae_train_data.json
 model_type: AutoModelForCausalLM
-num_epochs: 3
-optim_args:
-  adam_beta1: 0.9
-  adam_beta2: 0.95
-  adam_epsilon: 1e-5
-optimizer: adamw_torch
+num_epochs: 1
+optimizer: adamw_bnb_8bit
 output_dir: miner_id_24
 pad_to_sequence_len: true
 resume_from_checkpoint: null
 s2_attention: null
 sample_packing: false
-save_steps: 50
+save_steps: 5
 save_strategy: steps
 sequence_len: 2048
 strict: false
 tf32: false
 tokenizer_type: AutoTokenizer
-torch_compile: false
 train_on_inputs: false
-trust_remote_code: true
-val_set_size: 50
+val_set_size: 0.05
 wandb_entity: rayonlabs-rayon-labs
 wandb_mode: online
 wandb_name: 0f6de495-2b3e-4109-828d-b91842a9e39d
 wandb_project: Public_TuningSN
 wandb_run: miner_id_24
 wandb_runid: 0f6de495-2b3e-4109-828d-b91842a9e39d
-warmup_raio: 0.03
-warmup_ratio: 0.04
-weight_decay: 0.01
+warmup_steps: 10
+weight_decay: 0.0
 xformers_attention: null
 
 ```
@@ -124,7 +101,7 @@ xformers_attention: null
 
 This model is a fine-tuned version of [unsloth/Meta-Llama-3.1-8B](https://huggingface.co/unsloth/Meta-Llama-3.1-8B) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.6324
+- Loss: 0.8763
 
 ## Model description
 
@@ -143,34 +120,32 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 0.0001
+- learning_rate: 0.002
 - train_batch_size: 1
 - eval_batch_size: 1
 - seed: 42
-- distributed_type: multi-GPU
-- num_devices: 4
-- gradient_accumulation_steps: 32
-- total_train_batch_size: 128
-- total_eval_batch_size: 4
-- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 4
+- optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
 - lr_scheduler_type: cosine
-- lr_scheduler_warmup_steps: 3
-- training_steps: 83
+- lr_scheduler_warmup_steps: 10
+- training_steps: 100
 
 ### Training results
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
-| 0.7046        | 0.0002 | 1    | 0.9375          |
-| 0.6112        | 0.0043 | 25   | 0.6832          |
-| 0.6283        | 0.0085 | 50   | 0.6453          |
-| 0.5699        | 0.0128 | 75   | 0.6324          |
+| 1.0718        | 0.0000 | 1    | 1.0175          |
+| 1.1737        | 0.0001 | 25   | 1.0244          |
+| 1.1765        | 0.0003 | 50   | 1.0543          |
+| 0.9795        | 0.0004 | 75   | 0.9441          |
+| 0.56          | 0.0006 | 100  | 0.8763          |
 
 
 ### Framework versions
 
 - PEFT 0.13.2
-- Transformers 4.45.2
-- Pytorch 2.4.1+cu124
+- Transformers 4.46.0
+- Pytorch 2.5.0+cu124
 - Datasets 3.0.1
 - Tokenizers 0.20.1

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8e588185b00e7e24e3ed2dccded1dccbd19b3679def86b1dd3efc2992586c9dc
-size 335706186
+oid sha256:ab68b37f5bace056be3f4f6814a6d913b009dfe9afc630df8d111a4a7752b76e
+size 84047370