End of training

README.md

@@ -65,7 +65,7 @@ lora_model_dir: null
 lora_r: 8
 lora_target_linear: true
 lr_scheduler: cosine
-max_steps: 1
+max_steps: 50
 micro_batch_size: 2
 mlflow_experiment_name: /tmp/6d8716b2dcd5dd55_train_data.json
 model_type: AutoModelForCausalLM
@@ -92,7 +92,7 @@ wandb_name: c7328a14-c445-42ae-9cc1-f30fbcce085b
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: c7328a14-c445-42ae-9cc1-f30fbcce085b
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -103,6 +103,8 @@ xformers_attention: null
 # d24684c5-8d5a-adf4-0ba0-57b762f59d7b
 
 This model is a fine-tuned version of [EleutherAI/pythia-70m-deduped](https://huggingface.co/EleutherAI/pythia-70m-deduped) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 22.6790
 
 ## Model description
 
@@ -129,14 +131,19 @@ The following hyperparameters were used during training:
 - total_train_batch_size: 8
 - 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: 2
-- training_steps: 1
+- lr_scheduler_warmup_steps: 10
+- training_steps: 50
 
 ### Training results
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
 | No log        | 0.0004 | 1    | 23.2700         |
+| 92.1238       | 0.0038 | 10   | 23.2460         |
+| 82.2434       | 0.0076 | 20   | 23.0463         |
+| 115.9972      | 0.0114 | 30   | 22.8223         |
+| 89.6675       | 0.0152 | 40   | 22.6928         |
+| 75.1007       | 0.0190 | 50   | 22.6790         |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9ec0f619a4ec48a5844c5d804abdd472da8cd2db92465cb1d1f84ce6ca7d158e
+oid sha256:d9b75302d3ab0df93610d4a845e3e2fe109998c6665d5185bb0f19472ac94e4b
 size 1590526