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/1252351227839360_train_data.json
 model_type: AutoModelForCausalLM
@@ -92,7 +92,7 @@ wandb_name: 6b68d715-277d-410f-8147-0e3954c97b2f
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 6b68d715-277d-410f-8147-0e3954c97b2f
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -103,6 +103,8 @@ xformers_attention: null
 # 0c6e6124-63ff-4a92-9d11-e0d594f5fc01
 
 This model is a fine-tuned version of [EleutherAI/pythia-410m-deduped](https://huggingface.co/EleutherAI/pythia-410m-deduped) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 3.6407
 
 ## 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        | 0.0002 | 1    | 4.1083          |
+| 17.8283       | 0.0017 | 10   | 4.0637          |
+| 16.7979       | 0.0034 | 20   | 3.8801          |
+| 14.0956       | 0.0051 | 30   | 3.7266          |
+| 15.509        | 0.0069 | 40   | 3.6544          |
+| 15.2501       | 0.0086 | 50   | 3.6407          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:92edb863f1f1434057a77ab794773ac8024c2688bde24bccb86c769cf0762f2d
+oid sha256:9a1b8006cc750884d9614057d878c454b10c1f2c6cd80906a412245b61a7234a
 size 12653258