End of training

README.md

@@ -66,7 +66,7 @@ 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/afd6e39c0f484e52_train_data.json
 model_type: AutoModelForCausalLM
@@ -93,7 +93,7 @@ wandb_name: 0625f4f2-3986-4a49-bfde-648574c5d967
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 0625f4f2-3986-4a49-bfde-648574c5d967
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -104,6 +104,8 @@ xformers_attention: null
 # 0625f4f2-3986-4a49-bfde-648574c5d967
 
 This model is a fine-tuned version of [NousResearch/Llama-3.2-1B](https://huggingface.co/NousResearch/Llama-3.2-1B) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 1.1971
 
 ## Model description
 
@@ -130,14 +132,19 @@ 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        | 0.0047 | 1    | 2.0055          |
+| 2.0283        | 0.0468 | 10   | 1.9290          |
+| 1.7649        | 0.0936 | 20   | 1.6194          |
+| 1.4338        | 0.1404 | 30   | 1.3618          |
+| 1.2174        | 0.1871 | 40   | 1.2221          |
+| 1.2708        | 0.2339 | 50   | 1.1971          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5beb6a3edf3def2eb835433b818f675c090163813dc1bcbda5be5b4bdcd8d55f
+oid sha256:9a47aa9d200d992b6a82d2e5f055d07abd2152468caa3f02851f337bce2b77b3
 size 22624970