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/3f2dea70a689bc2e_train_data.json
 model_type: AutoModelForCausalLM
@@ -91,7 +91,7 @@ wandb_name: 2650387b-5b9e-476b-85c6-ce62d4bde2e2
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 2650387b-5b9e-476b-85c6-ce62d4bde2e2
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -102,6 +102,8 @@ xformers_attention: null
 # afbdf303-5ec3-cb90-b17b-283c6f8ad70b
 
 This model is a fine-tuned version of [unsloth/Meta-Llama-3.1-8B-Instruct](https://huggingface.co/unsloth/Meta-Llama-3.1-8B-Instruct) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.8773
 
 ## Model description
 
@@ -128,14 +130,19 @@ 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        | 0.0187 | 1    | 2.0967          |
+| 1.9689        | 0.1869 | 10   | 1.5523          |
+| 1.1468        | 0.3738 | 20   | 1.0094          |
+| 1.0525        | 0.5607 | 30   | 0.9176          |
+| 0.9961        | 0.7477 | 40   | 0.8836          |
+| 0.9734        | 0.9346 | 50   | 0.8773          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0b01499a61144777f68f767ed41af1e0c7e0c42862ba9963fce50c7de8cfa430
+oid sha256:1d6f28d0930d058794559f4667576561ac8390fcf3b5b04dd8a9c195534b3748
 size 84047370