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/a2552944cccf89dd_train_data.json
 model_type: AutoModelForCausalLM
@@ -91,7 +91,7 @@ wandb_name: 730b1c45-901f-480e-bbc0-cbfe3241a83a
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 730b1c45-901f-480e-bbc0-cbfe3241a83a
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -102,6 +102,8 @@ xformers_attention: null
 # 730b1c45-901f-480e-bbc0-cbfe3241a83a
 
 This model is a fine-tuned version of [unsloth/Llama-3.2-3B](https://huggingface.co/unsloth/Llama-3.2-3B) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 3.7720
 
 ## 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        | 0.0010 | 1    | 10.3571         |
+| 9.7309        | 0.0105 | 10   | 9.7032          |
+| 6.6986        | 0.0210 | 20   | 5.8235          |
+| 4.4811        | 0.0314 | 30   | 4.2357          |
+| 4.1107        | 0.0419 | 40   | 3.8276          |
+| 3.7016        | 0.0524 | 50   | 3.7720          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dd230f85b8f4b108eb2f26ffb2f2d95a782b2f0278803301fa354dc575cef746
+oid sha256:8d585b14c5f3833940c8b96a6908f880764411374ec9be54c75d0d76cd2536fc
 size 48768810