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/4e62550040a6348e_train_data.json
 model_type: AutoModelForCausalLM
@@ -93,7 +93,7 @@ wandb_name: 0404c263-0247-42d3-9642-6e0ab1dc801f
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 0404c263-0247-42d3-9642-6e0ab1dc801f
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -104,6 +104,8 @@ xformers_attention: null
 # 0404c263-0247-42d3-9642-6e0ab1dc801f
 
 This model is a fine-tuned version of [elyza/Llama-3-ELYZA-JP-8B](https://huggingface.co/elyza/Llama-3-ELYZA-JP-8B) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.7538
 
 ## 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        | 0.0000 | 1    | 1.1074          |
+| 0.9234        | 0.0003 | 10   | 0.9549          |
+| 0.8243        | 0.0007 | 20   | 0.8327          |
+| 0.7362        | 0.0010 | 30   | 0.7774          |
+| 0.7301        | 0.0013 | 40   | 0.7576          |
+| 0.7898        | 0.0017 | 50   | 0.7538          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:36d04c857ce9d4c712c49edb074eb495d47204264df3e3035b0485c14bd2d2a3
+oid sha256:1114295aefc8ac9a0ec126972c3f4060354c3f4dd8654663ef7e2c7ef299115a
 size 84047370