End of training

README.md

@@ -64,7 +64,7 @@ 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/6d3f496dc688ac45_train_data.json
 model_type: AutoModelForCausalLM
@@ -91,7 +91,7 @@ wandb_name: 6bbb4b65-663e-464a-acb8-c37d9beb4345
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 6bbb4b65-663e-464a-acb8-c37d9beb4345
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -102,6 +102,8 @@ xformers_attention: null
 # ff14582a-6df7-4cb2-b4ba-8a949bed0113
 
 This model is a fine-tuned version of [Korabbit/llama-2-ko-7b](https://huggingface.co/Korabbit/llama-2-ko-7b) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 1.2167
 
 ## 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        | 0.0003 | 1    | 1.7010          |
+| 1.0425        | 0.0034 | 10   | 1.6531          |
+| 0.9458        | 0.0068 | 20   | 1.4235          |
+| 0.878         | 0.0102 | 30   | 1.3055          |
+| 0.8977        | 0.0136 | 40   | 1.2320          |
+| 0.8506        | 0.0170 | 50   | 1.2167          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bf053a4dcbca1ead1ef52a4020ea0adf6e3e8e6174f37299e5cf468509f58fec
+oid sha256:0163b830e8119ccb9cd5e9c820f9e98d93b71da340deceaa9235fb664b5b171f
 size 80115210