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/bb36bfc83b60b0b9_train_data.json
 model_type: AutoModelForCausalLM
@@ -89,7 +89,7 @@ wandb_name: 72a97be3-7171-471e-9c44-63e52b97f46d
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 72a97be3-7171-471e-9c44-63e52b97f46d
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -100,6 +100,8 @@ xformers_attention: null
 # e2089e4e-fb41-44ec-a355-ddffc329e1ab
 
 This model is a fine-tuned version of [oopsung/llama2-7b-koNqa-test-v1](https://huggingface.co/oopsung/llama2-7b-koNqa-test-v1) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 1.1734
 
 ## Model description
 
@@ -126,14 +128,19 @@ 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    | 1.4842          |
+| 1.3895        | 0.0004 | 10   | 1.4606          |
+| 1.3014        | 0.0009 | 20   | 1.3219          |
+| 1.2074        | 0.0013 | 30   | 1.2257          |
+| 1.1935        | 0.0017 | 40   | 1.1782          |
+| 1.1538        | 0.0022 | 50   | 1.1734          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5da177cc374704235eceb20449269e12e09cc28310d08c36b993d6fa8599d876
+oid sha256:21ef5d049812a9cc6f2e28b47a189319b9b44f9880bff6ddca95ee0090692bd7
 size 80115210