End of training

README.md

@@ -65,7 +65,7 @@ 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/031a4fca4dcf34bb_train_data.json
 model_type: AutoModelForCausalLM
@@ -90,7 +90,7 @@ wandb_name: a89e4a69-7134-4134-bd15-bd99406a233d
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: a89e4a69-7134-4134-bd15-bd99406a233d
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -101,6 +101,8 @@ xformers_attention: null
 # a89e4a69-7134-4134-bd15-bd99406a233d
 
 This model is a fine-tuned version of [zake7749/gemma-2-2b-it-chinese-kyara-dpo](https://huggingface.co/zake7749/gemma-2-2b-it-chinese-kyara-dpo) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 2.3137
 
 ## Model description
 
@@ -127,14 +129,19 @@ 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    | 3.6646          |
+| 3.3818        | 0.0005 | 10   | 3.0478          |
+| 2.4455        | 0.0009 | 20   | 2.4915          |
+| 2.4572        | 0.0014 | 30   | 2.3601          |
+| 2.3016        | 0.0019 | 40   | 2.3203          |
+| 2.2997        | 0.0024 | 50   | 2.3137          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ba49fdd4929180720d023b9d799fe7ae6fb01212688277fe778f075ff4dd66dc
+oid sha256:58bb7c4da529369d8fac93cc4aa1ae49b5acb379ec14cd03976183045edf4be6
 size 41664442