model_shp3_dpo1
This model is a fine-tuned version of meta-llama/Llama-2-7b-chat-hf on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 2.0149
- Rewards/chosen: -10.1791
- Rewards/rejected: -11.7765
- Rewards/accuracies: 0.5700
- Rewards/margins: 1.5974
- Logps/rejected: -384.5022
- Logps/chosen: -344.3792
- Logits/rejected: -0.9104
- Logits/chosen: -0.9161
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0005
- train_batch_size: 4
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 100
- training_steps: 1000
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.0698 | 2.67 | 100 | 1.0791 | -4.0199 | -4.9604 | 0.5900 | 0.9405 | -316.3408 | -282.7867 | -0.9293 | -0.9672 |
| 0.0004 | 5.33 | 200 | 1.5283 | -7.2654 | -8.3476 | 0.5800 | 1.0822 | -350.2130 | -315.2426 | -0.9409 | -0.9510 |
| 0.0001 | 8.0 | 300 | 1.9265 | -9.7069 | -11.2491 | 0.5700 | 1.5421 | -379.2276 | -339.6573 | -0.9224 | -0.9298 |
| 0.0001 | 10.67 | 400 | 1.9667 | -9.9054 | -11.4492 | 0.5700 | 1.5439 | -381.2295 | -341.6420 | -0.9181 | -0.9248 |
| 0.0001 | 13.33 | 500 | 1.9959 | -10.0523 | -11.6165 | 0.5700 | 1.5642 | -382.9025 | -343.1115 | -0.9137 | -0.9198 |
| 0.0 | 16.0 | 600 | 2.0035 | -10.1182 | -11.7116 | 0.5700 | 1.5934 | -383.8533 | -343.7699 | -0.9121 | -0.9185 |
| 0.0001 | 18.67 | 700 | 2.0159 | -10.1627 | -11.7547 | 0.5700 | 1.5920 | -384.2843 | -344.2155 | -0.9115 | -0.9169 |
| 0.0 | 21.33 | 800 | 2.0163 | -10.1740 | -11.7677 | 0.5700 | 1.5937 | -384.4142 | -344.3281 | -0.9103 | -0.9160 |
| 0.0 | 24.0 | 900 | 2.0220 | -10.1842 | -11.7817 | 0.5700 | 1.5976 | -384.5541 | -344.4297 | -0.9106 | -0.9160 |
| 0.0001 | 26.67 | 1000 | 2.0149 | -10.1791 | -11.7765 | 0.5700 | 1.5974 | -384.5022 | -344.3792 | -0.9104 | -0.9161 |
Framework versions
- PEFT 0.10.0
- Transformers 4.39.1
- Pytorch 2.2.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2
- Downloads last month
- 4
Model tree for guoyu-zhang/model_shp3_dpo1
Base model
meta-llama/Llama-2-7b-chat-hf