metadata
license: mit
base_model: HuggingFaceH4/mistral-7b-sft-beta
tags:
- trl
- dpo
- generated_from_trainer
model-index:
- name: zephyr-7b-dpo-full
results: []
zephyr-7b-dpo-full
This model is a fine-tuned version of HuggingFaceH4/mistral-7b-sft-beta on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.4920
- Rewards/chosen: -2.3074
- Rewards/rejected: -3.5196
- Rewards/accuracies: 0.7734
- Rewards/margins: 1.2122
- Logps/rejected: -609.3139
- Logps/chosen: -487.7755
- Logits/rejected: -0.7242
- Logits/chosen: -0.9597
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: 5e-07
- train_batch_size: 8
- eval_batch_size: 8
- seed: 2
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 2
- total_train_batch_size: 128
- total_eval_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 1
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.5392 | 0.11 | 100 | 0.6286 | -0.6554 | -0.9418 | 0.6523 | 0.2865 | -351.5352 | -322.5750 | -2.5756 | -2.5908 |
| 0.4524 | 0.23 | 200 | 0.5475 | -1.4831 | -2.1698 | 0.7227 | 0.6867 | -474.3327 | -405.3454 | -1.9678 | -1.9878 |
| 0.3976 | 0.34 | 300 | 0.5194 | -1.8541 | -2.8790 | 0.7617 | 1.0249 | -545.2501 | -442.4474 | -0.9783 | -1.1841 |
| 0.3892 | 0.45 | 400 | 0.5160 | -2.0795 | -3.1766 | 0.7773 | 1.0971 | -575.0087 | -464.9888 | -0.6002 | -0.8579 |
| 0.3964 | 0.57 | 500 | 0.4992 | -2.1896 | -3.3081 | 0.7656 | 1.1185 | -588.1666 | -476.0038 | -0.8012 | -1.0189 |
| 0.4149 | 0.68 | 600 | 0.4948 | -2.2061 | -3.3241 | 0.7461 | 1.1179 | -589.7601 | -477.6525 | -1.0527 | -1.2398 |
| 0.4004 | 0.79 | 700 | 0.4905 | -2.1723 | -3.3652 | 0.7695 | 1.1929 | -593.8731 | -474.2662 | -0.8519 | -1.0643 |
| 0.3887 | 0.91 | 800 | 0.4920 | -2.3074 | -3.5196 | 0.7734 | 1.2122 | -609.3139 | -487.7755 | -0.7242 | -0.9597 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.14.1