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: 0.1422
- Rewards/chosen: -1.3154
- Rewards/rejected: -2.2768
- Rewards/accuracies: 0.7617
- Rewards/margins: 0.9613
- Logps/rejected: -483.9327
- Logps/chosen: -386.7366
- Logits/rejected: -2.1695
- Logits/chosen: -2.2036
- Debug/policy Weights: 0.2815
- Debug/losses: 0.1397
- Debug/raw Losses: 0.4727
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: 42
- 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 | Debug/policy Weights | Debug/losses | Debug/raw Losses |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.1781 | 0.21 | 100 | 0.2007 | -0.6478 | -1.1693 | 0.7344 | 0.5214 | -373.1867 | -319.9806 | -2.6910 | -2.7080 | 0.3512 | 0.1953 | 0.5590 |
| 0.1616 | 0.42 | 200 | 0.1669 | -0.8830 | -1.6003 | 0.7109 | 0.7173 | -416.2844 | -343.4914 | -2.4277 | -2.4499 | 0.3174 | 0.1671 | 0.5079 |
| 0.1343 | 0.63 | 300 | 0.1368 | -1.5021 | -2.3715 | 0.7578 | 0.8695 | -493.4114 | -405.4042 | -2.2283 | -2.2618 | 0.2666 | 0.1365 | 0.4953 |
| 0.1398 | 0.84 | 400 | 0.1422 | -1.3154 | -2.2768 | 0.7617 | 0.9613 | -483.9327 | -386.7366 | -2.1695 | -2.2036 | 0.2815 | 0.1397 | 0.4727 |
Framework versions
- Transformers 4.39.3
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.15.2