|
--- |
|
tags: |
|
- trl |
|
- dpo |
|
- generated_from_trainer |
|
model-index: |
|
- name: zephyr-7b-dpo-full |
|
results: [] |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# zephyr-7b-dpo-full |
|
|
|
This model was trained from scratch on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.0224 |
|
- Rewards/chosen: -1.9945 |
|
- Rewards/rejected: -3.2919 |
|
- Rewards/accuracies: 0.7148 |
|
- Rewards/margins: 1.2974 |
|
- Logps/rejected: -640.8138 |
|
- Logps/chosen: -503.0325 |
|
- Logits/rejected: 0.3215 |
|
- Logits/chosen: 0.2841 |
|
|
|
## 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: 1e-06 |
|
- train_batch_size: 4 |
|
- eval_batch_size: 8 |
|
- seed: 4 |
|
- distributed_type: multi-GPU |
|
- num_devices: 8 |
|
- gradient_accumulation_steps: 4 |
|
- 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: 2 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen | |
|
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:| |
|
| 0.111 | 0.21 | 100 | 0.1080 | -0.3300 | -0.6434 | 0.7148 | 0.3134 | -375.9606 | -336.5851 | 0.4520 | 0.3976 | |
|
| 0.0697 | 0.42 | 200 | 0.0728 | -0.5844 | -1.2213 | 0.7422 | 0.6369 | -433.7567 | -362.0242 | 0.4101 | 0.3267 | |
|
| 0.055 | 0.63 | 300 | 0.0610 | -0.7945 | -1.5421 | 0.7266 | 0.7476 | -465.8376 | -383.0369 | 0.2780 | 0.2451 | |
|
| 0.0573 | 0.84 | 400 | 0.0566 | -0.8305 | -1.5952 | 0.7383 | 0.7647 | -471.1477 | -386.6394 | 0.2561 | 0.2348 | |
|
| 0.0215 | 1.05 | 500 | 0.0327 | -1.6150 | -2.8668 | 0.7305 | 1.2517 | -598.3008 | -465.0880 | 0.2419 | 0.2221 | |
|
| 0.0139 | 1.26 | 600 | 0.0260 | -1.8080 | -3.0895 | 0.7227 | 1.2815 | -620.5768 | -484.3871 | 0.2916 | 0.2601 | |
|
| 0.0125 | 1.47 | 700 | 0.0247 | -1.9121 | -3.1886 | 0.7305 | 1.2765 | -630.4850 | -494.7950 | 0.2947 | 0.2614 | |
|
| 0.0107 | 1.67 | 800 | 0.0226 | -1.9947 | -3.2951 | 0.7188 | 1.3004 | -641.1344 | -503.0576 | 0.3196 | 0.2841 | |
|
| 0.0106 | 1.88 | 900 | 0.0224 | -1.9945 | -3.2919 | 0.7148 | 1.2974 | -640.8138 | -503.0325 | 0.3215 | 0.2841 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.35.2 |
|
- Pytorch 2.1.2+cu121 |
|
- Datasets 2.14.6 |
|
- Tokenizers 0.14.1 |
|
|