File size: 4,385 Bytes
82a5799 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 |
---
library_name: peft
tags:
- trl
- dpo
- generated_from_trainer
base_model: NanQiangHF/llama3_8b_instruct_bwgenerator
model-index:
- name: llama3_8b_instruct_dpo_bwgenerator
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. -->
# llama3_8b_instruct_dpo_bwgenerator
This model is a fine-tuned version of [NanQiangHF/llama3_8b_instruct_bwgenerator](https://huggingface.co/NanQiangHF/llama3_8b_instruct_bwgenerator) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0706
- Rewards/chosen: -4.6241
- Rewards/rejected: -14.8342
- Rewards/accuracies: 0.9780
- Rewards/margins: 10.2101
- Logps/rejected: -216.1456
- Logps/chosen: -84.8191
- Logits/rejected: 0.9202
- Logits/chosen: 0.3552
## 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-06
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- 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 |
|:-------------:|:------:|:-----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
| 0.247 | 0.0719 | 1000 | 0.0906 | -3.7216 | -11.8877 | 0.9686 | 8.1662 | -186.6814 | -75.7941 | 0.8504 | 0.3080 |
| 0.083 | 0.1438 | 2000 | 0.0775 | -4.5564 | -14.1375 | 0.9764 | 9.5811 | -209.1791 | -84.1423 | 0.8989 | 0.3418 |
| 0.0623 | 0.2157 | 3000 | 0.0734 | -4.5379 | -14.4993 | 0.9770 | 9.9614 | -212.7973 | -83.9572 | 0.9082 | 0.3471 |
| 0.069 | 0.2876 | 4000 | 0.0713 | -4.5601 | -14.6450 | 0.9777 | 10.0850 | -214.2546 | -84.1790 | 0.9145 | 0.3514 |
| 0.0752 | 0.3595 | 5000 | 0.0706 | -4.4918 | -14.6244 | 0.9793 | 10.1326 | -214.0477 | -83.4960 | 0.9181 | 0.3533 |
| 0.0723 | 0.4313 | 6000 | 0.0710 | -4.6381 | -14.8167 | 0.9780 | 10.1787 | -215.9714 | -84.9590 | 0.9187 | 0.3542 |
| 0.0852 | 0.5032 | 7000 | 0.0705 | -4.6251 | -14.8143 | 0.9783 | 10.1893 | -215.9474 | -84.8290 | 0.9189 | 0.3542 |
| 0.0811 | 0.5751 | 8000 | 0.0706 | -4.6409 | -14.8406 | 0.9780 | 10.1997 | -216.2102 | -84.9870 | 0.9185 | 0.3538 |
| 0.0762 | 0.6470 | 9000 | 0.0699 | -4.6161 | -14.8083 | 0.9790 | 10.1921 | -215.8869 | -84.7398 | 0.9186 | 0.3541 |
| 0.0686 | 0.7189 | 10000 | 0.0703 | -4.6164 | -14.8042 | 0.9790 | 10.1878 | -215.8462 | -84.7421 | 0.9185 | 0.3537 |
| 0.061 | 0.7908 | 11000 | 0.0705 | -4.6191 | -14.8169 | 0.9793 | 10.1977 | -215.9726 | -84.7695 | 0.9207 | 0.3556 |
| 0.0786 | 0.8627 | 12000 | 0.0698 | -4.6080 | -14.7978 | 0.9793 | 10.1898 | -215.7822 | -84.6584 | 0.9195 | 0.3546 |
| 0.073 | 0.9346 | 13000 | 0.0706 | -4.6241 | -14.8342 | 0.9780 | 10.2101 | -216.1456 | -84.8191 | 0.9202 | 0.3552 |
### Framework versions
- PEFT 0.10.0
- Transformers 4.44.0
- Pytorch 2.3.0+cu121
- Datasets 2.14.7
- Tokenizers 0.19.1 |