|
--- |
|
license: mit |
|
library_name: peft |
|
tags: |
|
- trl |
|
- kto |
|
- generated_from_trainer |
|
base_model: HuggingFaceH4/zephyr-7b-beta |
|
model-index: |
|
- name: WeniGPT-QA-Zephyr-7B-5.0.0-KTO |
|
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. --> |
|
|
|
# WeniGPT-QA-Zephyr-7B-5.0.0-KTO |
|
|
|
This model is a fine-tuned version of [HuggingFaceH4/zephyr-7b-beta](https://huggingface.co/HuggingFaceH4/zephyr-7b-beta) on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.0062 |
|
- Rewards/chosen: 6.6430 |
|
- Rewards/rejected: -36.7537 |
|
- Rewards/margins: 43.3967 |
|
- Kl: 0.1669 |
|
- Logps/chosen: -144.6907 |
|
- Logps/rejected: -566.5795 |
|
|
|
## 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: 0.0002 |
|
- train_batch_size: 4 |
|
- eval_batch_size: 4 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 8 |
|
- total_train_batch_size: 32 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_ratio: 0.03 |
|
- training_steps: 786 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/margins | Kl | Logps/chosen | Logps/rejected | |
|
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:---------------:|:------:|:------------:|:--------------:| |
|
| 0.1437 | 0.38 | 50 | 0.0282 | 5.2842 | -20.1101 | 25.3943 | 0.0961 | -158.2786 | -400.1437 | |
|
| 0.0615 | 0.76 | 100 | 0.0222 | 5.7502 | -18.4500 | 24.2003 | 0.5886 | -153.6186 | -383.5430 | |
|
| 0.0346 | 1.14 | 150 | 0.0398 | 4.8839 | -41.3691 | 46.2529 | 0.3036 | -162.2825 | -612.7335 | |
|
| 0.0563 | 1.52 | 200 | 0.0212 | 6.1746 | -26.4848 | 32.6594 | 0.1584 | -149.3753 | -463.8907 | |
|
| 0.0533 | 1.9 | 250 | 0.0134 | 6.1913 | -29.0566 | 35.2479 | 0.4595 | -149.2076 | -489.6085 | |
|
| 0.0076 | 2.28 | 300 | 0.0161 | 6.3153 | -25.4861 | 31.8015 | 0.6193 | -147.9676 | -453.9040 | |
|
| 0.011 | 2.66 | 350 | 0.0120 | 6.3302 | -37.6836 | 44.0138 | 0.4913 | -147.8187 | -575.8787 | |
|
| 0.0049 | 3.04 | 400 | 0.0102 | 6.3273 | -29.9323 | 36.2596 | 0.4649 | -147.8484 | -498.3662 | |
|
| 0.0028 | 3.42 | 450 | 0.0083 | 6.5215 | -34.1028 | 40.6243 | 0.2949 | -145.9056 | -540.0707 | |
|
| 0.0087 | 3.8 | 500 | 0.0096 | 6.4117 | -35.2134 | 41.6251 | 0.0923 | -147.0044 | -551.1769 | |
|
| 0.004 | 4.18 | 550 | 0.0075 | 6.5708 | -37.6298 | 44.2006 | 0.1574 | -145.4131 | -575.3412 | |
|
| 0.0036 | 4.56 | 600 | 0.0068 | 6.6432 | -36.6865 | 43.3297 | 0.1629 | -144.6893 | -565.9077 | |
|
| 0.003 | 4.94 | 650 | 0.0064 | 6.6633 | -36.7249 | 43.3882 | 0.1661 | -144.4881 | -566.2917 | |
|
| 0.0016 | 5.32 | 700 | 0.0062 | 6.6430 | -36.7537 | 43.3967 | 0.1669 | -144.6907 | -566.5795 | |
|
| 0.0042 | 5.7 | 750 | 0.0062 | 6.6553 | -36.6367 | 43.2920 | 0.1671 | -144.5682 | -565.4096 | |
|
|
|
|
|
### Framework versions |
|
|
|
- PEFT 0.10.0 |
|
- Transformers 4.39.1 |
|
- Pytorch 2.1.0+cu118 |
|
- Datasets 2.18.0 |
|
- Tokenizers 0.15.2 |
