File size: 2,003 Bytes
b412e14 |
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 |
---
license: apache-2.0
library_name: peft
tags:
- trl
- kto
- generated_from_trainer
base_model: mistralai/Mixtral-8x7B-Instruct-v0.1
model-index:
- name: WeniGPT-QA-Zephyr-7B-5.0.1-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-5.0.1-KTO
This model is a fine-tuned version of [mistralai/Mixtral-8x7B-Instruct-v0.1](https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0146
- Eval/rewards/chosen: 6.5462
- Eval/rewards/rejected: -30.7776
- Eval/kl: 0.2505
- Eval/logps/chosen: -129.4441
- Eval/logps/rejected: -508.0271
- Eval/rewards/margins: 37.3238
## 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: 262
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | |
|:-------------:|:-----:|:----:|:---------------:|:-------:|
| 0.1177 | 0.38 | 50 | 0.0468 | 24.8637 |
| 0.0257 | 0.76 | 100 | 0.0236 | 30.5016 |
| 0.0141 | 1.14 | 150 | 0.0219 | 33.9185 |
| 0.0103 | 1.52 | 200 | 0.0146 | 37.3238 |
| 0.0084 | 1.9 | 250 | 0.0129 | 39.0837 |
### Framework versions
- PEFT 0.10.0
- Transformers 4.39.1
- Pytorch 2.1.0+cu118
- Datasets 2.18.0
- Tokenizers 0.15.1 |