metadata
license: mit
library_name: peft
tags:
- trl
- kto
- generated_from_trainer
base_model: HuggingFaceH4/zephyr-7b-beta
model-index:
- name: WeniGPT-Agents-Zephyr-1.0.9-KTO
results: []
WeniGPT-Agents-Zephyr-1.0.9-KTO
This model is a fine-tuned version of HuggingFaceH4/zephyr-7b-beta on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.5443
- Rewards/chosen: -157.2410
- Logps/chosen: -1859.8108
- Rewards/rejected: -156.3582
- Logps/rejected: -1831.8481
- Kl: 0.0
- Rewards/margins: 2.0475
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: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.03
- training_steps: 145
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Logps/chosen | Rewards/rejected | Logps/rejected | Kl | Rewards/margins |
|---|---|---|---|---|---|---|---|---|---|
| 0.5606 | 0.34 | 50 | 0.5443 | -151.0505 | -1797.9062 | -151.0357 | -1778.6233 | 0.0 | 2.7896 |
| 0.4995 | 0.68 | 100 | 0.5443 | -157.2410 | -1859.8108 | -156.3582 | -1831.8481 | 0.0 | 2.0475 |
Framework versions
- PEFT 0.10.0
- Transformers 4.39.1
- Pytorch 2.1.0+cu118
- Datasets 2.18.0
- Tokenizers 0.15.2