metadata
license: mit
base_model: HuggingFaceH4/zephyr-7b-beta
tags:
- trl
- sft
- generated_from_trainer
model-index:
- name: results
results: []
results
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: 0.6622
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: 2e-05
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- lr_scheduler_warmup_ratio: 0.03
- num_epochs: 2
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.5641 | 0.14 | 100 | 1.4610 |
| 1.1695 | 0.28 | 200 | 1.0388 |
| 1.0319 | 0.42 | 300 | 0.9440 |
| 0.905 | 0.56 | 400 | 0.8829 |
| 0.8655 | 0.7 | 500 | 0.8225 |
| 0.8329 | 0.85 | 600 | 0.8042 |
| 0.85 | 0.99 | 700 | 0.7728 |
| 0.7348 | 1.13 | 800 | 0.7426 |
| 0.6723 | 1.27 | 900 | 0.7197 |
| 0.6791 | 1.41 | 1000 | 0.6933 |
| 0.6576 | 1.55 | 1100 | 0.6864 |
| 0.6863 | 1.69 | 1200 | 0.6731 |
| 0.6328 | 1.83 | 1300 | 0.6652 |
| 0.6264 | 1.97 | 1400 | 0.6622 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1