metadata
base_model: mrs-7b
tags:
- alignment-handbook
- generated_from_trainer
datasets:
- gpt4_reward_with_format
model-index:
- name: zephyr-7b-sft-lora
results: []
zephyr-7b-sft-lora
This model is a fine-tuned version of zephyr-7b on the gpt4_reward. It achieves the following results on the evaluation set:
- Loss: 0.0911
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: 3
- eval_batch_size: 2
- seed: 42
- distributed_type: multi-GPU
- gradient_accumulation_steps: 64
- total_train_batch_size: 192
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.1001 | 1.0 | 219 | 0.1006 |
| 0.0969 | 2.0 | 439 | 0.0930 |
| 0.0795 | 2.99 | 657 | 0.0911 |
Framework versions
- Transformers 4.35.0
- Pytorch 2.1.1+cu121
- Datasets 2.14.6
- Tokenizers 0.14.1