https://github.com/jackfsuia/phi-3-dpo

a Phi-3 DPO training script. Phi-3 DPO训练脚本
https://github.com/jackfsuia/phi-3-dpo

Last synced: 3 days ago
JSON representation

a Phi-3 DPO training script. Phi-3 DPO训练脚本

• Host: GitHub
• URL: https://github.com/jackfsuia/phi-3-dpo
• Owner: jackfsuia
• Created: 2024-08-15T04:59:14.000Z (3 months ago)
• Default Branch: main
• Last Pushed: 2024-08-15T05:14:55.000Z (3 months ago)
• Last Synced: 2024-08-15T06:21:40.141Z (3 months ago)
• Language: Python
• Homepage:
• Size: 9.77 KB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Phi-3-DPO

a Phi-3 DPO training script. 一个用于Phi-3 DPO训练的脚本。

## Start

To start DPO training for Phi-3 model, 运行`dpo.py`即可

```

python dpo.py

```

or start DPO training with an evaluation of cross entropy loss on another eval dataset, run 假如想增加一个测试集求取交叉熵损失函数，运行`dpo-eval-another-loss.py`即可

```

python dpo-eval-another-loss.py

```