GitHub

手动构建Llama3模型,并使用RLHF方法训练.

运行环境:

torch==1.13.1+cu117

transformers==4.38.2

datasets==2.18.0

accelerate==0.28.0

2024年8月7日更新:添加 gemma2 模型.

Name		Name	Last commit message	Last commit date
Latest commit History 8 Commits
dataset		dataset
model		model
tokenizer		tokenizer
1.model.ipynb		1.model.ipynb
1.model_gemma2.ipynb		1.model_gemma2.ipynb
2.actor.ipynb		2.actor.ipynb
3.critic.ipynb		3.critic.ipynb
4.rlhf.ipynb		4.rlhf.ipynb
5.test.ipynb		5.test.ipynb
README.md		README.md
util.py		util.py

Provide feedback