Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

感觉好久没更新了 #78

Open
2catycm opened this issue Mar 13, 2024 · 4 comments
Open

感觉好久没更新了 #78

2catycm opened this issue Mar 13, 2024 · 4 comments

Comments

@2catycm
Copy link

2catycm commented Mar 13, 2024

No description provided.

@2catycm
Copy link
Author

2catycm commented Mar 31, 2024

我很喜欢你们的接口设计,我觉得隔壁的huggingface/peft库设计的不好

@2catycm
Copy link
Author

2catycm commented Mar 31, 2024

delta模型本身应该和原本的模型独立存在

  • 两边的参数独立保存。这点opendelta做的挺好的,可以单独看到delta的参数列表。
  • 为什么要独立存在呢?因为lora这样的模型,delta作为一个认知功能,虽然来自于某个backbone的基础上训练出来,但是可以操作到别的模型上面,影响其他模型的认知功能。而且这样freeze和unfreeze的逻辑很直观

@2catycm
Copy link
Author

2catycm commented Mar 31, 2024

但是opendelta和peft都不太支持视觉的模型,都是在支持nlp的peft,所以我最近自己重新写了一个,欢迎来互相交流一下。我用了Pytorch的hook特性,感觉可以更加简洁和通用一些。

@2catycm
Copy link
Author

2catycm commented Mar 31, 2024

opendelta的写法大概是在basemodel,basedelta里面有个insert module,然后我用torch的hook去实现类似的操作。
而peft大概喜欢用replace module,直接setattr替换掉

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant