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

GRPO zero3. RuntimeError: Inference tensors cannot be saved for backward. To work around you can make a clone to get a normal tensor and use it in autograd. #3237

Closed
Jintao-Huang opened this issue Feb 24, 2025 · 4 comments

Comments

@Jintao-Huang
Copy link
Collaborator

No description provided.

@Jintao-Huang
Copy link
Collaborator Author

解决方案:
pip install deepspeed==0.14.5

修改zero3.json,将stage3_prefetch_bucket_size从auto设置为0

@Jintao-Huang
Copy link
Collaborator Author

huggingface/trl#2953

@FANG-MING
Copy link

解决方案: pip install deepspeed==0.14.5

修改zero3.json,将stage3_prefetch_bucket_size从auto设置为0

试了一下,还是不行,改成zero2也不行

@Jintao-Huang
Copy link
Collaborator Author

Future-House/trl#7

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

2 participants