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.

Sign up for GitHub

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

Jump to bottom

finetune 65b-4bit OOM with larger cutoff_len #8

Open
better629 opened this issue May 16, 2023 · 0 comments
Open

finetune 65b-4bit OOM with larger cutoff_len #8

better629 opened this issue May 16, 2023 · 0 comments

Comments

@better629
Copy link

better629 commented May 16, 2023 •
edited
Loading

with default --cutoff_len=256 --mbatch_size 1 --batch_size 1 --lr 2e-4, it costs about 45315MiB(44.2G) in A100. Not 40G. And it costs 61843MiB with --cutoff_len=512.

and with larger cutoff_len like 768, it will cause OOM. but it's really indeed need larger cutoff_len.
@better629 better629 changed the title 65b-4bit OOM with larger cutoff_len finetune 65b-4bit OOM with larger cutoff_len May 16, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
1 participant
@better629