Reduce VRAM 11GB -> 6.5GB #28

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

Open

mykytahordia wants to merge 1 commit into hkproj:main from mykytahordia:opt/generation

mykytahordia commented Dec 17, 2025

Model barely fits into a standard consumer GPU because of double params loading. I did the following:

load state_dict to idle_device
release memory after generation

It reduced peak memory usage from 11GB to 6.5GB without any speed tradeoffs and made to_idle function actually work.


          Load weights to idle_device, release memory after generation

9176f88

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet