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

Q4 deep dive #38

Open

FL33TW00D opened this issue Jan 23, 2024 · 1 comment

Collaborator

FL33TW00D commented Jan 23, 2024

We need fast (& high quality) 4 bit quantisation.

You can see the kernel I have written here: https://github.com/FL33TW00D/wgpu-mm/blob/master/shaders/qgemm/qgemm_int4.wgsl

This is very naive, and also very slow.

Collaborator

sigma-andex commented May 1, 2024 •

edited

References for Q4K:

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