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

Suppose Gemma model shape #130

Closed
yzh119 opened this issue Feb 21, 2024 · 1 comment
Closed

Suppose Gemma model shape #130

yzh119 opened this issue Feb 21, 2024 · 1 comment

Comments

@yzh119
Copy link
Collaborator

yzh119 commented Feb 21, 2024 •
edited
Loading

Gemma uses head_dim=256 which is enabled in pip wheels by default. We should compile kernels for head_dim=256 and change some kernel parameters for best performance in this case.
@yzh119 yzh119 mentioned this issue Feb 22, 2024
Merged
yzh119 added a commit that referenced this issue Feb 25, 2024
@yzh119
feat: enable head_dim=256 for attention kernels (#132)
0372acc 
As mentioned in #130 , the kernels for `head_dim=256` are not compiled
by default, this PR expose these attention kernels to pip wheels and
adds unittests/benchmarks for `head_dim=256`.
@yzh119
Copy link
Collaborator Author

yzh119 commented Feb 25, 2024

Fixed in #132 .

@yzh119 yzh119 closed this as completed Feb 25, 2024
@yzh119 yzh119 mentioned this issue Feb 27, 2024
Closed
3 tasks
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
@yzh119