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

Fuse more ops & Simplify token mapping #1758

Merged
merged 3 commits into from
Oct 23, 2024
Merged

Conversation

merrymercy
Copy link
Contributor

@merrymercy merrymercy commented Oct 23, 2024

  • Use torch.compile to fuse more operators
  • Simplify token mapping
  • Skip softmax for greedy sampling

This should bring 1% - 2% improvement.

@merrymercy merrymercy changed the title Simplify token mapping Simplify token mapping & Fuse more ops Oct 23, 2024
@merrymercy merrymercy changed the title Simplify token mapping & Fuse more ops Fuse more ops & Simplify token mapping Oct 23, 2024
@merrymercy merrymercy requested a review from ByronHsu as a code owner October 23, 2024 01:56
@merrymercy merrymercy force-pushed the simplify-token-mapping branch from 99e0526 to 6b31970 Compare October 23, 2024 05:10
@merrymercy merrymercy merged commit ad4125d into main Oct 23, 2024
10 of 11 checks passed
@merrymercy merrymercy deleted the simplify-token-mapping branch October 23, 2024 06:20
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

Successfully merging this pull request may close these issues.

1 participant