Skip to content

Conversation

ZYHowell
Copy link
Collaborator

This PR makes the layout of Sequence parallel only adds padding tokens at the end of the last sp rank, and is after each request instead of at the end of all requests

Copy link
Owner

@ivanium ivanium left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I have tried this PR and tested it in my local branch #2 and it looks correct. So I think we can get this merged first before merging the sp attn kernel.

@ZYHowell ZYHowell merged commit 4b8203a into main Jul 28, 2024
@ZYHowell ZYHowell deleted the pr-new-sp-layout branch July 28, 2024 18:35
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.

2 participants