Skip to content

Enable interleaved sliding_window for gemma3 #1344

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

Open
wants to merge 34 commits into
base: habana_main
Choose a base branch
from

Conversation

jiminha
Copy link

@jiminha jiminha commented May 30, 2025

Enable interleaved sliding_window for gemma3

  • apply sliding_window mask for prompt attention
  • apply sliding_window for decode block information

Without this change, prompts longer than size of sliding_window have an accuracy issue.

@jiminha jiminha changed the base branch from main to habana_main May 30, 2025 04:57
@jiminha jiminha changed the title Enable interleaved sliding_window for gemma3 [WIP] Enable interleaved sliding_window for gemma3 May 30, 2025
@jiminha jiminha marked this pull request as draft May 30, 2025 16:43
@ssarkar2 ssarkar2 force-pushed the jha/sliding_window_gemma3 branch from aa86d92 to a9e5a7d Compare June 5, 2025 00:05
@ssarkar2 ssarkar2 force-pushed the jha/sliding_window_gemma3 branch from 1d44a03 to f783955 Compare June 5, 2025 05:56
@ssarkar2 ssarkar2 force-pushed the jha/sliding_window_gemma3 branch from 716e329 to be41114 Compare June 5, 2025 19:31
@jiminha jiminha changed the title [WIP] Enable interleaved sliding_window for gemma3 Enable interleaved sliding_window for gemma3 Jun 23, 2025
@jiminha jiminha marked this pull request as ready for review June 23, 2025 21:32
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.

4 participants