Skip to content

[V100/SM70] gemma-4 (31B + MTP) support: fully-FA hybrid attention (sliding-window + head_dim-512)#59

Closed
rivetphilbot wants to merge 25 commits into
1CatAI:mainfrom
rivetphilbot:feat/gemma4-mtp
Closed

[V100/SM70] gemma-4 (31B + MTP) support: fully-FA hybrid attention (sliding-window + head_dim-512)#59
rivetphilbot wants to merge 25 commits into
1CatAI:mainfrom
rivetphilbot:feat/gemma4-mtp

[FA_V100] Run gemma E2B/E4B KV-shared layers on FA (read target cache)

68d48fc
Select commit
Loading
Failed to load commit list.

Workflow runs completed with no jobs