Skip to content

fix(deepseek_v4): drop full-attention sharding for MoE-only strategy#1996

Open
adurham wants to merge 1 commit intoexo-explore:mainfrom
adurham:dsv4-moe-only-sharding
Open

fix(deepseek_v4): drop full-attention sharding for MoE-only strategy#1996
adurham wants to merge 1 commit intoexo-explore:mainfrom
adurham:dsv4-moe-only-sharding

Commits

Commits on Apr 27, 2026