Skip to content

[distributed] Use unit sharding for attention projections - #48855

Open
ArthurZucker wants to merge 1 commit into
replicate-modulo-tpfrom
unit-tp-plans
Open

[distributed] Use unit sharding for attention projections#48855
ArthurZucker wants to merge 1 commit into
replicate-modulo-tpfrom
unit-tp-plans

Commits

  1. Commits on Sep 16, 2026