Skip to content

CUDA backend: lower complex rotary embeddings and fused attention, and free repeated outputs once - #24403

Open
shoumikhin wants to merge 5 commits into
pytorch:mainfrom
shoumikhin:cuda-complex-ops
Open

shoumikhin wants to merge 5 commits into
pytorch:mainfrom
shoumikhin:cuda-complex-ops

Commits

  1. Commits on Nov 1, 2026