Repository navigation

feat(moore): call MATE flash attention natively - #987

Open

voltjia wants to merge 7 commits into

feat/moore-flash-attn-native

Open

feat(moore): call MATE flash attention natively#987
voltjia wants to merge 7 commits into
masterfrom
feat/moore-flash-attn-native

Commits on Oct 12, 2026
- feat(moore): support flash-attn via MooreThreads/mate v0.2.5
  
  spike-zhu
  authored and
  voltjia
  committed
  782ea57
- wip: call MATE flash attention natively
  voltjia
  committed
  4bf7254
- fix: compile Moore MATE glue with system CXX
  voltjia
  committed
  1cd4854
- fix: include linked headers in system CXX build
  voltjia
  committed
  9437ce5
- style: format Moore MATE provider
  voltjia
  committed
  56b250a
- fix: compile MATE providers separately
  voltjia
  committed
  f1a519d
- fix: isolate Moore MATE provider state
  voltjia
  committed
  e33050f