[Bugfix] Allow flashinfer_cutlass as a clamped NVFP4 MoE backend (#46492)

Signed-off-by: Zihua Wu <13583761+lucifer1004@users.noreply.github.com>
Signed-off-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>
This commit is contained in:
Gabriel Wu
2026-06-23 12:43:36 -07:00
committed by GitHub
co-authored by Claude Opus 4.8 Michael Goin
parent 0775b882ba
commit 0d4d164488
@@ -174,6 +174,7 @@ def select_nvfp4_moe_backend(
NVFP4_BACKENDS_WITH_CLAMP = {
NvFp4MoeBackend.FLASHINFER_TRTLLM,
NvFp4MoeBackend.FLASHINFER_CUTLASS,
NvFp4MoeBackend.MARLIN,
}