[Bugfix] Fixes for new marlin moe usage (#18017)

Signed-off-by: mgoin <mgoin64@gmail.com>
This commit is contained in:
Michael Goin
2025-05-12 23:50:04 -04:00
committed by GitHub
parent d8487ef557
commit 1df491c522
2 changed files with 5 additions and 4 deletions

View File

@@ -610,9 +610,9 @@ class GPTQMarlinMoEMethod(FusedMoEMethodBase):
activation: str = "silu",
) -> torch.Tensor:
assert activation == "silu", "Only SiLU activation is supported."
if apply_router_weight_on_input is not None:
if apply_router_weight_on_input:
raise NotImplementedError(
"Apply router weight on input is not supported for"
"Apply router weight on input is not supported for "
"fused Marlin MoE method.")
topk_weights, topk_ids = FusedMoE.select_experts(