Bump Flashinfer to v0.6.1 (#30993)

Signed-off-by: elvischenv <219235043+elvischenv@users.noreply.github.com>
2026-01-22 00:49:50 +08:00
parent 1861ae8aae
commit 808d6fd7b9
12 changed files with 20 additions and 73 deletions
--- a/vllm/model_executor/layers/quantization/mxfp4.py
+++ b/vllm/model_executor/layers/quantization/mxfp4.py
@@ -982,8 +982,7 @@ class Mxfp4MoEMethod(FusedMoEMethodBase):
                self.intermediate_size,  # padded to multiple of 256
                layer.ep_rank * layer.local_num_experts,  # local_expert_offset
                self.num_experts,  # local num experts
-                None,
-                None,
+                None,  # routed_scaling_factor
                1 if layer.renormalize else 0,  # routing_method_type, renormalize
                True,  # do finalize
                tune_max_num_tokens=max(self.max_capture_size, 1),