We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent 5ab56cb commit 847150aCopy full SHA for 847150a
vllm/model_executor/layers/fused_moe/layer.py
@@ -437,9 +437,9 @@ def apply(
437
custom_routing_function: Optional[Callable] = None,
438
scoring_func: str = "softmax",
439
e_score_correction_bias: Optional[torch.Tensor] = None,
440
+ apply_router_weight_on_input: bool = False,
441
activation: str = "silu",
442
) -> torch.Tensor:
-
443
assert activation == "silu"
444
assert global_num_experts == layer.w13_weight.shape[0]
445
assert expert_map is None
@@ -466,6 +466,7 @@ def apply(
466
self.c_strides1,
467
self.ab_strides2,
468
self.c_strides2,
469
+ apply_router_weight_on_input=apply_router_weight_on_input,
470
)
471
472
0 commit comments