This commit is contained in:
Lianmin Zheng
2025-08-09 16:00:10 -07:00
committed by GitHub
parent a886564a18
commit ef48d5547e
9 changed files with 120 additions and 75 deletions

View File

@@ -449,7 +449,9 @@ def grouped_topk_cpu(
routed_scaling_factor: Optional[float] = None,
num_token_non_padded: Optional[torch.Tensor] = None,
expert_location_dispatch_info: Optional[ExpertLocationDispatchInfo] = None,
apply_routed_scaling_factor_on_output: Optional[bool] = False,
):
assert not apply_routed_scaling_factor_on_output
assert expert_location_dispatch_info is None
return torch.ops.sgl_kernel.grouped_topk_cpu(
hidden_states,