diff --git a/vllm/_custom_ops.py b/vllm/_custom_ops.py index bf2d8a865a..4de43e96a8 100644 --- a/vllm/_custom_ops.py +++ b/vllm/_custom_ops.py @@ -2246,6 +2246,18 @@ def topk_sigmoid( e_score_correction_bias: torch.Tensor | None = None, routed_scaling_factor: float = 1.0, ) -> None: + if current_platform.is_xpu(): + # xpu doesn't support routed_scaling_factor currently, will revert + # in next vllm-xpu-kernels bumpup + torch.ops._moe_C.topk_sigmoid( + topk_weights, + topk_ids, + token_expert_indices, + gating_output, + renormalize, + e_score_correction_bias, + ) + return torch.ops._moe_C.topk_sigmoid( topk_weights, topk_ids,