vllm/csrc/moe/moe_ops.h
2024-02-05 17:38:02 -08:00

10 lines
197 B
C

#pragma once
#include <torch/extension.h>
void topk_softmax(
torch::Tensor& topk_weights,
torch::Tensor& topk_indices,
torch::Tensor& token_expert_indices,
torch::Tensor& gating_output);