Skip to content

Commit

Permalink
Fix template instantiation
Browse files Browse the repository at this point in the history
  • Loading branch information
vinx13 committed Dec 14, 2023
1 parent 7191fbd commit d489188
Show file tree
Hide file tree
Showing 3 changed files with 4 additions and 4 deletions.
2 changes: 1 addition & 1 deletion cutlass_kernels/moe_gemm/moe_gemm_kernels_fp16_fp16.cu
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,7 @@
namespace fastertransformer
{

template <>
template
void moe_gemm_bias_act(const half* A, const half* B, const half* weight_scales, const half* biases, half* C,
int64_t* total_rows_before_expert, int64_t total_rows, int64_t gemm_n, int64_t gemm_k, int num_experts,
std::optional<std::string> activation, cudaStream_t stream);
Expand Down
2 changes: 1 addition & 1 deletion cutlass_kernels/moe_gemm/moe_gemm_kernels_fp16_uint4.cu
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,7 @@
namespace fastertransformer
{

template <>
template
void moe_gemm_bias_act(const half* A, const cutlass::uint4b_t* B, const half* weight_scales, const half* biases,
half* C, int64_t* total_rows_before_expert, int64_t total_rows, int64_t gemm_n, int64_t gemm_k, int num_experts,
std::optional<std::string> activation, cudaStream_t stream);
Expand Down
4 changes: 2 additions & 2 deletions cutlass_kernels/moe_gemm/moe_gemm_kernels_fp16_uint8.cu
Original file line number Diff line number Diff line change
Expand Up @@ -19,9 +19,9 @@
namespace fastertransformer
{

template <>
template
void moe_gemm_bias_act(const half* A, const uint8_t* B, const half* weight_scales, const half* biases, half* C,
int64_t* total_rows_before_expert, int64_t total_rows, int64_t gemm_n, int64_t gemm_k, int num_experts,
std::optional<std::string> activation, cudaStream_t stream);

}
}

0 comments on commit d489188

Please sign in to comment.