1 2 3 4 5
#include "common.cuh" #define CUDA_PAD_REFLECT_1D_BLOCK_SIZE 256 void ggml_cuda_op_pad_reflect_1d(ggml_backend_cuda_context & ctx, ggml_tensor * dst);