whisper.cpp/ggml-cuda/clamp.cuh

6 lines
135 B
Plaintext
Raw Normal View History

#include "common.cuh"
#define CUDA_CLAMP_BLOCK_SIZE 256
void ggml_cuda_op_clamp(ggml_backend_cuda_context & ctx, ggml_tensor * dst);