Arrcttacsrks's picture
Upload llama.cpp/ggml/src/ggml-cuda/diagmask.cuh with huggingface_hub
2461a46 verified
raw
history blame
150 Bytes
#include "common.cuh"
#define CUDA_DIAG_MASK_INF_BLOCK_SIZE 32
void ggml_cuda_op_diag_mask_inf(ggml_backend_cuda_context & ctx, ggml_tensor * dst);