Datasets:

Modalities:
Text
Formats:
text
Size:
< 1K
ArXiv:
Libraries:
Datasets
llama.cpp / ggml /src /ggml-cuda /conv2d.cuh
dlxj
todo: 基于 CUDA 13.0 编译
2517be1
#pragma once
#include "common.cuh"
#define CUDA_CONV2D_BLOCK_SIZE 256
void ggml_cuda_op_conv2d(ggml_backend_cuda_context & ctx, ggml_tensor * dst);