char-fork/llama.cpp

git clone https://git.t4t.associates/char-fork/llama.cpp

Aman GuptaCUDA: add conv_2d_dw (#14265)9eaa51e7f

char/aliceai-t5-moe
155 B5 linesraw
1#pragma once
2#include "common.cuh"
3
4#define CUDA_CONV2D_DW_BLOCK_SIZE 256
5void ggml_cuda_op_conv2d_dw(ggml_backend_cuda_context & ctx, ggml_tensor * dst);