char-fork/llama.cpp

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

leejetggml: add ops for WAN video model (cuda && cpu) (#15669)0a1b3982c

char/aliceai-t5-moe
218 B6 linesraw
1#include "common.cuh"
2
3#define CUDA_IM2COL_BLOCK_SIZE 256
4
5void ggml_cuda_op_im2col(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
6void ggml_cuda_op_im2col_3d(ggml_backend_cuda_context & ctx, ggml_tensor * dst);