llama.cpp/ggml-cuda/fattn-tile-f16.cuh
2024-05-17 18:54:52 +02:00

4 lines
115 B
Plaintext

#include "common.cuh"
void ggml_cuda_flash_attn_ext_tile_f16(ggml_backend_cuda_context & ctx, ggml_tensor * dst);