6 lines
185 B
Plaintext
6 lines
185 B
Plaintext
#include "common.cuh"
|
|
|
|
void ggml_cuda_flash_attn_ext(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
|
|
|
|
bool ggml_cuda_flash_attn_ext_supported(int device, const ggml_tensor * dst);
|