6 lines
134 B
Plaintext
Vendored
6 lines
134 B
Plaintext
Vendored
#include "common.cuh"
|
|
|
|
#define CUDA_ALIBI_BLOCK_SIZE 32
|
|
|
|
void ggml_cuda_op_alibi(ggml_backend_cuda_context & ctx, ggml_tensor * dst);
|