update llama.cpp

This commit is contained in:
Michael Yang
2023-08-01 16:22:03 -07:00
parent da52f5bfdd
commit 7a1c3e62dc
18 changed files with 2603 additions and 493 deletions

View File

@@ -1,5 +1,5 @@
/**
* llama.cpp - git d91f3f0c55663719ea03b76311e8c36ed55eb0e2
* llama.cpp - git c574bddb368424b5996cbee2ec45ec050967d404
*
* MIT License
*
@@ -1196,7 +1196,18 @@ extern "C" {
int mode,
int n_ctx);
// custom RoPE, in-place, returns view(a)
// custom RoPE
GGML_API struct ggml_tensor * ggml_rope_custom(
struct ggml_context * ctx,
struct ggml_tensor * a,
int n_past,
int n_dims,
int mode,
int n_ctx,
float freq_base,
float freq_scale);
// in-place, returns view(a)
GGML_API struct ggml_tensor * ggml_rope_custom_inplace(
struct ggml_context * ctx,
struct ggml_tensor * a,