CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-08 10:07:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

context : reuse built_attn_mha

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov
2025-02-24 11:18:40 +02:00
parent a5a85a3bc0
commit 4a1054b552
5 changed files with 109 additions and 167 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
src/llama-context.h
View File

@@ -261,17 +261,25 @@ public:
ggml_tensor * build_attn(
ggml_context * ctx0,
ggml_cgraph * gf,
ggml_tensor * wo,
ggml_tensor * wo_b,
ggml_tensor * q_cur,
ggml_tensor * k_cur,
ggml_tensor * v_cur,
ggml_tensor * kq_b,
int32_t n_tokens,
float kq_scale,
int il) override;
protected:
virtual ggml_tensor * build_attn_mha(
ggml_context * ctx0,
ggml_cgraph * gf,
ggml_tensor * q,
ggml_tensor * k,
ggml_tensor * v,
ggml_tensor * kq_b,
ggml_tensor * kq_mask,
bool v_trans,
float kq_scale);
virtual ggml_tensor * build_inp_self_k_shift(
ggml_context * ctx0);
@@ -472,13 +480,10 @@ public:
ggml_tensor * build_attn(
ggml_context * ctx0,
ggml_cgraph * gf,
ggml_tensor * wo,
ggml_tensor * wo_b,
ggml_tensor * q_cur,
ggml_tensor * k_cur,
ggml_tensor * v_cur,
ggml_tensor * kq_b,
int32_t n_tokens,
float kq_scale,
int il) override;