CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-13 10:57:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama : add ggml_cont to trigger bug with Metal

Browse Source
This commit is contained in:
Georgi Gerganov
2023-09-04 19:46:52 +03:00
parent bd33e5ab92
commit ad80e5a4a7
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
llama.cpp
View File

@@ -2418,11 +2418,11 @@ static struct ggml_cgraph * llm_build_llama(
// split cached V into n_head heads
struct ggml_tensor * V =
ggml_view_3d(ctx0, kv_self.v,
ggml_cont(ctx0, ggml_view_3d(ctx0, kv_self.v,
n_past + N, n_embd_head, n_head_kv,
ggml_element_size(kv_self.v)*n_ctx,
ggml_element_size(kv_self.v)*n_ctx*n_embd_head,
ggml_element_size(kv_self.v)*n_ctx*n_embd_gqa*il);
ggml_element_size(kv_self.v)*n_ctx*n_embd_gqa*il));
offload_func_v(V);
ggml_set_name(V, "V");