CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-01 09:01:57 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama : fix command-r inference when omitting outputs (#6367)

Browse Source
This commit is contained in:
compilade
2024-03-28 08:05:54 -04:00
committed by GitHub
parent 28cb9a09c4
commit 0308f5e3d7
1 changed files with 3 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
llama.cpp
View File

@@ -9152,8 +9152,9 @@ struct llm_build_context {
if (il == n_layer - 1) {
// skip computing output for unused tokens
struct ggml_tensor * inp_out_ids = build_inp_out_ids();
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
inpL = ggml_get_rows(ctx0, inpL, inp_out_ids);
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
inpL = ggml_get_rows(ctx0, inpL, inp_out_ids);
ffn_inp = ggml_get_rows(ctx0, ffn_inp, inp_out_ids);
}
struct ggml_tensor * attn_out = cur;