vrr/prima.cpp
2
0
Fork 0
mirror of https://github.com/Lizonghang/prima.cpp.git synced 2025-09-07 07:59:29 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix q6k and q80

Browse source
This commit is contained in:
Zonghang Li 2025-04-16 08:55:07 +04:00
parent b59d6d9cdf
commit 5a5f103833
1 changed files with 1 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
common/profiler.cpp
View file

@ -370,7 +370,7 @@ float device_inp_embd_delay(struct llama_model * model, enum ggml_type src0t, in
case GGML_TYPE_Q6_K: case GGML_TYPE_Q6_K:
case GGML_TYPE_Q8_K: case GGML_TYPE_Q8_K:
case GGML_TYPE_Q8_0: case GGML_TYPE_Q8_0:
QK_K = 256; QK_K = ggml_blck_size(src0t);
matrix_B = malloc((embd_size / QK_K) * ggml_type_size(src0t)); matrix_B = malloc((embd_size / QK_K) * ggml_type_size(src0t));
break; break;
default: default: