tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-08-15 04:33:06 -04:00
Code Issues Packages Projects Releases Wiki Activity

k-quants : fix zero-weight guard in Q6_K (ref #3040)

Browse Source
This commit is contained in:
Georgi Gerganov
2023-09-06 12:40:57 +03:00
parent ea2c85d5d2
commit 178b1850eb
1 changed files with 1 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
k_quants.c
View File

@@ -1089,6 +1089,7 @@ void quantize_row_q6_K_reference(const float * restrict x, block_q6_K * restrict
if (!max_abs_scale) { if (!max_abs_scale) {
memset(&y[i], 0, sizeof(block_q6_K)); memset(&y[i], 0, sizeof(block_q6_K));
y[i].d = ggml_fp32_to_fp16(0.f); y[i].d = ggml_fp32_to_fp16(0.f);
x += QK_K;
continue; continue;
} }