vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Fixed bug in q4_K quantization added with the 64-block addition

Browse source
This commit is contained in:
Iwan Kawrakow 2023-06-23 17:46:11 +03:00
parent 558a19427b
commit 333ffcc5ba
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

@ -707,6 +707,7 @@ void quantize_row_q4_K_reference(const float * restrict x, block_q4_K * restrict
uint8_t * q = y[i].qs;
for (int j = 0; j < QK_K; j += 64) {
for (int l = 0; l < 32; ++l) q[l] = L[j + l] | (L[j + l + 32] << 4);
q += 32;
}
x += QK_K;