ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

ggml-quants, llama : removed excess checks (#7274)

Browse source
This commit is contained in:
Herman Semenov 2024-05-17 07:08:49 +00:00 • committed by GitHub
parent e18bc6aaf3
commit 359cbe3f46
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 4 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml-quants.c
View file

@ -1986,7 +1986,7 @@ static void quantize_row_q3_K_impl(const float * restrict x, block_q3_K * restri
for (int j = 0; j < QK_K/16; ++j) {
if (quant_weights) {
const float * qw = quant_weights ? quant_weights + QK_K * i + 16*j : NULL;
const float * qw = quant_weights + QK_K * i + 16*j;
for (int l = 0; l < 16; ++l) weight[l] = qw[l] * sqrtf(sigma2 + x[16*j+l]*x[16*j+l]);
} else {
for (int l = 0; l < 16; ++l) weight[l] = x[16*j+l]*x[16*j+l];