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

Revert Q5_K back to make_qkx1_quants

Browse source
This commit is contained in:
Iwan Kawrakow 2023-08-20 09:02:43 +03:00
parent 404e43cc3b
commit 9f78d4cdf9
1 changed files with 9 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

17
k_quants.c
View file

@ -731,6 +731,7 @@ void quantize_row_q4_K_reference(const float * restrict x, block_q4_K * restrict
float max_scale = 0; // as we are deducting the min, scales are always positive float max_scale = 0; // as we are deducting the min, scales are always positive
float max_min = 0; float max_min = 0;
for (int j = 0; j < QK_K/32; ++j) { for (int j = 0; j < QK_K/32; ++j) {
//scales[j] = make_qkx1_quants(32, 15, x + 32*j, L + 32*j, &mins[j], 9, 0.5f);
float sum_x2 = 0; float sum_x2 = 0;
for (int l = 0; l < 32; ++l) sum_x2 += x[32*j + l] * x[32*j + l]; for (int l = 0; l < 32; ++l) sum_x2 += x[32*j + l] * x[32*j + l];
float av_x = sqrtf(sum_x2/32); float av_x = sqrtf(sum_x2/32);
@ -888,8 +889,8 @@ void quantize_row_q5_K_reference(const float * restrict x, block_q5_K * restrict
uint8_t L[QK_K]; uint8_t L[QK_K];
float mins[QK_K/32]; float mins[QK_K/32];
float scales[QK_K/32]; float scales[QK_K/32];
float weights[32]; //float weights[32];
uint8_t Laux[32]; //uint8_t Laux[32];
#else #else
int8_t L[QK_K]; int8_t L[QK_K];
float scales[QK_K/16]; float scales[QK_K/16];
@ -902,12 +903,12 @@ void quantize_row_q5_K_reference(const float * restrict x, block_q5_K * restrict
float max_scale = 0; // as we are deducting the min, scales are always positive float max_scale = 0; // as we are deducting the min, scales are always positive
float max_min = 0; float max_min = 0;
for (int j = 0; j < QK_K/32; ++j) { for (int j = 0; j < QK_K/32; ++j) {
//scales[j] = make_qkx1_quants(32, 31, x + 32*j, L + 32*j, &mins[j], 9, 0.5f); scales[j] = make_qkx1_quants(32, 31, x + 32*j, L + 32*j, &mins[j], 9, 0.5f);
float sum_x2 = 0; //float sum_x2 = 0;
for (int l = 0; l < 32; ++l) sum_x2 += x[32*j + l] * x[32*j + l]; //for (int l = 0; l < 32; ++l) sum_x2 += x[32*j + l] * x[32*j + l];
float av_x = sqrtf(sum_x2/32); //float av_x = sqrtf(sum_x2/32);
for (int l = 0; l < 32; ++l) weights[l] = av_x + fabsf(x[32*j + l]); //for (int l = 0; l < 32; ++l) weights[l] = av_x + fabsf(x[32*j + l]);
scales[j] = make_qkx2_quants(32, 31, x + 32*j, weights, L + 32*j, &mins[j], Laux, -0.5f, 0.1f, 15, false); //scales[j] = make_qkx2_quants(32, 31, x + 32*j, weights, L + 32*j, &mins[j], Laux, -0.5f, 0.1f, 15, false);
float scale = scales[j]; float scale = scales[j];
if (scale > max_scale) { if (scale > max_scale) {
max_scale = scale; max_scale = scale;