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

Fixed type as per reviewer comment

Browse source
This commit is contained in:
Iwan Kawrakow 2023-04-19 18:57:07 +02:00
parent 49beb2cdb8
commit 96d84438bc
1 changed files with 3 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

6
ggml.c
View file

@ -1177,8 +1177,8 @@ static float kquantize_q4_with_bounds(int n, int nmin, int nmax, const float * r
} }
static void quantize_row_q4_2_rmse(const float * restrict x, block_q4_2 * restrict y, int k) { static void quantize_row_q4_2_rmse(const float * restrict x, block_q4_2 * restrict y, int k) {
#define kCandiateCount 8 #define CANDIDATE_COUNT 8
static const float candidates[kCandiateCount] = { +8.7f, +8.3f, +8.1f, +7.8f, +7.3f, +7.0f, +6.3f, +5.7f }; static const float candidates[CANDIDATE_COUNT] = { +8.7f, +8.3f, +8.1f, +7.8f, +7.3f, +7.0f, +6.3f, +5.7f };
assert(k % QK4_2 == 0); assert(k % QK4_2 == 0);
int8_t L[QK4_2]; int8_t L[QK4_2];
@ -1187,7 +1187,7 @@ static void quantize_row_q4_2_rmse(const float * restrict x, block_q4_2 * restri
for (int i = 0; i < nb; i++) { for (int i = 0; i < nb; i++) {
float scale = kquantize_q4_with_bounds(QK4_2, -8, 7, x, kCandiateCount, candidates, L); float scale = kquantize_q4_with_bounds(QK4_2, -8, 7, x, CANDIDATE_COUNT, candidates, L);
y[i].d = GGML_FP32_TO_FP16(scale); y[i].d = GGML_FP32_TO_FP16(scale);
for (int l = 0; l < QK4_2; l += 2) { for (int l = 0; l < QK4_2; l += 2) {