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

Fix buffer size

Browse source
This commit is contained in:
Henrik Forstén 2024-01-02 20:07:34 +02:00
parent 7e03872f1e
commit 6683eb4960
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml-quants.c
View file

@ -1336,7 +1336,7 @@ static void quantize_q_k_1(const float * x, int bits, int scale_bits, int block_
float mins[QK_K/block_size]; float mins[QK_K/block_size];
for (int j = 0; j < QK_K/block_size; j++) { for (int j = 0; j < QK_K/block_size; j++) {
uint8_t q[QK_K/block_size]; uint8_t q[block_size];
// First find least squares solution for min and scale for each block. // First find least squares solution for min and scale for each block.
quantize_1(&x[block_size*j], block_size, bits, q, &mins[j], &scales[j]); quantize_1(&x[block_size*j], block_size, bits, q, &mins[j], &scales[j]);
// Flip the sign because quantize_1 assumes that min is added, but min // Flip the sign because quantize_1 assumes that min is added, but min