Skip to content

Commit

Permalink
ggml : do not crash when quantizing q4_x_x with an imatrix (#9192)
Browse files Browse the repository at this point in the history
  • Loading branch information
slaren authored Aug 26, 2024
1 parent 06658ad commit 7d787ed
Showing 1 changed file with 6 additions and 21 deletions.
27 changes: 6 additions & 21 deletions ggml/src/ggml-aarch64.c
Original file line number Diff line number Diff line change
Expand Up @@ -337,33 +337,18 @@ static size_t quantize_q4_0_nr_bl(const float * restrict src, void * restrict ds
}

size_t quantize_q4_0_4x4(const float * restrict src, void * restrict dst, int64_t nrow, int64_t n_per_row, const float * quant_weights) {
if (!quant_weights) {
return quantize_q4_0_nr_bl(src, dst, nrow, n_per_row, 4, 4);
}
else {
assert(false);
return 0;
}
UNUSED(quant_weights);
return quantize_q4_0_nr_bl(src, dst, nrow, n_per_row, 4, 4);
}

size_t quantize_q4_0_4x8(const float * restrict src, void * restrict dst, int64_t nrow, int64_t n_per_row, const float * quant_weights) {
if (!quant_weights) {
return quantize_q4_0_nr_bl(src, dst, nrow, n_per_row, 4, 8);
}
else {
assert(false);
return 0;
}
UNUSED(quant_weights);
return quantize_q4_0_nr_bl(src, dst, nrow, n_per_row, 4, 8);
}

size_t quantize_q4_0_8x8(const float * restrict src, void * restrict dst, int64_t nrow, int64_t n_per_row, const float * quant_weights) {
if (!quant_weights) {
return quantize_q4_0_nr_bl(src, dst, nrow, n_per_row, 8, 8);
}
else {
assert(false);
return 0;
}
UNUSED(quant_weights);
return quantize_q4_0_nr_bl(src, dst, nrow, n_per_row, 8, 8);
}

void ggml_gemv_q4_0_4x4_q8_0(int n, float * restrict s, size_t bs, const void * restrict vx, const void * restrict vy, int nr, int nc) {
Expand Down

0 comments on commit 7d787ed

Please sign in to comment.