zhangyiss/mlx
1
0
Fork 0
mirror of https://github.com/ml-explore/mlx.git synced 2025-07-27 11:31:21 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

Fix typo in qmm check (#940)

Browse Source
This commit is contained in:
Angelos Katharopoulos 2024-03-31 19:15:44 -07:00 committed by GitHub
parent 433c0206b0
commit 9cbff5ec1d
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
mlx/backend/metal/kernels/quantized.metal
View File

@ -729,7 +729,7 @@ template <typename T, const int BM, const int BK, const int BN, const int group_
const device uint32_t * w_local = w + offset_row * N_w + offset_col; const device uint32_t * w_local = w + offset_row * N_w + offset_col;
threadgroup T * Ws_local = Ws + offset_row * BN + offset_col * el_per_int; threadgroup T * Ws_local = Ws + offset_row * BN + offset_col * el_per_int;
if (y_row + offset_row < K) { if (k + offset_row < K) {
uint32_t wi = *w_local; uint32_t wi = *w_local;
T scale = scales_block[offset_row * groups_per_block + offset_col / (group_size / el_per_int)]; T scale = scales_block[offset_row * groups_per_block + offset_col / (group_size / el_per_int)];
T bias = biases_block[offset_row * groups_per_block + offset_col / (group_size / el_per_int)]; T bias = biases_block[offset_row * groups_per_block + offset_col / (group_size / el_per_int)];