L-electron-Rare/mlx
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[CUDA] Implement BlockMaskedMM (#3299)

Browse Source 
Co-authored-by: Cheng <[email protected]>
This commit is contained in:
Long Yixing
2026-03-27 06:57:26 +09:00
committed by GitHub
co-authored by Cheng
parent df7f7db943
commit 0ff1115a46
7 changed files with 475 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/tests/cuda_skip.py
-2
View File
@@ -1,7 +1,5 @@
cuda_skip = {
"TestLayers.test_quantized_embedding",
# Block masked matmul NYI
"TestBlas.test_block_masked_matmul",
# Gather matmul NYI
"TestBlas.test_gather_matmul",
"TestBlas.test_gather_matmul_grad",