• Joined on 2024-11-24
squall created repository squall/vllm 2024-11-24 18:26:40 +08:00
squall pushed to main at squall/flash-attention 2024-11-24 17:05:03 +08:00
7153673c1a Fix swiglu backwards return type (#1337)
641db759ab [CI] Pytorch 2.5.1 does not support python 3.8
7435839e3d Update README for FA3
241c682c9f [CI] Switch back to CUDA 12.4
c555642172 Bump to v2.7.0
Compare 10 commits »
squall created branch main in squall/flash-attention 2024-11-24 17:04:58 +08:00
squall created repository squall/flash-attention 2024-11-24 17:03:49 +08:00
squall created branch main in squall/cutlass 2024-11-24 17:03:18 +08:00
squall pushed to main at squall/cutlass 2024-11-24 17:03:18 +08:00
b0e09d7cd3 Fix cutlass python library with cuda 12.6.2.post1 (#1942)
8aa95dbb88 Fix the racing condition of mixed-input gemm when writing the registers (#1931)
d656afbd2a fix undefined in device code error (#1880)
32e3c38aef remove restriction of stride == kernel in nhwc_pooling (#1896)
9004ed2d1b Update publications (#1912)
Compare 10 commits »
squall created repository squall/cutlass 2024-11-24 17:02:06 +08:00
squall pushed to main at squall/torch_ext 2024-11-24 15:46:35 +08:00
4da12fd0c2 多维的问题也实验了一下,看起来还不错的样子。
bf81e39d83 简单修改一下。
a43baa8b7f test multi dimension matrix multiply
2285b8b6f2 修改一下
8ae9fb6819 简单提交一下。
squall created branch main in squall/torch_ext 2024-11-24 15:46:34 +08:00
squall created repository squall/torch_ext 2024-11-24 15:45:26 +08:00