Skip to content

Releases: Dao-AILab/flash-attention

v2.3.2

09 Oct 00:22
Compare
Choose a tag to compare
Bump to v2.3.2

v2.3.1.post1

04 Oct 05:21
Compare
Choose a tag to compare
[CI] Use official Pytorch 2.1, add CUDA 11.8 for Pytorch 2.1

v2.3.1

04 Oct 02:57
Compare
Choose a tag to compare
Bump to v2.3.1

v2.3.0

27 Sep 05:09
Compare
Choose a tag to compare
Bump to v2.3.0

v2.2.5

24 Sep 08:05
Compare
Choose a tag to compare
Bump to v2.2.5

v2.2.4.post1

22 Sep 06:57
Compare
Choose a tag to compare
Re-enable compilation for Hopper

v2.2.4

21 Sep 06:40
Compare
Choose a tag to compare
Bump to v2.2.4

v2.2.3.post2

18 Sep 05:17
Compare
Choose a tag to compare
Don't compile for Pytorch 2.1 on CUDA 12.1 due to nvcc segfaults

v2.2.3.post1

17 Sep 23:40
Compare
Choose a tag to compare
Set block size to 64 x 64 for kvcache to avoid nvcc segfaults

v2.2.3

16 Sep 08:47
Compare
Choose a tag to compare
Bump to v2.2.3