 rocking
|
77ad12d24e
[AMD ROCm] Support variable length of page attention (#1431)
|
vor 1 Woche |
 Adam Louly
|
df1a744887
Add how to import FA3 to documentation (#1112)
|
vor 1 Monat |
 Michael Melesse
|
b518517cb8
[AMD] Triton Backend for ROCm (#1203)
|
vor 1 Monat |
 Tri Dao
|
7435839e3d
Update README for FA3
|
vor 2 Monaten |
|
Tri Dao
|
c555642172
Bump to v2.7.0
|
vor 2 Monaten |
 juejuezi
|
e371bea04f
feat: change minimal supported CUDA version to 11.7 (#1206)
|
vor 4 Monaten |
 Garrett Byrd
|
16025d8cc9
Clearer install instructions for CUDA and ROCm backends (#1147)
|
vor 5 Monaten |
|
Tri Dao
|
5ca83a9c71
Clean up softcapping bwd a bit
|
vor 6 Monaten |
|
rocking
|
d8f104e97a
Support AMD ROCm on FlashAttention 2 (#1010)
|
vor 6 Monaten |
|
Tri Dao
|
7f67966cc7
FA3 initial code release
|
vor 6 Monaten |
|
Tri Dao
|
da11d1b853
Bump v2.6.0
|
vor 6 Monaten |
|
Tri Dao
|
320fb59487
Update citation
|
vor 7 Monaten |
|
Tri Dao
|
2406f28805
Enable headdim 256 backward on consumer GPUs (Ampere, Ada)
|
vor 11 Monaten |
 Tao He
|
204c3c6d1b
Fixes an error in comment (#785)
|
vor 1 Jahr |
|
Tri Dao
|
54e80a3829
Implement page KV cache
|
vor 1 Jahr |
 Erich Schubert
|
99ea4baa1d
Typo in README (#760)
|
vor 1 Jahr |
|
Tri Dao
|
732654583c
Implement deterministic backward (thanks to Meituan)
|
vor 1 Jahr |
|
Tri Dao
|
50d144c906
Mention Alibi in README
|
vor 1 Jahr |
|
Tri Dao
|
7f31e7c16a
Bump to v2.3.2
|
vor 1 Jahr |
|
Tri Dao
|
5a83425442
Change constexpr int to constexpr static int
|
vor 1 Jahr |
|
Tri Dao
|
3a9fe7b0fa
Add change log
|
vor 1 Jahr |
|
Tri Dao
|
aa4fd2d166
Clarify that Windows is not supported right now
|
vor 1 Jahr |
|
Tri Dao
|
0c04943fa2
Require CUDA 11.6+, clean up setup.py
|
vor 1 Jahr |
 Jeffrey Quesnelle
|
1d817a8ffc
fix citation in README (#501)
|
vor 1 Jahr |
|
Tri Dao
|
45ba93cd96
Add newlines to README
|
vor 1 Jahr |
|
Tri Dao
|
9e5e8bc91e
Change causal mask to be aligned to bottom-right instead of top-left
|
vor 1 Jahr |
|
Tri Dao
|
d30f2e1cd5
Bump to v2.0.4
|
vor 1 Jahr |
 Ian Timmis
|
cbf982afa5
README syntax highlighting (#365)
|
vor 1 Jahr |
|
Tri Dao
|
d1a3b52f17
Add instruction about limiting number of ninja jobs
|
vor 1 Jahr |
|
Tri Dao
|
b4cc152e97
Make sure dout is contiguous
|
vor 1 Jahr |
