Skip to content

Releases: lucidrains/native-sparse-attention-pytorch

0.0.44

21 Feb 18:15
Compare
Choose a tag to compare
when doing interpolation of importance score, remask to 0 for illegal…

0.0.43

21 Feb 17:31
Compare
Choose a tag to compare
default to one mem kv for compressed attn

0.0.42

21 Feb 17:12
Compare
Choose a tag to compare

Full Changelog: 0.0.41...0.0.42

0.0.41

21 Feb 16:21
Compare
Choose a tag to compare
ready to be compared with full attention.

0.0.40

21 Feb 15:54
Compare
Choose a tag to compare
oops

0.0.39

21 Feb 15:48
Compare
Choose a tag to compare
do the differential topk gating in a more suboptimal way, but accommo…

0.0.38

21 Feb 15:22
Compare
Choose a tag to compare

Full Changelog: 0.0.36...0.0.38

0.0.37

21 Feb 15:21
Compare
Choose a tag to compare
account for learned memory key values in flex compress mask, also cle…

0.0.36

21 Feb 14:57
Compare
Choose a tag to compare
refactor compressed pathway with gqa

0.0.35

21 Feb 14:44
Compare
Choose a tag to compare
deviate from the paper and allow for interpolation of the compressed …