Skip to content

Releases: lucidrains/MEGABYTE-pytorch

0.3.3

09 Sep 23:08
Compare
Choose a tag to compare

What's Changed

  • chore: update flash attention config by @eegli in #18

New Contributors

  • @eegli made their first contribution in #18

Full Changelog: 0.3.2...0.3.3

0.3.2

07 Sep 12:42
Compare
Choose a tag to compare

Full Changelog: 0.3.1...0.3.2

0.3.1

07 Sep 12:09
Compare
Choose a tag to compare

Full Changelog: 0.3.0...0.3.1

0.3.0

03 May 02:13
Compare
Choose a tag to compare

Full Changelog: 0.2.1...0.3.0

0.2.1

15 Jun 20:12
Compare
Choose a tag to compare
make sure it supports greater than 2 hierarchies

0.2.0

15 Jun 19:38
Compare
Choose a tag to compare
move closer to what the paper did, with local and global token embedd…

…ings not shared

0.1.7

15 Jun 18:13
Compare
Choose a tag to compare
switch to rotary embeddings, as they did in the paper

0.1.6

14 Jun 16:50
Compare
Choose a tag to compare
evidence is emerging that decoders generate implicit absolute and rel…

…ative positions without help

0.1.4

31 May 01:01
Compare
Choose a tag to compare
address https://github.com/lucidrains/MEGABYTE-pytorch/issues/10

0.1.2

29 May 17:42
Compare
Choose a tag to compare
learn on the very first start token