0.3.10

fix some bug that popped up in a new environment

0.3.9

Full Changelog: 0.3.8...0.3.9

0.3.8

Full Changelog: 0.3.7...0.3.8

0.3.7

backwards cuda causal mask hack now relies on padding once and indexi…

0.3.6

final micro optimization to ring flash attn cuda forward

0.3.5

successful tests on 8 A40s on runpod.io for quarter million tokens

0.3.4

Full Changelog: 0.3.3...0.3.4

0.3.3

Full Changelog: 0.3.2...0.3.3

0.3.2

Full Changelog: 0.3.1...0.3.2

0.3.1

Full Changelog: 0.3.0...0.3.1