0.9.0

latest transformer research shows bias is not needed and could be har…

0.8.13

remove init, because empiricism

0.8.12

remove global linear attention feature

0.8.11

small bug with init

0.8.10

fix tests for pytorch 1.8.1 marking inplace operations as unsafe

0.8.9

be able to have fine control over the dimensions per type in the fina…

0.8.8

Merge pull request #13 from lucidrains/vector-gating

allow one to set their own path to cache directory with environmental…

0.8.7

Merge pull request #12 from lucidrains/vector-gating

vector gating

0.8.6

allow for customization of dimension per typeN by passing in hidden_f…

0.8.4

throw EGNN into the same file as main SE3 Transformer code, and add o…