0.0.21

cleanup

0.0.19

cleanup

0.0.18

cleanup

0.0.17

one more linear in mlp attention

0.0.16

one more linear in mlp attention

0.0.15

do the same for dot product attention, start experiment

0.0.14

need to separately project xi and xj in mlp attention before first te…

0.0.12

fix a bug with mlp attention (attending on wrong dimension), also mak…

0.0.11

first pass at MLP attention, seems to be training, but not better tha…

0.0.9a

oops