Releases: lucidrains/g-mlp-pytorch
Releases · lucidrains/g-mlp-pytorch
0.1.5
0.1.4
correctly attribute
0.1.2
deploy multiheadedness
0.1.1
give absolute positional parameters to circulant matrix
0.1.0
release ability to use circulant matrix
0.0.18
nonsquare patch sizes too
0.0.17
allow for non-square images
0.0.16
correct tiny attention
0.0.15
add ability to add nonlinearity to spatial gate
0.0.14
small initial epsilon for SGU weights