0.5.1

allow for mean pool with efficient version too

0.5.0

offer a way to use mean pooling of last layer

0.4.0

fix the dimension per head to be independent of dim and heads, to mak…

0.3.0

simplify mlp head

0.2.7

remove float(-inf) as masking value

0.2.6

allow for training on different image sizes, provided images are smal…

0.2.5

more efficient

0.2.4

remove dropout from last linear to logits

0.2.3

assert minimum number of patches

0.2.2

dropouts are more specific and aggressive in the paper, thanks for le…