0.0.22

add the general logic for the zero shot mask weighting of loss

0.0.21

another round of feedbacks, update todos

0.0.20

handle the training of multiple concepts; save the inference part for…

0.0.19

prepare two more functions that make life easy

0.0.18

tweak

0.0.17

more feedback; authors used uncentered covariance matrix, detail in a…

0.0.16

cleanup for yet another round of review

0.0.15

clarified with author that superclass output is initialized using the…

0.0.14

lower the effective learning rate of the concept embedding through th…

0.0.12

account for more feedback