0.0.11

query chunking feature for sparse 3dna

0.0.9

add first stability improvement from cogview paper

0.0.8

sparse 3dna without padding the queries too

0.0.7

batch process video frame decoding

0.0.6

pass keyword arguments prepended with vq_ to VectorQuantize instantia…

…tion, for features from improved VQGan paper as well as orthogonal regularization

0.0.5

add adaptive weight from vqgan paper

0.0.4

do axial positional embedding manually

0.0.3

avoid softmax and torch multinomial

0.0.2

make sure queries are not padded

0.0.1

complete video generation code