1.12.12

Full Changelog: 1.12.11...1.12.12

1.12.11

address https://github.com/lucidrains/vector-quantize-pytorch/issues/101

1.12.10

just do lfq in f32 for now

1.12.9

only use the frac of per sample probs taken for the codebook entropy …

1.12.8

disable autocast within lfq and fsq

1.12.7

able to use only a fraction of the probs for per sample entropy regul…

1.12.6

per sample entropy needs to account for variable length sequences as …

1.12.5

address https://github.com/lucidrains/vector-quantize-pytorch/issues/96

1.12.4

1.12.4

1.12.1

allow for freezing codebooks for residual VQ forward - https://github…