1.6.7

helper method and accessor off soundstream trainer

1.6.6

patch

1.6.5

sanity check, clearly demonstrate tokenization of audio ability using…

1.6.4

address https://github.com/lucidrains/audiolm-pytorch/issues/242

1.6.3

allow for local attention to attend to nothing

1.6.2

increase default codebook size

1.6.1

allow for grouped residual LFQ

1.6.0

integrate residual lookup free quantization

1.5.7

address padding in acoustic tokens when sampling from fine transforme…

1.5.6

address https://github.com/lucidrains/audiolm-pytorch/issues/235