Releases: lucidrains/phenaki-pytorch
Releases · lucidrains/phenaki-pytorch
0.5.0
Full Changelog: 0.4.2...0.5.0
0.4.2
update to working lfq
0.4.1
fix lookup of codes when using LFQ
0.4.0
start adopting finite scalar / lookup free quantization for vae
0.3.1
update ema
0.3.0
bet on new attention stabilization technique
0.2.0
adopt dual patchnorm design
0.1.1
make cvivit required
0.1.0
greatly simplify token critic training, as seeing good results elsewh…
0.0.72
token critic should be able to see the prime as well