1.6.1

add a wrapper for beta schedule functions to enforce zero terminal sn…

1.6.0

enforce zero terminal snr for cosine schedule

1.5.8

go back to using np.cov for fid score calculation

1.5.7

address torch cov in fid calculation https://github.com/lucidrains/de…

1.5.6

fixes for learned gaussian diffusion

1.5.5

release trainer for 1d

1.5.4

fix logits in classifier free guidance example, thanks to @youngwoo-yoon

1.5.3

fix min snr logic, thanks to @justinlovelace

1.5.2

adopt min snr for simple diffusion as well

1.5.1

remove all traces of p2 paper