Skip to content

Understanding SoftclipTransform #10

Answered by francois-rozet
arnauqb asked this question in Q&A

You must be logged in to vote

Hello @arnauqb 👋

transforms is the sequence of transformations f 0 , f 1 , f n from the feature space towards the latent space. In order words, z = f n f 1 f 0 ( x ) . When sampling, the order of the transformations is reversed, meaning that x = f 0 1 f 1 1 f n 1 ( z ) .

Therefore, if you want your samples to be bounded, the inverse of the first transformation f 0 1 should be the SoftclipTransform, but, in your snippet, you made the last transformation f n the SoftclipTransform.

The easiest solution is to insert the inverse SoftclipTransform at the front of the transforms list.

class MyFlow(zuko.flows.MAF):
    def __init__(

Replies: 1 comment 2 replies

You must be logged in to vote
2 replies
@felixdivo

@francois-rozet

Answer selected by arnauqb
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Category
Q&A
Labels
None yet
3 participants
Converted from issue

This discussion was converted from issue #9 on January 27, 2023 12:35.