How do I reverse rearrange with ellipsis composed? e.g. "b ... -> b (...)"

[drscotthawley]

Francois Fleuret suggested that it'd be nice if there were an function whereby einops.rearrange could be easily 'inverted' i.e. undone or "transformed back".

I replied that a "wrapper" function or class shouldn't be too hard, and wrote one at the following link, which includes a few examples:
https://gist.github.com/drscotthawley/81865a5c5e729b769486efb9c3f2249d

Whether such a functionality just remains as an "external wrapper" that users can add-on, or somehow gets added to the einops codebase (maybe not as this class, but something similar) is up to you, but wanted to share it here to add to the conversation!

One way to include it into the existing codebase could be to rename einops.rearrange to einops._rearrange and then have the new einops.rearrange =RearrangeWrapper() where RearrangeWrapper's sub-methods call _rearrange (as shown in my gist example). If that sounds interesting then I could submit a PR.

[arogozhnikov]

implicit storage of information for reversal has flaws, and you already figured some of them

I've implemented this:
https://github.com/arogozhnikov/einops/blob/master/docs/4-pack-and-unpack.ipynb

• more restricted, but addresses a scenario originally asked in that discussion.
• can revert more than one tensor
• inverse can have different pattern compared to forward

[lucidrains]

@arogozhnikov love it! https://github.com/lucidrains/audiolm-pytorch/blob/main/audiolm_pytorch/hubert_kmeans.py#L46-L55

[arogozhnikov]

👍 that's a good example @lucidrains

PS: now I remember I actually encountered a number of cases when I wanted to apply sklearn models to tensors not matrices.