Hacker News Reader

Would be much better if these changes were a branch on the starting repo rather than entire refactors with many changes.

There are pros and cons to each of these Stable Diffusion hacks (usually time/memory/usability tradeoffs) so it's not correct to merge it back.

And in this case it seems like mostly an implementation detail - it could just be a flag to enable, or enabled whenever it detects running on colab.

Which is why it would be a branch, no?

did you even peek at the link ? There's a PR on diffusers, and it's mentioned on the front page https://github.com/huggingface/diffusers/pull/532#issuecomme...