Nerogar
Nerogar
> For masking, I currently I find it difficult to reason what the result of a auto-mask run will be as there is no preview or easy way to test...
Please try to update. The new diffusers version should fix this.
the default `runwayml/stable-diffusion-v1-5` and `runwayml/stable-diffusion-inpainting` repositories for SD1.5 don't exist anymore. You can still train this model if you have the checkpoint or diffusers model downloaded. Just select the downloaded...
I found a completely unrelated bug while analyzing this issue, but couldn't reproduce the reported bug. There were a lot of changes since this report, so it's possible it was...
Please update and try again, this should be fixed now
Please try to update. It should be fixed now.
Just FYI: I think this is a great idea that can be really useful. But I see a small issue with the data loader and I haven't thought of a...
Adding it to the dataloader is pretty simple. But then you still need to find some way to select useful timesteps. And I think that can be pretty complicated. -...
Do you think there is a use case for using different units for `save_after_unit` and `skip_save_unit`? If not, what do you think about adding only a new value so we...
> Further noticed that the datatype of save_every is float, but the default setting is an int. Did not change that, because I'm not sure this wasn't intentional and might...