Nathan Lambert

@patrickvonplaten this isn't associated wit ha specific model. My thoughts: * I feel like the `layers_per_block` arg is just really useful for people trying to build new stuff in the...

@patrickvonplaten I made this non breaking changes and more intuitive. The dance diffusion test confuses me because none of the real logic changes, it seems like a weight initialization issue,...

I want to make a colab comparing the prediction for training on one scheduler to start (make sure it works). Then we need to make it on more schedulers. Though...

@Pie31415 Another really useful thing would be to just verify the implementation from the original papers and links above. This is a pretty tricky port so I will do this...

FYI [here](https://github.com/huggingface/diffusers/pull/1121#issuecomment-1309265676) are some examples for DDPM butterfly generations from @bglick13 , we want to see similar results on DDIM than maybe merge this initial version.

Update for the diffusers team (@patrickvonplaten , @anton-l , @patil-suraj ). We updated DDIM now (promising results), and I'll add tests / fix merge issues this afternoon.

@patrickvonplaten this should be go to go. Now, this leaves only `DPMSolverMultistepSchedulerTest` with predict_epsilon rather than `prediction_type`, but I think it is okay until v-prediction expands in the library. Lots...

Added more deprecating across the board. I tried to address @patrickvonplaten's comment above, but would like a double check on that!

The code isn't as clear, but you can see some details on model parametrization in the SD 2.0 code [here](https://github.com/CompVis/stable-diffusion/blob/21f890f9da3cfbeaba8e2ac3c425ee9e998d5229/ldm/models/diffusion/dpm_solver/dpm_solver.py#L194). The option `parametrization = "v"` doesn't show up in the...

Closing this as the changes were integrated into #1505 and other earlier PRs. The added type of v-prediction for DDIM will be addressed in #1010.