Use K. Crowson's denoising model to save VRAM, improve generations

[afiaka87]

Katherine just trained a new checkpoint meant to be used in tandem with existing unconditional checkpoints from Open AI. I intend to add this functionality here eventually but am going to be very busy for the next week or so.

[fpinnola]

I am running into Pytorch oom issue using RTX 3070 with 8GB of dedicated VRAM trying to run with image_size>128. Are there any solutions without this implementation? Otherwise this implementation may help.