(1) We released the 50 diffusion steps model (instead of 1000 steps) which runs 20X faster with comparable results. (2) Calling CLIP just once and caching the result runs 2X faster for all models.

Recently, diffusion models, exemplified by the denoising diffusion probabilistic model (DDPM), have garnered significant attention for their remarkable performance and straightforward architecture.

Upgrades to python 3.9, and adds support for NVIDIA 50xx series (Blackwell) GPUs. Adds support for AMD GPUs, as well as integrated GPUs on Windows.