Overview
🤗 Diffusers is a modular, open-source library built on PyTorch for state-of-the-art diffusion models, enabling image, audio, and 3D structure generation. The library prioritizes usability, simplicity, and customizability, offering pretrained models and interchangeable noise schedulers. It supports both simple inference and custom training workflows. Diffusers includes diffusion pipelines for quick inference with minimal code, various noise schedulers to control diffusion speed and output quality, and pretrained models combinable with schedulers for building custom diffusion systems. It facilitates the creation of end-to-end diffusion-based applications, offering flexibility in model architecture and training techniques. The library is designed for developers to easily integrate and experiment with different diffusion models and algorithms.