Machine Learning Library
33.4k 2026-04-18
huggingface/diffusers
A modular PyTorch library for state-of-the-art diffusion models, enabling easy generation of images, audio, and more.
Core Features
State-of-the-art diffusion pipelines for inference with minimal code.
Interchangeable noise schedulers for diverse generation speeds and output quality.
Pretrained models as building blocks for custom end-to-end diffusion systems.
Designed with a focus on usability, simplicity, and customizability.
Supports generation of images, audio, and 3D structures of molecules.
Quick Start
pip install --upgrade diffusers[torch]Detailed Introduction
🤗 Diffusers is a comprehensive PyTorch library designed to democratize access to state-of-the-art diffusion models. It provides a modular toolbox for both rapid inference and custom model training, emphasizing ease of use, clear design, and high customizability. The library offers pre-built pipelines, flexible schedulers, and foundational models, empowering developers to generate high-quality images, audio, and even molecular structures with just a few lines of code. This project fosters innovation in generative AI by providing a robust and flexible platform for researchers and practitioners alike.