# openai/improved-diffusion

Release for Improved Denoising Diffusion Probabilistic Models

Repository: https://github.com/openai/improved-diffusion
Canonical: https://ross.abutalabs.com/products/improved-diffusion
Language: Python
License: MIT
License Family: permissive
Last push: 2024-07-18T02:45:29+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2032, "days_push": 776, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3844, forks 547 (observed 2026-08-28T04:08:24.057507+00:00)

## What it is
The official codebase for OpenAI's Improved Denoising Diffusion Probabilistic Models paper, providing a Python package for training and sampling diffusion-based image generation models. It includes training scripts, dataset preparation utilities for ImageNet, LSUN, and CIFAR-10, and support for distributed GPU training.

## Use cases
- train a diffusion model for image generation
- reproduce improved DDPM research results
- generate images from a trained denoising diffusion model
- train a class-conditional image generation model on my own dataset
- experiment with cosine noise schedules and learned sigmas in diffusion models
- run distributed training of generative image models on multiple GPUs

## When to choose
- you want to train or fine-tune diffusion models from the improved DDPM paper
- you need a research-grade reference implementation of DDPM training and sampling
- you want to experiment with diffusion hyperparameters like noise schedules and learned sigmas

## When to avoid
- you need a production-ready image generation API or pretrained model serving
- you want modern diffusion frameworks with schedulers like Diffusers
- you need text-to-image or latent diffusion capabilities

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, image-processing, llm-training
- domain: deep-learning, machine-learning, image-processing, artificial-intelligence
- platform: python
- tags: diffusion-models, ddpm, generative-models, image-generation, research-code, openai, gpu, linux

## Member repositories
- openai/improved-diffusion (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:24.057507+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:26:12.537725+00:00, confidence not recorded.
  - readme: https://github.com/openai/improved-diffusion (fetched 2026-08-28T04:08:24.057507+00:00, sha 19c59486dbde)
- Data as of 2026-08-30T08:39:29.467469+00:00.