# jannerm/diffuser

Code for the paper "Planning with Diffusion for Flexible Behavior Synthesis"

Repository: https://github.com/jannerm/diffuser
Canonical: https://ross.abutalabs.com/products/diffuser
Homepage: https://diffusion-planning.github.io
Language: Python
License: MIT
License Family: permissive
Last push: 2024-07-18T05:31:52+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": 1584, "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 1304, forks 201 (observed 2026-08-28T04:04:18.524879+00:00)

## What it is
Official code for the ICML 2022 paper 'Planning with Diffusion for Flexible Behavior Synthesis', which plans by denoising trajectories with diffusion models. It supports training diffusion models and value-function guided sampling on D4RL locomotion, maze2d, and Kuka block-stacking environments.

## Use cases
- train diffusion models for trajectory planning in RL environments
- plan behavior with value-function guided diffusion sampling
- reach goals via inpainting in Maze2D environments
- run block-stacking experiments with a Kuka arm
- sample from pretrained diffusion planners in a Colab notebook
- reproduce results from the Diffuser paper

## When to choose
- you want to experiment with diffusion-based planning for control and RL
- you need a reference implementation of the Diffuser paper on D4RL benchmarks
- you want to condition trajectory generation on rewards or goal constraints

## When to avoid
- you need a production-ready RL training framework with active support
- you want a general-purpose diffusion model library rather than RL planning code
- you need environments or tooling outside D4RL-style MuJoCo tasks

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, simulation
- domain: reinforcement-learning, machine-learning, artificial-intelligence
- platform: python
- tags: diffusion-models, reinforcement-learning, trajectory-planning, d4rl, research-code, icml-2022, gpu

## Member repositories
- jannerm/diffuser (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:18.524879+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 2026-08-30T04:51:08.248313+00:00, confidence not recorded.
  - readme: https://github.com/jannerm/diffuser (fetched 2026-08-28T04:04:18.524879+00:00, sha 810a0021fa1c)
  - homepage: https://diffusion-planning.github.io (fetched 2026-08-29T12:09:08.471588+00:00, sha 192c44a35da6)
- Data as of 2026-08-30T08:39:29.467469+00:00.