# lucidrains/transfusion-pytorch

Pytorch implementation of Transfusion, "Predict the Next Token and Diffuse Images with One Multi-Modal Model", from MetaAI

Repository: https://github.com/lucidrains/transfusion-pytorch
Canonical: https://ross.abutalabs.com/products/transfusion-pytorch
Language: Python
License: MIT
License Family: permissive
Topics: artificial-intelligence, attention, deep-learning, multi-modal, transformers, flow-matching
Last push: 2026-08-25T16:15:57+00:00

## Health v2 (maintenance only)
Score: 75/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 56, longevity 52
- inputs: {"age_days": 740, "days_push": 8, "days_rel": 218, "gap_med": 0.0, "n_releases_24m": 127}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1395, forks 75 (observed 2026-08-28T04:04:36.460392+00:00)

## What it is
A PyTorch implementation of Transfusion, MetaAI's approach to predicting the next token and diffusing images with a single multi-modal model. It substitutes diffusion with flow matching and supports extending to any number of modalities.

## Use cases
- train a single model that generates both text and images
- implement next-token prediction combined with image diffusion
- experiment with flow matching for multimodal generation
- build a multimodal model handling text, images, and audio
- research multi-modality transformer architectures
- generate interleaved text and image sequences

## When to avoid
- you need a production-ready, fully trained multimodal model out of the box
- you want a simple inference-only text-to-image tool without training
- you are not working in PyTorch
- you need a maintained product with extensive documentation and support

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, llm-training, transformers
- domain: artificial-intelligence, deep-learning, machine-learning, large-language-models
- platform: python
- tags: multimodal, flow-matching, diffusion, text-to-image, research-implementation, pytorch

## Member repositories
- lucidrains/transfusion-pytorch (main) score 75

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:36.460392+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 2026-08-30T04:39:23.797870+00:00, confidence not recorded.
  - readme: https://github.com/lucidrains/transfusion-pytorch (fetched 2026-08-28T04:04:36.460392+00:00, sha 70f2f0d24c65)
- Data as of 2026-08-30T08:39:29.467469+00:00.