Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
MMaDA - Open-Sourced Multimodal Large Diffusion Language Models (dLLMs with block diffusion, mixed-CoT, unified RL)
| Date | Stars |
|---|---|
| 2026-07-31 | 1661 |
| 2026-08-04 | 1661 |
| 2026-08-06 | 1661 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
<div align="center">
<br>
<img src="assets/title.png" width="166">
<h3>Multimodal Large Diffusion Language Models (NeurIPS 2025)</h3></div>
<p align="center">
<a href="https://arxiv.org/abs/2505.15809">
<img
src="https://img.shields.io/badge/MMaDA-Paper-red?logo=arxiv&logoColor=red"
alt="MMaDA Paper on arXiv"
/>
</a>
<a href="https://huggingface.co/spaces/Gen-Verse/MMaDA">
<img
src="https://img.shields.io/badge/MMaDA%20Demo-Hugging%20Face%20Space-blue?logo=huggingface&logoColor=blue"
alt="MMaDA on Hugging Face"
/>
</a>
<a href="https://huggingface.co/Gen-Verse/MMaDA-8B-Base">
<img
src="https://img.shields.io/badge/MMaDA--8B--Base-Hugging%20Face%20Model-orange?logo=huggingface&logoColor=yellow"
alt="MMaDA on Hugging Face"
/>
</a>
<a href="https://huggingface.co/Gen-Verse/MMaDA-8B-MixCoT">
<img
src="https://img.shields.io/badge/MMaDA--8B--MixCoT-Hugging%20Face%20Model-orange?logo=huggingface&logoColor=yellow"
alt="MMaDA on Hugging Face"
/>
</a>
</a>
<a href="https://huggingface.co/tyfeld/MMaDA-Parallel-M">
<img
src="https://img.shields.io/badge/MMaDA--Parallel-Hugging%20Face%20Model-orange?logo=huggingface&logoColor=yellow"
alt="MMaDA on Hugging Face"
/>
</a>
</a>
<a href="https://github.com/Gen-Verse/dLLM-RL">
<img
src="https://img.shields.io/badge/RL%20Infra-TraceRL%20Framework-pink?logo=huggingface&logoColor=brown"
alt="MMaDA on Hugging Face"
/>
</a>
</p>
## 🌌 Introduction
MMaDA is a new family of **multimodal diffusion foundation models** designed to achieve superior performance across diverse domains such as textual reasoning, multimodal understanding, and text-to-image generation. MMaDA is distinguished by three key innovations:
1. MMaDA adopts a **unified diffusion architecture** with a shared probabilistic formulation and a modality-agnostic design, eliminating the need for modality-specific components.
2. MMaDA introduces a **mixed long chain-of-thought (CoT) fine-tuning** strategy that curates a unified CoT format across modalities.
3. MMaDA adopts a unified policy-gradient-based RL algorithm, which we call **UniGRPO**, tailored for diffusion foundation models. Utilizing diversified reward modeling, **UniGRPO** unifies post-training across both reasoning and generation tasks, ensuring consistent performance improvements.
<div align="center" style="width: 600px; margin: auto;">
<img src="assets/showcase0.8.gif" alt="MMaDA decoding demo" width="550" />
<p style="font-style: italic; font-size: 14px; color: #555; margin-top: 6px;">
MMaDA's decoding demo. This video showcases how a diffusion foundation model generates text and image.<br>
The "Text Generation" part uses a semi-autoregressive sampling method, while the "Multimodal Generation" part adopts non-autoregressive diffusion denoising.
</p>
</div>
<!--
## Decoding Demo
We demonstrate the decoding process of MMaDA with a teaser video to show how a diffusion model generates text and image. The "Text Generation" part adopts a "semi-autoregressive" sampling method and the "MultiModal Generation" part adopts a non-autoregressive sampling method which is purely diffusion denoising.
<!-- <div style="display: flex; justify-content: center; flex-wrap: wrap;">
<img src="assets/showcase0.8.gif" style="width: 90%" />
</div> -->
## 📰 Latest Updates
* **[2025-11-13]** We release **[MMaDA-Parallel](https://arxiv.org/abs/2511.09611)**, a new class of multimodal dLLMs for Thinking-Aware Image Editing and Generation.
* **[2025-09-09]** We open source a comprehensive RL framework for dLLMs, **[dLLM-RL](https://github.com/Gen-Verse/dLLM-RL)** with released SOTA instruct and long-CoT models **[TraDo-8B-Instruct](https://huggingface.co/Gen-Verse/TraDo-8B-Instruct), [TraDo-4B-Instruct](https://huggingface.co/Gen-Verse/TraDo-4B-Instruct), and [TraDo-8B-Thinking](https://huggingExcerpt of 10,898 characters
Read on GitHub
31
Ling Yang
24
1
1
1
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:8e8325d6d9653685, topic:diffusion-models