---
title: README
emoji: 📊
colorFrom: yellow
colorTo: purple
sdk: static
pinned: false
---

# 🚀 MAIR Lab @ HUST
**🌐 Multimodal Artificial Intelligence Research Lab**
**🏛️ School of Computer Science and Technology, Huazhong University of Science and Technology**

MAIR Lab is a multimodal artificial intelligence research group led by **Prof. Zhiyuan Ma** at the School of Computer Science and Technology, Huazhong University of Science and Technology.

We focus on frontier research topics including **multimodal large language models**, **unified understanding and generation paradigms**, **controllable multimodal generation**, **world models**, and **AI for Science**. Our long-term goal is to explore the fundamental capabilities and boundaries of next-generation intelligent systems.


## 🔬 Research Interests
Our research spans several closely related directions:

* **🧠 Multimodal Large Language Models**
  Building foundation models that can perceive, reason, generate, and interact across modalities.
* **🔄 Unified Understanding and Generation**
  Exploring new paradigms that bridge multimodal perception, content generation, reasoning, and decision-making.
* **🎨 Controllable Multimodal Generation**
  Developing methods for controllable, reliable, and human-aligned generation of images, videos, and other modalities.
* **🌍 World Models**
  Investigating how intelligent systems can model, predict, and intervene in the physical and digital world.
* **🧪 AI for Science**
  Applying artificial intelligence to accelerate scientific discovery and solve challenging scientific problems.

## 💡 Our Vision
We believe that the next generation of AI systems should go beyond passive perception and isolated generation. They should be able to understand complex multimodal environments, reason over structured knowledge, generate controllable content, and interact with the world in a reliable and generalizable manner.

At MAIR Lab, we aim to develop models, algorithms, and systems that push multimodal AI toward more general, trustworthy, and impactful intelligence.

## 🛠️ Open Source
This huggingface organization hosts our research projects, open-source models, datasets, demos, and related resources. More code-bases can also be found at our Github organization: https://github.com/MAIR-Lab-HUST

We are committed to reproducible and accessible research. We hope our open-source efforts can help the community better understand, evaluate, and extend our work.

## 🌱 Join Us
We are always looking for **self-motivated researchers** who are interested in conducting interesting and impactful research with us.

Students and collaborators with interests in multimodal AI, foundation models, generative models, world models, agents, or AI for Science are welcome to reach out.

We especially value:

* 🔥 Strong research curiosity
* 🚀 Self-motivation and long-term commitment
* 💻 Solid engineering and implementation ability
* 🧩 Clear thinking and independent problem-solving
* 🌌 Willingness to explore ambitious and open-ended research problems

## 📬 Contact
For collaboration, internship, or student opportunities, please feel free to contact us.

* **👤 Contact:** Prof. Zhiyuan Ma
* **📧 Email:** [mzyth@hust.edu.cn](mailto:mzyth@hust.edu.cn)
* **🏷️ Lab:** MAIR Lab
* **🏛️ Institution:** Huazhong University of Science and Technology
* **💻 School:** School of Computer Science and Technology