Yufan Dang

Yufan Dang

Junior undergrad at Tsinghua University, major in software & MPS

My name is Yufan Dang (党余凡). I’m a undergrad at Tsinghua university(2021-2025) , double major in MPS(math and physics science) & Software. Now as a research intern at THUNLP, I’m learning around LLM & agent, and engaged in ChatDev as a main contributor.

I have a passion for sharing tech blogs on my blog site, where I discuss the courses I am currently enrolled in and share insights into cutting-edge papers and news. Aside from research and tons of math and physics, I love volleyball 🏐/ gym workingout 🏋️‍♀️/ tennis 🎾/ hiking 🏞️ and I’m a violin enthusiast.

Feel free to reach me : dangyf2003@gmail.com.

Research Interest:

LLM agents 🤖 I'm focused on developing LLM-based autonomous agents capable of effectively and efficiently tackling multi-step and complex s. Previously, my focus was primarily on software development, but now I'm eager to explore broader applications. Particularly, I'm intrigued by the concept of LLM-based multi-agent systems。
Learning Theory / Reinforcement Learning / Machine Learning 🎨 Learning ability stands as the true differentiator between intelligence and any other attribute. There is ample room for exploration in this domain, particularly in: - **Learning Theory / Machine Learning**: Continuously advancing our understanding of how learning occurs and devising strategies to optimize learning processes. - **Reinforcement Learning**: Harnessing the power of reinforcement learning algorithms to enable agents to learn from interactions with their environment and make informed decisions.

🍺 News

  • [2024.6.14] Our two papers related with multi-agent collaboration from the pserspective of intra-team and cross-team organization has been released on arXiv first.
  • [2024.5.16] Our two papers have been accepted to ACL 2024, main conference 🥳. Thanks to my best mentor and all co-authors!
  • [2024.5.6] Following our prior work on ECL, we’re excited to announce the publication of our latest endeavor, Iterative Experience Refinement of Software-Developing Agents, available on arXiv:2405.04219, 2024.
  • [2024.1.25] Code and data around Experiential Co-Learning of Software-Developing Agents has released in ECL.
  • [2023.12.29] Our recent work Experiential Co-Learning of Software-Developing Agents has released on arXiv:2307.07924, 2023.

💡 Publication

Multi-Agent Software Development through Cross-Team Collaboration. InarXiv 2406.08979.
Zhuoyun Du, Chen Qian, Wei Liu, Zihao Xie, Yifei Wang, Yufan Dang, Weize Chen, Cheng Yang.

Scaling Large-Language-Model-based Multi-Agent Collaboration. In arXiv 2406.07155.
Chen Qian$^{†}$, Zihao Xie$^{†}$, Yifei Wang$^{†}$, Wei Liu, Yufan Dang, Zhuoyun Du, Weize Chen, Cheng Yang, Zhiyuan Liu, Maosong Sun.

Experiential Co-Learning of Software-Developing Agents. ACL 2024 main conference.
Chen Qian$^{†}$, Yufan Dang$^{†}$, Jiahao Li, Wei Liu, Weize Chen, Cheng Yang, Zhiyuan Liu, Maosong Sun.

Communicative Agents for Software Development. ACL 2024 main conference.
Chen Qian, Xin Cong, Wei Liu, Cheng Yang, Weize Chen, Yusheng Su, Yufan Dang , Jiahao Li, Juyuan Xu, Dahai Li, Zhiyuan Liu, Maosong Sun.

Iterative Experience Refinement of Software-Developing Agents. In arXiv:2405.04219, 2024.
Chen Qian$^{†}$, Jiahao Li$^{†}$, Yufan Dang, Wei Liu, YiFei Wang, Zihao Xie, Weize Chen, Cheng Yang, Yingli Zhang, Zhiyuan Liu, Maosong Sun.

$^{†}$: Equal Contribution

🐣 Research Experience

THUNLP Natural Language Processing Lab at Tsinghua University
Research Intern, Jul 2023 – Present

  • Participating in the ChatDev project: optimizing dialogue-related mechanisms of the AI, conducting multiple testing, improving the project’s frontend, and carrying out daily operational maintenance

  • Currently involved in agent-oriented co-evolution research, accumulating experience and facilitating the mutual evolution of capabilities

Development and Application of Driving Safety AI Detection System
Student Research Training Program in Autonomous Department, Tsinghua University, Nov 2022 – Jun 2023

  • Focus on multimodal perception based on deep learning in self-driving
  • Researched the state-of-the-art single terminal and multimodal perception network and got to know techniques about computer vision
  • Used Pytorch to replicate advanced algorithms in communication-efficient collaborative perception and reproduced consistent experimental results