I recently completed my PhD at East China Normal University (ECNU) in the School of Data Science and Engineering, where I was advised by Ming Gao and Xiang Li.

My current research interests focus on the evaluation and analysis of large language models.

Industry Experience

I have gained substantial industry experience through several internships:

  • Alibaba (2021): Interned with the NLP group, gaining insights into industry-level NLP applications.
  • Meituan (2022): Continued my journey with the NLP group, further honing my skills.
  • Microsoft Xiaoice (Jan-Jun 2023): Focused on enhancing the reasoning capabilities of large language models.
  • Ant Group (Jun-Oct 2023): Researched techniques to mitigate hallucinations in large language models.
  • Shanghai AI Lab (Oct 2023-May 2024): Concentrated on developing AI Agents for operating systems.
  • Meituan (Since Jul 2024): Full-time researcher.

Join My Team

I'm currently on the lookout for interns interested in evaluation and analysis of large language models. If you're ready to dive into this exciting field, shoot me an email at hccngu@163.com. Let's explore the future of AI together!

Personal Interests

When I'm not doing NLP, I like to work out, play guitar, and read.


Updates

  • September 2024: We published a paper titled "Length Desensitization in Directed Preference Optimization", addressing the length bias in DPO. We hope it garners significant interest!
  • March 2024: We released a self-improving embodied conversational agent, OS-Copilot, seamlessly integrated into the operating system to automate our daily tasks. Friends who are interested can try it out~
  • March 2024: In March, we published a survey paper titled "A Survey of Neural Code Intelligence: Paradigms, Advances, and Beyond". We hope it attracts a lot of interest and attention!
  • March 2024: 2 papers were accepted at COLING 2024. Check out the publications page for more info!
  • October 2023: 1 paper accepted at EMNLP 2023. Check out publications page for more info!
  • May 2023: We've created an open-source project, Viscacha, aiming to release a comprehensive Chinese information extraction dataset. We welcome everyone to pay close attention to it~
  • May 2023: 1 paper accepted at ACL 2023. Check out publications page for more info!
  • February 2023: Our team, DataIsPower, won the runner-up prize in the pre-trained language model application tuning algorithm category at the Guangdong-Hong Kong-Macau Greater Bay Area International Algorithm Competition, with a prize money of ¥200,000. We are so grateful!!!
  • January 2023: 1 paper accepted at DASFAA 2023. Check out publications page for more info! Also, I attended DASFAA 2023 in person :)