My name is Zhihan Zhang (张智涵). I am a fourth-year Ph.D. student in Computer Science and Engineering at the University of Notre Dame, advised by Dr. Meng Jiang. I am currently a member of Dr. Jiang’s Data Mining towards Decision Making (DM²) Lab. Prior to my Ph.D. career, I received my Bachelor’s degree in Computer Science at Peking University, where I had the fortune of working with Dr. Yunfang Wu and Dr. Xu Sun. My research is mainly related to instructing LLMs to perform tasks for human users, such as instruction tuning, evaluation of instruction-following capabilities, etc.

For my past education and internship experience, please refer to Experience. For the full list of my publications, please refer to Publications or check my Google Scholar page.

News

• Oct 2024: Several co-authored preprints came out! Check out our latest works:
  1. A state-of-the-art MLLM in the domain of text-rich images, trained on a multi-image text-rich corpus with 1M instruction data (link)
  2. A new pipeline using process-supervised verifiers to make LLMs identify and revise their own reasoning errors (link)
  3. A novel method to build autonomous LLM agents to solve software engineering tasks with a repository-level code graph (link)
  4. A new benchmark that evaluates MLLMs’ capabilities in multi-chart understanding and reasoning (link)

• Sept 2024: 3 papers got accepted by EMNLP 2024! These include two main conference papers: a first-authored paper working on LMs & Math Reasoning and a co-authored paper working on LMs self-correcting their mistakes. And one findings paper: a co-authored paper working on evaluating instruction-following capabilities of LMs.

• Jun 2024: Several new preprints came out! Check out our latest works on math & multimodal reasoning:
  • A first-authored paper incorporated reflection into LLM fine-tuning and improved their deep reasoning abilities.
  • A co-authored paper utilized backward verification methods to make LLMs identify and correct their own reasoning errors.
  • A co-authored paper released a benchmark on LLMs’ math abilities in multi-turn interactions and open-ended generation.
  • A co-authored paper improved visual understanding capacity of MLLMs for more accurate multimodal math reasoning.

• Jun 2024: I am joining Amazon as a full-time research intern in summer 2024.

• May 2024: One first-authored paper is accepted to ACL 2024 main conference! We studied the challenges in cross-lingual instruction-tuning and proposed a simple yet effective method to improve LLMs’ proficiency in low-resource languages. Code is available on Github.

• Feb 2024: I am joining Tecent America as a full-time research intern in spring 2024.

Contact

• Email: zzhang23 [at] nd.edu
• Office: 355 Fitzpatrick Hall of Engineering
• Location: University of Notre Dame, Notre Dame, IN 46556