Yifei is a (𝓍-𝟤𝟢𝟤𝟥)-Year PhD at University of Pennsylvania advised by Prof. Mark Yatskar, affiliated with Penn NLP. In the past, he has also worked with Prof. Lyle Ungar, Prof. João Sedoc, and Prof. Chris Callison-Burch.
His research spans LLMs and multimodality, with a primary focus on:
(1) Scalable Evaluation and Reward
(2) Deep Research Agents
(3) Applications and Utilities in Expert-level AI and Scientific Discovery
Selected Works
ResearchQA: Evaluating Scholarly Question Answering at Scale Across 75 Fields with Survey-Mined Questions and Rubrics
Li S. Yifei *, Allen Chang *, Chaitanya Malaviya, Mark Yatskar
ResearchQA proposes a scalable query-rubric evaluation framework for scholarly long-form question answering via data synthesis, benchmarking frontier LLM systems including deep research agents by ensembled LLM judges. By distilling the research expertise from survey papers, it yields 21K queries and 160K specific rubric items across 75 research fields, which are validated by 31 Ph.D. annotators.