I am a Software Engineering student at Fudan University. My current research focuses on trustworthy evaluation of large language models, medical and real-world benchmarks, and scientific-intelligence systems for evaluating novelty and reasoning quality.
I am especially interested in evaluation pipelines that are more robust to contamination, closer to real use cases, and better aligned with the kinds of claims we actually make about modern models.
OpenNovelty studies whether language models can judge the novelty of open-ended ideas rather than only solve fixed-answer tasks. It introduces an open-domain benchmark for comparing LLM judgments of novelty ...
LLMEval-Fair proposes a dynamic evaluation framework that samples unseen test sets from a large question bank, combines contamination-resistant curation with anti-cheating design, and studies almost 50 front...
LLMEval-Med is a physician-validated clinical benchmark built from real-world electronic health records and expert-designed scenarios. It targets the weaknesses of existing medical LLM evaluations by moving ...