Open-Endedness Bench: Measuring Epistemic Process from Agent Records
Under review at ICLR 2027
I'm an undergraduate student majoring in Mathematics and Computer Science at the University of Michigan and Electrical & Computer Engineering at Shanghai Jiao Tong University. My interests lie in AI for Scientific Research, especially Autoresearch, where AI agents carry out the research loop themselves, from forming hypotheses to running experiments, and LLM Evaluation, which measures whether what these agents claim is actually earned.
My recent work, Open-Endedness Bench, evaluates research agents by reading their full records and checking each conclusion against what actually ran, rather than grading only the final score. I want to keep building evaluations like this, so that progress in automated research is something we can trust.
I am always glad to talk about research. Feel free to reach out by email.
Under review at ICLR 2027
Under review at CHI 2027
arXiv preprint, 2026
Full list on Google Scholar