Reproduce “BioProBench: A Corpus and Benchmark for Biological Protocol Reasoning in Autonomous Science” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yuyang Liu, Liuzhenghao Lv, Xiancheng Zhang, Jingya Wang Li Yuan, Yonghong Tian
BioProBench tests whether models can answer, order, correct, reason about, and reconstruct biological protocols from provided information. Built from 22,413 protocols across five repositories, it contains 523,784 instances. ProAgent retrieves relevant protocol context and raises Claude-3.7-Sonnet’s generation Step Recall from 42.8% to 62.24%, but this remains incomplete text reconstruction, not evidence of safe laboratory execution.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.