Reproduce “ConformalNL2LTL: Translating Natural Language Instructions into Temporal Logic Formulas with Conformal Correctness Guarantees” 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
David Smith Sundarsingh, Jun Wang, Jyotirmoy V. Deshmukh, Yiannis Kantaros
ConformalNL2LTL translates natural-language robot instructions into Linear Temporal Logic formulas, building each formula step by step and using conformal prediction to assess uncertainty. Ambiguous choices can prompt an auxiliary model or person. On tested in-distribution tasks, it recorded 99.24% translation accuracy at a 99% target, with human help on 0.378% of response steps; this does not establish performance under distribution shift.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.