Search papers, labs, and topics across Lattice.
Affiliation:
3
0
5
Qualitative analysis shows that S4L captures implicit causal relations, deontic modality, and exception structure more reliably than the baselines, demonstrating that structured reasoning prompts can substantially improve the reliability of natural-language-to-logic translation for legal and safety-critical applications.
High-quality model outputs can be misleading, with trustworthiness plummeting under degraded conditions, revealing critical flaws in current evaluation methods.
LLMs can now translate legal cases into logical formulas with significantly improved accuracy and stability, thanks to a novel retrieval-augmented in-context learning approach that carefully balances exemplar diversity and mitigates entity bias.
