Repairing LLM-generated specifications with deterministic diagnosis
In brief
- LLMs frequently generate formal specifications with semantic errors.
- DRAGO cuts regressions from 11.8% to 1.3% using deterministic diagnosis.
- Accuracy improves by 12.9 points over zero-shot formalization.
Large language models (LLMs) often produce semantically incorrect formal specifications from natural language problems. The study introduces DRAGO, a repair framework using formal, deterministic diagnosis to reduce regressions and improve answer accuracy across multiple benchmarks.
This summary gives you the essentials. The full article, its figures and quotes are on cs.AI updates on arXiv.org.
Read the original articleAI-generated summary of the source, checked by rules. Report an error