VeriTrans: Fine-Tuned LLM-Assisted NL-to-PL Translation via a Deterministic Neuro-Symbolic Pipeline
Xuan Liu, Dheeraj Kodakandla, Kushagra Srivastva, Mahfuza Farooque

TL;DR
VeriTrans is a reliable, fine-tuned LLM system that translates natural language requirements into solver-ready logic, ensuring high correctness and auditability through a deterministic neuro-symbolic pipeline with validation gating.
Contribution
It introduces a deterministic neuro-symbolic pipeline with validator-gated acceptance for reliable NL-to-logic translation, enabling auditability and reproducibility in critical workflows.
Findings
Achieves 94.46% correctness on SAT/UNSAT classification.
Improves fidelity by 1-1.5 percentage points with compact fine-tuning.
Exposes a reliability-coverage trade-off with a thresholded acceptance policy.
Abstract
\textbf{VeriTrans} is a reliability-first ML system that compiles natural-language requirements into solver-ready logic with validator-gated reliability. The pipeline integrates an instruction-tuned NLPL translator, round-trip reconstruction (PLNL) used as a high-precision acceptance gate, and canonical PLCNF compilation, all executed via fixed API configuration (temperature; fine-tuning runs use seed) and per-item artifact logging (prompts, outputs, hashes) to support auditability and replay-driven debugging. On \textbf{SatBench} (2{,}100 specifications), VeriTrans achieves 94.46\% SAT/UNSAT correctness and 87.73\% median round-trip similarity. Compact fine-tuning on 100--150 curated examples improves fidelity by about 1--1.5\,pp without increasing latency (mean 25.8\,s/spec on our 201-spec runtime subset). A thresholded acceptance policy on the…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
