Truth-Aware Decoding: A Program-Logic Approach to Factual Language Generation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Alpay, Faruk, Alakkad, Hamdi
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916997795676160
author Alpay, Faruk
Alakkad, Hamdi
author_facet Alpay, Faruk
Alakkad, Hamdi
contents This paper introduces Truth-Aware Decoding (TAD), a verification-oriented decoding scheme that aligns neural language generation with knowledge bases. Situated in the tradition of probabilistic program semantics for sequence models, TAD augments modern instruction-tuned systems with a lattice of semantic guards that operate at decode time. Our contributions are fourfold: (i) a constraint-based semantics that renders oracle filtering as a program-logic judgment, (ii) a proof that greedy selection enjoys local likelihood dominance under sound and complete guards (Theorem 2.7), (iii) an entropy-style invariant that quantifies factual risk via knowledge-aware safe mass, and (iv) a multi-agent operational calculus with verified Lean artefacts to certify implementation behaviour. Numerical and algorithmic case studies confirm that the resulting guardrails reduce hallucinations without sacrificing throughput, yielding a pragmatic bridge between large-scale empirical models and formal verification.
format Preprint
id arxiv_https___arxiv_org_abs_2510_07331
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Truth-Aware Decoding: A Program-Logic Approach to Factual Language Generation
Alpay, Faruk
Alakkad, Hamdi
Artificial Intelligence
Logic in Computer Science
68N15, 68Q55, 68Q60, 03B35
D.3.1; F.3.1; F.3.2
This paper introduces Truth-Aware Decoding (TAD), a verification-oriented decoding scheme that aligns neural language generation with knowledge bases. Situated in the tradition of probabilistic program semantics for sequence models, TAD augments modern instruction-tuned systems with a lattice of semantic guards that operate at decode time. Our contributions are fourfold: (i) a constraint-based semantics that renders oracle filtering as a program-logic judgment, (ii) a proof that greedy selection enjoys local likelihood dominance under sound and complete guards (Theorem 2.7), (iii) an entropy-style invariant that quantifies factual risk via knowledge-aware safe mass, and (iv) a multi-agent operational calculus with verified Lean artefacts to certify implementation behaviour. Numerical and algorithmic case studies confirm that the resulting guardrails reduce hallucinations without sacrificing throughput, yielding a pragmatic bridge between large-scale empirical models and formal verification.
title Truth-Aware Decoding: A Program-Logic Approach to Factual Language Generation
topic Artificial Intelligence
Logic in Computer Science
68N15, 68Q55, 68Q60, 03B35
D.3.1; F.3.1; F.3.2
url https://arxiv.org/abs/2510.07331