NCO: A Versatile Plug-in for Handling Negative Constraints in Decoding
2026-05-11 • Computation and Language
Computation and LanguageArtificial Intelligence
AI summaryⓘ
The authors address the problem of stopping language models from generating bad or private information while they write. Previous methods either fixed the text after it was made or slowed down the writing by checking many rules at once, which can be too slow. They created a new method called NCO that quickly spots and controls unwanted words or patterns as the model writes, without getting overwhelmed by too many rules. Their method works smoothly with common ways to make text and helps keep outputs clean and private.
Large Language ModelsDecodingRegex ConstraintsFinite AutomatonPattern MatchingProfanity SuppressionPersonally Identifiable InformationBeam SearchSampling MethodsSoft Masking
Authors
Hyundong Jin, Yo-Sub Han
Abstract
Controlling Large Language Models (LLMs) to prevent the generation of undesirable content, such as profanity and personally identifiable information (PII), has become increasingly critical. While earlier approaches relied on post-processing or resampling, recent research has shifted towards constrained decoding methods that control outputs during generation to mitigate high computational costs and quality degradation. However, preventing multiple forbidden hard constraints or regex constraints from appearing anywhere in the output is computationally challenging. A straightforward solution is to convert these constraints into a single automaton that tracks all forbidden patterns during decoding, but this often becomes impractically large. Standard regex engines also do not readily support the operations needed to build such a constraint, such as complement and intersection. In order to address these limitations, we propose NCO, a decoding strategy that performs online pattern matching over finite hard constraints and regex constraints, reducing computational overhead without inducing state explosion. NCO is fully compatible with standard inference strategies, including various sampling methods and beam search, while also supporting soft masking for probabilistic suppression. We empirically demonstrate its effectiveness across practical tasks, including PII and profanity suppression. Our implementation is available at https://github.com/hyundong98/NCO-Decoding.git .