Learning and Enforcing Context-Sensitive Control for LLMs
It eliminates the need for manual specification of context-sensitive constraints in LLM generation, a key bottleneck for ensuring output validity.
This work introduces a framework that automatically learns context-sensitive constraints from LLM interactions, enabling even small 1B-parameter LLMs to achieve perfect constraint adherence, outperforming larger models and state-of-the-art reasoning models.
Controlling the output of Large Language Models (LLMs) through context-sensitive constraints has emerged as a promising approach to overcome the limitations of Context-Free Grammars (CFGs) in guaranteeing generation validity. However, such constraints typically require manual specification -- a significant barrier demanding specialized expertise. We introduce a framework that automatically learns context-sensitive constraints from LLM interactions through a two-phase process: syntactic exploration to gather diverse outputs for constraint learning, followed by constraint exploitation to enforce these learned rules during generation. Experiments demonstrate that our method enables even small LLMs (1B parameters) to learn and generate with perfect constraint adherence, outperforming larger counterparts and state-of-the-art reasoning models. This work represents the first integration of context-sensitive grammar learning with LLM generation, eliminating manual specification while maintaining generation validity.