Neuro-Symbolic Verification on Instruction Following of LLMs

2026-01-25

AI Summary

This work addresses the challenge of implicit violations in large language models (LLMs) that lead to instruction deviation and error propagation in reasoning chains. To mitigate this, we propose NSVIF, a novel framework that introduces neural-symbolic methods into general-purpose instruction-following verification for the first time. NSVIF formulates the problem as a unified constraint satisfaction task, jointly leveraging logical reasoning and semantic analysis to verify instruction adherence. The approach is agnostic to both instruction format and model architecture and provides interpretable feedback. We also introduce VIFBENCH, a fine-grained annotated benchmark for evaluation. Experimental results demonstrate that NSVIF significantly outperforms existing LLM verification methods and effectively enhances instruction-following performance through feedback without requiring model fine-tuning.

Technology Category

• Natural Language Processing: Safety and Robustness
• Computer Vision: Visual Reasoning & Symbolic Representations
• Machine Learning: Neuro-Symbolic Learning

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

A fundamental problem of applying Large Language Models (LLMs) to important applications is that LLMs do not always follow instructions, and violations are often hard to observe or check. In LLM-based agentic workflows, such violations can propagate and amplify along reasoning chains, causing task failures and system incidents. This paper presents NSVIF, a neuro-symbolic framework for verifying whether an LLM's output follows the instructions used to prompt the LLM. NSVIF is a universal, general-purpose verifier; it makes no assumption about the instruction or the LLM. NSVIF formulates instruction-following verification as a constraint-satisfaction problem by modeling user instructions as constraints. NSVIF models both logical and semantic constraints; constraint solving is done by a unified solver that orchestrates logical reasoning and semantic analysis. To evaluate NSVIF, we develop VIFBENCH, a new benchmark for instruction-following verifiers with fine-grained data labels. Experiments show that NSVIF significantly outperforms LLM-based approaches and provides interpretable feedback. We also show that feedback from NSVIF helps improve LLMs'instruction-following capability without post-training.

Problem

• instruction following
• Large Language Models
• verification
• constraint satisfaction
• neuro-symbolic

Innovation

• Neuro-Symbolic Verification
• Instruction Following
• Constraint Satisfaction
• Interpretable Feedback
• LLM Verification