LSRIF: Logic-Structured Reinforcement Learning for Instruction Following

TL;DR

This paper introduces LSRIF, a framework that explicitly models logical structures in instructions to improve large language models' ability to follow complex, structured instructions and enhance reasoning capabilities.

Contribution

The paper presents a novel logic-structured training framework and dataset that explicitly incorporate logical dependencies, improving instruction following and reasoning in language models.

Findings

Significant improvements in instruction-following accuracy

Enhanced reasoning capabilities in models trained with LSRIF

Parameter updates focus on attention layers and logical operators

Abstract

Instruction-following is critical for large language models, but real-world instructions often contain logical structures such as sequential dependencies and conditional branching. Existing methods typically construct datasets with parallel constraints and optimize average rewards, ignoring logical dependencies and yielding noisy signals. We propose a logic-structured training framework LSRIF that explicitly models instruction logic. We first construct a dataset LSRInstruct with constraint structures such as parallel, sequential, and conditional types, and then design structure-aware rewarding method LSRIF including average aggregation for parallel structures, failure-penalty propagation for sequential structures, and selective rewards for conditional branches. Experiments show LSRIF brings significant improvements in instruction-following (in-domain and out-of-domain) and general reasoning. Analysis reveals that learning with explicit logic structures brings parameter updates in attention layers and sharpens token-level attention to constraints and logical operators.