IFDECORATOR: Wrapping Instruction Following Reinforcement Learning with Verifiable Rewards
It addresses training challenges in instruction-following AI for developers and researchers, offering an incremental improvement over existing RLVR methods.
The paper tackles inefficiency and over-optimization in Reinforcement Learning with Verifiable Rewards (RLVR) for instruction following in large language models by introducing IFDecorator, a framework that improves training robustness and sample efficiency, achieving 87.43% accuracy on IFEval and reducing reward hacking rates.
Reinforcement Learning with Verifiable Rewards (RLVR) improves instruction following capabilities of large language models (LLMs), but suffers from training inefficiency due to inadequate difficulty assessment. Moreover, RLVR is prone to over-optimization, where LLMs exploit verification shortcuts without aligning to the actual intent of user instructions. We introduce Instruction Following Decorator (IFDecorator}, a framework that wraps RLVR training into a robust and sample-efficient pipeline. It consists of three components: (1) a cooperative-adversarial data flywheel that co-evolves instructions and hybrid verifications, generating progressively more challenging instruction-verification pairs; (2) IntentCheck, a bypass module enforcing intent alignment; and (3) trip wires, a diagnostic mechanism that detects reward hacking via trap instructions, which trigger and capture shortcut exploitation behaviors. Our Qwen2.5-32B-Instruct-IFDecorator achieves 87.43% accuracy on IFEval, outperforming larger proprietary models such as GPT-4o. Additionally, we demonstrate substantial improvements on FollowBench while preserving general capabilities. Our trip wires show significant reductions in reward hacking rates. We will release models, code, and data for future research.