Search papers, labs, and topics across Lattice.
This paper introduces AutoVerifier, a novel residual-guided non-parametric optimization method designed to enhance the accuracy of reference-based answer verification in reasoning models. By learning from recurring verifier errors and framing implicit assumptions as inductive biases, AutoVerifier effectively adapts its verification strategies through a system of rule cards that are validated before being implemented. Experimental results show that AutoVerifier significantly outperforms existing state-of-the-art verifiers across four benchmark datasets, highlighting its effectiveness in improving verification accuracy in reinforcement learning contexts.
AutoVerifier learns from its mistakes, transforming verifier errors into reusable strategies that dramatically boost verification accuracy.
Reference-based verifiers are important for evaluating reasoning models and providing accurate outcome rewards in reinforcement learning with verifiable rewards. To improve verification accuracy, prior work has explored rule-based, model-based, and tool-augmented verifiers for checking answer equivalence across diverse answer forms. However, the equivalence of answer forms such as $1+3.14$ and $1+\pi$ may depend on the question and scoring criterion. We frame such implicit assumptions as verifier inductive biases. To address this challenge, we propose AutoVerifier, a residual-guided non-parametric optimization method that learns these biases from recurring verifier errors. Specifically, AutoVerifier records these biases in rule cards and promotes them to code modules or prompt guidance only after replay validation detects no direct regressions, keeping accepted updates auditable, editable, and reusable. Experiments on four verifier benchmarks demonstrate that AutoVerifier outperforms state-of-the-art verifiers by a large margin.
