PRIME: A Process-Outcome Alignment Benchmark for Verifiable Reasoning in Mathematics and Engineering
This work addresses a critical gap in verifiable reasoning for AI systems in STEM domains, though it is incremental as it builds on existing RLVR frameworks.
The paper tackles the problem of verifiers in reinforcement learning with verifiable rewards (RLVR) often missing errors in derivation processes by focusing only on final outcomes, introducing the PRIME benchmark for process-outcome alignment verification in mathematics and engineering. It shows that using PRIME-selected verifiers in a process-aware RLVR paradigm improves performance by 8.29%, 9.12%, and 7.31% on specific benchmarks and validates PRIME as a reliable predictor with a strong linear correlation (R² > 0.92).
While model-based verifiers are essential for scaling Reinforcement Learning with Verifiable Rewards (RLVR), current outcome-centric verification paradigms primarily focus on the consistency between the final result and the ground truth, often neglecting potential errors in the derivation process. This leads to assigning positive rewards to correct answers produced from incorrect derivations. To bridge this gap, we introduce PRIME, a benchmark for evaluating verifiers on Process-Outcome Alignment verification in Mathematics and Engineering. Curated from a comprehensive collection of college-level STEM problems, PRIME comprises 2,530 high-difficulty samples through a consistency-based filtering pipeline. Through extensive evaluation, we find that current verifiers frequently fail to detect derivation flaws. Furthermore, we propose a process-aware RLVR training paradigm utilizing verifiers selected via PRIME. This approach substantially outperforms the outcome-only verification baseline, achieving absolute performance gains of 8.29%, 9.12%, and 7.31% on AIME24, AIME25, and Beyond-AIME, respectively, for the Qwen3-14B-Base model. Finally, we demonstrate a strong linear correlation ($R^2 > 0.92$) between verifier accuracy on PRIME and RLVR training effectiveness, validating PRIME as a reliable predictor for verifier selection.