SHAPE: Stage-aware Hierarchical Advantage via Potential Estimation for LLM Reasoning

arXiv:2604.06636v1 Announce Type: cross Abstract: Process supervision has emerged as a promising approach for enhancing LLM reasoning, yet existing methods fail to distinguish meaningful progress from mere verbosity, leading to limited reasoning capabilities and unresolved token inefficiency. To address this, we propose Stage-aware Hierarchical Advantage via Potential Estimation (SHAPE), a framework that formalizes reasoning as a trajectory through a state space of empirical solvability. SHAPE introduces a hierarchical credit assignment mechanism: at the segment level, it employs a stage-aware advantage function to prioritize efficient breakthroughs in low-potential states; at the token level, it utilizes entropy-driven redistribution to sharpen execution signals. Extensive experiments in math reasoning across three base models and five benchmarks demonstrate that SHAPE achieves an average accuracy gain of 3% with 30% reduced token consumption.

Related

• Efficient PRM Training Data Synthesis via Formal Verification
• The Detection-Extraction Gap: Models Know the Answer Before They Can Say It
• On the Step Length Confounding in LLM Reasoning Data Selection
• Rectifying LLM Thought from Lens of Optimization

Source: arXiv cs.AI | 2026-04-10