Large Reasoning Models (LRMs) increasingly rely on reasoning traces with complex internal structures. However, existing work lacks a unified answer to three fundamental questions: (1) what defines high-quality reasoning, (2) how to reliably evaluate long, implicitly structured reasoning traces, and (3) how to use such evaluation signals for reasoning optimization. To address these challenges, we provide a unified perspective. (1) We introduce the ME2 principle to characterize reasoning quality along macro- and micro-level concerning efficiency and effectiveness. (2) Built on this principle, we model reasoning traces as directed acyclic graphs (DAGs) and develop a DAG-based pairwise evaluation method, capturing complex reasoning structures. (3) Based on this method, we construct the TRM-Preference dataset and train a Thinking Reward Model (TRM) to evaluate reasoning quality at scale. Experiments show that thinking rewards serve as an effective optimization signal. At test time, selecting better reasoning leads to better outcomes (up to 19.3% gain), and during RL training, thinking rewards enhance reasoning and performance (up to 3.9% gain) across diverse tasks.
@article{arxiv.2602.08498,
title = {Characterizing, Evaluating, and Optimizing Complex Reasoning},
author = {Haoran Zhang and Yafu Li and Zhi Wang and Zhilin Wang and Shunkai Zhang and Xiaoye Qu and Yu Cheng},
journal= {arXiv preprint arXiv:2602.08498},
year = {2026}
}
Comments
Code and data are available at \url{https://github.com/zzzhr97/TRM}