Towards a Mechanistic Understanding of Large Reasoning Models: A Survey of Training, Inference, and Failures

TL;DR

This survey reviews recent advances in understanding the internal mechanisms of Large Reasoning Models, focusing on training, reasoning processes, and failures, to promote transparency and guide future research.

Contribution

It systematically organizes current findings on LRMs' training, reasoning, and failures, highlighting gaps and proposing a roadmap for mechanistic interpretability research.

Findings

Insights into training dynamics of LRMs

Analysis of reasoning mechanisms in LRMs

Identification of unintended behaviors and failures

Abstract

Reinforcement learning (RL) has catalyzed the emergence of Large Reasoning Models (LRMs) that have pushed reasoning capabilities to new heights. While their performance has garnered significant excitement, exploring the internal mechanisms driving these behaviors has become an equally critical research frontier. This paper provides a comprehensive survey of the mechanistic understanding of LRMs, organizing recent findings into three core dimensions: 1) training dynamics, 2) reasoning mechanisms, and 3) unintended behaviors. By synthesizing these insights, we aim to bridge the gap between black-box performance and mechanistic transparency. Finally, we discuss under-explored challenges to outline a roadmap for future mechanistic studies, including the need for applied interpretability, improved methodologies, and a unified theoretical framework.