Steering Large Reasoning Models towards Concise Reasoning via Flow Matching

TMLR Paper6595 Authors

Abstract: Large Reasoning Models (LRMs) excel at complex reasoning tasks, but their efficiency is often hampered by overly verbose outputs. Prior steering methods attempt to address this issue by applying a single, global vector to hidden representations—an approach grounded in the restrictive \textit{linear representation hypothesis}. In this work, we introduce FlowSteer, a nonlinear steering method that goes beyond uniform linear shifts by learning a complete \textit{transformation between the distributions} associated with verbose and concise reasoning. This transformation is learned via \textit{Flow Matching} as a velocity field, enabling precise, input-dependent control over the model's reasoning process. By aligning steered representations with the distribution of concise-reasoning activations, FlowSteer yields more compact reasoning than the linear shifts. Across diverse reasoning benchmarks, FlowSteer demonstrates strong task performance and token efficiency compared to leading inference-time baselines. Our work demonstrates that modeling the full distributional transport with generative techniques offers a more effective and principled foundation for controlling LRMs.

Submission Type: Regular submission (no more than 12 pages of main content)

Assigned Action Editor: ~Emanuele_Sansone1

Submission Number: 6595