Steering Large Reasoning Models towards Concise Reasoning via Flow Matching
This work addresses efficiency issues in reasoning models for AI applications, offering a novel method that is more effective than prior linear approaches.
The paper tackles the problem of overly verbose outputs in Large Reasoning Models by introducing FlowSteer, a nonlinear steering method that learns a transformation between verbose and concise reasoning distributions via Flow Matching, resulting in strong task performance and improved token efficiency across benchmarks.
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 linear representation hypothesis. In this work, we introduce FlowSteer, a nonlinear steering method that goes beyond uniform linear shifts by learning a complete transformation between the distributions associated with verbose and concise reasoning. This transformation is learned via 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.