Wait, We Don't Need to "Wait"! Removing Thinking Tokens Improves Reasoning Efficiency

TL;DR

This paper introduces NoWait, a method that disables explicit self-reflection tokens in reasoning models, significantly reducing reasoning steps and improving efficiency across multiple multimodal reasoning benchmarks without losing performance.

Contribution

It demonstrates that removing thinking tokens like 'Wait' and 'Hmm' enhances reasoning efficiency without sacrificing accuracy, providing a simple, effective, and versatile inference technique.

Findings

Reduces chain-of-thought length by up to 51%

Maintains model utility across diverse tasks

Offers a plug-and-play solution for multimodal reasoning

Abstract

Recent advances in large reasoning models have enabled complex, step-by-step reasoning but often introduce significant overthinking, resulting in verbose and redundant outputs that hinder efficiency. In this study, we examine whether explicit self-reflection, signaled by tokens such as "Wait" and "Hmm", is necessary for advanced reasoning. We propose NoWait, a simple yet effective approach that disables explicit self-reflection by suppressing these tokens during inference. Extensive experiments on ten benchmarks across textual, visual, and video reasoning tasks show that NoWait reduces chain-of-thought trajectory length by up to 27%-51% in five R1-style model series, without compromising model utility. NoWait thus offers a plug-and-play solution for efficient and utility-preserving multimodal reasoning.