Efficient Reasoning via Chain of Unconscious Thought

TL;DR

This paper introduces Chain of Unconscious Thought (CoUT), a novel reasoning paradigm inspired by human unconscious thought, which significantly improves token efficiency in large reasoning models without sacrificing accuracy.

Contribution

The paper proposes CoUT, a new reasoning method that internalizes reasoning in hidden layers and employs token-efficient strategies, reducing token usage by nearly half while maintaining performance.

Findings

CoUT reduces token usage by 47.62% compared to Chain of Thought (CoT).

Models utilizing CoUT maintain comparable accuracy to traditional methods.

Extensive experiments validate the effectiveness of CoUT in improving reasoning efficiency.

Abstract

Large Reasoning Models (LRMs) achieve promising performance but compromise token efficiency due to verbose reasoning processes. Unconscious Thought Theory (UTT) posits that complex problems can be solved more efficiently through internalized cognitive processes. Inspired by UTT, we propose a new reasoning paradigm, termed Chain of Unconscious Thought (CoUT), to improve the token efficiency of LRMs by guiding them to mimic human unconscious thought and internalize reasoning processes. Concretely, we first prompt the model to internalize the reasoning by thinking in the hidden layer. Then, we design a bag of token-efficient strategies to further help models reduce unnecessary tokens yet preserve the performance. Our work reveals that models may possess beneficial unconscious thought, enabling improved efficiency without sacrificing performance. Extensive experiments demonstrate the effectiveness of CoUT. Remarkably, it surpasses CoT by reducing token usage by 47.62% while maintaining comparable accuracy, as shown in Figure 1. The code of CoUT is available at this link: https://github.com/Rohan-GRH/CoUT