Think in Blocks: Adaptive Reasoning from Direct Response to Deep Reasoning

TL;DR

This paper introduces the Think in Blocks framework, enabling large language models to adaptively adjust their reasoning process length based on task complexity by partitioning reasoning into a tunable number of blocks.

Contribution

It establishes a block-structured paradigm, trains an adaptive model through a multi-stage pipeline, and allows dynamic control of reasoning depth during inference.

Findings

Model predicts reasoning budget and partitions reasoning accordingly.

Adaptive reasoning improves efficiency and performance.

Dynamic control of reasoning length during deployment.

Abstract

Large Language Models (LLMs) with chains-of-thought have demonstrated strong performance on an increasing range of tasks, particularly those involving complex logical reasoning. However, excessively long chains can lead to overthinking, causing computational waste and slower responses. This raises a question: can LLMs dynamically adjust the length of their reasoning processes based on task complexity? To address this, we propose the Think in Blocks framework, which enables adaptive reasoning-from zero to deep reasoning-by partitioning the reasoning process into a tunable number of blocks. Our main contributions are: (1) Establishing an explicit block-structured paradigm in which the model first predicts an integer reasoning budget-the number of blocks-and then partitions its reasoning accordingly; (2) Training an adaptive model through a three-stage pipeline-Supervised Fine-Tuning, reward-guided Direct Preference Optimization, and Reinforcement Learning-that adjusts its reasoning depth to problem difficulty; (3) Exploiting the explicit block count to dynamically control reasoning depth at inference time, allowing flexible adjustment of chain-of-thought length during deployment.