Chain of Draft: Thinking Faster by Writing Less

TL;DR

Chain of Draft (CoD) is a new reasoning paradigm for LLMs that generates concise intermediate thoughts, reducing verbosity and computational cost while maintaining or improving accuracy in complex reasoning tasks.

Contribution

We introduce Chain of Draft, a novel approach inspired by human cognition, that produces minimalistic intermediate reasoning steps to enhance efficiency and effectiveness of LLM reasoning.

Findings

CoD matches or exceeds CoT accuracy across tasks.

CoD uses only 7.6% of tokens compared to CoT.

Significant reductions in cost and latency achieved.

Abstract

Large Language Models (LLMs) have demonstrated remarkable performance in solving complex reasoning tasks through mechanisms like Chain-of-Thought (CoT) prompting, which emphasizes verbose, step-by-step reasoning. However, humans typically employ a more efficient strategy: drafting concise intermediate thoughts that capture only essential information. In this work, we propose Chain of Draft (CoD), a novel paradigm inspired by human cognitive processes, where LLMs generate minimalistic yet informative intermediate reasoning outputs while solving tasks. By reducing verbosity and focusing on critical insights, CoD matches or surpasses CoT in accuracy while using as little as only 7.6% of the tokens, significantly reducing cost and latency across various reasoning tasks. Our code and data are available at https://github.com/sileix/chain-of-draft.