SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2502.09601 · ACL · 2025

CoT-Valve: Length-Compressible Chain-of-Thought Tuning

Xinchao Wang, Xinyin Ma, Gongfan Fang, Runpeng Yu, Guangnian Wan

arXiv · PDF · Open in the Atlas

Code that ran

We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. The repositories linked to it are listed below.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Chain-of-Thought significantly enhances a model's reasoning capability, but it also comes with a considerable increase in inference costs due to long chains. With the observation that the reasoning path can be easily compressed under easy tasks but struggles on hard tasks, we explore the feasibility of elastically controlling the length of reasoning paths with only one model, thereby reducing the inference overhead of reasoning models dynamically based on task difficulty. We introduce a new tuning and inference strategy named CoT-Valve, designed to allow models to generate reasoning chains of varying lengths. To achieve this, we propose to identify a direction in the parameter space that, when manipulated, can effectively control the length of generated CoT. Moreover, we show that this property is valuable for compressing the reasoning chain. We construct datasets with chains from long to short for the same questions and explore two enhanced strategies for CoT-Valve: (1) a precise length-compressible CoT tuning method, and (2) a progressive chain length compression approach. Our experiments show that CoT-Valve successfully enables controllability and compressibility of the chain and shows better performance than the prompt-based control. We applied this method to QwQ-32B-Preview, reducing reasoning chains on GSM8K from 741 to 225 tokens with a minor performance drop (95.07% to 94.92%) and on AIME from 6827 to 4629 tokens, with only one additional incorrect answer. 1. John starts with €100. 2. He buys a roast for €17: €100 -€17 = €83 remaining 3. Then he buys vegetables for €11:\n €83 -€11 = €72 remaining So, after his purchases, John has €72 left. John starts with €100. He makes two purchases: a roast for €17 and vegetables for €11. To find out how much money he has left, we need to subtract the total amount he spent from his initial amount. First, <Omitted>Yes, that seems correct. John has €72 remaining.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2502.09601")
get_code_for_paper("2502.09601")
have("2502.09601")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2502.09601.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2502.09601.svg)](https://syntology.ai/paper/2502.09601)

Connect an agent — have() is free.