Group Think: Multiple Concurrent Reasoning Agents Collaborating at Token Level Granularity

TL;DR

This paper introduces Group Think, a novel approach where a single large language model acts as multiple concurrent reasoning agents at the token level, enabling dynamic collaboration, reducing latency, and improving reasoning quality.

Contribution

The paper proposes a new concurrent reasoning paradigm called Group Think, allowing a single LLM to simulate multiple agents collaborating at token granularity, with a simple modification for existing models.

Findings

Reduces reasoning latency compared to turn-based methods.

Improves reasoning quality through dynamic token-level collaboration.

Efficiently utilizes idle computational resources, especially on edge devices.

Abstract

Recent advances in large language models (LLMs) have demonstrated the power of reasoning through self-generated chains of thought. Multiple reasoning agents can collaborate to raise joint reasoning quality above individual outcomes. However, such agents typically interact in a turn-based manner, trading increased latency for improved quality. In this paper, we propose Group Think--a single LLM that acts as multiple concurrent reasoning agents, or thinkers. With shared visibility into each other's partial generation progress, Group Think introduces a new concurrent-reasoning paradigm in which multiple reasoning trajectories adapt dynamically to one another at the token level. For example, a reasoning thread may shift its generation mid-sentence upon detecting that another thread is better positioned to continue. This fine-grained, token-level collaboration enables Group Think to reduce redundant reasoning and improve quality while achieving significantly lower latency. Moreover, its concurrent nature allows for efficient utilization of idle computational resources, making it especially suitable for edge inference, where very small batch size often underutilizes local~GPUs. We give a simple and generalizable modification that enables any existing LLM to perform Group Think on a local GPU. We also present an evaluation strategy to benchmark reasoning latency and empirically demonstrate latency improvements using open-source LLMs that were not explicitly trained for Group Think. We hope this work paves the way for future LLMs to exhibit more sophisticated and more efficient collaborative behavior for higher quality generation.