CMAT: A Multi-Agent Collaboration Tuning Framework for Enhancing Small Language Models

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Liang, Xuechen, He, Yangfan, Tao, Meiling, Xia, Yinghui, Wang, Jianhui, Shi, Tianyu, Wang, Jun, Yang, JingSong
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866912327360577536
author Liang, Xuechen
He, Yangfan
Tao, Meiling
Xia, Yinghui
Wang, Jianhui
Shi, Tianyu
Wang, Jun
Yang, JingSong
author_facet Liang, Xuechen
He, Yangfan
Tao, Meiling
Xia, Yinghui
Wang, Jianhui
Shi, Tianyu
Wang, Jun
Yang, JingSong
contents Open large language models (LLMs) have significantly advanced the field of natural language processing, showcasing impressive performance across various tasks.Despite the significant advancements in LLMs, their effective operation still relies heavily on human input to accurately guide the dialogue flow, with agent tuning being a crucial optimization technique that involves human adjustments to the model for better response to such guidance.Addressing this dependency, our work introduces the TinyAgent model, trained on a meticulously curated high-quality dataset. We also present the Collaborative Multi-Agent Tuning (CMAT) framework, an innovative system designed to augment language agent capabilities through adaptive weight updates based on environmental feedback. This framework fosters collaborative learning and real-time adaptation among multiple intelligent agents, enhancing their context-awareness and long-term memory. In this research, we propose a new communication agent framework that integrates multi-agent systems with environmental feedback mechanisms, offering a scalable method to explore cooperative behaviors. Notably, our TinyAgent-7B model exhibits performance on par with GPT-3.5, despite having fewer parameters, signifying a substantial improvement in the efficiency and effectiveness of LLMs.
format Preprint
id arxiv_https___arxiv_org_abs_2404_01663
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle CMAT: A Multi-Agent Collaboration Tuning Framework for Enhancing Small Language Models
Liang, Xuechen
He, Yangfan
Tao, Meiling
Xia, Yinghui
Wang, Jianhui
Shi, Tianyu
Wang, Jun
Yang, JingSong
Computation and Language
Artificial Intelligence
Computational Complexity
Open large language models (LLMs) have significantly advanced the field of natural language processing, showcasing impressive performance across various tasks.Despite the significant advancements in LLMs, their effective operation still relies heavily on human input to accurately guide the dialogue flow, with agent tuning being a crucial optimization technique that involves human adjustments to the model for better response to such guidance.Addressing this dependency, our work introduces the TinyAgent model, trained on a meticulously curated high-quality dataset. We also present the Collaborative Multi-Agent Tuning (CMAT) framework, an innovative system designed to augment language agent capabilities through adaptive weight updates based on environmental feedback. This framework fosters collaborative learning and real-time adaptation among multiple intelligent agents, enhancing their context-awareness and long-term memory. In this research, we propose a new communication agent framework that integrates multi-agent systems with environmental feedback mechanisms, offering a scalable method to explore cooperative behaviors. Notably, our TinyAgent-7B model exhibits performance on par with GPT-3.5, despite having fewer parameters, signifying a substantial improvement in the efficiency and effectiveness of LLMs.
title CMAT: A Multi-Agent Collaboration Tuning Framework for Enhancing Small Language Models
topic Computation and Language
Artificial Intelligence
Computational Complexity
url https://arxiv.org/abs/2404.01663