Enregistré dans:
Détails bibliographiques
Auteurs principaux: Wang, Ruiyi, Yu, Haofei, Zhang, Wenxin, Qi, Zhengyang, Sap, Maarten, Neubig, Graham, Bisk, Yonatan, Zhu, Hao
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:https://arxiv.org/abs/2403.08715
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866909182749310976
author Wang, Ruiyi
Yu, Haofei
Zhang, Wenxin
Qi, Zhengyang
Sap, Maarten
Neubig, Graham
Bisk, Yonatan
Zhu, Hao
author_facet Wang, Ruiyi
Yu, Haofei
Zhang, Wenxin
Qi, Zhengyang
Sap, Maarten
Neubig, Graham
Bisk, Yonatan
Zhu, Hao
contents Humans learn social skills through both imitation and social interaction. This social learning process is largely understudied by existing research on building language agents. Motivated by this gap, we propose an interactive learning method, SOTOPIA-$π$, improving the social intelligence of language agents. This method leverages behavior cloning and self-reinforcement training on filtered social interaction data according to large language model (LLM) ratings. We show that our training method allows a 7B LLM to reach the social goal completion ability of an expert model (GPT-4-based agent), while improving the safety of language agents and maintaining general QA ability on the MMLU benchmark. We also find that this training paradigm uncovers some difficulties in LLM-based evaluation of social intelligence: LLM-based evaluators overestimate the abilities of the language agents trained specifically for social interaction.
format Preprint
id arxiv_https___arxiv_org_abs_2403_08715
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle SOTOPIA-$π$: Interactive Learning of Socially Intelligent Language Agents
Wang, Ruiyi
Yu, Haofei
Zhang, Wenxin
Qi, Zhengyang
Sap, Maarten
Neubig, Graham
Bisk, Yonatan
Zhu, Hao
Computation and Language
Humans learn social skills through both imitation and social interaction. This social learning process is largely understudied by existing research on building language agents. Motivated by this gap, we propose an interactive learning method, SOTOPIA-$π$, improving the social intelligence of language agents. This method leverages behavior cloning and self-reinforcement training on filtered social interaction data according to large language model (LLM) ratings. We show that our training method allows a 7B LLM to reach the social goal completion ability of an expert model (GPT-4-based agent), while improving the safety of language agents and maintaining general QA ability on the MMLU benchmark. We also find that this training paradigm uncovers some difficulties in LLM-based evaluation of social intelligence: LLM-based evaluators overestimate the abilities of the language agents trained specifically for social interaction.
title SOTOPIA-$π$: Interactive Learning of Socially Intelligent Language Agents
topic Computation and Language
url https://arxiv.org/abs/2403.08715