Role-Play Zero-Shot Prompting with Large Language Models for Open-Domain Human-Machine Conversation

TL;DR

This paper introduces a role-play zero-shot prompting method using multilingual LLMs like Vicuna to create open-domain conversational agents that outperform fine-tuned models in human evaluations, offering a cost-effective alternative.

Contribution

It demonstrates that role-play zero-shot prompting with multilingual LLMs can effectively generate conversational agents without fine-tuning, surpassing traditional methods in certain languages.

Findings

Outperforms fine-tuned models in French human evaluations

Uses multilingual LLMs with role-play prompting for open-domain conversation

Provides a cost-effective alternative to fine-tuning

Abstract

Recently, various methods have been proposed to create open-domain conversational agents with Large Language Models (LLMs). These models are able to answer user queries, but in a one-way Q&A format rather than a true conversation. Fine-tuning on particular datasets is the usual way to modify their style to increase conversational ability, but this is expensive and usually only available in a few languages. In this study, we explore role-play zero-shot prompting as an efficient and cost-effective solution for open-domain conversation, using capable multilingual LLMs (Beeching et al., 2023) trained to obey instructions. We design a prompting system that, when combined with an instruction-following model - here Vicuna (Chiang et al., 2023) - produces conversational agents that match and even surpass fine-tuned models in human evaluation in French in two different tasks.