Prompted LLMs as Chatbot Modules for Long Open-domain Conversation

Gibbeum Lee, Volker Hartmann, Jongho Park, Dimitris Papailiopoulos, Kangwook Lee
5/8/2023
cs.CLcs.AIcs.LG

No official code found. Best implementation: krafton-ai/MPC (35 stars, Apache-2.0, last commit 2023-05-24).

Abstract

In this paper, we propose MPC (Modular Prompted Chatbot), a new approach for creating high-quality conversational agents without the need for fine-tuning. Our method utilizes pre-trained large language models (LLMs) as individual modules for long-term consistency and flexibility, by using techniques such as few-shot prompting, chain-of-thought (CoT), and external memory. Our human evaluation results show that MPC is on par with fine-tuned chatbot models in open-domain conversations, making it an effective solution for creating consistent and engaging chatbots.

View on arXivView PDF

Code Implementations(1)

krafton-ai/MPCCommunity100%
Apache-2.0

The git repository of Modular Prompted Chatbot paper

354PythonMay 5, 20233 years agoApache-2.0

Similar Papers

DialogueForge: LLM Simulation of Human-Chatbot Dialogue

85% match2025

Dialog Flow Induction for Constrainable LLM-Based Chatbots

85% match1 repo102024

Multi-User Chat Assistant (MUCA): a Framework Using LLMs to Facilitate Group Conversations

85% match2024

On Overcoming Miscalibrated Conversational Priors in LLM-based Chatbots

85% match2024

From LLM to Conversational Agent: A Memory Enhanced Architecture with Fine-Tuning of Large Language Models

85% match2024

Is There a Case for Conversation Optimized Tokenizers in Large Language Models?

85% match2025

Cite this paper

@article{lee2023prompted,
  title  = {Prompted LLMs as Chatbot Modules for Long Open-domain Conversation},
  author = {Gibbeum Lee and Volker Hartmann and Jongho Park and Dimitris Papailiopoulos and Kangwook Lee},
  year   = {2023},
  eprint = {2305.04533},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2305.04533v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2305.04533.svg)](https://codeofpaper.com/paper/2305.04533)

Discussion