Hugging GPT: Solving AI Tasks with Chat GPT and its Friends in Hugging Face

By Y. Shen et al

Published on Dec. 3, 2023

Read the original document by opening this link in a new tab.

Table of Contents

Abstract
1 Introduction
2 Related Works
3 HuggingGPT
3.1 Task Planning
3.2 Model Selection
3.3 Task Execution
3.4 Response Generation
4 Summary

Summary

HuggingGPT is a collaborative system for solving AI tasks, composed of a large language model (LLM) and numerous expert models from ML communities. Its workflow includes four stages: task planning, model selection, task execution, and response generation. The system utilizes LLM as the controller to route user requests to expert models, effectively combining language comprehension capabilities with expertise from expert models. By organizing cooperation among models through the LLM, HuggingGPT can address tasks in any modality or domain. The design of task planning enables automatic generation of task procedures to solve complex problems. Model selection is flexible, allowing the system to integrate diverse expert models from AI communities. Overall, HuggingGPT aims to provide generalized AI solutions and advance towards artificial general intelligence.