JARVIS / HuggingGPT icon

JARVIS / HuggingGPT

Microsoft's research project on using LLMs as controllers for orchestrating models on Hugging Face Hub.

Visit Website Agent Frameworks
agent-framework open-source microsoft research multi-agent

Overview

Microsoft’s research project on using LLMs as controllers for orchestrating models on Hugging Face Hub.

Details

JARVIS (also known as HuggingGPT) is Microsoft’s research project on using LLMs as controllers for orchestrating models on Hugging Face Hub, the LLM plans which models to use for different parts of a task and orchestrates their execution. JARVIS demonstrated how LLMs can serve as orchestrators for ML model ecosystems. JARVIS is influential as a research demonstration of agentic LLM orchestration of broader AI capabilities.

Tags

agent-framework, open-source, microsoft, research, multi-agent