7B 파라미터로 GPT-4o를 이긴 AI 에이전트: Microsoft Fara-7B
Microsoft의 7B 파라미터 AI 에이전트 Fara-7B가 GPT-4o를 능가하는 성능으로 디바이스에서 직접 웹을 조작합니다. 프라이버시와 효율성을 동시에 잡은 새로운 접근법을 소개합니다.7B 파라미터로 GPT-4o를 이긴 AI 에이전트: Microsoft Fara-7B
Microsoft의 7B 파라미터 AI 에이전트 Fara-7B가 GPT-4o를 능가하는 성능으로 디바이스에서 직접 웹을 조작합니다. 프라이버시와 효율성을 동시에 잡은 새로운 접근법을 소개합니다.Microsoft introduces Magentic-One in AutoGen v0.4 for orchestrating multi AI agent workflows #AI #Microsoft #AutoGen #AIAgents #MagenticOne #AITools #Automation
Microsoft researchers are introducing AutoGen, a framework for simplifying the orchestration, optimization, and automation of workflows for large language model (LLM) applications—potentially transforming and extending what LLMs can do. Learn more.
🎯 #MagenticOne: New generalist multi-agent system developed by #Microsoft for handling complex web and file-based tasks
🤖 Core Features:
• Lead Orchestrator agent manages task planning, delegation, and progress tracking using Task & Progress Ledgers
• #WebSurfer agent controls Chrome browser for web interactions and navigation
• #FileSurfer agent handles local file operations across various formats
• #Coder and Computer Terminal agents for code writing and execution
💡 Technical Details:
• Built on #AutoGen framework
• Default model: #GPT4 (model-agnostic design)
• Runs in Docker containers for safety
• Supports human-in-the-loop supervision
⚙️ Implementation:
• Available as #opensource project
• Requires Docker and Playwright
• Supports Azure AD and #OpenAI configurations
• Currently being ported to AutoGen AgentChat
🔗 Learn more: https://github.com/microsoft/autogen/tree/main/python/packages/autogen-magentic-one