Sign in to confirm you’re not a bot
This helps protect our community. Learn more

Deep Dive into inner workings of AI Agent frameworks

Microsoft Reactor
106K subscribers
<__slot-el>
<__slot-el>
734 views Streamed 4 months ago
This talk will delve into the core principles and mechanisms that modern AI agent frameworks like autogeny and crewel rely on. We'll explore some interesting papers like: • Chain-of-Thought (CoT) • React • Program-Aided Language Models (PAL), and many more resources. Goal is to learn how these frameworks enable Large Language Models (LLMs) to function as autonomous agents. Getting deeper into these abstractions. (https://aka.ms/AgentArchitecture) The session will focus on: Theoretical with strong examples. Moving beyond high-level abstractions to examine the underlying algorithms and methodologies that enable LLMs to exhibit agent-like behaviors. What will the attendees learn from this session? Attendees will learn the idea and logic behind papers like CoT, ReAct, and PAL, understanding how these approaches transform LLMs into effective autonomous agents. They'll gain insights into implementing these concepts in their own agenti …...more
...more
Deep Dive into inner workings of AI Agent frameworks
This talk will delve into the core principles and mechanisms that modern AI agent frameworks like autogeny and crewel rely on. We'll explore some interesting papers like: • Chain-of-Thought (CoT) • React • Program-Aided Language Models (PAL), and many more resources. Goal is to learn how these frameworks enable Large Language Models (LLMs) to function as autonomous agents. Getting deeper into these abstractions. (https://aka.ms/AgentArchitecture) The session will focus on: Theoretical with strong examples. Moving beyond high-level abstractions to examine the underlying algorithms and methodologies that enable LLMs to exhibit agent-like behaviors. What will the attendees learn from this session? Attendees will learn the idea and logic behind papers like CoT, ReAct, and PAL, understanding how these approaches transform LLMs into effective autonomous agents. They'll gain insights into implementing these concepts in their own agentic projects. Speaker Bio: Sai Yashwanth is an AI research engineer at an early stage startup (based out of london). He is also an author currently writing a book on AI agents with manning publications. He has experience building AI agents and working closely with early stage AI startups like composio (summer intern), and vuhosi (current). Builds autonomous agents in his free time (Check out devyan). Social Handle of the speaker: Twitter: https://x.com/yashwanthsai29 Linkedin:   / saiyashwanth29   Pre-requisites: Basic interactions with llms like chatgpt [eventID:24121]

Follow along using the transcript.

Microsoft Reactor

106K subscribers
Live chat replay is not available for this video.