Lecture 03 (Part 1): Agentic AI Frameworks & Autogen
Link to lecture recording on YouTube
Date: 2024-09-23
Speaker: Chi Wang
Speaker’s social profile: Website / Google Scholar / GitHub / LinkedIn / X (Twitter)
Education:
- Ph.D. in Computer Science, 2009-2014, University of Illinois Urbana-Champaign
- B.E. in Computer Sciences and Technology, 2005-2009, Tsinghua University
Work:
- Senior Staff Research Scientist, Google Deepmind
Notes
What are future AI applications like?
- Generative: generate content like text & image
- Agentic: execute complex tasks on behalf of human
Article1 by Berkeley talking about how we are observing more and more AI results shifting from using a simple language model to building compound AI systems
[Incomplete, work in progress]
How do we empower every developer to build them?
References
- Berkeley Artificial Intelligence Research (BAIR): The Shift from Models to Compound AI Systems.
Lecture 03 (Part 2): Building a Multimodal Knowledge Assistant
Link to lecture recording (from 37:05) on YouTube
Date: 2024-09-23
Speaker: Jerry Liu
Speaker’s social profile: Website / Google Scholar / GitHub / LinkedIn / X (Twitter)
Education:
- Bachelor’s Degree in Computer Science, 2013-2017, Princeton University
Work:
Notes
LlamaIndex helps any developer build context-augmented LLM apps from prototype to production over enterprise data
Limitations of basic RAG:
- naive data processing, primitive retrieval interface
- poor query understanding / planning
- no function calling or tool use
- stateless, no memory
[Incomplete, work in progress]
References