Agentic Conversations (formally mlops.community) Democratizing AI // Yujian Tang // #163
46 snips
Jun 27, 2023 Vector databases have emerged as key infrastructure for generative AI, but when should an app use one instead of relying on prompts or search? The conversation explores LLM costs, data quality, chunking, access controls, and real-world enterprise uses, alongside open-source tools and efforts to make vector technology easier for more people to adopt.
Chapters
Transcript
Episode notes
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Introduction
00:00 • 5min
How to Make Money With Snacks
05:10 • 4min
The Rise of Vector Databases
09:00 • 3min
The Trade-Off of Using a Vector Database for Prompts
12:18 • 2min
The Cost of Scaling a Large Language Model
14:16 • 6min
How to Use a Vector Store to Build a CNN With Open Source Documents
20:09 • 3min
How to Maintain the Validity of Questions That Haven't Been Asked
23:29 • 4min
How to Set Up Rules Around Using Vector Databases
27:36 • 4min
The Importance of Knowing When Your Data Looks Like
31:32 • 2min
The Importance of a Vector Database
33:12 • 2min
Common Gaps in Generative AI
35:30 • 4min
The Importance of Knowing Your Data
39:38 • 2min
Milvis: The Open Source Factor Database
41:52 • 4min
The Power of Zealous Cloud
45:48 • 3min
How to Implement Role Based Access Controls in Your Organization
48:29 • 2min
GPT Cash: A Data Mesh Platform for Data Products
50:32 • 2min
The Rise of the Milvis Community
52:13 • 2min