A gentle introduction to Vector Databases
summary
Explains that vector databases store embeddings plus original objects, use approximate nearest neighbor indexes to search billions of vectors, and support CRUD and metadata filtering; shows a working pipeline using weaviate with a multilingual model, batch-ingesting 100,000 Wikipedia paragraphs, running semantic queries (gazelle/cheetah for “fast animals”) and piping results to OpenAI for generative retrieval; demo runs end-to-end and returns ranked passages and LLM-generated answers.
AI-written summary · about
chapters (16)
- 0:00intro
- 0:22preface
- 1:00about zain
- 1:42from keyword search to semantic search
- 4:31process, understand and search through unstructured data...
- 5:01in a scalable and secure way
- 5:20use ml to understand the context
- 5:57machine learning models
- 9:14vector representations
- 10:54vector databases
- 17:16weaviate: an open-source vector db
- 18:01a vector search pipeline
- 19:40weaviate is modular: flexible search pipelines
- 21:17using vector search to provide llms context
- 25:12demo
- 43:49reach out
related talks
41mGetting Started With Vector Databases
24mIntroduction to Vector Search with Weaviate
50mA fun and absurd introduction to Vector Databases
01h 46mVector Databases (for GenAI & beyond)
27mUsing Weaviate for building Generative AI apps
48mA fun and absurd introduction to Vector Databases
57mUsing Vector Databases for Multimodal Embeddings and Search
50mWhat Comes After ChatGPT? Vector Databases - the Simple & Powerful Future of ML?