From Documents to Vectors: Understanding ChatGPT's Technical Architecture with OpenAI Plugins
Discover AI via YouTube
Earn a Michigan Engineering AI Certificate — Stay Ahead of the AI Revolution
NY State-Licensed Certificates in Design, Coding & AI — Online
Overview
Google, IBM & Meta Certificates – 40% Off
One plan covers every Professional Certificate on Coursera.
Unlock All Certificates
Learn about the technical intricacies of OpenAI's API and plugins in this 20-minute video that explores real-time data retrieval with ChatGPT. Explore vector databases and their role in storing document representations, comparing open-source options like Weaviate and Milvus with commercial solutions such as Azure's Cognitive Search. Master the fundamentals of OpenAI's retrieval plugin, built using FastAPI, and understand its four main API endpoints: upsert, upsert-file, query, and delete. Dive into the technical architecture where plugins consist of an API, API schema, and manifest JSON file, while learning how ADA-002 embeddings enable sophisticated semantic search capabilities. Discover the integration possibilities with various systems including Google's Search API, YouTube's Data API, and ScholarAI for archive pre-print server access. Grasp the concept of hybrid search, which combines neural and index search methods for enhanced query results on GPT-4, and understand the importance of security through authentication tokens for both OpenAI and vector database providers.
Syllabus
Introduction
API Retrieval
API Endpoints
Vector Database
Hybrid Search
Cognitive Search
Term Frequency
Log E Function
Summary
Outro
Taught by
Discover AI