Master Vector Databases

Create embeddings using OpenAI embeddings model and explore how text becomes high-dimensional vectors, guided by a practical Colab example. Understand API keys, pricing, and the model text-embedding-002.

Learn to use SQLite for vector search by creating a simple database, performing create, insert, and select operations, and storing vectors as vector data in Python in Google Colab.

master vector databases by using sqlite as a vector storage, converting numpy embeddings to bytes in blob fields, then retrieving and deserializing with numpy.frombuffer.

Explore how to store and deserialize vectors in binary format, then find the nearest vector to a numpy array using SQL calculation; learn about vector search extensions for semantic search.

Enable vector semantic search in SQLite with the sqlite-vss extension. Use LangChain and syntax transformer embeddings to split text into chunks and perform similarity queries.

Explore chroma, an open source AI native vector database with built-in embedding capabilities, showing how to create collections, add documents with metadata and IDs, and perform semantic search with filtering.

Explore how to manage vector databases by working with collections: count items, retrieve all items, and add or upsert documents with metadata and embeddings.

Store matrix movie quotes in a Chroma vector database, creating Neo, Morpheus, and Trinity collections, and explore adding, editing, deleting, and querying with cosine distance.

Learn to add and manage document associated embeddings in a vector database, creating collections, attaching embeddings and metadata, and querying by text or IDs with cosine distance.

Learn to query data with where filters in a vector database, filtering by speaker to retrieve quotes, while also performing updates, upserts, and deletions by IDs or conditional where clauses.

Learn to build a persistent chroma vector store, embed multiple documents with OpenAI embeddings, and run retrieval QA across them using LangChain in Google Colab.

Learn how to set up ChromaDB, create a retriever, and perform multi-document QA with LangChain and GPT-3.5 turbo, controlling outputs and source documents.

Explore Faiss and the Facebook AI similarity search library for efficient vector search, high-dimensional indexing, and fast nearest-neighbor queries using l2 distance and diverse index methods.

Explore how to perform similarity search to find nearest neighbors by comparing a random query vector against a set of data vectors, using L2 (euclidean) distance.

Discover Pinecone, a fully managed, highly scalable vector database that stores billions of vectors, enables fast live vector search, embedding model integration, and metadata filters for precise queries.

Create a pinecone account, build an index, and review the dashboard to manage API keys, environments, and upsert or add records to vectors.

Explore creating and configuring a pinecone index with dimension, matrix, and distance metrics. Understand pod types and sizes, capacity planning, and that an index's pod type cannot be changed.

Master index management for vector databases using pinecone in a Google Colab workflow, including creating, describing, listing, querying, and deleting indices with cosine as the metric.

Insert vector data into the index with upsert, formatting as a list of id and vector tuples, optionally adding metadata and using pandas for zip-based upserts.

Learn how to query vector data in Pinecone, retrieving nearest neighbors with similarity scores, optionally include values, and metadata filters like sector automobile to refine results.

Learn how to upsert vector data in batches to a pinecone index, generating IDs and 64-d vectors, and chunking data for efficient batch uploads, with parallel upserts previewed.

Learn to upsert data in parallel using asynchronous upserts, configuring pool threads and batch sizes, and retrieve async results for vector data in a Pinecone index.

Ensure vector IDs are strings; numeric IDs trigger errors during upsert, so convert IDs to strings (for example by using uuid or timestamp) to enable correct updates or inserts.

Learn to generate text embeddings with syntax transformer and OpenAI, create a 384-dim vector index in Pinecone using cosine similarity, and upsert and query vector data.

Learn to build semantic search over news articles using embeddings and metadata filtering with Pinecone, including data preparation, upserting vectors, and filtered queries.

Explore the open source Qdrant vector database, a cloud-native, scalable solution with a free 1 GB tier, docker quickstart, and a Python library for OpenAI embeddings.

Connect to the quadrant vector database by obtaining an API key, then call the cluster's HTTP endpoints with simple get or post requests to explore collections and vectors.

Connect to a fresh qdrant cluster via Python, install and import the qdrant client and LangChain components, configure host and API key, and prep for creating a collection.

Create or recreate a vector collection with a Python client by specifying the name, vector size, and cosine distance, then verify it on the cluster.

Create a vector store with LangChain by configuring a collection and cosine distance, then wire OpenAI embeddings and API keys to store and index vectors.

Add documents to the cloud vector store by creating embeddings, splitting long text into overlapping chunks with a text splitter, and uploading to a collection.

Query the document by embedding it into a vector store with OpenAI, using LangChain retrievers and a retrieval QA chain to extract answers and key points.