Embedding

ELI5 — The Vibe Check

An embedding is turning words, sentences, or entire documents into lists of numbers (vectors) that capture their meaning. The magic is that similar meanings end up as similar numbers — so 'dog' and 'puppy' will be close together in number-space while 'dog' and 'democracy' will be far apart. It's how AI understands meaning.

Real Talk

An embedding is a dense, fixed-dimensional vector representation of discrete data (text, images, etc.) learned by a neural network. In the embedding space, semantic similarity corresponds to vector proximity, enabling operations like similarity search and clustering. Text embeddings are produced by encoder models and used for retrieval, classification, and RAG.

Show Me The Code

from openai import OpenAI
client = OpenAI()
response = client.embeddings.create(
    input="The quick brown fox",
    model="text-embedding-3-small"
)
vector = response.data[0].embedding  # list of 1536 floats

When You'll Hear This

"Generate an embedding for each document, then compare cosine similarity." / "Embeddings power our semantic search."