Skip to main content

Embeddings

Also known as: Vector Embeddings, Semantic Embeddings, Text Embeddings

In one sentence

Collections of numerical representations that capture meaning. When you have multiple embeddings, you can compare them to find similar content, power search systems, and enable AI to understand relationships between concepts.

Explain like I'm 12

Imagine turning every book in a library into a GPS coordinate. Books about similar topics would be close together on the map. Embeddings let computers 'see' which ideas are related by checking how close their coordinates are.

In context

When you search in ChatGPT, it converts your question into embeddings and compares them against embeddings of its knowledge to find relevant information. RAG systems store document embeddings in vector databases, then retrieve the closest matches when you ask questions. Recommendation engines use embeddings to find 'customers who liked X also liked Y'.

See also

Related Guides

Learn more about Embeddings in these guides: