Vector Spacesmath

A set of vectors closed under addition and scalar multiplication, with notions of basis, dimension, and norm.

Embeddings are points in a high-dimensional vector space; distance and angle there encode semantic relations.

Where it appears

Descend

Prerequisites

Dependencies

Foundations

Historical evolution

Hilbert spaces formalized infinite-dimensional analogs; the setting for all embedding geometry.

Implementation details

Embedding dimension (1536 in upstream OpenAI, 256 after random-projection reduction).

Sources

View in graph →