Dimensionality

Dimensionality is the number of dimensions in an embedding vector—for example, 768 or 1024. Each dimension is a single numeric value; together they represent the semantic meaning of the text. The embedding model determines dimensionality.

Impact

• Storage — Higher dimensionality increases vector index size
• Similarity — Cosine similarity and other metrics work on vectors of fixed dimensionality
• Model lock-in — Switching embedding model usually requires re-indexing; dimensions must match for comparison

DocLD uses Pinecone with llama-text-embed-v2, which defines the dimensionality for your vectors.

Related Concepts

Dimensionality is determined by the embedding model. Vector search and similarity score operate on vectors of the same dimensionality. Vector index storage scales with dimensionality.