Embeddings

Distance measures such as cosine similarity or Euclidean distance can compare embeddings for semantic relatedness. A vector is a fixed-length sequence of floating-point numbers. Embeddings are mathematical representations of data expressed as lists of numbers. Embeddings allow semantic similarity to map onto spatial proximity. Each embedding number encodes some aspect of the data's meaning or characteristics. Machine learning models produce embeddings as numeric representations in a standard output format. Vectors can encode semantic text meaning, visual image features, or user preference signals.