Embedding Model

An embedding model is a neural network model that encodes discrete text or multimodal inputs into continuous high-dimensional real vectors. Geometric distance in feature space directly reflects semantic similarity.