Multi-Vector (Late Interaction) Embedding Models with Sentence Transformers

Sentence Transformers now supports multi-vector, late-interaction embedding models, adding a retrieval approach where queries and documents are represented by multiple token-level vectors instead of a single embedding. This matters because late interaction can improve search quality over dense single-vector embeddings while still fitting into the Sentence Transformers ecosystem for training and deployment.

Source: huggingface.co

← All news