Amazon Bedrock Knowledge Bases now supports binary vector embeddings to build RAG applications

Amazon Bedrock Knowledge Bases now supports binary vector embeddings for building Retrieval Augmented Generation (RAG) applications. This feature is available with Titan Text Embeddings V2 model and Cohere Embed models. Amazon Bedrock Knowledge Bases offers fully-managed RAG workflows to create highly accurate, low latency, secure and customizable retrieval-augmented-generation (RAG) applications by incorporating contextual information from an organization's data sources.

Binary vector embeddings represent document embeddings as binary vectors, with each dimension encoded as a single binary digit (0 or 1). Binary embeddings in RAG applications offer significant benefits in storage efficiency, computational speed, and scalability. They are particularly useful for large-scale information retrieval, resource-constrained environments, and real-time applications.

This new capability is currently supported with Amazon OpenSearch Serverless as vector store. It is supported in all Amazon Bedrock Knowledge Bases regions where Amazon Opensearch Serverless and Amazon Titan Text Embeddings V2 or Cohere Embed are available.

For more information, please refer to the documentation.