Configuring IBM® watsonx.data Milvus in IBM watsonx.ai
If you are using IBM watsonx.ai, you can integrate Milvus vector database in watsonx.data through the Milvus connector in platform connections.
watsonx.data on IBM Software Hub
watsonx.data Developer edition
IBM watsonx.ai Prompt Lab
IBM watsonx.ai Prompt Lab offers an interface for experimenting with various foundation models through engineered prompts. By incorporating retrieval-augmented generation (RAG) techniques, you can enhance model accuracy and relevance by adding grounding documents that contain context-specific information. These grounding documents, supported in formats like DOCX, PDF, PPTX, and TXT, are first converted into text embeddings by using pre-trained embedding models. Then, these embeddings are indexed by using Milvus vector database for efficient searching during prompt processing, ensuring more reliable and up-to-date responses.
For more information about Prompt Lab, see Prompt Lab.
- Create a Milvus Service in watsonx.data: Set up Milvus in the watsonx.data environment. For more information, see Adding a Milvus service.
- Create the Milvus Connection in platform connection: Establish a connection to the Milvus service through the platform connection settings in watsonx.ai. For more information, Milvus connection.
To have a Milvus index as the grounding data in Prompt Lab, select or create a Milvus index from the Milvus connection. For more information, see Creating a vector index.
IBM watsonx.ai notebook
If you want to run an IBM watsonx.ai notebook to complete the integration with Milvus database in watsonx.data, you can use the Milvus connector and avoid entering the credentials directly into the notebook. For a sample notebook to implement a RAG use case with watsonx.data Milvus and Milvus connector, see sample notebook.