Three options now obtainable in preview allow you to take pleasure in simplified administration of the client-managed key (CMK) encrypted workspace, make the most of our content material filtering system alongside fashions you could entry by Models as a Service, and usher in indexes from Cosmos DB for MongoDB vCore and Pinecone for use in RAG workflows. Service-side encryption of metadata (Simplified CMK structure): You can now make the most of the client-managed key-encryption workspace’s new structure, which is able to finally cut back prices and restrict the chance of operating into Azure coverage conflicts. Protect your self and your customers from probably dangerous content material: You can now guard towards probably dangerous content material for you and your customers, and make the most of the flexibleness to show content material filtering on or off primarily based on the wants of what you are promoting. Use RAG patterns throughout a number of information sources in AzureML & AI Studio: You will quickly be capable to deliver your personal indexes with out having to create them from scratch, permitting you to satisfy clients the place information is already obtainable and supplying you with the flexibleness to make use of extra index varieties.
https://azure.microsoft.com/hu-hu/updates/azure-machine-learning-public-preview-for-build-2024/
