Databricks ML & GenAI Announcements May 2024

lara_rachidi — Tue, 30 Jul 2024 15:44:48 GMT

The Foundation Model Training API (formerly “Finetuning”) is in Gated Public Preview: Training a foundation model is an easy way for you to get high-quality models for a specific task using your own proprietary datasets. You can achieve better performance with a smaller model, requiring less prompt engineering and saving money when serving in production. Databricks now provides a “just works” experience to customers, removing a lot of complexity involved in training such as infrastructure configuration, fault tolerance, hyperparameter selection, and more. Link to blog post: https://docs.databricks.com/en/large-language-models/foundation-model-training/index.html Databricks Vector Search is GA: Vector Search enables developers to improve the accuracy of their Retrieval Augmented Generation (RAG) and generative AI applications through similarity search over unstructured documents such as PDFs, Office Documents, Wikis, and more. This enriches the LLM queries with context and domain knowledge, improving accuracy, and quality of results. New capabilities were added in GA: PrivateLink and IP access lists, support for CMK, improved audit logs and cost attribution tracking. Also, you can now save generated embeddings as a Delta table. Link to blog post: https://www.databricks.com/blog/announcing-mosaic-ai-vector-search-general-availability-databricks

article Databricks ML & GenAI Announcements May 2024 in Databricks TV

Databricks ML & GenAI Announcements May 2024