Three features now available in public preview enable you to choose which models/datasets are right for you without having to run any job, seamlessly read/write data from a Fabric datastore; and Model-as-a-Service with inference APIs and hosted fine-tuning.
Model Benchmarks for Curated Models: You can now view the model leaderboard table with models from the Model Catalog, publicly available datasets, and pre-populated metrics.
Create a Fabric OneLake Lakehouse Datastore type: You can now directly consume data assets from a Fabric Lakehouse in AzureML by creating a OneLake Lakehouse datastore type.
Model-as-a-Service with inference APIs and hosted fine-tuning: It enables machine learning professionals to easily integrate foundation models such as Llama 2 from Meta as an application programming interface (API) endpoint to their applications and fine-tune models without having to manage the underlying GPU infrastructure.
