migrate-from-model-serving

To migrate an MLflow ResponsesAgent from Model Serving to Databricks Apps, run the migration command with your desired app name. The process transforms existing predict and predict_stream methods into invoke and stream decorators for the new architecture.

Yes, you can convert Model Serving synchronous predict methods to asynchronous Databricks Apps. The migration supports both synchronous and asynchronous operation modes based on your preference when configuring the target app.

Yes, you need Databricks CLI profiles configured before migrating MLflow agents to Databricks Apps. The migration process requires these profiles alongside MLflow and Databricks SDK dependencies to successfully execute the transition.

Model Serving uses predict and predict_stream methods for synchronous operations, while Databricks Apps uses invoke and stream decorators. This migration enables a more scalable architecture and enhances deployment flexibility for MLflow agents.

Yes, the MLflow agent migration to Databricks Apps supports streaming responses. It transforms the existing predict_stream methods from Model Serving into dedicated stream decorators within the Databricks Apps architecture.