This repository provides a MLflow Plugin that provides backend implementation for running MLFlow project on Databricks Container without uploading project files to DBFS.
User can simply install this plugin by running pip install mlflow-databricks-container
inside mlflow environment. MLflow will register this plugin as an entrypoint with databricks-container
backend.
To run an MLflow project, use the command:
mlflow run <uri> -b databricks-container --backend-config <json-new-cluster-spec>