MLflow

Provides a brief overview of MLflow in HPE Ezmeral Unified Analytics Software.

MLflow is an open-source platform that manages the end-to-end machine learning lifecycle, including experimentation, reproducibility, deployment, and a central model registry. You can train your ML model and run ML experiments in a fully managed and secured unified environment provided by HPE Ezmeral Unified Analytics Software. To learn more, see open-source MLflow documentation.

The model management framework with MLflow integration in HPE Ezmeral Unified Analytics Software is offered with the following capabilities.
Notebook Integration
Build and Train ML models using MLFlow APIs with an underlying tracking server.
Experiment Tracking
Track experiments and compare the output parameters for various runs.
MLflow Models
Enables users to log all parameters, save artifacts, load models, and deploy models.
Model Artifacts
Log params and save model artifacts to HPE Ezmeral Data Fabric Object Store.
MLflow Registry

A centralized model store, set of APIs, and UI, to collaboratively manage the full lifecycle of an MLflow Model.

Exploring MLflow in HPE Ezmeral Unified Analytics Software

HPE Ezmeral Unified Analytics Software includes sample files and data that you can access through the notebook server instance.

To access the sample files in your notebook server instance:
  1. Sign in to HPE Ezmeral Unified Analytics Software.
  2. In the left navigation pane, click Notebooks.
  3. Connect to your notebook server instance.
  4. To access the sample files, navigate to the mlflow folder in the /<username> directory.
    TIP
    If the /user directory does not contain the sample files, copy the sample files from the /shared/mlflow folder to the /username directory. The /shared directory is accessible to all users. Editing or running examples from the /shared directory is not advised. The /username directory is specific to you and cannot be accessed by other users.