obsideo-cloud-mlflow 0.1.0b1

MLflow artifact store plugin — stores artifacts with mlvault encrypted cloud storage

mlvault-mlflow

MLflow artifact store plugin for mlvault — store your MLflow artifacts with encrypted cloud storage.

Install

pip install mlvault mlvault-mlflow
mlvault init

Usage

Set the artifact store when creating an MLflow experiment:

import mlflow

mlflow.create_experiment("my-experiment", artifact_location="mlvault://my-project")
mlflow.set_experiment("my-experiment")

Then log artifacts as normal:

with mlflow.start_run() as run:
    mlflow.log_artifact("checkpoint.pt")
    run_id = run.info.run_id

After training, push to cloud storage:

mlvault commit <mlflow_run_id>

See the mlvault README for full documentation.

License

MIT