Features

• Fetch connection strings and tell MLflow how to connect with MFlux.ai

• Download and upload objects/datasets from/to the MFlux.ai cloud service

Quickstart

Installation

pip install mflux-ai

Basic usage

import mflux_ai

mflux_ai.init("INSERT_YOUR_PROJECT_TOKEN_HERE")

# MLflow now knows how to connect with your project server, hosted on MFlux.ai

Store and retrieve datasets

my_dataset = np.zeros(shape=(10000, 100), dtype=np.float32)
dataset_filename = "my-dataset.pkl"

mflux_ai.put_dataset(my_dataset, dataset_filename)

my_loaded_dataset = mflux_ai.get_dataset(dataset_filename)

assert_array_equal(my_dataset, my_loaded_dataset)

v0.7.0 (2020-01-14)

• Implement mflux_ai.get_best_run(), which returns the best run in a model group defined in MFlux.ai.

v0.6.0 (2019-12-13)

• Move functions from mflux_ai.mflux_ai to mflux_ai.core. The API stays backwards-compatible for now, but warnings are shown if the old API is used. This change was applied to get cleaner and more consistent import statements.

v0.5.3 (2019-09-23)

• Improve the performance and the support for special characters in object names in mflux_ai.get_dataset by unpickling in memory instead of using a temporary file on disk.

• Don’t expose non-public variables and imports on the top-level package

v0.5.2 (2019-09-20)

• Improve the performance of mflux_ai.put_dataset by pickling in memory instead of using a temporary file on disk.

v0.5.1 (2019-09-12)

• Add support for MLflow authentication

• Improve the performance of mflux_ai.put_dataset

• Correctly reset the MinIO client when init completes successfully

• Specify the desired API version and let the user know if an upgrade is needed

v0.4.0 (2019-09-01)

• Mark mflux_ai.set_env_vars() as deprecated. Use mflux_ai.init() instead.

• Remove support for Python 3.4

• Add support for secure MinIO connections

v0.3.0 (2019-08-16)

• Add a function init that will eventually replace set_env_vars

• Check if the provided project token is valid.

v0.2.1 (2019-08-16)

• Set licence to Apache License 2.0

• Transition from pre-alpha to alpha.

v0.2.0 (2019-08-14)

• Add convenience functions for storing and retrieving datasets

v0.1.1 (2019-08-14)

• First release on PyPI. Has support for setting environment variables for MLflow based on an MFlux.ai project token.