cloudml-hypertune

What it is and what it does

cloudml-hypertune is a helper library for reporting metrics to Google CloudML Engine's hyperparameter tuning service. It provides a simple HyperTune class that writes metric data—including hyperparameter tags, metric values, and global step counts—to a local JSON file that the CloudML Engine service reads and processes.

The package has no runtime dependencies and is designed as a lightweight bridge between your training code and Google's hyperparameter tuning infrastructure. It is abandoned, with its last release in December 2019 and no commits since June 2023, making it suitable only for legacy projects already committed to this integration pattern.

Use it for:

• Report custom metrics during hyperparameter tuning runs on Google CloudML Engine to guide the tuning algorithm.
• Log training progress (loss, accuracy, etc.) at specific global steps for CloudML Engine to evaluate hyperparameter configurations.
• Integrate metric reporting into existing TensorFlow or other ML training scripts running on CloudML Engine infrastructure.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Reports hyperparameter tuning metrics to Google CloudML Engine's hyperparameter tuning service, writing metric data to a local file for the service to consume.

No, unless you are maintaining a legacy Google CloudML Engine hyperparameter tuning project that already depends on this library. The package is abandoned (last release December 2019, no commits since June 2023), targets obsolete Python versions (2.7, 3.5), and modern Google Cloud ML alternatives have superseded it. For new projects, use Google's current ML training and tuning services instead.

Install

cloudml-hypertune on PyPI

pip install cloudml-hypertune

uv add cloudml-hypertune

poetry add cloudml-hypertune

Installing cloudml-hypertune

Before you install

High install friction due to a development-stage tarball release (0.1.0.dev6). Package is abandoned as of 2023-06-02, with no maintenance for over 2 years. Suitable only for legacy Google CloudML Engine projects already locked to this library.

License in practice

Apache 2.0 permissive license allows commercial and private use with minimal restrictions, making licensing itself not a barrier to adoption.

Quickstart

pip install cloudml-hypertune

import hypertune

hpt = hypertune.HyperTune()
hpt.report_hyperparameter_tuning_metric(
    hyperparameter_metric_tag='my_metric_tag',
    metric_value=0.987,
    global_step=1000)

Requires Google CloudML Engine hyperparameter tuning service infrastructure; metrics are written to /tmp/hypertune/output.metric in JSON format.

Verify before relying

• Whether this package works with modern Python versions (classifiers list Python 2.7 and 3.5 only, both EOL)
• Current compatibility with Google CloudML Engine API if the service has evolved since 2019
• Whether /tmp/hypertune/output.metric path is still the correct metric reporting location for current CloudML Engine

Package facts

Evidence: cloudml-hypertune-0.1.0.dev6.tar.gz

Keywords: ml, hyperparameter, tuning

Tags

More Scientific/Engineering packages