# hyperopt.dev

> AutoML tuning suggestions with model training and data kept on your computer.

The Python package is `tuning-client`; the import is `tuning_client`. Public service benchmarks are sequential.

## Documentation

- [hyperopt.dev — Hyperparameter Tuning with Local Training](https://hyperopt.dev/index.md): Compare model configurations with the hyperopt.dev AutoML service. Train locally, report validation losses, and get tuning suggestions through the Python SDK.
- [Python SDK Quickstart | hyperopt.dev](https://hyperopt.dev/docs/quickstart.md): Install tuning-client, run the breast-cancer demo, compare repeated runs, and resolve common installation and authentication errors.
- [How Hyperparameter Tuning Works | hyperopt.dev](https://hyperopt.dev/docs/how-it-works.md): Learn how hyperopt.dev allocates training observations, what data is sent, how model ranking works, and the limits of the current demo.
- [Hyperparameter Tuning Benchmarks and Methodology | hyperopt.dev](https://hyperopt.dev/benchmarks.md): Compare six tuning schedulers across 11 demo datasets. Read the ranking table and benchmark methodology.
- [Privacy and Data Handling | hyperopt.dev](https://hyperopt.dev/privacy.md): How hyperopt.dev handles account records, optional demo events, cookies, feedback, and data retention.
- [Free Beta, Usage Limits and Pricing | hyperopt.dev](https://hyperopt.dev/pricing.md): The hyperopt.dev public beta has no billing. See current session limits and join the waitlist for future paid plans without a purchase commitment.

## Results and package

- [Benchmark data](https://hyperopt.dev/benchmarks/results.json): Per-run results, sample counts, and source hashes.
- [Tuning Client on PyPI](https://pypi.org/project/tuning-client/): SDK, optional demos, and license notices.