Skip to content
Asynchronous Distributed Hyperparameter Optimization.
Python Shell
Branch: master
Clone or download
Latest commit acb106a Jul 23, 2019
Type Name Latest commit message Commit time
Failed to load latest commit information.
conda Add new `status` command (#222) Jul 23, 2019
docs Add documentation for `status` and `list` (#228) Jul 23, 2019
src/orion Bring back space.{un}pack_point but deprecate them Jul 23, 2019
.gitattributes Adapt codebase to new name Orion (#61) Apr 20, 2018
.gitignore Fix Experiment tests Jul 23, 2019
.travis.yml Add tests for python 3.6 on OSX Jul 8, 2019
LICENSE Remove `Team` from `Epistímio Team` Oct 25, 2018 Add Roadmap Jul 8, 2019
README.rst Fix proper install command in README Jul 8, 2019 Update Roadmap for release v0.1.5 Jul 23, 2019
codecov.yml Adapt codebase to new name Orion (#61) Apr 20, 2018
dev-requirements.txt Add flake8 == 3.5.0 to dev-requirements.txt May 15, 2018
setup.cfg Fix version returned by versioneer Nov 5, 2018 Add new `status` command (#222) Jul 23, 2019
tox.ini Add documentation for `status` and `list` (#228) Jul 23, 2019 Add versioneer Oct 9, 2017



Current PyPi Version Supported Python Versions BSD 3-clause license Documentation Status Codecov Report Travis tests

Oríon is an asynchronous framework for black-box function optimization.

Its purpose is to serve as a meta-optimizer for machine learning models and training, as well as a flexible experimentation platform for large scale asynchronous optimization procedures.

Core design value is the minimum disruption of a researcher's workflow. It allows fast and efficient tuning, providing minimum simple non-intrusive (not even necessary!) helper client interface for a user's script.

So if ./ --mini-batch=50 looks like what you execute normally, now what you have to do looks like this:

orion -n experiment_name ./ --mini-batch~'randint(32, 256)'

Check out user's guide-101 for the simplest of demonstrations!


As simple and as complex you want

  • Simple and natural, but also explicit and verbose, search domain definitions
  • Minimal and non-intrusive client interface for reporting target function values
  • Database logging (currently powered by MongoDB)
  • Flexible configuration
  • Explicit experiment termination conditions
  • Algorithms algorithms algorithms: Skopt's bayesian optimizers are at hand without writing. Random search is the default. only a single line of code.
  • More algorithms: Implementing and distributing algorithms is as easy as possible! Check developer's guide-101. Expect algorithm plugins to pop out quickly!
  • Came up with an idea? Your intuition is still at play: Help your optima hunter now by a command line interface.
  • And other many more already there or coming soon!


Install Oríon by running:

pip install orion.core

For more information read the full installation docs.

Contribute or Ask

Do you have a question or issues? Do you want to report a bug or suggest a feature? Name it! Please contact us by opening an issue in our repository below:

Start by starring and forking our Github repo!

Thanks for the support!


You can find our roadmap here:


The project is licensed under the BSD license.

You can’t perform that action at this time.