Configuration

Configuration

A Configuration is an dict-like object, going from the name of selected hyperparameters to the values.

a: 6
b: dog
6

Underneath the hood, there is some vectorized representation of the configuration, which in this case may look like np.array([0.32, 1]) which stands for {"a": 3.2, "b": "dog"}. This vectorized representation can be useful for optimizer numerical optimization algorithms. You can access it with configuration.get_array().

Tip

All Configuration have a reference to the underlying ConfigurationSpace which can be access with Configuration.config_space.

For more, please check out the API documentation for Configuration.