num_experiments

num_experiments: <int> (Optional)

Description

The number of experiments to run during AutoML. AutoML selects hyperparameter configurations via random sampling without replacement (best effort). If the candidate search space is smaller than num_experiments, or constraints/pruning reduce the viable pool, Kumo may reuse a configuration or run fewer unique experiments.

• num_experiments must be > 0.

Supported Task Types

• All

Default Values (Classification Tasks)

Default Values (Other Tasks)