Skip to main contentnum_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
Default Values (Classification Tasks)
run_mode | Default Value |
|---|
| FAST | 2 |
| NORMAL | 4 |
| BEST | 8 |
Default Values (Other Tasks)
run_mode | Default Value |
|---|
| FAST | 4 |
| NORMAL | 8 |
| BEST | 16 |
