Skip to content
Crystallize Documentation

Optimizers

module crystallize.experiments.optimizers

Section titled “module crystallize.experiments.optimizers”

class Objective

Section titled “class Objective”

Defines the optimization goal.

method Objective.__init__

Section titled “method Objective.__init__”
__init__(
    metric: 'Union[str, List[str]]',
    direction: 'Union[str, List[str]]'
) → None

class BaseOptimizer

Section titled “class BaseOptimizer”

The abstract base class for all optimization strategies.

method BaseOptimizer.__init__

Section titled “method BaseOptimizer.__init__”
__init__(objective: 'Objective')

method BaseOptimizer.ask

Section titled “method BaseOptimizer.ask”
ask() → list[Treatment]

Suggest one or more Treatments for the next trial.

method BaseOptimizer.get_best_treatment

Section titled “method BaseOptimizer.get_best_treatment”
get_best_treatment() → Treatment

Return the best treatment found after all trials.

method BaseOptimizer.tell

Section titled “method BaseOptimizer.tell”
tell(objective_values: 'dict[str, float]') → None

Provide the aggregated objective value(s) for the last trial.