matchcake.ml.optimizer_strategies package¶

Submodules¶

matchcake.ml.optimizer_strategies.adam_strategy module¶

class matchcake.ml.optimizer_strategies.adam_strategy.AdamStrategy¶

Bases: AdamWStrategy

NAME: str = 'Adam'¶

REQUIRES_GRAD = True¶

__init__()¶

set_parameters(parameters, **hyperparameters)¶

matchcake.ml.optimizer_strategies.adamw_strategy module¶

class matchcake.ml.optimizer_strategies.adamw_strategy.AdamWStrategy¶

Bases: OptimizerStrategy

NAME: str = 'AdamW'¶

REQUIRES_GRAD = True¶

REQUIRES_HYPERPARAMETERS = ['learning_rate', 'max_grad_norm']¶

__init__()¶

set_parameters(parameters, **hyperparameters)¶

step(closure: Callable[[List[Parameter] | None], TensorLike], callback: Callable[[], Any] | None = None) → TensorLike¶

matchcake.ml.optimizer_strategies.genetic_strategy module¶

class matchcake.ml.optimizer_strategies.genetic_strategy.GeneticStrategy¶

Bases: OptimizerStrategy

NAME: str = 'Genetic'¶

OPTIONAL_HYPERPARAMETERS = ['init_range_low', 'init_range_high', 'num_parents_mating', 'sol_per_pop', 'parent_selection_type', 'keep_parents', 'crossover_type', 'mutation_type', 'mutation_percent_genes']¶

__init__()¶

fitness_func(ga_instance, solution, solution_idx)¶

get_initial_population()¶

on_generation(ga_instance)¶

optimize(*, n_iterations: int, closure: Callable[[List[Parameter] | None], TensorLike], callback: Callable[[], Any] | None = None, **hyperparameters) → List[Parameter]¶

set_parameters(parameters, **hyperparameters)¶

step(closure: Callable[[List[Parameter] | None], TensorLike], callback: Callable[[], Any] | None = None) → TensorLike¶

matchcake.ml.optimizer_strategies.optimizer_strategy module¶

class matchcake.ml.optimizer_strategies.optimizer_strategy.OptimizerStrategy¶

Bases: ABC

NAME: str = 'OptimizerStrategy'¶

OPTIONAL_HYPERPARAMETERS = []¶

REQUIRES_GRAD = False¶

REQUIRES_HYPERPARAMETERS = []¶

__init__()¶

check_required_hyperparameters(hyperparameters)¶

jac(vector, closure)¶

optimize(*, n_iterations: int, closure: Callable[[List[Parameter] | None], TensorLike], callback: Callable[[], Any] | None = None, **hyperparameters) → List[Parameter]¶

property params_vector¶

set_optional_hyperparameters(hyperparameters, default=None)¶

set_parameters(parameters, **hyperparameters)¶

abstract step(closure: Callable[[List[Parameter] | None], TensorLike], callback: Callable[[], Any] | None = None) → TensorLike¶

vector_to_parameters(vector)¶

class matchcake.ml.optimizer_strategies.optimizer_strategy.ScipyOptimizerStrategy¶

Bases: OptimizerStrategy

NAME: str = 'ScipyOptimizerStrategy'¶

REQUIRES_HYPERPARAMETERS = []¶

get_callback_func(base_callback)¶

optimize(*, n_iterations: int, closure: Callable[[List[Parameter] | None], TensorLike], callback: Callable[[], Any] | None = None, **hyperparameters) → List[Parameter]¶

set_parameters(parameters, **hyperparameters)¶

step(closure: Callable[[List[Parameter] | None], TensorLike], callback: Callable[[], Any] | None = None) → TensorLike¶

matchcake.ml.optimizer_strategies.random_strategy module¶

class matchcake.ml.optimizer_strategies.random_strategy.RandomStrategy¶

Bases: OptimizerStrategy

NAME: str = 'Random'¶

REQUIRES_HYPERPARAMETERS = []¶

__init__()¶

set_parameters(parameters, **hyperparameters)¶

step(closure: Callable[[List[Parameter] | None], TensorLike], callback: Callable[[], Any] | None = None) → TensorLike¶

matchcake.ml.optimizer_strategies.scipy_strategies module¶

class matchcake.ml.optimizer_strategies.scipy_strategies.COBYLAStrategy¶

Bases: ScipyOptimizerStrategy

NAME: str = 'COBYLA'¶

class matchcake.ml.optimizer_strategies.scipy_strategies.L_BFGSStrategy¶

Bases: ScipyOptimizerStrategy

NAME: str = 'L-BFGS'¶

class matchcake.ml.optimizer_strategies.scipy_strategies.SLSQPStrategy¶

Bases: ScipyOptimizerStrategy

NAME: str = 'SLSQP'¶

matchcake.ml.optimizer_strategies.simulated_annealing_strategy module¶

class matchcake.ml.optimizer_strategies.simulated_annealing_strategy.SimulatedAnnealingStrategy¶

Bases: OptimizerStrategy

DEFAULT_TEMPERATURE = 10.0¶

NAME: str = 'SimulatedAnnealing'¶

OPTIONAL_HYPERPARAMETERS = ['seed', 'temperature']¶

REQUIRES_HYPERPARAMETERS = ['learning_rate']¶

__init__()¶

optimize(*, n_iterations: int, closure: Callable[[List[Parameter] | None], TensorLike], callback: Callable[[], Any] | None = None, **hyperparameters) → List[Parameter]¶

set_parameters(parameters, **hyperparameters)¶

step(closure: Callable[[List[Parameter] | None], TensorLike], callback: Callable[[], Any] | None = None) → TensorLike¶

Module contents¶

matchcake.ml.optimizer_strategies.get_optimizer_strategy(name: str | OptimizerStrategy | None) → OptimizerStrategy¶