hyper_opt_naswot

Hyperparameter optimizer working without training.

class HyperOptNASWOT(params: Parameters, data=None, use_pkl_checkpoints=False)

Bases: HyperOpt

Hyperparameter optimizer that does not require training networks.

Networks are analysed using the Jacobian.

Parameters:

• params (mala.common.parametes.Parameters) – Parameters used to create this hyperparameter optimizer.

• data (mala.datahandling.data_handler.DataHandler) – DataHandler holding the data for the hyperparameter optimization.

perform_study(trial_list=None)

Perform the study, i.e. the optimization.

This is done by sampling a certain subset of network architectures. Currently it is mandatory to provide a trial_list, although it will be optional later on.

Parameters:

trial_list (list) – A list containing trials from either HyperOptOptuna or HyperOptOAT.

Returns:

best_trial_loss – Loss of the best trial.

Return type:

float

set_optimal_parameters()

Set the optimal parameters found in the present study.

The parameters will be written to the parameter object with which the hyperparameter optimizer was created.

property best_trial

Get the best trial determined in this NASWOT run.

This property is read only, and will be recomputed.

property best_trial_index

Get the index and loss of best trial determined in this NASWOT run.

This property is read only, and will be recomputed.

Returns:

best_trial_index – A list containing [0] the best trial index and [1] the best trial loss.

Return type:

list