Run Optuna Neural HPO Study

Execute one persisted Optuna study against the repository neural trainers.

scripts.training.run_optuna_neural_hpo_study.parse_command_line_arguments()

Parse command-line arguments.

Return type:

Namespace

scripts.training.run_optuna_neural_hpo_study.configure_gpu_visibility(gpu_id_text)

Pin the current process to one requested GPU before importing torch.

Parameters:

gpu_id_text (str)

Return type:

None

scripts.training.run_optuna_neural_hpo_study.resolve_objective_metric(metrics_snapshot_dictionary, metric_name)

Resolve one scalar objective metric from the saved metrics snapshot.

Parameters:

• metrics_snapshot_dictionary (dict[str, Any])

• metric_name (str)

Return type:

float

scripts.training.run_optuna_neural_hpo_study.build_trial_training_config(base_training_config, study_config_dictionary, sampled_parameter_dictionary, trial_number, shared_training_infrastructure)

Build one prepared training configuration for a single Optuna trial.

Parameters:

• base_training_config (dict[str, Any])

• study_config_dictionary (dict[str, Any])

• sampled_parameter_dictionary (dict[str, Any])

• trial_number (int)

• shared_training_infrastructure (Any)

Return type:

dict[str, Any]

scripts.training.run_optuna_neural_hpo_study.save_trial_result_snapshot(study_output_root, trial_number, payload)

Persist one per-trial result snapshot under the study output root.

Parameters:

• study_output_root (Path)

• trial_number (int)

• payload (dict[str, Any])

Return type:

None

scripts.training.run_optuna_neural_hpo_study.main()

Run the Optuna study execution entry point.

Return type:

None