Results package

Submodules

Results.extract_from_configuration module

Results.extract_from_configuration.create_model(data_config: dict, model_config: dict, dataset: WeatherDataset, device: device) → ACW

Creates the model from the model configurations.

Args:

model_config (dict): Model configuration

Results.extract_from_configuration.get_dataset(data_config: dict, device: device)

Gets the dataset in the proper format.

Args:

data_config (dict): Dataset configuration

Results.extract_from_configuration.get_training_components(model: ACW, training_config: dict, data_config: dict)

Returns the loss function, optimizer and scheduler.

Args:

model (nn.Module): The ML model training_config (dict): Training configuration

Results.extract_from_configuration.str_to_class(classname: str)

Translates class names from configuration files to classes.

Results.parser module

class Results.parser.ConfigManager(config_path)

Bases: object

Manages the configuration files, containing model, data and training details.

get_data_config()

Retrieve data configuration.

get_model_config()

Retrieve model configuration.

get_output_config()

Retrieve output configuration.

get_training_config()

Retrieve training configuration.

load_config()

Load configuration from a YAML file.

save_config(output_path)

Save configuration to a YAML file.

Results.run_configuration module

Results.run_configuration.run_configuration(config_name, verbose=False, useConfigName=False, device=None, max_batches=None)

Runs a configuration file.

Results.run_configuration.run_folder(folder_name: str, skip_files=[], verbose=False, useConfigName=False, device=None, max_batches=None)

Runs each configuration within a folder.

Results.save_results module

class Results.save_results.DataSaver(config, run_number=None, run_name=None)

Bases: object

Class responsible for saving (1) trained models with (2) their configuration and (3) training and evaluation metrics.

end_run(model)

get_base_dir()

get_run_dir()

log_metrics(log_dict)

save_config(output_dir)

Save configuration as a YAML file.

save_metrics(output_dir)

Save metrics as a JSON file.

save_model(output_dir, model)

Save PyTorch model in the TorchScript format.

wandb_summary(metric)