utils

convert_string_to_list(s)

Converts a Pandas Series containing stringified lists into actual lists.

Parameters:

Name	Type	Description	Default
s	Series	Series with stringified lists.	required

Returns:

Type	Description
list[list[float]]	list[list[float]]: List of lists with float values.

Source code in src/run/single_run/utils.py

def convert_string_to_list(s: pd.Series) -> list[list[float]]:
    """
    Converts a Pandas Series containing stringified lists into actual lists.

    Args:
        s (pd.Series): Series with stringified lists.

    Returns:
        list[list[float]]: List of lists with float values.
    """
    return s.apply(ast.literal_eval).tolist()

get_config(config_path)

Load the config for testing.

Source code in src/run/single_run/utils.py

def get_config(config_path: Path) -> Args:
    """
    Load the config for testing.
    """
    output_dir = to_absolute_path(str(config_path))
    overrides = OmegaConf.load(join(output_dir, '.hydra/overrides.yaml'))
    hydra_config = OmegaConf.load(join(output_dir, '.hydra/hydra.yaml'))

    # getting the config name from the previous job.
    config_name = hydra_config.hydra.job.config_name

    # compose a new config from scratch
    cfg = compose(config_name, overrides=overrides)
    updated_cfg = instantiate(cfg, _convert_='object')

    return updated_cfg

instantiate_config(cfg)

Instantiate the config object with the appropriate datamodule and model.

Parameters:

Name	Type	Description	Default
cfg	dict	The configuration object.	required

Returns:

Name	Type	Description
Args	Args	The instantiated configuration object.

Source code in src/run/single_run/utils.py

def instantiate_config(cfg: DictConfig) -> Args:
    """
    Instantiate the config object with the appropriate datamodule and model.

    Args:
        cfg (dict): The configuration object.

    Returns:
        Args: The instantiated configuration object.
    """
    args: Args = instantiate(config=cfg, _convert_='object')
    args.data.full_dataset_name = args.data.dataset_name
    args.model.full_model_name = args.model.model_name
    args.model.max_time_limit = args.model.max_time
    args.model.is_ml = args.model.base_model_name in MLModelNames
    args.model.use_class_weighted_loss = (
        args.model.use_class_weighted_loss
        if len(list(args.data.class_names)) > 1
        else False
    )
    return _configure_model_backbone(args)

update_cfg_with_wandb(cfg)

Update the configuration object with the wandb config. This function will overwrite the config with the wandb config if the wandb config is not empty. Args:

cfg (Args): The configuration object.

Returns: Args: The updated configuration object.

Source code in src/run/single_run/utils.py

def update_cfg_with_wandb(cfg: Args) -> Args:
    """
    Update the configuration object with the wandb config.
    This function will overwrite the config with the wandb config if the
    wandb config is not empty.
    Args:

        cfg (Args): The configuration object.
    Returns:
        Args: The updated configuration object.
    """

    logger.info('Overwriting args with wandb config')

    def validate_and_setattr(
        obj: object, attr_name: str, value: str | int | float
    ) -> None:
        if not hasattr(obj, attr_name):
            raise AttributeError(
                f"Attribute '{attr_name}' does not exist in {obj.__class__.__name__}",
            )
        setattr(obj, attr_name, value)

    for key, value in wandb.config.items():
        if isinstance(value, dict):
            if not hasattr(cfg, key):
                raise AttributeError(
                    f"Attribute '{key}' does not exist in {cfg.__class__.__name__}",
                )
            sub_cfg = getattr(cfg, key)

            for sub_key, sub_value in value.items():
                if isinstance(sub_value, dict):
                    if not hasattr(sub_cfg, sub_key):
                        raise AttributeError(
                            f"Attribute '{sub_key}' does not exist in {key}",
                        )
                    sub_sub_cfg = getattr(sub_cfg, sub_key)

                    for sub_sub_key, sub_sub_value in sub_value.items():
                        logger.info(
                            f'Setting cfg.{key}.{sub_key}.{sub_sub_key} to {sub_sub_value}',
                        )
                        validate_and_setattr(sub_sub_cfg, sub_sub_key, sub_sub_value)
                else:
                    logger.info(f'Setting cfg.{key}.{sub_key} to {sub_value}')
                    validate_and_setattr(sub_cfg, sub_key, sub_value)
        else:
            logger.info(f'Setting cfg.{key} to {value}')
            validate_and_setattr(cfg, key, value)

    return _configure_model_backbone(cfg)