How to use the ludwig.utils.data_utils.load_from_file function in ludwig

]
    output_features = [category_feature(vocab_size=2, reduce_input='sum')]

    # Generate test data
    rel_path = generate_data(input_features, output_features, csv_filename)
    input_features[0]['encoder'] = 'parallel_cnn'
    exp_dir_name = run_experiment(
        input_features,
        output_features,
        data_csv=rel_path
    )
    output_feature_name = get_output_feature_name(exp_dir_name)
    experiment_source_data_name = csv_filename.split('.')[0]
    ground_truth = experiment_source_data_name + '.hdf5'

    ground_truth_train_split = load_from_file(ground_truth, output_feature_name,
                                              ground_truth_split=0)
    ground_truth_val_split = load_from_file(ground_truth, output_feature_name,
                                              ground_truth_split=1)
    ground_truth_test_split = load_from_file(ground_truth, output_feature_name)

    test_df, train_df, val_df = obtain_df_splits(csv_filename)
    target_predictions_from_train = train_df[output_feature_name]
    target_predictions_from_val = val_df[output_feature_name]
    target_predictions_from_test = test_df[output_feature_name]
    gtm_name = experiment_source_data_name + '.json'
    ground_truth_metadata = load_json(gtm_name)
    ground_truth_loaded_train_split = np.asarray([
        ground_truth_metadata[output_feature_name]['str2idx'][train_row]
        for train_row in target_predictions_from_train
    ])
    ground_truth_loaded_val_split = np.asarray([

ground_truth,
        ground_truth_split,
        threshold_output_feature_names,
        **kwargs
):
    """Load model data from files to be shown by
    confidence_thresholding_2thresholds_2d_cli

    :param probabilities: Path to experiment probabilities file
    :param ground_truth: Path to ground truth file
    :param ground_truth_split: Type of ground truth split - train, val, test
    :param threshold_output_feature_names: Name of the output feature to visualizes
    :param kwargs: model configuration arguments
    :return None:
    """
    gt1 = load_from_file(
        ground_truth,
        threshold_output_feature_names[0],
        ground_truth_split
    )
    gt2 = load_from_file(
        ground_truth,
        threshold_output_feature_names[1],
        ground_truth_split
    )
    probabilities_per_model = load_data_for_viz(
        'load_from_file', probabilities, dtype=float
    )
    confidence_thresholding_2thresholds_2d(
        probabilities_per_model, [gt1, gt2], threshold_output_feature_names,
        **kwargs
    )

ground_truth_metadata,
        ground_truth_split,
        output_feature_name,
        **kwargs
):
    """Load model data from files to be shown by compare_classifiers_from_pred

    :param predictions: Path to experiment predictions file
    :param ground_truth: Path to ground truth file
    :param ground_truth_metadata: Path to ground truth metadata file
    :param ground_truth_split: Type of ground truth split - train, val, test
    :param output_feature_name: Name of the output feature to visualize
    :param kwargs: model configuration arguments
    :return None:
    """
    gt = load_from_file(ground_truth, output_feature_name, ground_truth_split)
    metadata = load_json(ground_truth_metadata)
    predictions_per_model_raw = load_data_for_viz(
        'load_from_file', predictions, dtype=str
    )
    predictions_per_model = [
        np.ndarray.flatten(pred) for pred in predictions_per_model_raw
    ]
    compare_classifiers_performance_from_pred(
        predictions_per_model, gt, metadata, output_feature_name, **kwargs
    )

ground_truth,
        ground_truth_split,
        output_feature_name,
        **kwargs
):
    """Load model data from files to be shown by
    confidence_thresholding_data_vs_acc_cli.

    :param probabilities: Path to experiment probabilities file
    :param ground_truth: Path to ground truth file
    :param ground_truth_split: Type of ground truth split - train, val, test
    :param output_feature_name: Name of the output feature to visualize
    :param kwargs: model configuration arguments
    :return None:
    """
    gt = load_from_file(ground_truth, output_feature_name, ground_truth_split)
    probabilities_per_model = load_data_for_viz(
        'load_from_file', probabilities, dtype=float
    )
    confidence_thresholding_data_vs_acc(
        probabilities_per_model, gt, **kwargs
    )

probabilities,
        ground_truth,
        ground_truth_split,
        output_feature_name,
        **kwargs
):
    """Load model data from files to be shown by compare_classifiers_from_prob.

    :param probabilities: Path to experiment probabilities file
    :param ground_truth: Path to ground truth file
    :param ground_truth_split: Type of ground truth split - train, val, test
    :param output_feature_name: Name of the output feature to visualize
    :param kwargs: model configuration arguments
    :return None:
    """
    gt = load_from_file(ground_truth, output_feature_name, ground_truth_split)
    probabilities_per_model = load_data_for_viz(
        'load_from_file', probabilities, dtype=float
    )
    compare_classifiers_performance_from_prob(
        probabilities_per_model, gt, **kwargs
    )

probabilities,
        ground_truth,
        ground_truth_split,
        output_feature_name,
        **kwargs
):
    """Load model data from files to be shown by confidence_thresholding.

    :param probabilities: Path to experiment probabilities file
    :param ground_truth: Path to ground truth file
    :param ground_truth_split: Type of ground truth split - train, val, test
    :param output_feature_name: Name of the output feature to visualize
    :param kwargs: model configuration arguments
    :return None:
    """
    gt = load_from_file(ground_truth, output_feature_name, ground_truth_split)
    probabilities_per_model = load_data_for_viz(
        'load_from_file', probabilities, dtype=float
    )
    confidence_thresholding(
        probabilities_per_model, gt, **kwargs
    )

probabilities,
        ground_truth,
        ground_truth_split,
        output_feature_name,
        **kwargs
):
    """Load model data from files to be shown by binary_threshold_vs_metric_cli.

    :param probabilities: Path to experiment probabilities file
    :param ground_truth: Path to ground truth file
    :param ground_truth_split: Type of ground truth split - train, val, test
    :param output_feature_name: Name of the output feature to visualize
    :param kwargs: model configuration arguments
    :return None:
    """
    gt = load_from_file(ground_truth, output_feature_name, ground_truth_split)
    probabilities_per_model = load_data_for_viz(
        'load_from_file', probabilities, dtype=float
    )
    binary_threshold_vs_metric(
        probabilities_per_model, gt, **kwargs
    )

predictions,
        ground_truth,
        ground_truth_split,
        output_feature_name,
        **kwargs
):
    """Load model data from files to be shown by compare_classifiers_predictions

    :param predictions: Path to experiment predictions file
    :param ground_truth: Path to ground truth file
    :param ground_truth_split: Type of ground truth split - train, val, test
    :param output_feature_name: Name of the output feature to visualize
    :param kwargs: model configuration arguments
    :return None:
    """
    gt = load_from_file(ground_truth, output_feature_name, ground_truth_split)
    predictions_per_model = load_data_for_viz(
        'load_from_file', predictions, dtype=str
    )
    compare_classifiers_predictions(predictions_per_model, gt, **kwargs)

ground_truth,
        ground_truth_split,
        threshold_output_feature_names,
        **kwargs
):
    """Load model data from files to be shown by
    confidence_thresholding_2thresholds_3d_cli

    :param probabilities: Path to experiment probabilities file
    :param ground_truth: Path to ground truth file
    :param ground_truth_split: Type of ground truth split - train, val, test
    :param threshold_output_feature_names: Names of the output features to visualize
    :param kwargs: model configuration arguments
    :return None:
    """
    gt1 = load_from_file(
        ground_truth,
        threshold_output_feature_names[0],
        ground_truth_split
    )
    gt2 = load_from_file(
        ground_truth,
        threshold_output_feature_names[1],
        ground_truth_split
    )
    probabilities_per_model = load_data_for_viz(
        'load_from_file', probabilities, dtype=float
    )
    confidence_thresholding_2thresholds_3d(
        probabilities_per_model, [gt1, gt2], threshold_output_feature_names,
        **kwargs
    )

probabilities,
        ground_truth,
        ground_truth_split,
        output_feature_name,
        **kwargs
):
    """Load model data from files to be shown by compare_classifiers_subset.

    :param probabilities: Path to experiment probabilities file
    :param ground_truth: Path to ground truth file
    :param ground_truth_split: Type of ground truth split - train, val, test
    :param output_feature_name: Name of the output feature to visualize
    :param kwargs: model configuration arguments
    :return None:
    """
    gt = load_from_file(ground_truth, output_feature_name, ground_truth_split)
    probabilities_per_model = load_data_for_viz(
        'load_from_file', probabilities, dtype=float
    )
    compare_classifiers_performance_subset(
        probabilities_per_model, gt, **kwargs
    )