tfma.load_eval_results

Loads results for multiple models or multiple data sets.

tfma.load_eval_results(
    output_paths: Union[str, List[str]],
    output_file_format: Optional[str] = 'tfrecord',
    mode: str = constants.MODEL_CENTRIC_MODE,
    model_name: Optional[str] = None
) -> view_types.EvalResults

Used in the notebooks

Used in the tutorials
TensorFlow Model Analysis

Args
output_paths	A single path or list of output paths of completed tfma runs.
output_file_format	Optional file extension to filter files by.
mode	The mode of the evaluation. Currently, tfma.DATA_CENTRIC_MODE and tfma.MODEL_CENTRIC_MODE are supported.
model_name	Filters to only return results for given model. If unset all models are returned.

Returns
An EvalResults containing the evaluation results serialized at output_paths. This can be used to construct a time series view.