corpus.py 文件源码

def randomly_grouped_by(key_from_example: Callable[[LabeledExample], Any], training_share: float = .9) -> Callable[
        [List[LabeledExample]], Tuple[List[LabeledExample], List[LabeledExample]]]:
        def split(examples: List[LabeledExample]) -> Tuple[List[LabeledExample], List[LabeledExample]]:
            examples_by_directory = group(examples, key=key_from_example)
            directories = examples_by_directory.keys()

            # split must be the same every time:
            random.seed(42)
            keys = set(random.sample(directories, int(training_share * len(directories))))

            training_examples = [example for example in examples if key_from_example(example) in keys]
            test_examples = [example for example in examples if key_from_example(example) not in keys]

            return training_examples, test_examples

        return split