INNER CODE UNIT · Python

that

HendrikStrobelt/detecting-fake-text · backend/api.py:29

        Function that GLTR interacts with to check the probabilities of words

        Params:
        - in_text: str -- The text that you want to check
        - topk: int -- Your desired truncation of the head of the distribution

        Output:
        - payload: dict -- The wrapper for results in this function, described below

        Payload values
        ==============
        bpe_strings: list of str -- Each individual token in the text
        real_topk: list of tuples -- (ranking, prob) of each token
        pred_topk: list of list of tuple -- (word, prob) for all topk
        """
        raise NotImplementedError

    def postprocess(self, token):

View source record →