Class BertData

data_utils/corpora.py:404–415

class BertData(PromptReader):
    is_json = False
    PATH = '/dataset/fd5061f6/english_data/wikibook'

    def process_line(self, data, tokenizer, tokenize):
        if data:
            prompt, text = "", data
            prompt, text = self.process_sample(prompt, tokenizer, tokenize), self.process_sample(text, tokenizer,
                                                                                                 tokenize)
            return [prompt], [text]
        else:
            return [], []


class Pile(PromptReader):