def createSeedPrompts()

in 1_synthetic-qa-generation/reasoningplaning/evolve.py [0:0]

• 20 lines of code
• 7 McCabe index (conditional complexity)

    def createSeedPrompts(self):
        """
        Turn self.seed_data into a list of strings of text self.source_text_list
        Each text string can represent as little as a word, or as much as document.
        Just has to be representative of some concept or body of text.

        :return: None
        """
        if isinstance(self.seed_data, str) and os.path.exists(self.seed_data):
            data = load_dataset("json", data_files=self.seed_data)
            self.seed_text_dict = dict()
            for d in data['train']:
                s = ""
                if isinstance(self.column_names, str):
                    s = d[self.column_names]
                else:
                    assert False, "column_names must be a str"
                    for col in self.column_names:
                        s += d[col] + "\n"
                # self.seed_text_dict.append(s.strip())
                self.seed_text_dict[s.strip()] = {
                        "idx": d["idx"],
                        GRND_TRUTH_COL: d[GRND_TRUTH_COL]
                    }
                if int(d["idx"]) > self.maxIdx:
                    self.maxIdx = int(d["idx"])
            assert self.seed_text_dict, "data import failed, got empty list"
            self.maxIdx = self.maxIdx + 10