def build_all_evaluation_batch_list()

in E2E_TOD/dataclass.py [0:0]

• 45 lines of code
• 6 McCabe index (conditional complexity)

    def build_all_evaluation_batch_list(self, ref_bs, ref_act, ref_db, input_contain_db, eva_batch_size, eva_mode):
        '''
            ref_bs: whether using reference belief state to perform generation
                    if with reference belief state, then we also use reference db result
                    else generating belief state to query the db
            ref_act: whether using reference dialogue action to perform generation
                    if true: it always means that we also use reference belief state
                    if false: we can either use generated belief state and queried db result or
                              use reference belief state and reference db result
            eva_mode: 'dev' or 'test'; perform evaluation either on dev set or test set
            eva_batch_size: size of each evaluated batch
        '''
        if eva_mode == 'dev':
            data_list = self.dev_data_list
            eva_num = self.dev_num
        elif eva_mode == 'test':
            data_list = self.test_data_list
            eva_num = self.test_num
        else:
            raise Exception('Wrong Evaluation Mode!!!')

        all_bs_input_id_list, all_da_input_id_list, all_nlg_input_id_list, all_parse_dict_list = \
        [], [], [], []
        for item in data_list:
            one_bs_input_id_list, one_da_input_id_list, one_nlg_input_id_list, one_parse_dict = \
            self.parse_one_eva_instance(item, ref_bs, ref_act, ref_db, input_contain_db)
            all_bs_input_id_list.append(one_bs_input_id_list)
            all_da_input_id_list.append(one_da_input_id_list)
            all_nlg_input_id_list.append(one_nlg_input_id_list)
            all_parse_dict_list.append(one_parse_dict)
        assert len(all_bs_input_id_list) == len(all_da_input_id_list)
        assert len(all_da_input_id_list) == len(all_nlg_input_id_list)
        assert len(all_nlg_input_id_list) == len(all_parse_dict_list)
        bs_batch_list = self.build_batch_list(all_bs_input_id_list, eva_batch_size)
        da_batch_list = self.build_batch_list(all_da_input_id_list, eva_batch_size)
        nlg_batch_list = self.build_batch_list(all_nlg_input_id_list, eva_batch_size)
        parse_dict_batch_list = self.build_batch_list(all_parse_dict_list, eva_batch_size)

        batch_num = len(bs_batch_list)
        final_batch_list = []
        for idx in range(batch_num):
            one_final_batch = [bs_batch_list[idx], da_batch_list[idx], nlg_batch_list[idx], parse_dict_batch_list[idx]]
            if len(bs_batch_list[idx]) == 0: 
                continue
            else:
                final_batch_list.append(one_final_batch)
        return final_batch_list