def build_valid_dataloader()

in empchat/datasets/loader.py [0:0]

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

    def build_valid_dataloader(self, shuffle=True, test=False):
        if self.dataset_name == "dailydialog":
            splitname = "valid" if not test else "test"
            dataset = DDDataset(
                splitname=splitname,
                dic=self.temp_dict,
                data_folder=self.opt.dailydialog_folder,
                maxlen=self.opt.max_sent_len,
                history_len=self.opt.max_hist_len,
            )
        elif self.dataset_name == "empchat":
            splitname = "valid" if not test else "test"
            dataset = EmpDataset(
                splitname,
                self.temp_dict,
                data_folder=self.opt.empchat_folder,
                maxlen=self.opt.max_sent_len,
                reactonly=self.opt.reactonly,
                history_len=self.opt.max_hist_len,
                fasttext=self.opt.fasttext,
                fasttext_type=self.opt.fasttext_type,
                fasttext_path=self.opt.fasttext_path,
            )
        elif self.dataset_name == "reddit":
            dataset = self.build_reddit_dataset(999)
        else:
            raise ValueError("Dataset name unrecognized!")
        return DataLoader(
            dataset,
            batch_size=self.opt.batch_size,
            collate_fn=self.batchify,
            num_workers=0,
            shuffle=shuffle,
            pin_memory=self.opt.cuda,
        )