Update: loading script TODO test

utcd.py

@@ -134,9 +134,11 @@ class Utcd(datasets.GeneratorBasedBuilder):
         splits = ['train', 'eval', 'test'] if self.config.normalize_aspect else ['train', 'test']
         dnms = self._get_dataset_names()
         dir_nm = self.config.to_dir_name()
-        split2paths = {s: [os_join(dir_nm, f'{dnm}_split', f'{s}.json') for dnm in dnms] for s in splits}
+        # TODO: update root dataset naming version & dataset split naming
+        split2paths = {s: [os_join(f'{dir_nm}_split', dnm, f'{s}.json') for dnm in dnms] for s in splits}
 
-        dl_manager.download_and_extract('datasets.zip')
+        fls = dl_manager.download_and_extract('datasets.zip')
+        mic(fls, type(fls))
         return [
             datasets.SplitGenerator(name=_split2hf_split[s], gen_kwargs=dict(filepath=split2paths[s])) for s in splits
         ]