Initial commit

docbank.py

@@ -170,6 +170,7 @@ class Docbank(datasets.GeneratorBasedBuilder):
         # By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
         urls = _URLS[self.config.name]
         data_dir = dl_manager.download_and_extract(urls)
+        print(data_dir)
         train_txts = glob(data_dir + '/train/txt/*.csv')
         train_data = [(txt, data_dir + '/train/img/' + os.path.basename(txt)[:-4] + '.jpg') for txt in train_txts]
         test_txts = glob(data_dir + '/test/txt/*.csv')