keshan commited on
Commit
c226be5
1 Parent(s): 573b8b7

Update wit-dataset.py

Browse files

fixing filepath variable

Files changed (1) hide show
  1. wit-dataset.py +3 -4
wit-dataset.py CHANGED
@@ -1,8 +1,7 @@
1
  import datasets
2
 
3
  logger = datasets.logging.get_logger(__name__)
4
- _DESCRIPTION = """\
5
- Wikipedia-based Image Text (WIT) Dataset is a large multimodal multilingual dataset.
6
  WIT is composed of a curated set of 37.6 million entity rich image-text examples with 11.5 million unique images across 108 Wikipedia languages.
7
  Its size enables WIT to be used as a pretraining dataset for multimodal machine learning models.
8
  """
@@ -193,8 +192,8 @@ class WIT(datasets.GeneratorBasedBuilder):
193
  data_fields = list(self._info().features.keys())
194
  path_idx = data_fields.index("image_url")
195
  # ToDO: Remove after debugging..
196
- print(path_to_data)
197
- with open(path_to_data, encoding="utf-8") as f:
198
  lines = f.readlines()
199
  headline = line[0]
200
 
 
1
  import datasets
2
 
3
  logger = datasets.logging.get_logger(__name__)
4
+ _DESCRIPTION = """\\nWikipedia-based Image Text (WIT) Dataset is a large multimodal multilingual dataset.
 
5
  WIT is composed of a curated set of 37.6 million entity rich image-text examples with 11.5 million unique images across 108 Wikipedia languages.
6
  Its size enables WIT to be used as a pretraining dataset for multimodal machine learning models.
7
  """
 
192
  data_fields = list(self._info().features.keys())
193
  path_idx = data_fields.index("image_url")
194
  # ToDO: Remove after debugging..
195
+ print(filepath)
196
+ with open(filepath, encoding="utf-8") as f:
197
  lines = f.readlines()
198
  headline = line[0]
199