eladsegal commited on
Commit
5349227
1 Parent(s): 327ffe0

Update scrolls.py

Browse files
Files changed (1) hide show
  1. scrolls.py +15 -4
scrolls.py CHANGED
@@ -158,6 +158,7 @@ _QUALITY_CITATION = """\
158
  }
159
  """
160
 
 
161
  class ScrollsConfig(datasets.BuilderConfig):
162
  """BuilderConfig for Scrolls."""
163
 
@@ -186,13 +187,16 @@ class ScrollsConfig(datasets.BuilderConfig):
186
  self.citation = citation
187
  self.url = url
188
 
 
189
  class QualityConfig(ScrollsConfig):
190
  def __init__(self, **kwargs):
191
  super().__init__(**kwargs)
192
  self.hard_only = False
193
 
 
194
  class Scrolls(datasets.GeneratorBasedBuilder):
195
  """The SuperGLUE benchmark."""
 
196
  features = ["id", "pid", "input", "output"]
197
  DEFAULT_WRITER_BATCH_SIZE = 1000 # because Narrative QA is a rather large dataset
198
  BUILDER_CONFIGS = [
@@ -251,7 +255,7 @@ class Scrolls(datasets.GeneratorBasedBuilder):
251
  data_url="https://scrolls-tau.s3.us-east-2.amazonaws.com/quality.zip",
252
  citation=_QUALITY_DESCRIPTION,
253
  url="https://github.com/nyu-mll/quality",
254
- )
255
  ]
256
 
257
  def _info(self):
@@ -268,6 +272,12 @@ class Scrolls(datasets.GeneratorBasedBuilder):
268
  dl_dir = dl_manager.download_and_extract(self.config.data_url)
269
  task_name = _get_task_name_from_data_url(self.config.data_url)
270
  dl_dir = os.path.join(dl_dir, task_name)
 
 
 
 
 
 
271
  return [
272
  datasets.SplitGenerator(
273
  name=datasets.Split.TRAIN,
@@ -286,7 +296,7 @@ class Scrolls(datasets.GeneratorBasedBuilder):
286
  datasets.SplitGenerator(
287
  name=datasets.Split.TEST,
288
  gen_kwargs={
289
- "data_file": os.path.join(dl_dir, "test.jsonl") if self.config.data_files is None else self.config.data_files["test"],
290
  "split": datasets.Split.TEST,
291
  },
292
  ),
@@ -298,11 +308,12 @@ class Scrolls(datasets.GeneratorBasedBuilder):
298
  row = json.loads(line)
299
 
300
  if self.config.name == "quality":
301
- if self.config.hard_only and not row["is_hard"]:
 
302
  continue
303
 
304
  yield row["pid"], row
305
 
306
 
307
  def _get_task_name_from_data_url(data_url):
308
- return data_url.split("/")[-1].split(".")[0]
 
158
  }
159
  """
160
 
161
+
162
  class ScrollsConfig(datasets.BuilderConfig):
163
  """BuilderConfig for Scrolls."""
164
 
 
187
  self.citation = citation
188
  self.url = url
189
 
190
+
191
  class QualityConfig(ScrollsConfig):
192
  def __init__(self, **kwargs):
193
  super().__init__(**kwargs)
194
  self.hard_only = False
195
 
196
+
197
  class Scrolls(datasets.GeneratorBasedBuilder):
198
  """The SuperGLUE benchmark."""
199
+
200
  features = ["id", "pid", "input", "output"]
201
  DEFAULT_WRITER_BATCH_SIZE = 1000 # because Narrative QA is a rather large dataset
202
  BUILDER_CONFIGS = [
 
255
  data_url="https://scrolls-tau.s3.us-east-2.amazonaws.com/quality.zip",
256
  citation=_QUALITY_DESCRIPTION,
257
  url="https://github.com/nyu-mll/quality",
258
+ ),
259
  ]
260
 
261
  def _info(self):
 
272
  dl_dir = dl_manager.download_and_extract(self.config.data_url)
273
  task_name = _get_task_name_from_data_url(self.config.data_url)
274
  dl_dir = os.path.join(dl_dir, task_name)
275
+
276
+ data_files = {} if self.config.data_files is not None else None
277
+ if data_files is not None:
278
+ for split, paths in self.config.data_files.items():
279
+ data_files[split] = paths[0]
280
+
281
  return [
282
  datasets.SplitGenerator(
283
  name=datasets.Split.TRAIN,
 
296
  datasets.SplitGenerator(
297
  name=datasets.Split.TEST,
298
  gen_kwargs={
299
+ "data_file": os.path.join(dl_dir, "test.jsonl") if data_files is None else data_files["test"],
300
  "split": datasets.Split.TEST,
301
  },
302
  ),
 
308
  row = json.loads(line)
309
 
310
  if self.config.name == "quality":
311
+ is_hard = row.pop("is_hard", False)
312
+ if self.config.hard_only and is_hard:
313
  continue
314
 
315
  yield row["pid"], row
316
 
317
 
318
  def _get_task_name_from_data_url(data_url):
319
+ return data_url.split("/")[-1].split(".")[0]