baber commited on
Commit
6b06755
1 Parent(s): 5adfba7

Update agieval.py

Browse files

added few_shot split

Files changed (1) hide show
  1. agieval.py +63 -21
agieval.py CHANGED
@@ -16,6 +16,7 @@
16
  import datasets
17
  import json
18
  import ast
 
19
  import csv
20
 
21
  _CITATION = """\
@@ -169,6 +170,7 @@ class AgiEval(datasets.GeneratorBasedBuilder):
169
  "question": datasets.Value("string"),
170
  "options": datasets.features.Sequence(datasets.Value("string")),
171
  "label": datasets.ClassLabel(num_classes=4, names=["A", "B", "C", "D"]),
 
172
  }
173
  )
174
  elif self.config.name == "math_agieval":
@@ -185,6 +187,7 @@ class AgiEval(datasets.GeneratorBasedBuilder):
185
  {"question": datasets.Value("string"),
186
  "options": datasets.features.Sequence(datasets.Value("string")),
187
  "label": datasets.ClassLabel(num_classes=5, names=["A", "B", "C", "D", "E"]),
 
188
  }
189
  )
190
 
@@ -218,30 +221,68 @@ class AgiEval(datasets.GeneratorBasedBuilder):
218
  return splits
219
 
220
  def _generate_examples(self, filepath, split):
 
 
 
 
 
221
  if split == "few_shot":
222
- with open(filepath, mode="r", encoding="utf-8") as csv_file:
223
- reader = csv.DictReader(csv_file)
224
-
225
- # Extract data for the current dataset configuration
226
- print(reader.keys())
227
- for key, row in enumerate(reader):
228
- data_str = row[self.config.name]
229
-
230
- # Skip if there's no data for this key (can happen due to alternating samples and explanations)
231
- if pd.isnull(data_str):
232
- continue
233
-
234
- # Convert the string representation into a dictionary
235
- data = json.loads(data_str.replace("'", "\""))
236
-
237
- # Depending on the dataset, yield the correct format
238
- if self.config.name in ["aqua_rat", "sat_math", "sat_en"]:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
239
  yield key, {
 
 
 
 
 
 
 
 
 
 
240
  "question": data["question"],
241
  "options": data["options"],
242
- "label": data.get("label", None),
243
- "solution": data.get("solution", None)
 
 
 
 
 
 
 
244
  }
 
 
245
  else:
246
  with open(filepath, encoding="utf-8") as f:
247
  for key, row in enumerate(f):
@@ -260,6 +301,7 @@ class AgiEval(datasets.GeneratorBasedBuilder):
260
  "question": data["question"],
261
  "options": data["options"],
262
  "label": data["label"],
 
263
  }
264
  elif self.config.name == "math_agieval":
265
  if not data.get("level"):
@@ -271,7 +313,7 @@ class AgiEval(datasets.GeneratorBasedBuilder):
271
  "answer": data["answer"],
272
  "solution": data["other"]["solution"],
273
  "level": data["level"],
274
- "type": data["type"]
275
  }
276
 
277
  elif self.config.name == "sat_en":
@@ -287,5 +329,5 @@ class AgiEval(datasets.GeneratorBasedBuilder):
287
  "question": data["question"],
288
  "options": data["options"],
289
  "label": data["label"],
 
290
  }
291
-
 
16
  import datasets
17
  import json
18
  import ast
19
+ import pandas as pd
20
  import csv
21
 
22
  _CITATION = """\
 
170
  "question": datasets.Value("string"),
171
  "options": datasets.features.Sequence(datasets.Value("string")),
172
  "label": datasets.ClassLabel(num_classes=4, names=["A", "B", "C", "D"]),
173
+ "solution": datasets.Value("string"),
174
  }
175
  )
176
  elif self.config.name == "math_agieval":
 
187
  {"question": datasets.Value("string"),
188
  "options": datasets.features.Sequence(datasets.Value("string")),
189
  "label": datasets.ClassLabel(num_classes=5, names=["A", "B", "C", "D", "E"]),
190
+ "solution": datasets.Value("string"),
191
  }
192
  )
193
 
 
221
  return splits
222
 
223
  def _generate_examples(self, filepath, split):
224
+ # Mapping for column names in CSV to dataset names
225
+ names = {'aqua_rat': 'aqua-rat', 'sat_en': 'sat-en', 'sat_math': 'sat-math',
226
+ 'lsat_ar': 'lsat-ar', 'lsat_lr': 'lsat-lr', 'lsat_rc': 'lsat-rc',
227
+ 'logiqa': 'logiqa-en', 'math_agieval': 'math'}
228
+
229
  if split == "few_shot":
230
+ # Load the data from the CSV
231
+ df = pd.read_csv(filepath, keep_default_na=False)
232
+
233
+ # Extract samples and explanations
234
+ samples = df[df.index % 2 == 0].reset_index(drop=True)
235
+ explanations = df[df.index % 2 != 0].reset_index(drop=True)
236
+
237
+ for key in range(samples.shape[0]):
238
+ try:
239
+ data = ast.literal_eval(samples[names[self.config.name]][key])
240
+ explanation_row = explanations[names[self.config.name]][key]
241
+ if self.config.name in ["aqua_rat", "sat_math"]:
242
+ yield key, {
243
+ "question": data["question"],
244
+ "options": data["options"],
245
+ "label": data["label"],
246
+ "solution": str(explanation_row),
247
+ }
248
+ elif self.config.name == "logiqa":
249
+ yield key, {
250
+ "passage": data["passage"],
251
+ "question": data["question"],
252
+ "options": data["options"],
253
+ "label": data["label"],
254
+ "solution": str(explanation_row),
255
+ }
256
+ elif self.config.name == "math_agieval":
257
+ if not data.get("level"):
258
+ data["level"] = data['other']['level']
259
+ if not data.get("type"):
260
+ data["type"] = data['other']['type']
261
  yield key, {
262
+ "question": data["question"],
263
+ "answer": data["answer"],
264
+ "level": data["level"],
265
+ "type": data["type"],
266
+ "solution": str(explanation_row),
267
+ }
268
+
269
+ elif self.config.name == "sat_en":
270
+ yield key, {
271
+ "passage": data["passage"],
272
  "question": data["question"],
273
  "options": data["options"],
274
+ "label": data["label"],
275
+ "solution": str(explanation_row),
276
+ }
277
+ elif self.config.name in ['lsat_lr', 'lsat_rc', 'lsat_ar']:
278
+ yield key, {
279
+ "question": data["question"],
280
+ "options": data["options"],
281
+ "label": data["label"],
282
+ "solution": str(explanation_row),
283
  }
284
+ except:
285
+ pass
286
  else:
287
  with open(filepath, encoding="utf-8") as f:
288
  for key, row in enumerate(f):
 
301
  "question": data["question"],
302
  "options": data["options"],
303
  "label": data["label"],
304
+ "solution": data["label"],
305
  }
306
  elif self.config.name == "math_agieval":
307
  if not data.get("level"):
 
313
  "answer": data["answer"],
314
  "solution": data["other"]["solution"],
315
  "level": data["level"],
316
+ "type": data["type"],
317
  }
318
 
319
  elif self.config.name == "sat_en":
 
329
  "question": data["question"],
330
  "options": data["options"],
331
  "label": data["label"],
332
+ "solution": data["label"],
333
  }