Datasets:

Modalities:
Text
Libraries:
Datasets
Jon Gauthier commited on
Commit
9d3d980
·
1 Parent(s): 70f4226

Get prediction evaluation working

Browse files
Files changed (2) hide show
  1. syntaxgym.py +12 -1
  2. test.py +8 -0
syntaxgym.py CHANGED
@@ -185,7 +185,18 @@ class SyntaxGymMetric(datasets.Metric):
185
  # up the aggregation output
186
  print("Warning: exceeded ", token)
187
 
188
- region_totals = {c: dict(totals) for c, totals in region_totals.items()}
 
 
 
 
 
 
 
 
 
 
 
189
 
190
  def get_region_edges(self, item_number, condition_name):
191
  """
 
185
  # up the aggregation output
186
  print("Warning: exceeded ", token)
187
 
188
+ region_totals = {(condition_name, region_number): float(total)
189
+ for condition_name, totals in region_totals.items()
190
+ for region_number, total in totals.items()}
191
+
192
+ results = {
193
+ "prediction_results": [
194
+ p.formula(region_totals) for p in self.predictions
195
+ ],
196
+
197
+ "region_totals": region_totals
198
+ }
199
+ return results
200
 
201
  def get_region_edges(self, item_number, condition_name):
202
  """
test.py CHANGED
@@ -1,3 +1,6 @@
 
 
 
1
  import datasets
2
  import numpy as np
3
  import transformers
@@ -18,6 +21,11 @@ model = transformers.AutoModelForCausalLM.from_pretrained(model_ref)
18
  model.eval()
19
 
20
 
 
 
 
 
 
21
  for item in dataset["test"]:
22
  # TODO full preprocessing setup
23
  condition_names = item["conditions"]["condition_name"]
 
1
+ import itertools
2
+ from typing import List
3
+
4
  import datasets
5
  import numpy as np
6
  import transformers
 
21
  model.eval()
22
 
23
 
24
+ # all_sentences: List[List[str]] = [item["conditions"]["content"] for item in dataset["test"]]
25
+ # all_sentences_flat = list(itertools.chain.from_iterable(all_sentences))
26
+
27
+ tokenized = tokenizer(all_sentences_flat,
28
+ return_tensors="pt", padding=True, return_offsets_mapping=True)
29
  for item in dataset["test"]:
30
  # TODO full preprocessing setup
31
  condition_names = item["conditions"]["condition_name"]