Unverified Commit 0513f8d2 authored by Victor SANH's avatar Victor SANH Committed by GitHub
Browse files

correct label extraction + add note on discrepancies on trained MNLI model and HANS (#6221)

parent 3c289fb3
......@@ -255,7 +255,11 @@ class HansProcessor(DataProcessor):
return self._create_examples(self._read_tsv(os.path.join(data_dir, "heuristics_evaluation_set.txt")), "dev")
def get_labels(self):
"""See base class."""
"""See base class.
Note that we follow the standard three labels for MNLI
(see :class:`~transformers.data.processors.utils.MnliProcessor`)
but the HANS evaluation groups `contradiction` and `neutral` into `non-entailment` (label 0) while
`entailment` is label 1."""
return ["contradiction", "entailment", "neutral"]
def _create_examples(self, lines, set_type):
......@@ -268,7 +272,7 @@ class HansProcessor(DataProcessor):
text_a = line[5]
text_b = line[6]
pairID = line[7][2:] if line[7].startswith("ex") else line[7]
label = line[-1]
label = line[0]
examples.append(InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label, pairID=pairID))
return examples
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment