def find_mention_head(mention: CementSpan,
predictor: Predictor) -> Optional[int]:
tokens = mention.get_tokens()
predicted_results = predictor.predict(sentence=tokens)
# sanity check
if (
len(tokens) != len(predicted_results['words'])
or not all([a == b for a, b in zip(tokens, predicted_results['words'])])
):
logger.warning(f'Tokenizations do not match: {tokens} - {predicted_results["words"]}')
return None
for i in range(len(tokens)):
if predicted_results['predicted_dependencies'][i] == 'root':
logger.info(f'Mention: {mention} - has head: {predicted_results["words"][i]}')
return i
return 0
def predict_outputs(model: Model,
test_data: List[Dict[str, str]],
predictor: Predictor,
tgt_attr: str) -> List[Dict[str, str]]:
assert tgt_attr in ["positive", "negative"]
outputs = []
for datum in test_data:
predictions = predictor.predict(datum["source"], tgt_attr)['predictions']
final_predictions = predictions
if any(isinstance(el, list) for el in predictions):
final_predictions = predictions[0]
tokens = [model.vocab.get_token_from_index(i, 'tokens') for i in final_predictions]
predicted_sentence = " ".join(tokens)
evaluation = {
"input": datum["source"],
"prediction": predicted_sentence,
"gold": datum["target"]
}
outputs.append(evaluation)
log_predictions(outputs)
return outputs
def predict(
self,
premise='',
hypothesis='',
):
cache_key = '{}->{}'.format(premise, hypothesis)
if cache_key in self._cache:
return self._cache[cache_key]['label_probs']
result = self.predictor.predict(
premise=premise,
hypothesis=hypothesis,
)
# [entailment, contradiction, neutral]
self._cache[cache_key] = result
self.save()
return result['label_probs']
def save(self):
pickle.dump(self._cache, open(self.cache_path, "wb"))
if __name__ == '__main__':
predictor = Predictor()
print(
predictor.predict(
hypothesis=
'The information is straight-forward, and probably easy to verify.',
premise='The sentence provides coherent, verifiable information.'))
