Join TensorFlow at Google I/O, May 11-12 Register now

cos_e

  • Description:

Common Sense Explanations (CoS-E) allows for training language models to automatically generate explanations that can be used during training and inference in a novel Commonsense Auto-Generated Explanation (CAGE) framework.

Split Examples
'train' 9,741
'validation' 1,221
  • Feature structure:
FeaturesDict({
    'abstractive_explanation': Text(shape=(), dtype=tf.string),
    'answer': Text(shape=(), dtype=tf.string),
    'choices': Sequence(Text(shape=(), dtype=tf.string)),
    'extractive_explanation': Text(shape=(), dtype=tf.string),
    'id': Text(shape=(), dtype=tf.string),
    'question': Text(shape=(), dtype=tf.string),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
abstractive_explanation Text tf.string
answer Text tf.string
choices Sequence(Text) (None,) tf.string
extractive_explanation Text tf.string
id Text tf.string
question Text tf.string
  • Citation:
@inproceedings{rajani2019explain,
     title = "Explain Yourself! Leveraging Language models for Commonsense Reasoning",
    author = "Rajani, Nazneen Fatema  and
      McCann, Bryan  and
      Xiong, Caiming  and
      Socher, Richard",
      year="2019",
    booktitle = "Proceedings of the 2019 Conference of the Association for Computational Linguistics (ACL2019)",
    url ="https://arxiv.org/abs/1906.02361"
}