Skip to content

Latest commit

 

History

History
44 lines (32 loc) · 2.83 KB

README.md

File metadata and controls

44 lines (32 loc) · 2.83 KB

News

! New work on e-SNLI: Make Up Your Mind! Adversarial Generation of Inconsistent Natural Language Explanations. Accepted as a short paper at ACL, 2020.

! New dataset of visual textual entailment with natural language explanations taken from e-SNLI: e-SNLI-VE-2.0: Corrected Visual-Textual Entailment with Natural Language Explanations. At the IEEE CVPR Workshop on Fair, Data Efficient and Trusted Computer Vision, 2020

! If are also interested in feature-based explanations besides natural language explanations, check out our new work on verifying post-hoc explanatory methods: Can I Trust the Explainer? Verifying Post-hoc Explanatory Methods. At NeurIPS 2019 Workshop on Safety and Robustness in Decision Making, 2020.

e-SNLI

There are 2 splits for the train set due to the github sie restrictions, please simply merge them.

Clarification on the two potentially confusing headers:

  • Sentence1_marked_1: is the premise (Sentence2 for hypothesis) were words between star (*) were highlighted by the annotators. The annotators had to click on every word individually to highlight it. The punctuation has not been separated from the words, hence highlighting a word automatically included any punctuation near it. Please use only this header to retrieve the highlighted words simply by retrieving the words between stars without space between them, i.e., things like w1 w2 w3 only w1 and w3 were highlighted.

Please ignore the fields Sentence_Highlighted_ and retrieve the highlighted words from the Sentence_marked_ fields as stated above.

Trained models

Trained models can be downloaded at:

Dependancies

  • Python 2.7
  • Pytorch 0.3.1
  • NLTK >= 3

Bibtex

If you use this dataset or code in your work, please cite our paper:

@incollection{NIPS2018_8163,
title = {e-SNLI: Natural Language Inference with Natural Language Explanations},
author = {Camburu, Oana-Maria and Rockt\"{a}schel, Tim and Lukasiewicz, Thomas and Blunsom, Phil},
booktitle = {Advances in Neural Information Processing Systems 31},
editor = {S. Bengio and H. Wallach and H. Larochelle and K. Grauman and N. Cesa-Bianchi and R. Garnett},
pages = {9539--9549},
year = {2018},
publisher = {Curran Associates, Inc.},
url = {http://papers.nips.cc/paper/8163-e-snli-natural-language-inference-with-natural-language-explanations.pdf}
}