Analysing Neural Language Models: Contextual Decomposition Reveals Default Reasoning in Number and Gender Assignment

Jaap Jumelet, Willem H. Zuidema, Dieuwke Hupkes. Analysing Neural Language Models: Contextual Decomposition Reveals Default Reasoning in Number and Gender Assignment. In Mohit Bansal, Aline Villavicencio, editors, Proceedings of the 23rd Conference on Computational Natural Language Learning, CoNLL 2019, Hong Kong, China, November 3-4, 2019. pages 1-11, Association for Computational Linguistics, 2019. [doi]

@inproceedings{JumeletZH19,
  title = {Analysing Neural Language Models: Contextual Decomposition Reveals Default Reasoning in Number and Gender Assignment},
  author = {Jaap Jumelet and Willem H. Zuidema and Dieuwke Hupkes},
  year = {2019},
  doi = {10.18653/v1/K19-1001},
  url = {https://doi.org/10.18653/v1/K19-1001},
  researchr = {https://researchr.org/publication/JumeletZH19},
  cites = {0},
  citedby = {0},
  pages = {1-11},
  booktitle = {Proceedings of the 23rd Conference on Computational Natural Language Learning, CoNLL 2019, Hong Kong, China, November 3-4, 2019},
  editor = {Mohit Bansal and Aline Villavicencio},
  publisher = {Association for Computational Linguistics},
  isbn = {978-1-950737-72-7},
}