diagNNose: A Library for Neural Activation Analysis

Open Access
Authors
Publication date 2020
Host editors
  • A. Alishahi
  • Y. Belinkov
  • G. Chrupała
  • D. Hupkes
  • Y. Pinter
Book title Proceedings of the Third BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP
Book subtitle BlackboxNLP2020
ISBN (electronic)
  • 9781952148866
Event BlackboxNLP 2020
Pages (from-to) 342-350
Publisher Stroudsburg, PA: The Association for Computational Linguistics
Organisations
  • Interfacultary Research - Institute for Logic, Language and Computation (ILLC)
Abstract In this paper we introduce diagNNose, an open source library for analysing the activations of deep neural networks. diagNNose contains a wide array of interpretability techniques that provide fundamental insights into the inner workings of neural networks. We demonstrate the functionality of diagNNose with a case study on subject-verb agreement within language models.diagNNose is available at https://github.com/i-machine-think/diagnnose.
Document type Conference contribution
Language English
Published at https://doi.org/10.18653/v1/2020.blackboxnlp-1.32
Other links https://github.com/i-machine-think/diagnnose https://slideslive.com/38940638
Downloads
2020.blackboxnlp-1.32 (Final published version)
Permalink to this page
Back