Generalised Contextual Decomposition

In the 2019 CONLL paper Analysing neural language models: Contextual Decomposition Reveals Default Reasoning in Number and Gender Assignment, we propose Generalised Contextual Decomposition. We show its use through two case studies with a pretrained neural language model. You can find:

• The paper here

• The code here

If you have any questions, please contact Jaap Jumelet!