This repository contains code and data for "Assessing Social and Intersectional Biases in Contextualized Word Representations" presented at NeurIPS 2019, which is work in collaboration with L. Elisa Celis.
This repository owes a large debt to https://github.com/W4ngatang/sent-bias for a significant part of the framework and design, upon which this work was done.