From n-gram to Attention: How Model Architectures Learn and Propagate Bias in Language Modeling.

Mohsinul Kabir, Tasfia Tahsin, Sophia Ananiadou

Browse the full EMNLP paper archive.