Chandler May, Alex Wang, Shikha Bordia, Samuel R. Bowman, Rachel Rudinger
TL;DR本研究在Word Embedding Association Test的基础上，对句子编码进行了偏差测量，实验包括了包括ELMo和BERT在内的多种方法，并提议了未来的研究方向。
Abstract
The word embedding association test shows that GloVe and word2vec word embeddings exhibit human-like implicit biases based on gender, race, and other social constructs (Caliskan et al., 2017). Meanwhile, research on learning reusable text representations has begun to explore sentence-l