In recent years, pretrained word embeddings have proved useful for multimodal
neural machine translation (NMT) models to address the shortage of available
datasets. However, the integration of pretrained word embeddings
本研究使用 Word Embeddings Association Test (WEAT)、Clustering 和 Sentence Embeddings Association Test (SEAT) 等方法,衡量荷兰语词嵌入中的性别偏见,并使用 Hard-Debias 和 Sent-Debias 调控方法,探索性别偏见对下游任务的影响。结果表明,传统和上下文嵌入中存在性别偏见,研究人员提供了翻译荷兰语数据集和减轻偏误的嵌入。