A Predictive Factor Analysis of Social Biases and Task-Performance in Pretrained Masked Language Models
This work addresses the problem of understanding and mitigating social biases in language models for researchers and practitioners, but it is incremental as it builds on prior studies by systematically analyzing known factors.
The study investigated how various factors like model size, training data, and tokenization influence social biases and task performance in pretrained masked language models, analyzing 39 models to identify key overlooked factors such as tokenization and training objectives.
Various types of social biases have been reported with pretrained Masked Language Models (MLMs) in prior work. However, multiple underlying factors are associated with an MLM such as its model size, size of the training data, training objectives, the domain from which pretraining data is sampled, tokenization, and languages present in the pretrained corpora, to name a few. It remains unclear as to which of those factors influence social biases that are learned by MLMs. To study the relationship between model factors and the social biases learned by an MLM, as well as the downstream task performance of the model, we conduct a comprehensive study over 39 pretrained MLMs covering different model sizes, training objectives, tokenization methods, training data domains and languages. Our results shed light on important factors often neglected in prior literature, such as tokenization or model objectives.