Gender Encoding Patterns in Pretrained Language Model Representations
This addresses gender bias in language models, which poses social and ethical challenges, but is incremental as it builds on existing bias analysis methods.
The study analyzed how gender biases are encoded in pretrained language models using an information-theoretic approach, finding that debiasing techniques often have limited efficacy and can inadvertently increase internal bias while reducing it in outputs.
Gender bias in pretrained language models (PLMs) poses significant social and ethical challenges. Despite growing awareness, there is a lack of comprehensive investigation into how different models internally represent and propagate such biases. This study adopts an information-theoretic approach to analyze how gender biases are encoded within various encoder-based architectures. We focus on three key aspects: identifying how models encode gender information and biases, examining the impact of bias mitigation techniques and fine-tuning on the encoded biases and their effectiveness, and exploring how model design differences influence the encoding of biases. Through rigorous and systematic investigation, our findings reveal a consistent pattern of gender encoding across diverse models. Surprisingly, debiasing techniques often exhibit limited efficacy, sometimes inadvertently increasing the encoded bias in internal representations while reducing bias in model output distributions. This highlights a disconnect between mitigating bias in output distributions and addressing its internal representations. This work provides valuable guidance for advancing bias mitigation strategies and fostering the development of more equitable language models.