Gender-tuning: Empowering Fine-tuning for Debiasing Pre-trained Language Models

TL;DR

Gender-tuning is a novel fine-tuning approach that effectively reduces societal biases in pre-trained language models without additional resources, while enhancing their performance on downstream tasks.

Contribution

It introduces a resource-efficient debiasing method that integrates MLM objectives into fine-tuning, outperforming existing baselines in bias reduction and task performance.

Findings

Outperforms state-of-the-art bias mitigation methods.

Improves downstream task performance.

Works with any PLM using only downstream datasets.

Abstract

Recent studies have revealed that the widely-used Pre-trained Language Models (PLMs) propagate societal biases from the large unmoderated pre-training corpora. Existing solutions require debiasing training processes and datasets for debiasing, which are resource-intensive and costly. Furthermore, these methods hurt the PLMs' performance on downstream tasks. In this study, we propose Gender-tuning, which debiases the PLMs through fine-tuning on downstream tasks' datasets. For this aim, Gender-tuning integrates Masked Language Modeling (MLM) training objectives into fine-tuning's training process. Comprehensive experiments show that Gender-tuning outperforms the state-of-the-art baselines in terms of average gender bias scores in PLMs while improving PLMs' performance on downstream tasks solely using the downstream tasks' dataset. Also, Gender-tuning is a deployable debiasing tool for any PLM that works with original fine-tuning.