arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2306.04597cs.CLcs.LG
收藏

语言模型的性别改造:通过少样本数据干预缓解性别偏见

Language Models Get a Gender Makeover: Mitigating Gender Bias with Few-Shot Data Interventions

Carnegie Mellon University(卡内基梅隆大学)

Himanshu Thakur, Atishay Jain, Praneetha Vaddamanu, Paul Pu Liang, Louis-Philippe Morency

AI总结:

针对预训练语言模型传播性别偏见且重新训练成本高的问题,本文提出仅用10个去偏样本进行微调的少样本数据干预方法,在尽量保持语言建模能力的同时取得优于现有基线的去偏效果。

摘要:

预训练大型语言模型中存在的社会偏见是一个关键问题,因为已有研究表明这些模型会在无数下游应用中传播偏见,导致其对特定人群不公平。由于从头对这些模型进行大规模重新训练既耗时又耗费算力,此前已有多种方法被提出,用于对预训练模型进行去偏。尽管当前大多数最先进的去偏方法都聚焦于改变训练机制,但在本文中,我们提出将数据干预策略作为一种强大而简单的技术,以减少预训练模型中的性别偏见。具体而言,我们通过实验表明,仅用10个去偏(经干预)训练样本对预训练模型进行微调,就可显著降低其偏向任何性别的倾向。由于我们提出的方法只需要少量训练样本,因此这种少样本去偏方法具有很高的可行性和实用性。通过大量实验,我们表明所提出的去偏技术优于具有竞争力的最先进基线,同时语言建模能力损失极小。

英文摘要:

Societal biases present in pre-trained large language models are a critical issue as these models have been shown to propagate biases in countless downstream applications, rendering them unfair towards specific groups of people. Since large-scale retraining of these models from scratch is both time and compute-expensive, a variety of approaches have been previously proposed that de-bias a pre-trained model. While the majority of current state-of-the-art debiasing methods focus on changes to the training regime, in this paper, we propose data intervention strategies as a powerful yet simple technique to reduce gender bias in pre-trained models. Specifically, we empirically show that by fine-tuning a pre-trained model on only 10 de-biased (intervened) training examples, the tendency to favor any gender is significantly reduced. Since our proposed method only needs a few training examples, our few-shot debiasing approach is highly feasible and practical. Through extensive experimentation, we show that our debiasing technique performs better than competitive state-of-the-art baselines with minimal loss in language modeling ability.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑