Unmasking the Mask -- Evaluating Social Biases in Masked Language Models

2021-04-15 · Masahiro Kaneko, Danushka Bollegala

Masked Language Models (MLMs) have shown superior performances in numerous downstream NLP tasks when used as text encoders. Unfortunately, MLMs also demonstrate significantly worrying levels of social biases. We show that the previously proposed evaluation metrics for quantifying the social biases in MLMs are problematic due to following reasons: (1) prediction accuracy of the masked tokens itself tend to be low in some MLMs, which raises questions regarding the reliability of the evaluation metrics that use the (pseudo) likelihood of the predicted tokens, and (2) the correlation between the prediction accuracy of the mask and the performance in downstream NLP tasks is not taken into consideration, and (3) high frequency words in the training data are masked more often, introducing noise due to this selection bias in the test cases. To overcome the above-mentioned disfluencies, we propose All Unmasked Likelihood (AUL), a bias evaluation measure that predicts all tokens in a test case given the MLM embedding of the unmasked input. We find that AUL accurately detects different types of biases in MLMs. We also propose AUL with attention weights (AULA) to evaluate tokens based on their importance in a sentence. However, unlike AUL and AULA, previously proposed bias evaluation measures for MLMs systematically overestimate the measured biases, and are heavily influenced by the unmasked tokens in the context.

📄 PDF Abstract BibTeX arXiv:2104.07496

Code (2)

kanekomasahiro/evaluate_bias_in_mlm 공식 구현 pytorch

zalkikar/mlm-bias pytorch

Tasks

Selection biasSentence

AI-Powered Facial Mask Removal Is Not Suitable For Identification

2026-03-29 · Emily A Cooper, Hany Farid arxiv

Recently, crowd-sourced online criminal investigations have used generative-AI to enhance low-quality visual evidence. In one high-profile case, social-media users circulated an "AI-unmasked" image of a federal agent inv…

Measuring Social Biases in Masked Language Models by Proxy of Prediction Quality

2024-02-21 · Rahul Zalkikar, Kanchan Chandra

Innovative transformer-based language models produce contextually-aware token embeddings and have achieved state-of-the-art performance for a variety of natural language tasks, but have been shown to encode unwanted bias…

Language Modeling Language Modelling Masked Language Modeling

Where-to-Unmask: Ground-Truth-Guided Unmasking Order Learning for Masked Diffusion Language Models

2026-02-10 · Hikaru Asano, Tadashi Kozuno, Kuniaki Saito, Yukino Baba arxiv

Masked Diffusion Language Models (MDLMs) generate text by iteratively filling masked tokens, requiring two coupled decisions at each step: which positions to unmask (where-to-unmask) and which tokens to place (what-to-un…

Reinforcement Learning Logical Reasoning

Evaluating Short-Term Temporal Fluctuations of Social Biases in Social Media Data and Masked Language Models

2024-06-19 · Yi Zhou, Danushka Bollegala, Jose Camacho-Collados

Social biases such as gender or racial biases have been reported in language models (LMs), including Masked Language Models (MLMs). Given that MLMs are continuously trained with increasing amounts of additional data coll…

valid

Re-evaluating Confidence Remasking in Masked Diffusion Language Models

2026-06-10 · Stipe Frkovic, Metod Jazbec, Dan Zhang, Christian A. Naesseth 외 arxiv

Masked diffusion language models (dLLMs) have recently emerged as a competitive alternative to autoregressive language models, with the promise of faster inference via parallel token generation. A notable limitation of t…