English

FairPy: A Toolkit for Evaluation of Prediction Biases and their Mitigation in Large Language Models

Computation and Language 2025-04-16 v2 Artificial Intelligence Machine Learning

Abstract

Recent studies have demonstrated that large pretrained language models (LLMs) such as BERT and GPT-2 exhibit biases in token prediction, often inherited from the data distributions present in their training corpora. In response, a number of mathematical frameworks have been proposed to quantify, identify, and mitigate these the likelihood of biased token predictions. In this paper, we present a comprehensive survey of such techniques tailored towards widely used LLMs such as BERT, GPT-2, etc. We additionally introduce Fairpy, a modular and extensible toolkit that provides plug-and-play interfaces for integrating these mathematical tools, enabling users to evaluate both pretrained and custom language models. Fairpy supports the implementation of existing debiasing algorithms. The toolkit is open-source and publicly available at: \href{https://github.com/HrishikeshVish/Fairpy}{https://github.com/HrishikeshVish/Fairpy}

Keywords

Cite

@article{arxiv.2302.05508,
  title  = {FairPy: A Toolkit for Evaluation of Prediction Biases and their Mitigation in Large Language Models},
  author = {Hrishikesh Viswanath and Tianyi Zhang},
  journal= {arXiv preprint arXiv:2302.05508},
  year   = {2025}
}