arXiv：cs.CL· Irina Proskurina, Guillaume Metzler, Julien Velcin·· 3 小时前
Fair-GPTQ：面向大语言模型的偏见感知量化方法
Fair-GPTQ: Bias-Aware Quantization for Large Language Models
AI 导读
Fair-GPTQ 是首个在量化目标中显式加入群体公平性约束的量化方法，通过引导舍入操作学习减少受保护群体的偏见生成。该方法在零样本基准上保留至少 90% 的基线准确率，相对半精度模型降低不公平性，同时保持 4-bit 量化的内存与速度优势。
正文
Abstract:The high memory demands of generative language models have drawn attention to quantization, which reduces memory usage by mapping model weights to lower-precision integers. However, recent empirical studies show that, while efficient, quantization can increase the likelihood of generating biased outputs and degrade performance on fairness benchmarks. In this work, we draw new links between quantization and model fairness by adding explicit group-fairness constraints to the quantization objective and introduce Fair-GPTQ, the first quantization method explicitly designed to reduce unfairness in large language models. The added constraints guide the learning of the rounding operation toward less-biased text generation for protected groups. Specifically, we focus on stereotype generation involving occupational bias and discriminatory language spanning gender, race, and religion. Fair-GPTQ has minimal impact on performance, preserving at least 90% of baseline accuracy on zero-shot benchmarks, reduces unfairness relative to a half-precision model, and retains the memory and speed benefits of 4-bit quantization.
| Comments: | Accepted for publication in TACL. Pre-MIT Press publication version |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2509.15206 [cs.CL] |
| (or arXiv:2509.15206v4 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2509.15206 arXiv-issued DOI via DataCite |
Submission history
From: Irina Proskurina [view email]
[v1]
Thu, 18 Sep 2025 17:56:16 UTC (448 KB)
[v2]
Mon, 2 Feb 2026 14:23:42 UTC (414 KB)
[v3]
Mon, 6 Jul 2026 11:20:29 UTC (5,287 KB)
[v4]
Thu, 8 Oct 2026 13:57:23 UTC (2,038 KB)
来源：arXiv：cs.CL · arxiv.org