Investigating the Impact of Quantization Methods on the Safety and Reliability of Large Language Models

TL;DR

This paper evaluates how different quantization methods affect the safety and reliability of large language models, revealing trade-offs and the need for safety-aware compression strategies.

Contribution

It introduces OpenMiniSafety, a new safety dataset, and provides a comprehensive evaluation of quantization impacts on LLM safety across multiple models and benchmarks.

Findings

Quantization can degrade safety alignment of LLMs.

No single quantization method consistently outperforms others in safety.

Precision-specific quantization methods excel at their target bit-widths.

Abstract

Large Language Models (LLMs) are powerful tools for modern applications, but their computational demands limit accessibility. Quantization offers efficiency gains, yet its impact on safety and trustworthiness remains poorly understood. To address this, we introduce OpenMiniSafety, a human-curated safety dataset with 1.067 challenging questions to rigorously evaluate model behavior. We publicly release human safety evaluations for four LLMs (both quantized and full-precision), totaling 4.268 annotated question-answer pairs. By assessing 66 quantized variants of these models using four post-training quantization (PTQ) and two quantization-aware training (QAT) methods across four safety benchmarks including human-centric evaluations we uncover critical safety performance trade-offs. Our results show both PTQ and QAT can degrade safety alignment, with QAT techniques like QLORA or STE performing less safely. No single method consistently outperforms others across benchmarks, precision settings, or models, highlighting the need for safety-aware compression strategies. Furthermore, precision-specialized methods (e.g., QUIK and AWQ for 4-bit, AQLM and Q-PET for 2-bit) excel at their target precision, meaning that these methods are not better at compressing but rather different approaches.