# Safety-Preserving PTQ via Contrastive Alignment Loss

arXiv: https://arxiv.org/abs/2511.07842 · HTML: https://mlsanity.com/paper/2511.07842/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Safety alignment, refusal and guardrails, one of 12 directions in LLM safety and security.
- Builds on 19 of its 32 references that are in the map.
- First posted on arXiv 11 Nov 2025.

## Placed in

- LLM safety and security / [Safety alignment, refusal and guardrails](https://mlsanity.com/safety/safety-alignment-refusal-and-guardrails/)

## Builds on (most cited first, 19 of 19)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge](https://mlsanity.com/paper/1803.05457/) (2018)
- [HellaSwag: Can a Machine Really Finish Your Sentence?](https://mlsanity.com/paper/1905.07830/) (2019)
- [Pointer Sentinel Mixture Models](https://mlsanity.com/paper/1609.07843/) (2016)
- [PIQA: Reasoning about Physical Commonsense in Natural Language](https://mlsanity.com/paper/1911.11641/) (2019)
- [A Broad-Coverage Challenge Corpus for Sentence Understanding through Inference](https://mlsanity.com/paper/1704.05426/) (2017)
- [BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questions](https://mlsanity.com/paper/1905.10044/) (2019)
- [Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answering](https://mlsanity.com/paper/1809.02789/) (2018)
- [GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers](https://mlsanity.com/paper/2210.17323/) (2022)
- [CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge](https://mlsanity.com/paper/1811.00937/) (2018)
- [The LAMBADA dataset: Word prediction requiring a broad discourse context](https://mlsanity.com/paper/1606.06031/) (2016)
- [QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMs](https://mlsanity.com/paper/2404.00456/) (2024)
- [SpinQuant: LLM quantization with learned rotations](https://mlsanity.com/paper/2405.16406/) (2024)
- [DuQuant: Distributing Outliers via Dual Transformation Makes Stronger Quantized LLMs](https://mlsanity.com/paper/2406.01721/) (2024)
- [FlatQuant: Flatness Matters for LLM Quantization](https://mlsanity.com/paper/2410.09426/) (2024)
- [OstQuant: Refining Large Language Model Quantization with Orthogonal and Scaling Transformations for Better Distribution Fitting](https://mlsanity.com/paper/2501.13987/) (2025)
- [Investigating the Impact of Quantization Methods on the Safety and Reliability of Large Language Models](https://mlsanity.com/paper/2502.15799/) (2025)

## Built on by (most cited first, 3 of 3)

- [Safety in Self-Evolving LLM Agent Systems: Threats, Amplification, and Case Studies](https://mlsanity.com/paper/2606.23075/) (2026)
- [Quality Is Not a Safety Proxy Under Quantization](https://mlsanity.com/paper/2606.10154/) (2026)
- [SafeTune: A Unified Faithful Library for Auditing and Repairing Safety Drift in Fine-Tuned LLMs](https://mlsanity.com/paper/2609.22153/) (2026)

## Abstract (arXiv)

Post-Training Quantization (PTQ) has become the de-facto standard for efficient LLM deployment, yet its optimization objective remains fundamentally incomplete. Standard PTQ methods minimize reconstruction error (e.g., MSE or KL divergence) without accounting for behavioral alignment--a critical property instilled through safety fine-tuning. We demonstrate that this objective mismatch introduces a systematic vulnerability: models can maintain low perplexity yet exhibit significant degradation in safety alignment, revealing that perplexity alone is an insufficient and often misleading proxy for deployment readiness. To address this, we propose Contrastive Alignment Quantization (CAQ), which extends the PTQ objective design space by integrating a Contrastive Alignment Loss (CAL). CAL introduces a principled push-pull mechanism that jointly optimizes distributional fidelity and behavioral alignment: it steers the quantized model toward its safe, instruction-tuned counterpart while diverging from the unaligned, pre-trained reference. CAQ requires no specialized safety datasets, relying solely on standard calibration data, and introduces negligible computational overhead over existing transformation-based PTQ pipelines. We show that CAQ enables robust 4-bit (W4A4) quantization across diverse model families--including LLaMA, Qwen, and Mistral--achieving superior safety alignment where state-of-the-art PTQ methods fail, without sacrificing general capabilities. Anonymized code is available in the supplementary material.