SERQ: Saliency-Aware Low-Rank Error Reconstruction for LLM Quantization

TL;DR

SERQ introduces a saliency-aware low-rank error reconstruction technique for low-bit LLM quantization, significantly improving accuracy and efficiency in W4A4 and W4A8 settings with minimal latency overhead.

Contribution

It proposes a novel saliency-aware low-rank error reconstruction method that enhances low-bit LLM inference accuracy while maintaining computational efficiency.

Findings

Outperforms prior error reconstruction methods in W4A8 and W4A4 settings.

Achieves higher accuracy than state-of-the-art rotation-based W4A4 approaches.

Reduces calibration complexity significantly.

Abstract

Post-training quantization (PTQ) has emerged as a prevailing technique for deploying large language models (LLMs) efficiently in terms of both memory and computation, across edge devices and server platforms. Existing PTQ methods primarily aim to reduce precision in weights and activations by mitigating quantization errors caused by channel-wise outlier activations (e.g., pre-quantization scaling, online transformations, or low-rank error reconstruction). Among these approaches, error reconstruction with low-rank adaptation (LoRA) has proven particularly effective, as it introduces a lightweight auxiliary computation path without requiring heavy optimization or additional online layers. However, prior studies reveal severe accuracy degradation under W4A4 settings, and conventional low-rank adaptations rely on two sequential factors, necessitating intermediate quantization during inference and thereby limiting low-precision efficiency. In this work, we propose SERQ, a saliency-aware error reconstruction method for low-bit LLM inference that employs a single low-rank compensation matrix. SERQ preserves efficient 4-bit matrix multiplication in linear layers by jointly mitigating quantization errors arising from both activation and weight saliency through three stages: (1) static activation flattening, (2) saliency-aware error reconstruction, and (3) offline weight permutation. The method incurs additional computation only for low-rank error reconstruction via a single decomposition, while all other operations are performed offline, thereby keeping latency overhead minimal. Empirically, SERQ outperforms prior error reconstruction methods under both W4A8 and W4A4 settings, and achieves higher accuracy than state-of-the-art rotation-based W4A4 approaches, while substantially reducing calibration complexity.