Summary
SliderQuant is a new post-training quantization (PTQ) framework for large language models introduced in an arXiv paper (2603.25284) by Shigeng Wang, Chao Li, Yangyuxuan Kang, Jiawei Fan, Zhonghong Ou, and colleagues. The authors approach PTQ from a previously overlooked angle: mainstream sequential quantization frameworks treat all layers uniformly, which may be suboptimal under challenging bit-width settings. Through empirical study of per-layer quantization impact, they observe that shallow and deep layers are generally more sensitive to quantization than middle layers, and within these, the first and last layers are the most sensitive, exhibiting noticeably larger quantization error than other layers. Based on these findings, SliderQuant adopts a simple adaptive sliding quantization concept. Extensive experiments across various LLMs show that the method outperforms existing PTQ approaches in both weight-only quantization and weight-activation quantization settings.
Paper Overview
- Field: Machine Learning
- Authors: Shigeng Wang, Chao Li, Yangyuxuan Kang, Jiawei Fan, Zhonghong Ou, et al.
- Published: 2026-03-26
- arXiv: 2603.25284
Abstract
This paper addresses post-training quantization (PTQ) of large language models (LLMs) from a previously overlooked perspective: given a pre-trained high-precision LLM, mainstream sequential quantization frameworks treat all layers uniformly, but this may not be optimal under challenging bit-width settings.
The authors empirically investigate the quantization impact of different layers and observe that:
1. Shallow and deep layers are generally more sensitive to quantization than middle layers.
2. Within shallow/deep layers, the most sensitive ones are the first and last layers, whose quantization error is significantly larger than that of other layers.
Based on these findings, the authors propose SliderQuant, a novel PTQ framework that adopts a simple adaptive sliding quantization concept.
Extensive experiments on various LLMs demonstrate that the proposed method outperforms existing PTQ approaches in both weight-only quantization and weight-activation quantization.
*Auto-collected on 2026-03-29.*
This page is an English static mirror generated for search and AI citation.
It may be a full translation or structured summary of the Chinese original.
Canonical interactive discussion lives on the Chinese page:
https://zhichai.net/topic/177169404