AI Paper Atlas

Research map / Architectures and efficiency / Quantization

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

Yeonsik Park, Hyeonseong Kim, Seungkyu Choi

ICLR 2026 · Architectures and efficiency › Quantization · approach: Post-training quantization of LLMs · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use SERQ single saliency-aware low-rank matrix instead of L2QER two-factor SVD reconstructionSERQ · WikiText2 W4A4 INT4, LLaMA-3 8Bperplexity: 11.44 -> 7.75Improved
use SERQ instead of L2QER under W4A8SERQ (RTN) · WikiText2 W4A8, LLaMA-3 8Bperplexity: 7.16 -> 6.71Improved
use SERQ instead of SpinQuant/QuaRot rotation methodsSERQ · MMLU W4A4, LLaMA-3 8Baccuracy: 49.93 -> 53.8Improved
use SERQ instead of SpinQuant on LLaMA-2 13BSERQ · WikiText2 W4A4, LLaMA-2 13Bperplexity: 5.2 -> 5.2No clear effect
use SERQ instead of SpinQuant on LLaMA-3.2 1B MMLUSERQ · MMLU W4A4, LLaMA-3.2 1Baccuracy: 26.38 -> 26.34No clear effect
use SERQ instead of L2QER on generationSERQ · GSM8K 5-shot W4A4, LLaMA-3 8Baccuracy: 7.96 -> 23.65Improved
use MXFP4 with L2QER instead of INT4 L2QERL2QER · MMLU W4A4, LLaMA-3 8Baccuracy: 38.33 -> 53.82Improved
increase low-rank size from 0 to 256 salient rows (ablation)SERQ · WikiText2 W4A4, LLaMA-3 8Bperplexity: 9.8 -> 7.98Improved
add static activation flattening to SERQ (ablation)SERQ · WikiText2 W4A4 INT RTN, Qwen-2.5 3Bperplexity: 10.83 -> 9.57Improved
add static activation flattening to SERQ (ablation)SERQ · WikiText2 W4A4 INT RTN, LLaMA-2 7Bperplexity: 6.05 -> 6.03No clear effect
add error reconstruction to static flattening (ablation)SERQ · WikiText2 W4A4 INT RTN, LLaMA-3.2 1Bperplexity: 56.92 -> 13.49Improved
change calibration data size or dataset (Wiki vs Pile, 32-512 samples) (ablation)SERQ · WikiText2 W4A4, LLaMA-3 8Bperplexity: 8.07 -> 7.98No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy