AI Paper Atlas

Research map / Architectures and efficiency / Quantization

NeUQI: Near-Optimal Uniform Quantization Parameter Initialization for Low-Bit LLMs

Li Lin, Xinyu Hu, Xiaojun Wan

ICML 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
replace Min-Max-style integer-zero-point initialization with NeUQI (float zero-point, optimized scale)GPTQ-style PTQ on LLaMA 2 7B · 2-bit channel-wise, Wiki2perplexity: 6953 -> 17.14 (GPTQ vs NeUQI)Improved
relax the integer zero-point constraint (NeUQI vs Int-Search with the same loss) (ablation)NeUQI · LLaMA 2 7B 2-bit channel-wise, Wiki2perplexity: 26.26 -> 17.14Improved
relax the integer zero-point constraint (NeUQI vs Int-Search) (ablation)NeUQI · LLaMA 2 70B 2-bit channel-wise, average accuracyaccuracy: 61.61 -> 65.13Improved
use NeUQI instead of other init methods at same group-wise bit budget (group 128 vs integer-constrained group 64) (ablation)LLaMA 2 7B 2-bit · average accuracyaccuracy: 49.73 -> 51.33 (best group-64 baseline GPTAQ vs NeUQI)Improved
add NeUQI initialization to lightweight distillation2-bit g128 LLaMA 2 7B · Wiki2perplexity: 14.95 -> 8.38 (GPTQ vs NeUQI distillation)Improved
replace EfficientQAT initialization with NeUQIEfficientQAT on LLaMA 2 7B · Wiki2perplexity: 7.19 -> 6.81Improved
only allow floating zero-point in EfficientQAT without NeUQI initialization (ablation)EfficientQAT on LLaMA 2 7B · average accuracyaccuracy: 59.50 -> 59.37No clear effect
freeze scale in first stage of EfficientQAT with NeUQI init (ablation)EfficientQAT+NeUQI on LLaMA 3 8B · Wiki2perplexity: 9.14 -> 8.87Improved
add coarse-to-fine scale search (ablation)NeUQI · LLaMA 2 7B 2-bit, relative runtimerelative runtime: 0.548 -> 0.027 for Q projection, loss 1.00197 -> 1.00193Improved
add transition-point reduction (ablation)NeUQI zero-point search · LLaMA 2 7B 2-bitruntime: nearly 50% runtime reductionImproved
add NeUQI 3-bit to larger model versus smaller BF16 model at same memoryQwen 2.5 14B · average accuracy vs 3B BF16accuracy: 64.99 -> 69.12Improved
combine NeUQI with Hadamard transform at W4A4LLaMA 2 7B · Wiki2perplexity: 6.02 -> 5.99 (MagR vs NeUQI)No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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