AI Paper Atlas

Research map / Architectures and efficiency / Quantization

Dissecting Quantization Error: A Concentration-Alignment Perspective

Marco Federici, Boris van Breugel, Paul N. Whatmough, Markus Nagel

ICML 2026 · Architectures and efficiency › Quantization · approach: Post-training quantization of ViTs · 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
add alignment-optimizing block transform before Hadamard (CAT block) vs Hadamard-only QuaRotW4A4 RTN · WikiText-2, Llama 3 8Bperplexity: 10.99 -> 8.29Improved
train CAT (block) transform on calibration dataCAT (block) W4A4 RTN · WikiText-2, Llama 3 8Bperplexity: 8.29 -> 7.33Improved
train CAT (block) transformCAT (block) W4A4 RTN · 0-shot avg, Llama 3 8Baccuracy (%): 67.77 -> 69.66Improved
CAT (block) trained instead of FlatQuantW4A4 RTN · WikiText-2 / 0-shot, Llama 3 8Bperplexity / accuracy: 7.61 -> 7.33 ppl; 68.34 -> 69.66 accMixed
replace RTN with GPTQ weight quantization (ablation)CAT (block) trained and FlatQuant · WikiText-2, Llama 3 8Bperplexity: 7.33 -> 7.25 (CAT)No clear effect
replace RTN with GPTQ weight quantization (ablation)QuaRot · WikiText-2, Llama 3 8Bperplexity: 10.99 -> 8.21Improved
increase CAT block size (ablation)CAT (block) · Llama 3.2 1B, WikiText-2alignment, SQNR, perplexityImproved
apply Hadamard or other rotations (ablation)rotation-based transforms · alignment on LLM layersalignment [dB]No clear effect
apply channel scaling (SmoothQuant) to move outliers into weights (ablation)W4A4 · LLM layersactivation/weight concentrationMixed
use asymmetric instead of symmetric activation quantization (ablation)uniform integer quantization · LLM layersconcentrationNo clear effect
change calibration set size or dataset for CAT (ablation)CAT (block) · calibration sensitivityaccuracyNo clear effect
use SmoothQuant at W4A4 without rotationSmoothQuant RTN · WikiText-2, Llama 3 8Bperplexity: 321.22 -> 233.61 vs no transform; still far above FP 6.14Hurt

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