<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Quantization | About Hyeongheon</title><link>https://chahh9808.github.io/tags/quantization/</link><atom:link href="https://chahh9808.github.io/tags/quantization/index.xml" rel="self" type="application/rss+xml"/><description>Quantization</description><generator>Hugo Blox Builder (https://hugoblox.com)</generator><language>en-us</language><lastBuildDate>Tue, 06 Oct 2026 13:48:33 +0000</lastBuildDate><image><url>https://chahh9808.github.io/media/icon_hu68170e94a17a2a43d6dcb45cf0e8e589_3079_512x512_fill_lanczos_center_3.png</url><title>Quantization</title><link>https://chahh9808.github.io/tags/quantization/</link></image><item><title>Test-Time Adaptation of Quantized ViTs via Single-Pass Quantizer-Aligned Recalibration</title><link>https://chahh9808.github.io/post/quar/</link><pubDate>Tue, 06 Oct 2026 13:48:33 +0000</pubDate><guid>https://chahh9808.github.io/post/quar/</guid><description>&lt;h2 id="position">Position&lt;/h2>
&lt;p>First author&lt;/p>
&lt;h2 id="project-goals--works">Project Goals &amp;amp; Works&lt;/h2>
&lt;ol>
&lt;li>Improve the robustness of quantized vision transformers under distribution shift while keeping the model and quantizers frozen.&lt;/li>
&lt;li>Recalibrate activations at late-block quantizer inputs and the classifier input using running channel statistics and source calibration statistics.&lt;/li>
&lt;li>Adapt within a single forward pass, without backpropagation, parameter updates, or storage of past samples.&lt;/li>
&lt;/ol>
&lt;h2 id="key-results">Key Results&lt;/h2>
&lt;ol>
&lt;li>On ImageNet-C at severity 5, improved ViT-B/16 accuracy over the strongest evaluated backpropagation-free baseline by 4.00 percentage points at W3A3 and 2.28 points at W8A8.&lt;/li>
&lt;li>Used about 170 KB of state and reduced latency by at least 46% compared with the evaluated two-forward-pass methods.&lt;/li>
&lt;li>Evaluated robustness across continual streams, label shifts, additional distribution shifts, and multiple architectures.&lt;/li>
&lt;/ol>
&lt;h2 id="links">Links&lt;/h2>
&lt;ul>
&lt;li>&lt;a href="https://chahh9808.github.io/publication/quar/">Publication details&lt;/a>&lt;/li>
&lt;li>&lt;a href="https://arxiv.org/abs/2610.08358">arXiv&lt;/a>&lt;/li>
&lt;li>&lt;a href="https://arxiv.org/pdf/2610.08358">Paper PDF&lt;/a>&lt;/li>
&lt;li>&lt;a href="https://github.com/chahh9808/QuAR">Code&lt;/a>&lt;/li>
&lt;/ul></description></item><item><title>Test-Time Adaptation of Quantized ViTs via Single-Pass Quantizer-Aligned Recalibration</title><link>https://chahh9808.github.io/publication/quar/</link><pubDate>Tue, 06 Oct 2026 13:48:33 +0000</pubDate><guid>https://chahh9808.github.io/publication/quar/</guid><description/></item></channel></rss>