AI Paper Atlas

Research map / Architectures and efficiency / Quantization

QuantVLA: Scale-Calibrated Post-Training Quantization for Vision-Language-Action Models

Jingxuan Zhang, Yunta Hsieh, Zhongwei Wan, Haokun Lin, Xin Wang, Ziqi Wang, Yingtie Lei, Mi Zhang

CVPR 2026 · Architectures and efficiency › Quantization · approach: Post-training quantization of ViTs · 23 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
quantize DiT attention projections too (all DiT linear layers) instead of only the MLP (ablation)QuantVLA layout on pi0.5 · LIBERO averagesuccess rate: 95.4% -> 76.3%Hurt
quantize DiT attention projections too instead of only the MLP (ablation)QuantVLA layout on GR00T N1.5 · LIBERO averagesuccess rate: 82.5% -> 70.0%Hurt
quantize only the DiT (all layers) instead of LLM only (ablation)W4A8 without ATM/OHB on pi0.5 · LIBERO averagesuccess rate: 96.5% -> 71.6%Hurt
add ATM and OHB calibration to the selective layout (ablation)QuantVLA on pi0.5 · LIBERO averagesuccess rate: 95.4% -> 97.6%Improved
add ATM and OHB calibration to the selective layout (ablation)QuantVLA on GR00T N1.5 · LIBERO averagesuccess rate: 82.5% -> 88.0%Improved
use QuantVLA instead of DuQuant on LLM+DiTpi0.5 W4A8 · LIBERO averagesuccess rate: 76.3% -> 97.6%Improved
use QuantVLA instead of DuQuant on LLM+DiTGR00T N1.5 W4A8 · LIBERO averagesuccess rate: 70.0% -> 88.0%Improved
quantize DiT MLP in addition to LLM (full QuantVLA vs LLM only)QuantVLA on pi0.5 · Memory (GB)success rate: 1.58 -> 1.28Improved
quantize DiT MLP in addition to LLM (full QuantVLA vs LLM only)QuantVLA on GR00T N1.5 · LIBERO averagesuccess rate: 87.0% -> 88.0%Improved
lower activation precision from W4A8 to W4A4QuantVLA on pi0.5 · LIBERO averagesuccess rate: 97.6% -> 95.3%Mixed
increase denoising steps from 8 to 16QuantVLA on GR00T N1.5 · LIBERO averagesuccess rate: 88.0% -> 88.5%No clear effect
use QuantVLA instead of SmoothQuantGR00T W4A8 · Pick-and-Cansuccesses out of 50: 16 / 50 -> 27 / 50Improved

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