machine learning

QuantWAMs: Calibrating at the Right Granularity for World Action Models

arXiv:2607.28405

summary

The paper proposes QuantWAMs, a post‑training quantization framework that tailors quantization decisions to the structure, rollout distribution, and task objectives of World Action Models, achieving memory reduction and speedups while maintaining accuracy in robotic manipulation tasks.

Abstract

World Action Models (WAMs) jointly predict future observations and actions, but their iterative denoising and closed-loop execution make efficient deployment costly. Existing post-training quantization (PTQ) methods are poorly suited to WAMs because they rely on open-loop objectives, homogeneous model assumptions, and calibration distributions that do not reflect deployment. We present QuantWAMs, a PTQ framework that aligns quantization decisions with the calibration context defined by model structure, rollout distribution, and task objective. QuantWAMs introduces three strategies: shared-basis outlier calibration, which pools activation evidence only across coordinate-compatible modules; co-training-objective saliency, which computes empirical-Fisher scores from the joint video--action gradient and assigns weight precision at a calibration-stable layer granularity; and fixed-intervention rollout auditing, which revises denoising-step protection schedules using reachable closed-loop states without changing the precision budget. We evaluate QuantWAMs on Fast-WAM and LingBot-VA across RoboTwin 2.0, LIBERO, and real-robot manipulation with an AgiBot G2. Under a W4A4-dominant setting, the reported simulation means differ from FP16 by 0.2--0.7 percentage points. Real-robot trials further establish deployment feasibility on three manipulation tasks. For the targeted video and action blocks, QuantWAMs reduces peak weight-and-activation memory to about 29\% of FP16 and provides 1.4--1.6 block-level speedups.

13 pages, 6 figures

Topics & keywords

#post-training quantization#world action models#robotic manipulation#model calibration#closed-loop inferenceshared-basis outlier calibrationempirical Fisher scoresfixed-intervention rollout auditingW4A4 quantizationFast-WAMLingBot-VA
QuantWAMs: Calibrating at the Right Granularity for World Action Models · wovepaper