aiqbench analysis
This research introduces SteerQuant, a novel 4-bit quantization framework for world-action models that selectively directs numerical errors away from critical action computations to preserve control reliability. By integrating an adaptive inference engine called Rudder, the method achieves significant speedups in denoising and real-world robotic execution while maintaining task success rates comparable to full-precision models. The work addresses a key bottleneck in deploying advanced AI agents on resource-constrained hardware without sacrificing performance.
SteerQuant: Steering Quantization Error with Action-Guided Scaling in World-Action Models
arXiv:2609.39056v1 Announce Type: new Abstract: World-action models (WAMs) jointly generate future world states and actions through iterative denoising, using shared weights to process heterogeneous semantic streams of video, proprioceptive, and action tokens. Quantization reduces inference cost, but comparable numerical errors in different streams can have markedly different effects on final actions, making numerical accuracy alone insufficient for reliable control. We introduce SteerQuant, a 4-bit quantization framework for WAMs that steers errors toward computations with less influence on final actions. It maps how each stream's quantization errors affect final actions and uses this map to guide shared channel scaling. Activation scaling is further calibrated for each stream and denoising step to accommodate changes in activation ranges and action impact. This adapts quantization to different stream requirements without duplicating weights or increasing bit-widths for selected streams. To reduce the extra kernel launches and memory traffic introduced by scaling, we develop Rudder, a 4-bit inference engine for WAMs that fuses scaling and output compensation into low-bit kernels. Under W4A8 and W4A4, SteerQuant maintains mean LIBERO success within 0.8 percentage points of full precision, while delivering up to $2.23\times$ denoising speedup over BF16 across three WAMs with reduced peak GPU memory usage. On a real dual-arm robot, W4A8 deployment achieves a $1.35\times$ end-to-end inference speedup while maintaining average task success relative to BF16.
Abstract:World-action models (WAMs) jointly generate future world states and actions through iterative denoising, using shared weights to process heterogeneous semantic streams of video, proprioceptive, and action tokens. Quantization reduces inference cost, but comparable numerical errors in different streams can have markedly different effects on final actions, making numerical accuracy alone insufficient for reliable control. We introduce SteerQuant, a 4-bit quantization framework for WAMs that steers errors toward computations with less influence on final actions. It maps how each stream's quantization errors affect final actions and uses this map to guide shared channel scaling. Activation scaling is further calibrated for each stream and denoising step to accommodate changes in activation ranges and action impact. This adapts quantization to different stream requirements without duplicating weights or increasing bit-widths for selected streams. To reduce the extra kernel launches and memory traffic introduced by scaling, we develop Rudder, a 4-bit inference engine for WAMs that fuses scaling and output compensation into low-bit kernels. Under W4A8 and W4A4, SteerQuant maintains mean LIBERO success within 0.8 percentage points of full precision, while delivering up to $2.23\times$ denoising speedup over BF16 across three WAMs with reduced peak GPU memory usage. On a real dual-arm robot, W4A8 deployment achieves a $1.35\times$ end-to-end inference speedup while maintaining average task success relative to BF16.
Robotics (cs.RO)
arXiv:2609.39056 [cs.RO]
(or arXiv:2609.39056v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2609.39056 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 307 words
Related coverage
MolmoMotion: Language-guided 3D motion forecasting
Hugging Face · Jun 17, 2026
[AINews] Claude Opus 5.5, the new default model for AINews — and everybody cuts prices 40-50%
Latent Space · Sep 23, 2026
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · Sep 25, 2026
The Continual Constitutional Convention: How language models revise the documents that govern them
AI Alignment Forum · Oct 13, 2026