analiza aiqbench
WISE is a pre-execution selector that samples several candidate rollouts from a frozen world action model and executes the one that scores best on action validity, inverse-dynamics consistency, and task progress. On the reported benchmarks it raises success from 6.0% to 24.0% on RoboLab and from 75.1% to 78.3% on LIBERO, with four candidates per decision. For readers, the main points are that the gain comes from inference-time selection rather than retraining, that the largest jumps are on a harder benchmark where absolute success remains low, and that the added sampling and scoring cost is the tradeoff to weigh.
Seeing Through the Illusion: Test-Time Scaling for World Action Models
arXiv:2611.03585v1 Announce Type: new Abstract: World action models (WAMs) jointly generate actions and imagined futures, but a plausible prediction need not be realized by its co-generated action. We introduce WISE (\emph{World Action Inference Scaling through Evaluation}), a pre-execution selector that scores each sampled rollout on action validity and temporal regularity, inverse-dynamics consistency between the proposed action and the imagined motion, and instruction-conditioned task progress, and executes the highest-scoring candidate while keeping the WAM frozen. With four candidates per decision, WISE raises closed-loop success from 6.0\% to 24.0\% on RoboLab with Cosmos~3 Edge and from 75.1\% to 78.3\% mean success on LIBERO with DreamZero, and exceeds the best existing selector in aggregate by 4.0 and 0.9 points. Component ablations and agreement analyses show that the three scores rank candidates differently and that task progress carries the largest weight in the fused decision.
Abstract:World action models (WAMs) jointly generate actions and imagined futures, but a plausible prediction need not be realized by its co-generated action. We introduce WISE (\emph{World Action Inference Scaling through Evaluation}), a pre-execution selector that scores each sampled rollout on action validity and temporal regularity, inverse-dynamics consistency between the proposed action and the imagined motion, and instruction-conditioned task progress, and executes the highest-scoring candidate while keeping the WAM frozen. With four candidates per decision, WISE raises closed-loop success from 6.0\% to 24.0\% on RoboLab with Cosmos~3 Edge and from 75.1\% to 78.3\% mean success on LIBERO with DreamZero, and exceeds the best existing selector in aggregate by 4.0 and 0.9 points. Component ablations and agreement analyses show that the three scores rank candidates differently and that task progress carries the largest weight in the fused decision.
Robotics (cs.RO)
arXiv:2611.03585 [cs.RO]
(or arXiv:2611.03585v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2611.03585 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.
Text integral sub licența deschisă a publicației · 229 cuvinte