Can Test-Time Scaling Improve World Foundation Model?
This work addresses the practical challenge of scaling world foundation models for applications like autonomous driving and robotics, offering an incremental improvement in inference efficiency.
The paper tackles the computational and data constraints of world foundation models by proposing SWIFT, a test-time scaling framework that improves inference efficiency without retraining or enlarging the model, demonstrating its effectiveness on the COSMOS model with test-time scaling laws.
World foundation models, which simulate the physical world by predicting future states from current observations and inputs, have become central to many applications in physical intelligence, including autonomous driving and robotics. However, these models require substantial computational resources for pretraining and are further constrained by available data during post-training. As such, scaling computation at test time emerges as both a critical and practical alternative to traditional model enlargement or re-training. In this work, we introduce SWIFT, a test-time scaling framework tailored for WFMs. SWIFT integrates our extensible WFM evaluation toolkit with process-level inference strategies, including fast tokenization, probability-based Top-K pruning, and efficient beam search. Empirical results on the COSMOS model demonstrate that test-time scaling exists even in a compute-optimal way. Our findings reveal that test-time scaling laws hold for WFMs and that SWIFT provides a scalable and effective pathway for improving WFM inference without retraining or increasing model size. Project page: https://scalingwfm.github.io/.