Metraipolis › archive › dc_015da2482583
SparseWAM: Driving World-Action Models are Worth 16 Register Tokens — arXiv:2611.00044
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.00044
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T11:39:47.996Z
- SHA-256 of the archived text
65a74d3085faf8eef5368899753bb900e3bbf66c4f077d6c997003981ae7a0c4- Stable URL of this page
https://metraipolis.com/documento/dc_015da2482583
How to cite
SparseWAM: Driving World-Action Models are Worth 16 Register Tokens — arXiv:2611.00044 — arXiv (en), https://arxiv.org/abs/2611.00044 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_015da2482583 (Metraipolis archive dc_015da2482583, sha256 65a74d3085faf8eef5368899753bb900e3bbf66c4f077d6c997003981ae7a0c4)
BibTeX
@misc{dc015da2482583,
title = {SparseWAM: Driving World-Action Models are Worth 16 Register Tokens — arXiv:2611.00044},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.00044},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_015da2482583; sha256 65a74d3085faf8eef5368899753bb900e3bbf66c4f077d6c997003981ae7a0c4},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.00044. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
World Action Models (WAMs) transfer scene-dynamics priors to planning by jointly forecasting future scenes and ego trajectories. However, WAMs built on video generators must generate dense future videos at inference, incurring prohibitive latency, while efficient variants discard the world branch at deployment and thus forgo test-time imagination. Even when dense video latents are replaced with patch-level DINO or JEPA features, these representations still involve long token sequences and encode decision-irrelevant information. We therefore propose SparseWAM, which forecasts in a sparse register space of only 16 tokens per frame, compressed from frozen DINOv2 features and shaped by joint planning-and-scoring pretraining to concentrate decision-relevant information. Within this space, SparseWAM employs a modality-decoupled Mixture-of-Transformers with a causal attention mask that lets action tokens condition on imagined future registers, retaining explicit imagine-then-act reasoning at deployment. Its compactness further enables low-cost multimodal sampling. As a result, our SparseWAM achieves 92.6 PDMS on NAVSIM v1, 90.1 EPDMS on NAVSIM v2, and 39.6 on NavHard at only 164 ms latency, roughly 1/5 that of DriveVA, demonstrating the efficiency of the sparse register space. Code will be released upon acceptance. Tema degli esploratori: statistica e scienza dei dati (elenco arXiv cs.CV, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2611.00044 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-11-03T11:06:37.362Z dalla pagina https://arxiv.org/list/cs.CV/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).