Metraipolis › archive › dc_50c19e741ca6

Training-Free Predictive Spatio-Temporal Token Pruning for Few-Step Autoregressive Video Diffusion — arXiv:2611.07551

Source: arXiv (en)
Original: https://arxiv.org/abs/2611.07551
License (as archived): CC0-1.0
Archived: 2026-11-06T14:53:23.553Z
SHA-256 of the archived text: 7ea8902c35a751819b7991a58c96142b8c7801bd110dc60293bcaa38fafad604
Stable URL of this page: https://metraipolis.com/documento/dc_50c19e741ca6

How to cite

Training-Free Predictive Spatio-Temporal Token Pruning for Few-Step Autoregressive Video Diffusion — arXiv:2611.07551 — arXiv (en), https://arxiv.org/abs/2611.07551 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_50c19e741ca6 (Metraipolis archive dc_50c19e741ca6, sha256 7ea8902c35a751819b7991a58c96142b8c7801bd110dc60293bcaa38fafad604)

BibTeX

@misc{dc50c19e741ca6,
  title = {Training-Free Predictive Spatio-Temporal Token Pruning for Few-Step Autoregressive Video Diffusion — arXiv:2611.07551},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.07551},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_50c19e741ca6; sha256 7ea8902c35a751819b7991a58c96142b8c7801bd110dc60293bcaa38fafad604},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.07551. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Few-step autoregressive (AR) video diffusion models generate videos chunk by chunk and have made efficient video generation practical. However, further accelerating these models toward real-time generation remains challenging, as few-step distillation leaves little cross-step redundancy for existing training-free caching methods to exploit. We revisit this regime and identify two token-level redundancy patterns: cross-frame redundancy from static content shared across adjacent frames, and cross-step redundancy from tokens that converge early during denoising. Crucially, we find that early clean estimates along the model's own denoising trajectory provide an effective online signal for locating both patterns, requiring no retraining, auxiliary predictor, or reference video. Based on these findings, we propose PreSTP (Predictive Spatio-Temporal Token Pruning), a training-free framework that constructs a unified token-wise redundancy mask from cross-frame similarity and cross-step convergence. Tokens marked as redundant skip subsequent computation under a consistent last-computed-value restoration rule, and a content-adaptive mask reuse strategy further reduces dense probing cost. On 5-second generation, PreSTP achieves 1.52x-3.20x speedups and a better efficiency-fidelity trade-off than coarse-grained caching baselines such as TeaCache and FlowCache. On 30-second generation, it reaches up to 1.96x speedup while improving VBench-Long and suppressing long-horizon drift. Notably, it continues to reduce inference cost even in the extreme one-step regime, where cross-step redundancy vanishes entirely and coarse step-wise caching is no longer applicable.

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/2611.07551 — 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 2026-11-06T14:25:42.610Z dalla pagina https://arxiv.org/list/cs.CV/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).