Metraipolis › archive › dc_f78dd6fe2f62
Later Is Better: Token Reduction for ViTs Under Distribution Shift — arXiv:2610.07758
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.07758
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T11:21:09.249Z
- SHA-256 of the archived text
344b0dacc2ae9bca15848b0ba38d8c34e0860b523b93083a2a6579588c72ee6e- Stable URL of this page
https://metraipolis.com/documento/dc_f78dd6fe2f62
How to cite
Later Is Better: Token Reduction for ViTs Under Distribution Shift — arXiv:2610.07758 — arXiv (en), https://arxiv.org/abs/2610.07758 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f78dd6fe2f62 (Metraipolis archive dc_f78dd6fe2f62, sha256 344b0dacc2ae9bca15848b0ba38d8c34e0860b523b93083a2a6579588c72ee6e)
BibTeX
@misc{dcf78dd6fe2f62,
title = {Later Is Better: Token Reduction for ViTs Under Distribution Shift — arXiv:2610.07758},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.07758},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f78dd6fe2f62; sha256 344b0dacc2ae9bca15848b0ba38d8c34e0860b523b93083a2a6579588c72ee6e},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.07758. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Training-free token reduction accelerates vision transformers by removing redundant tokens across layers, recovering most of the original accuracy at a fraction of the compute. These methods, however, are designed and evaluated primarily on clean data, and under real-world distribution shift their accuracy gap to the uncompressed model widens with the removal rate. We show that this gap is governed by the reduction schedule, the depth profile of removal, usually left fixed as an implementation detail. Concretely, we introduce a one-parameter late-concentrated power-law schedule that consistently improves out-of-distribution accuracy over flat at no extra inference cost. On ImageNet-C with DeiT-S, the late schedule closes 83% of that gap at a 26% compute reduction (+1.17pp), and 99% of it at a lighter 7% reduction (+0.26pp). The gain cannot be attributed to retaining more tokens or using extra compute: held to flat's compute, the late schedule removes more tokens in total and leaves fewer tokens at the end, yet still wins. Single-layer probes point to a mechanism: earlier reductions perturb features that pass through more remaining layers, front-loading reduction error in depth. The effect is broad, holding across five token-reduction methods (ToMe, EViT, ATS, ATC, PiToMe), nine backbones, all ImageNet-C corruption types, eight further shift suites, and two further modalities, video and vision-language QA. It is also specific to shift, still positive on clean and rising monotonically to ~4x that at the highest severity 5. The schedule keeps its gain under six test-time adaptation methods, and needs no per-input or per-domain tuning. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.07758 — 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 2026-10-07T11:15:24.763Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).