Metraipolis › archive › dc_d75ad71fc64d

TileShare: Training-Free Acceleration of Overlapping Tiled ViT Decoding — arXiv:2610.32233

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.32233
License (as archived)
CC0-1.0
Archived
2026-10-26T20:14:16.414Z
SHA-256 of the archived text
164c89ec244349871f950fc2e6c92aa5036d54df3609341ab5b8440b5bb53b4d
Stable URL of this page
https://metraipolis.com/documento/dc_d75ad71fc64d

How to cite

TileShare: Training-Free Acceleration of Overlapping Tiled ViT Decoding — arXiv:2610.32233 — arXiv (en), https://arxiv.org/abs/2610.32233 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d75ad71fc64d (Metraipolis archive dc_d75ad71fc64d, sha256 164c89ec244349871f950fc2e6c92aa5036d54df3609341ab5b8440b5bb53b4d)
BibTeX
@misc{dcd75ad71fc64d,
  title = {TileShare: Training-Free Acceleration of Overlapping Tiled ViT Decoding — arXiv:2610.32233},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.32233},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d75ad71fc64d; sha256 164c89ec244349871f950fc2e6c92aa5036d54df3609341ab5b8440b5bb53b4d},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.32233. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Fixed-window Vision Transformer (ViT) decoders run at deployment resolution as overlapping tiles: every token in an overlap band is processed by several full decoder instances. In the 2.4B-parameter MiniMax-H3 video decoder, about half of all per-token computation is duplicated. Removing the tiles fails because attention, global tokens and coordinates are bound to them. TileShare keeps each released tile's attention and computes every other operation once per token. The released computation changes at one kind of step only: an overlap token's per-tile results are blended into one shared state with the released weights. Interventions show that this blend costs quality where attention reads the shared states, mainly in the first few layers, which TileShare therefore keeps per tile. Without training, keeping six layers per tile decodes 1.38-1.51x faster than batched and compiled tiling on two GPU types (up to 1.96x with FP8) and is equivalent to the released decoder in PSNR, LPIPS and DISTS on DAVIS videos. On unseen UVG videos with the setting frozen beforehand, it matches PSNR and DISTS and improves LPIPS. TileShare carries over to two further decoders and is again confirmed on unseen data. It is equivalent in PSNR and DISTS on MAGI-1. On the ViTok-v2 image autoencoder at 2K-4K, it matches overlapping tiling in PSNR, improves LPIPS and DISTS and decodes 1.5-1.8x faster. How many layers stay per tile is measured once per model (for decoders, without labels); we report it for 14 decoders and encoders. Optionally, two GPU-hours of self-distillation let MiniMax-H3 share every layer without a loss in PSNR, and a coordinate field trained while the network stays frozen brings fully shared CLIP-based open-vocabulary segmentation to parity with sliding windows on seven benchmarks.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.32233 — 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 2026-10-26T19:49:03.474Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).