Metraipolis › archive › dc_f088dbffbdb2

MergeOver: Post-Training Token Merging for Recursive Vision Transformers — arXiv:2608.13141

Source: arXiv (en)
Original: https://arxiv.org/abs/2608.13141
License (as archived): CC0-1.0
Archived: 2026-10-19T04:27:53.377Z
SHA-256 of the archived text: ffb6aaceff3935dded644e4be88666f37b699b438a6e4433e5149bd65cf29570
Stable URL of this page: https://metraipolis.com/documento/dc_f088dbffbdb2

How to cite

MergeOver: Post-Training Token Merging for Recursive Vision Transformers — arXiv:2608.13141 — arXiv (en), https://arxiv.org/abs/2608.13141 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f088dbffbdb2 (Metraipolis archive dc_f088dbffbdb2, sha256 ffb6aaceff3935dded644e4be88666f37b699b438a6e4433e5149bd65cf29570)

BibTeX

@misc{dcf088dbffbdb2,
  title = {MergeOver: Post-Training Token Merging for Recursive Vision Transformers — arXiv:2608.13141},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.13141},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f088dbffbdb2; sha256 ffb6aaceff3935dded644e4be88666f37b699b438a6e4433e5149bd65cf29570},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.13141. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Vision Transformers (ViTs) demonstrate exceptional performance in computer vision but suffer from large parameter counts and quadratic computational complexity, severely limiting their deployment on resource-constrained edge hardware. While recursive weight-sharing reduces parameter counts and token merging mitigates computational and memory bottlenecks, integrating these two paradigms without costly retraining is non-trivial, leaving this intersection largely unexplored. We propose MergeOver, a post-training approach that integrates Token Merging (ToMe) into the recursively weight-shared Sliced Recursive Transformer (SReT). Through an Unmerge tracking stack, constraint-safe merge-rate adjustment, and synchronised token-mass tracking across spatial permutations, MergeOver resolves the spatial and merging constraints of this integration. We further employ a stage-wise single-shot schedule that performs token reduction at the first block of each stage and maintains a fixed sequence length throughout its subsequent recursive iterations. Benchmarked on ImageNet-1K, our selected configuration reduces top-1 accuracy by 1.47 percentage points. On the GPU, it reduces peak activation memory by 37.3% and 38.4% at batch sizes 1 and 16, while throughput decreases by 21.7% at batch size 1 but increases by 21.7% at batch size 16. On a Raspberry Pi 5 (ARM CPU), it reduces latency by 2.4% and 17.6% at batch sizes 1 and 16. These results show that MergeOver can recover a meaningful part of the throughput and memory cost that recursive weight-sharing introduces, without retraining, and provides a baseline for combining token merging with hierarchical recursive transformers.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-14 in cs.LG).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2608.13141 — 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 2026-10-19T04:20:35.906Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-08-14?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).