Metraipolis › archive › dc_7bf1016f7418

RIG-RoPE: Relation-Stratified Multimodal Attention with Instance-Local Rotary Geometry and Representation-Aware Traversal Coordinates — arXiv:2608.05154

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.05154
License (as archived)
CC0-1.0
Archived
2026-10-31T10:37:49.184Z
SHA-256 of the archived text
088170d523a57ca3c7a101f1f38deda9263bf4de38550420058ee7bfa5c40e32
Stable URL of this page
https://metraipolis.com/documento/dc_7bf1016f7418

How to cite

RIG-RoPE: Relation-Stratified Multimodal Attention with Instance-Local Rotary Geometry and Representation-Aware Traversal Coordinates — arXiv:2608.05154 — arXiv (en), https://arxiv.org/abs/2608.05154 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7bf1016f7418 (Metraipolis archive dc_7bf1016f7418, sha256 088170d523a57ca3c7a101f1f38deda9263bf4de38550420058ee7bfa5c40e32)
BibTeX
@misc{dc7bf1016f7418,
  title = {RIG-RoPE: Relation-Stratified Multimodal Attention with Instance-Local Rotary Geometry and Representation-Aware Traversal Coordinates — arXiv:2608.05154},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.05154},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7bf1016f7418; sha256 088170d523a57ca3c7a101f1f38deda9263bf4de38550420058ee7bfa5c40e32},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.05154. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Multimodal rotary positional encodings apply temporal, height, and width phases to interleaved text, image, and video tokens. This creates two ambiguities: cross-instance spatial displacement depends on preprocessing chart choices unless registration is declared, and scalar advance across visual blocks is often inherited from coordinate extrema rather than defined at the representation level. We introduce RIG-RoPE, combining instance-local rotary geometry, relation-stratified attention, and representation-aware traversal coordinates. RIG-RoPE normalizes relation-homogeneous scores separately, allocates mass with a common H/W-neutral LogSumExp statistic, and uses traversal extent that is additive over ordered slices and sublinear over parallel spatial scale. Text advances by unit increments, image patches are simultaneous, and video accumulates over tokenizer temporal tokens. In a matched, inference-only Qwen2-VL-2B checkpoint experiment, native and RIG text-only paths were exactly equal. RIG was exactly invariant to a whole-chart single-image translation and to translating only the second instance of an unregistered image pair. Native attention remained sensitive to the latter, while an H/W-collapse control confirmed that RIG retained same-instance spatial effects; visual embeddings and all parameters were unchanged. Across three seeds of a frozen tiny task, RIG also had zero clean-to-Gauge logit change, whereas the raw-H/W baseline changed in every seed. Gauge-accuracy differences were +2/72, 0, and 0, failing the preregistered stability gate. These results support the specified activation and Gauge mechanisms, not stable task improvement, universality, or empirical superiority.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-08-07 in cs.CV).
Categorie arXiv: Computation and Language (cs.CL) ; Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2608.05154 — 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 2026-10-31T10:23:31.070Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-08-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).