Metraipolis › archive › dc_9e016bf82bc8

The Limits of Binding in Dual Encoders — arXiv:2608.15971

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.15971
License (as archived)
CC0-1.0
Archived
2026-10-18T19:01:23.836Z
SHA-256 of the archived text
53059a9009c4f38f5216f1fc55d7aa5252e194837ae41b7a2a77c6b3fa615c13
Stable URL of this page
https://metraipolis.com/documento/dc_9e016bf82bc8

How to cite

The Limits of Binding in Dual Encoders — arXiv:2608.15971 — arXiv (en), https://arxiv.org/abs/2608.15971 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_9e016bf82bc8 (Metraipolis archive dc_9e016bf82bc8, sha256 53059a9009c4f38f5216f1fc55d7aa5252e194837ae41b7a2a77c6b3fa615c13)
BibTeX
@misc{dc9e016bf82bc8,
  title = {The Limits of Binding in Dual Encoders — arXiv:2608.15971},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.15971},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_9e016bf82bc8; sha256 53059a9009c4f38f5216f1fc55d7aa5252e194837ae41b7a2a77c6b3fa615c13},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.15971. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Dual-encoder models such as CLIP score an image-caption pair by a single inner product of two independently computed unit vectors, and fail at binding, often scoring near chance when asked to distinguish "a red car and a blue dog" from "a blue car and a red dog". We give a mathematical account of when this failure is necessary and when it is contingent. Working within the ideal-encoder framework proposed by Kang et al., we first show the relevant axioms are satisfiable, so every impossibility must enter through an added, checkable hypothesis. We then prove three such obstructions. Depth: for recursive role-binding codes the swap margin obeys an exact law $m(D) = 2b^{-D}$ in the nesting depth D, with a finite-dimension version holding up to one explicitly flagged concentration estimate; the resolvable depth grows only logarithmically in the dimension and is single-digit at CLIP scale, the nesting depth of ordinary language. Objective: architecture-free throttle theorems showing that the contrastive objective's entire reward for binding is bounded by the rate at which training contrasts a caption against its own swap, a rate that vanishes at web scale, and that exactly reversed binding costs only that rate times the mean binding margin; both are verified in simulation. Geometry: a tight smoothness-binding frontier: the closer the two swap-related captions must embed to a shared paraphrase anchor, the smaller the binding margin can be, with an exact constant. Measuring its text-only diagnostic across 18 deployed text encoders, every model sits at roughly 25-35% of its ceiling, and the induced per-item ceiling tracks SugarCrepe's subset difficulty at r = 0.99. Binding failure in deployed dual encoders is thus not a dimension or smoothness limit today, but an incentive and code-structure limit, with a proved depth ceiling that remains once those are fixed.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-18 in cs.CL).
Categorie arXiv: Machine Learning (cs.LG) ; Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2608.15971 — 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 2026-10-18T18:06:57.926Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).