Metraipolis › archive › dc_b4edc8e87a76

Counting Events, Not Seeing Them: Diagnosing and Repairing the Readout Failure in Video LLMs — arXiv:2611.09949

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.09949
License (as archived)
CC0-1.0
Archived
2026-11-10T18:44:59.080Z
SHA-256 of the archived text
87bd18960efe3a97c39289ee2d1a7e326255f529e896c72e5939538bf6aefaef
Stable URL of this page
https://metraipolis.com/documento/dc_b4edc8e87a76

How to cite

Counting Events, Not Seeing Them: Diagnosing and Repairing the Readout Failure in Video LLMs — arXiv:2611.09949 — arXiv (en), https://arxiv.org/abs/2611.09949 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_b4edc8e87a76 (Metraipolis archive dc_b4edc8e87a76, sha256 87bd18960efe3a97c39289ee2d1a7e326255f529e896c72e5939538bf6aefaef)
BibTeX
@misc{dcb4edc8e87a76,
  title = {Counting Events, Not Seeing Them: Diagnosing and Repairing the Readout Failure in Video LLMs — arXiv:2611.09949},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.09949},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_b4edc8e87a76; sha256 87bd18960efe3a97c39289ee2d1a7e326255f529e896c72e5939538bf6aefaef},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.09949. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Video LLMs fail conspicuously at counting how many times an object crosses the screen, a failure attributed to missing object permanence. Using construction-grounded synthetic testbeds with independently validated labels, we show this attribution is incomplete: the count signal is present but unread - linear probes decode per-window presence at 93% while global count decoding collapses to a duration confound. Concurrent work reports the same representation-output dissociation for repetition counting; we show that in exit-and-reappear event counting it is repairable: two readout-directed repairs (episode-localization supervision and a discrete event-symbol side channel) raise same-object counting from 46.7% to 65.0% (oracle 66.7%) on a chance-33% benchmark across 7B-30B systems. Transfer to an independent factory costs nothing, and an audit of the measurement apparatus exposes four data-integrity failures that had obscured the boundary. A real-video audit (REALCOUNT, n=18, blind-audited) shows the boundary persists - best 55.6% at chance 33.3% - and symbol transfer fails zero-shot. A 1-epoch in-domain adaptation bench is negative (six levers at 28-50%), but a 3-epoch schedule shows this is budget-limited for the symbol channel: with a 3-epoch schedule and sequence-disjoint oracle timelines the reader reaches 94.4% on REALCOUNT, and with the deployed detector's timelines a 2-epoch run reaches 72.2%. The residual gap is detector evidence and the reader's match to detector noise, not reader capacity: a frozen-feature boundary probe finds the signal weak (AUC 0.62 over a 0.51 permutation control) and non-transferring across clips (0.77 within-clip vs. 0.62 across windows).

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Computation and Language (cs.CL); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2611.09949 — 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 2026-11-10T18:37:52.467Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).