Metraipolis › archive › dc_a4ec5d38d6df

REEL: Same Frames, Different Answers and Different Costs — arXiv:2610.38158

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.38158
License (as archived)
CC0-1.0
Archived
2026-11-01T02:40:19.233Z
SHA-256 of the archived text
a60cf0f83ce4245607d5c79a4fb781cd8af56f00bca06653c4f826d900dca4db
Stable URL of this page
https://metraipolis.com/documento/dc_a4ec5d38d6df

How to cite

REEL: Same Frames, Different Answers and Different Costs — arXiv:2610.38158 — arXiv (en), https://arxiv.org/abs/2610.38158 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a4ec5d38d6df (Metraipolis archive dc_a4ec5d38d6df, sha256 a60cf0f83ce4245607d5c79a4fb781cd8af56f00bca06653c4f826d900dca4db)
BibTeX
@misc{dca4ec5d38d6df,
  title = {REEL: Same Frames, Different Answers and Different Costs — arXiv:2610.38158},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.38158},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a4ec5d38d6df; sha256 a60cf0f83ce4245607d5c79a4fb781cd8af56f00bca06653c4f826d900dca4db},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.38158. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Video understanding keeps moving toward longer recordings, and the longer a recording is, the more questions can be asked about it. But processing it again for every question does not scale, so many systems watch the video once, build a memory, and answer later questions from it. Yet what to remember is decided before anyone knows what will be asked, and whatever is left out cannot be recovered later. Moreover, whether the savings pay off is judged by token counts, which fall when the input gets shorter but not when the model reuses what it has already computed. We present REEL, an evaluation framework that compares three saved representations built from exactly the same observed frames, so that a change in answers can come only from what the system remembered. Text writes a short record of what happened, Visual keeps the encoder outputs, and Prefix State also keeps the language-model prefix. Each is compared with full reprocessing (Frames) by paired correctness changes, cumulative GPU time, and storage size. Across six vision-language models and two benchmarks, Text and Frames differ in correctness on 20\% to 29\% of questions, even when their mean accuracy is almost the same. By contrast, Visual and Prefix State stay much closer to Frames. With unlimited storage, Prefix State is the cheapest in total GPU time among memories that keep almost all of the answers Frames gets right, for every model on both benchmarks. The trade-off is storage: Prefix State keeps up to about 21 GiB per video, whereas Text needs only kilobytes. Reusing computation and discarding information both look like efficiency, but only the second removes evidence that later questions may need, and neither mean accuracy nor token count tells them apart. REEL makes that difference measurable for any system that answers many questions from one saved memory.

Tema degli esploratori: statistica e scienza dei dati (elenco arXiv cs.CV, nuovi lavori del giorno).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2610.38158 — 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 2026-11-01T02:16:43.867Z dalla pagina https://arxiv.org/list/cs.CV/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).