Metraipolis › archive › dc_dcc1122b3387

Watch-Think-Interact: Bootstrapping Long-Horizon Multi-Turn Streaming Video Reasoning with Reinforcement Learning — arXiv:2609.37035

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.37035
License (as archived)
CC0-1.0
Archived
2026-10-07T16:23:26.474Z
SHA-256 of the archived text
0e561b759dcad35e292fb035209d8a80f5c0cbf99c027cfe7ff4b9cff5d3f323
Stable URL of this page
https://metraipolis.com/documento/dc_dcc1122b3387

How to cite

Watch-Think-Interact: Bootstrapping Long-Horizon Multi-Turn Streaming Video Reasoning with Reinforcement Learning — arXiv:2609.37035 — arXiv (en), https://arxiv.org/abs/2609.37035 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_dcc1122b3387 (Metraipolis archive dc_dcc1122b3387, sha256 0e561b759dcad35e292fb035209d8a80f5c0cbf99c027cfe7ff4b9cff5d3f323)
BibTeX
@misc{dcdcc1122b3387,
  title = {Watch-Think-Interact: Bootstrapping Long-Horizon Multi-Turn Streaming Video Reasoning with Reinforcement Learning — arXiv:2609.37035},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.37035},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_dcc1122b3387; sha256 0e561b759dcad35e292fb035209d8a80f5c0cbf99c027cfe7ff4b9cff5d3f323},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.37035. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Streaming video assistance requires models to answer asynchronous questions from an observed prefix under a fixed context budget. Existing approaches model response timing or compress history, but an online state formed before future questions are known can omit visual details before later questions reveal their relevance; the retained state alone cannot recover them. We introduce Watch-Think-Interact (WTI), a closed-loop framework for multi-question streaming video reasoning. WTI maintains compact natural-language memory entries tagged with source-video time ranges; these entries support direct reasoning when sufficient and otherwise anchor selective recall of finer visual evidence. For each question, WTI answers when current context and memory suffice, continues watching when required evidence has not appeared, or recalls a relevant past interval and decides again after incorporating the returned chunks, without replaying the full observed history. To train this behavior, we construct WTI-82K, comprising 82,335 timed questions across 4,812 causally aligned trajectories, and develop Stream-GDPO to optimize complete multi-question streaming rollouts using trajectory-level feedback for response timing, source-video recall, and memory updates. WTI achieves state-of-the-art aggregate performance among the compared open-source streaming baselines, reaching 83.3% on StreamingBench and 73.6% weighted overall accuracy on OVO-Bench.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-30 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.37035 — 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 2026-10-07T15:54:53.583Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).