cd /news/artificial-intelligence/temporal-backtracking-search-for-tes… · home topics artificial-intelligence article
[ARTICLE · art-27501] src=arxiv.org ↗ pub= topic=artificial-intelligence verified=true sentiment=↑ positive

Temporal Backtracking Search for Test-time Generative Video Reasoning

Researchers introduce Temporal Backtracking Search (TBS), a test-time scaling method for generative video reasoning that searches over the temporal axis rather than denoising steps. TBS transforms video generation into an iterative generate-verify-restart loop, achieving 22.7% success in out-of-distribution settings where one-shot generation collapses to 0.7%. The method demonstrates that video models' local reasoning competence far exceeds single-shot rollouts.

read1 min publishedJun 15, 2026

arXiv:2606.13861v1 Announce Type: new Abstract: While test-time scaling has revolutionized reasoning in large language models, generative video reasoning remains bottlenecked by a single-shot paradigm. We demonstrate that searching over denoising steps cannot rescue logically flawed rollouts because spatial trajectories commit early in the diffusion process. Root-level Best-of-N (BoN) sampling is similarly inefficient: reasoning errors cluster early in the temporal axis, and resampling blindly discards verified upstream progress. To unlock effective test-time scaling for video models, we introduce Temporal Backtracking Search (TBS), which shifts the search space to the temporal axis. TBS transforms video generation into an iterative generate-verify-restart loop via three core mechanisms: (1) variable-K conditioning to resume generation from arbitrary clean prefixes; (2) temporal process verification to localize failures and extract valid restart anchors; and (3) prefix-based search to reallocate compute toward extending correct trajectories rather than root resampling. Across algorithmic, navigation, and robotics domains, TBS Pareto-dominates matched-budget BoN. In a strict out-of-distribution setting where one-shot generation collapses (0.7% for BoN), TBS achieves 22.7%, with every solved episode stemming from a restarted branch. Ultimately, TBS reveals that the local reasoning competence of video models far exceeds what single-shot rollouts indicate, providing a scalable test-time framework to unlock it.

── more in #artificial-intelligence 4 stories · sorted by recency
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/temporal-backtrackin…] indexed:0 read:1min 2026-06-15 ·