aiqbench analysis
This paper addresses a critical gap in long-video understanding by introducing TRACE, a training-free agent that ensures answers are grounded in complete visual evidence rather than incomplete frame sampling. By auditing decoded frames against necessary event intervals, the method significantly improves accuracy and efficiency over standard uniform decoding strategies without requiring additional model training. This work advances the reliability of multimodal AI systems tasked with complex temporal reasoning over extended video sequences.
TRACE: Temporal Retrieval with Anchored and Convergent Evidence for Long-Horizon Video Understanding
arXiv:2608.22516v1 Announce Type: cross Abstract: A long-video answer is evidence-supported only when the frames decoded from the video cover every event the answer depends on. Existing evaluations score final-answer correctness or predicted evidence intervals, but the frames a method decodes before answering are rarely audited, so correct answers can still rest on incomplete observation. We introduce VES-Bench, a 600-question benchmark of Temporal Ordering and Event Counting items over 348 public long videos. Each item carries a jointly necessary set of evidence intervals, letting us audit at three strictness levels whether a method's decoded frames cover every one of them. We also propose TRACE, a training-free agent that grounds answers in raw visual clips, builds an evidence bundle round by round, and stops only when the answer stabilises as the bundle grows and a final pass over the same clips returns the same answer. Under a same-backbone audit, TRACE answers 50.7% of questions correctly with at least two decoded frames inside every evidence interval, at 98.7 frames per question: over 10 points above uniform decoding at 128 frames (40.2%), and within 2.6 points of uniform decoding at 256 frames at 0.39x its frame cost, while reaching the highest answer accuracy in the audit (63.5%). TRACE also stays competitive on Video-MME (86.1), LVBench (75.6), and LongVideoBench (75.1).
Abstract:A long-video answer is evidence-supported only when the frames decoded from the video cover every event the answer depends on. Existing evaluations score final-answer correctness or predicted evidence intervals, but the frames a method decodes before answering are rarely audited, so correct answers can still rest on incomplete observation. We introduce VES-Bench, a 600-question benchmark of Temporal Ordering and Event Counting items over 348 public long videos. Each item carries a jointly necessary set of evidence intervals, letting us audit at three strictness levels whether a method's decoded frames cover every one of them. We also propose TRACE, a training-free agent that grounds answers in raw visual clips, builds an evidence bundle round by round, and stops only when the answer stabilises as the bundle grows and a final pass over the same clips returns the same answer. Under a same-backbone audit, TRACE answers 50.7% of questions correctly with at least two decoded frames inside every evidence interval, at 98.7 frames per question: over 10 points above uniform decoding at 128 frames (40.2%), and within 2.6 points of uniform decoding at 256 frames at 0.39x its frame cost, while reaching the highest answer accuracy in the audit (63.5%). TRACE also stays competitive on Video-MME (86.1), LVBench (75.6), and LongVideoBench (75.1).
Accepted to EMNLP 2026 Main Conference. 19 pages, 5 figures, 6 tables. Project page: this https URL
Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
arXiv:2608.22516 [cs.CV]
(or arXiv:2608.22516v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2608.22516 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 328 words
Related coverage
Game Design Creativity evals
LessWrong · Oct 12, 2026
Temporal Evidence Forests for Long-Horizon Multi-Party Conversational Memory
arXiv · Oct 12, 2026
GRAVITY: Architecture-Agnostic Structured Anchoring for Long-Horizon Conversational Memory
arXiv · Oct 1, 2026
Models Can't Remember Their Training. Neither Can You.
LessWrong · Jul 19, 2026