Reproduce “PACE: Plan, Attend, Check, and Emit for Training-Free Response Timing in Streaming Video Understanding” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Siqi Guo, Yuxia Chen, Wenbo Zhu, Yangguang Ji, Zhenxiang Jiang, Teng Zhong, Yu Zhang
PACE addresses streaming video question answering by deciding when a reply is complete. It creates a question-specific watch plan, then checks drafts for stability, visual grounding, and event completion before emission. On ProactiveVideoQA, GPT-4.1-mini reaches average PAUC of 61.0, versus 55.2 without PACE, under matched chunking and frame budgets. The paper reports no matched measurement of runtime or total computation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.