Reproduce “Chat with AI: The Surprising Turn of Real-time Video Communication from Human to AI” 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
Jiangkai Wu, Zhiyuan Ren, Liming Liu, Xinggong Zhang
The authors propose context-aware video encoding that uses Mobile-CLIP to preserve question-relevant image patches at lower bitrate. In an earlier DeViBench test with Qwen2.5-Omni, accuracy—the fraction of questions answered correctly—fell from 0.93 to 0.87 as bitrate dropped from 850.1 to 432.7 Kbps, versus 0.73 to 0.33 for the baseline. Selected questions and one model limit generalization; no end-to-end latency reduction was measured.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.