Reproduce “MosaiChunk: Compositing Spatio-Temporal Memory for Autoregressive Video Generation” 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
Yiwen Zhang, Haocheng Xi, Michael Tian-Yue Liu, Alexei A. Efros, Hadar Averbuch-Elor, Qianqian Wang, Haiwen Feng
MosaiChunk retrieves selected sections of historical video memory and combines them with a recent sliding window for a frozen generator. A router selects relevant earlier sections within the far-memory budget. On H3-AR, median revisit CLIP similarity is 0.936, versus 0.841 for adapted whole-chunk retrieval and 0.751 for sliding-window generation; gains do not extend uniformly across camera paths.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.