Reproduce “StarWorld: A Cross-Layer World Model for Starlink Prediction and Streaming” 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
Bowen Zhang, Pengcheng Luo, Hongjun Xie, Shujin Jia, Genke Yang, Jian Chu
StarWorld samples user-observable Starlink network futures, runs prediction and streaming rollouts, and compares configurations on shared futures, with a Trust Gate for abstention. Its 60-second low-capacity threshold-weighted CRPS is 0.292 versus 0.324 for Causal Transformer. Pairwise QoE-effect error is 76% lower than Direct action across 89 test groups; this does not mean better experienced QoE or model internal network operation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.