Reproduce “Comparing Each Generation with Independent Siblings Gives Video World Model Post-Training a Proper Objective for Outcome Frequencies” 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
Xin Li, Yiwen Chen
SibPO trains video world models to represent conditional outcome frequencies by comparing each generation with the recorded future and independent sibling generations from the same input. On trained mechanisms, Cosmos-Predict2.5-2B achieves a PisaBench KS distance of 0.241, versus 0.276 with DPH-RL. Improvement is not universal: the CogVideoX comparison is not statistically distinguishable from zero.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.