New paper introduces "causal writability" in video models with depth limits
A new arXiv paper by Ziming Liu et al. proposes "causal writability," showing that physics rules inside video models can be rewritten via deep interventions, though only within clear depth boundaries.
2026-09-16 ~ 2026-09-16 · 2 related posts
- Video models 'commit' to physics at a depth boundary, new paper finds — ZimingLiu11 · 2026-09-16
1 near-duplicate retellings: ZimingLiu11