Causal writability in video models
A Chosen Future Can Still Be Rewritten: Causal Writability in Video Models
This research provides a new method to control video generation by identifying and correcting physically incorrect motion, which could improve the accuracy of video models.
Researchers found that video models can learn correct physical motion but may fail to use it.
They trained a model on videos where red masses oscillate slowly and blue masses quickly, then tested a red mass with fast motion.
Even when the model generated slow motion, a low-dimensional edit restored the correct fast motion.This ability, called causal writability, has a sharp depth boundary where edits are effective before a certain network depth.
A Chosen Future Can Still Be Rewritten: Causal Writability in Video Models
When a video model generates physically incorrect motion, did it fail to learn the correct motion, or did it learn it but fail to use it? We show the latter: the correct motion remains available inside the model and can still be made to control the generated video. We train on videos where red masses oscillate slowly and blue masses oscillate quickly, then test a red mass with fast observed motion. Even when the model generates slow motion in this conflicting case, a low-dimensional edit predicted from simple physical variables restores the correct fast motion. We call this ability causal writability. At fixed strength, we find a sharp depth boundary: the same edit changes the video before the boundary but not after it. This closure marks commitment for that write. The motion signal nevertheless remains, and a stronger downstream write can restore physical motion, while excessive gain overshoots. Early causal writability predicts which errors training later corrects: those errors are writable at more network depths than errors that persist. We reproduce both causal writability and its sharp closure in a pretrained 1.3B video model, supporting generality across model scale and training regime.
This report is categorized under AITOP's "Papers" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include video generation, causal writability, AI models, research paper.
Source attribution: curated from arXiv cs.LG (@Xingyun Wang, Haomin Zheng, Man Yuan, Leqian Yang, Ziming Liu).
