← 研踪 PaperTrailarXiv:2611.24512 · v1
Comparing Each Generation with Independent Siblings Gives Video World Model Post-Training a Proper Objective for Outcome Frequencies
首次提交 2026-11-18 22:34
版本更新 2026-11-18 22:34
雷达观测 2026-11-19 15:04
摘要
Video world models are used as simulators, so when a first frame leaves several physical outcomes open, repeated generations should produce each outcome as often as it occurs in the data. Physics post-training scores every generation against the single recorded future of its history, which makes single videos more faithful, but the expected score is linear in the model distribution and is maximized by always producing the most likely outcome. The log-likelihood of the recorded outcome is maximized by the data distribution, and its gradient weights agreement with the record by the inverse of how often the model produces the generated outcome, a frequency that other generations of the same history with independent noise, its siblings, can estimate. To train with this weight from a finite group, we propose SibPO, which performs Sibling-paired Policy Optimization: siblings estimate a truncated inverse-frequency weight for agreement with the record, and a second term subtracts the same weight times the agreement with the siblings, which makes the expected objective a strictly proper divergence over outcome cells at every group size. On PisaBench, SibPO lowers the KS distance to simulated landing times by 9.1-10.2% relative to the strongest method outside its reward family on three of four video models without degrading single videos, and it lowers the error of estimated robot policy success rates by 13%.
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-18 22:34 · 最新观测版本
Comparing Each Generation with Independent Siblings Gives Video World Model Post-Training a Proper Objective for Outcome Frequencies
Video world models are used as simulators, so when a first frame leaves several physical outcomes open, repeated generations should produce each outcome as often as it occurs in the data. Physics post-training scores every generation against the single recorded future of its history, which makes single videos more faithful, but the expected score is linear in the model distribution and is maximized by always producing the most likely outcome. The log-likelihood of the recorded outcome is maximized by the data distribution, and its gradient weights agreement with the record by the inverse of how often the model produces the generated outcome, a frequency that other generations of the same history with independent noise, its siblings, can estimate. To train with this weight from a finite group, we propose SibPO, which performs Sibling-paired Policy Optimization: siblings estimate a truncated inverse-frequency weight for agreement with the record, and a second term subtracts the same weight times the agreement with the siblings, which makes the expected objective a strictly proper divergence over outcome cells at every group size. On PisaBench, SibPO lowers the KS distance to simulated landing times by 9.1-10.2% relative to the strongest method outside its reward family on three of four video models without degrading single videos, and it lowers the error of estimated robot policy success rates by 13%.
查看此版本解读 →查看此版本原文 ↗