LY
Liuhaichen Yang, Zhengyang Zhong, Hanshang Zhu, Ningwei Bai, Qichen Yin, Zhi Han, Jiarui Qin, Zhuang Jiang, Chenchao Sheng, Hanbo Ma, Junkai Liu, Junkai Sun, Dongcheng Lyu, Bo Liu, Yi Dong, Zezhi Tang
· 1 min read
ResearcharXiv cs.AI
WAM-OPD: Joint Video-Action Supervision for World Action Model Post-Training with On-Policy Distillation
arXiv:2608.22364v2 Announce Type: replace
Abstract: World Action Models (WAMs) generate both future video and robot actions, offering two connected outputs for post-training supervision. How can a pretrained WAM learn from a stronger Teacher on the histories it encounters during execution? We present WAM-OPD, which collects Student rollout histories and queries a Teacher for paired video and action targets. The Student learns from both targets while retaining its one-step video and action generation at deployment. Across 12 RoboTwin 2.0 tasks, WAM-OPD improves average success from 33.8% to 65.7%; across four real-robot tasks, it improves average success from 51.4% to 64.6%. With the collected Student histories held fixed, joint video-action supervision achieves the highest observed success on all three ablation tasks, while either modality alone also improves performance. A separate comparison with Teacher-generated histories finds task-dependent differences between the two history sources. These results demonstrate the value of paired video-action supervision for improving WAM policies without increasing their deployed sampling budget.
Original source
This story was published by arXiv cs.AI and written by Liuhaichen Yang, Zhengyang Zhong, Hanshang Zhu, Ningwei Bai, Qichen Yin, Zhi Han, Jiarui Qin, Zhuang Jiang, Chenchao Sheng, Hanbo Ma, Junkai Liu, Junkai Sun, Dongcheng Lyu, Bo Liu, Yi Dong, Zezhi Tang. SyncAI.news shows a preview; the complete article is on the publisher's site.
Read the full story on arxiv.org
