no code implementations • 6 Jun 2023 • Minting Pan, Yitao Zheng, Wendong Zhang, Yunbo Wang, Xiaokang Yang
Pretraining RL models on offline video datasets is a promising way to improve their training efficiency in online tasks, but challenging due to the inherent mismatch in tasks, dynamics, and behaviors across domains.
1 code implementation • 27 Mar 2023 • Minting Pan, Xiangming Zhu, Yitao Zheng, Yunbo Wang, Xiaokang Yang
On top of our previous work, we further consider the sparse dependencies between controllable and noncontrollable states, address the training collapse problem of state decoupling, and validate our approach in transfer learning setups.
2 code implementations • 27 May 2022 • Minting Pan, Xiangming Zhu, Yunbo Wang, Xiaokang Yang
First, by optimizing the inverse dynamics, we encourage the world model to learn controllable and noncontrollable sources of spatiotemporal changes on isolated state transition branches.