Chain-of-Dynamics
A sequence representation composed of discrete motion tokens that describes the intermediate dynamic changes from the current world state to a future state. It does not depend on a specific embodiment and can serve as a general cross-embodiment action representation.
Timeline (last 90 days)
ThinkWorld's core innovation lies in representing each action thought as a Chain-of-Dynamics (CoD), that is, a sequence of discrete motion tokens that captures the intermediate dynamics from the current world to a future state and serves as a general action representation across embodiments.
ThinkWorld's training process consists of two steps: first, the CoD generation capability is initialized with action-annotated robot demonstration data; then the model predicts, for each CoD, the future state it induces, compares the prediction with the real observation, and uses reinforcement learning to increase the generation probability of those CoDs whose predictions are more accurate.
Reinforcing, in ThinkWorld, those CoDs whose predictions match the actual observations more closely is typically achieved through policy optimization in reinforcement learning, for example by using the negative of the prediction error as the reward and adopting REINFORCE or preference-optimization-style objectives to increase the generation probability of high-reward sequences.
All Facts (3)
ThinkWorld's core innovation lies in representing each action thought as a Chain-of-Dynamics (CoD), that is, a sequence of discrete motion tokens that captures the intermediate dynamics from the current world to a future state and serves as a general action representation across embodiments.
50%UnverifiedThinkWorld's training process consists of two steps: first, the CoD generation capability is initialized with action-annotated robot demonstration data; then the model predicts, for each CoD, the future state it induces, compares the prediction with the real observation, and uses reinforcement learning to increase the generation probability of those CoDs whose predictions are more accurate.
50%UnverifiedReinforcing, in ThinkWorld, those CoDs whose predictions match the actual observations more closely is typically achieved through policy optimization in reinforcement learning, for example by using the negative of the prediction error as the reward and adopting REINFORCE or preference-optimization-style objectives to increase the generation probability of high-reward sequences.
50%