Peng Wang

Joint modeling of time series with vastly different scales can be achieved without sacrificing accuracy, thanks to a novel self-Adaptive Scale-handling module.

ThinkingVLA achieves a breakthrough in robotic manipulation by seamlessly interleaving visual and textual reasoning, leading to superior performance in long-horizon tasks.

Recursive composition of verifiable environments can boost reasoning performance in RL by up to 3.1 points while using only a fraction of the original environments.

eMoT achieves 100% accuracy on the Game of 24 by evolving reasoning as dynamic memories, outperforming traditional models by up to 17.6%.