1 paper
Yuyuan Yang, Junkun Hong, Hongrong Wang +11
Embodied task planning demands vision-language models to generate action sequences that are both visually grounded and causally coherent over time. However, existing training parad…