Lixuan Zhang, Meina Kan, Shiguang Shan, Xilin Chen: PreLAR: World Model Pre-training with Learnable Action Representation. ECCV (23) 2024: 185-201