The field of robot learning and manipulation is rapidly advancing, with a focus on developing scalable and efficient methods for training robots to perform complex tasks. Recent research has explored the use of simulation environments, such as video world models, to evaluate policies and improve training efficiency. Other works have investigated the use of decoupled training recipes, diffusion-based methods, and visual sim-to-real frameworks to improve the performance of robots in various tasks, including grasping, manipulation, and locomotion. Notable papers in this area include: Scalable Policy Evaluation with Video World Models, which demonstrates the effectiveness of using video world models for policy evaluation. Decoupled Action Head: Confining Task Knowledge to Conditioning Layers, which proposes a decoupled training recipe for improving training efficiency. DiffuDepGrasp: Diffusion-based Depth Noise Modeling Empowers Sim2Real Robotic Grasping, which presents a diffusion-based method for sim-to-real robotic grasping. VIRAL: Visual Sim-to-Real at Scale for Humanoid Loco-Manipulation, which introduces a visual sim-to-real framework for humanoid loco-manipulation. In-N-On: Scaling Egocentric Manipulation with in-the-wild and on-task Data, which provides a scalable recipe for collecting and using egocentric data. DynaMimicGen: A Data Generation Framework for Robot Learning of Dynamic Tasks, which introduces a scalable dataset generation framework for dynamic tasks. InternData-A1: Pioneering High-Fidelity Synthetic Data for Pre-training Generalist Policy, which provides evidence that synthetic data alone can match the performance of real-robot pre-training. Dexterity from Smart Lenses: Multi-Fingered Robot Manipulation with In-the-Wild Human Demonstrations, which enables learning multi-fingered policies from in-the-wild human videos.