The world model team at Shengshu Technology has launched a half-month-long public live stream, showcasing the entire training process of Motubrain 2.0. This initiative aims to present the training procedures and research advancements in embodied AI models to a broader audience, while also delving into the practical applications of Physical AI. Throughout the live stream, viewers will witness four critical phases: pre-training, post-training, Agent task execution, and RSI exploration. The event is designed to demystify the training of embodied models, offering a comprehensive view of the entire workflow—from initial training and testing to iterative capability enhancements. This includes showcasing the inevitable setbacks, trial-and-error processes, and subsequent adjustments. By doing so, the team hopes to foster greater transparency within the embodied intelligence research community, share valuable experiences, challenges, and insights with fellow researchers, and provide additional reference points for the practical implementation of Physical AI. Following the live stream, Motubrain 2.0 will be officially released as open-source software.