Abstract page for arXiv paper 2610.28698: Scaling in Games: Continued Pre-Training For Embodied Agents In Diverse Virtual Worlds