AMD GPUOpen today announced Schola v2.1, an update to its platform designed for deep learning training and inference at scale. The release focuses on enhancing capabilities for AI development, particularly for large-scale deployments. Key changes include deeper integration with Unreal Engine and the addition of StateTree support, allowing for more sophisticated state management within AI applications. Furthermore, Schola v2.1 now features Kubernetes-oriented distributed training, which enables efficient scaling of workloads across clusters. The platform also incorporates stronger Minari workflows, which aid in dataset management and evaluation. According to AMD GPUOpen, these improvements aim to "streamline training and inference at scale" by providing a more robust and integrated stack for AI developers.