Announcing AMD Schola v2.1: state trees, scale, and a richer training stack
AMD Schola v2.1 is mostly about making reinforcement-learning workflows fit more naturally into real production pipelines. The update deepens Unreal Engine integration with StateTree support, so teams can connect gameplay state logic to training and inference without as much custom plumbing. It also expands the training stack with stronger Minari workflows, which should help with dataset handling and experiment iteration.
The other notable piece is scale: AMD is pushing distributed training in a Kubernetes-oriented direction, which matters if you’re running larger experiments or want to treat training like a managed service instead of a one-off local job. For game teams, that can mean faster iteration on AI behaviors, easier reproducibility, and fewer hand-built tools around the edges. It’s a niche update, but a useful one for studios already investing in UE-based AI tooling or ML-driven gameplay systems.
- what
- AMD announced Schola v2.1, adding StateTree support, Kubernetes-oriented distributed training, and stronger Minari workflows.
- who
- AMD GPUOpen is the source of the announcement; the update targets Unreal Engine users.
- when
- The article announces version 2.1; no specific release date is given in the provided text.
- impact
- Helps gameplay and tools teams train and deploy AI behavior at scale with less custom integration work.
Useful infra update for UE AI and scaling workflows
Follow Unreal Engine updates
See relevant stories in your personalized news feed.
Discussion