Interactive Training 2 introduces an open-source control plane for steering live model training through a shared protocol. Training applications declare exposed settings and actions; humans and automated controllers submit requests through the same interface, while the training loop validates and applies them at safe control points. A customized Aim workspace combines live metrics and controls with a chronological audit trail of requests and outcomes. The authors demonstrate the system across five natural-language-processing and reinforcement-learning workflows and release code and traces intended as a reusable foundation for auditable human- and agent-guided training.
No heat snapshots are available in the last 24 hours.