מנהל טירונות AI: שליטה סגורה סביב תהליך הטמעה

תקציר מקורי באנגליתarXiv:2606.29871v3 Announce Type: replace Abstract: We present the AI Training Manager, a bounded LLM-based metacognitive monitoring-and-control layer for machine-learning training. The Manager asynchronously observes structured telemetry from an active training run, assesses the current training regime, and selects adaptive interventions through a constrained, deterministically verified action interface. We evaluate the feasibility of this architecture using deliberately induced, interpretable failure modes in supervised and reinforcement learning tasks. For supervised learning, we induce a multi-objective overfitting failure in a GPT-2-style model trained on TinyStories. The Manager prevents the resulting late-run validation collapse, reducing final validation loss by 54.3% relative to t