Reinforcement Learning Training Pauses

Deliberate suspension or throttling of large reinforcement-learning runs on deployment-bound models pending security and alignment evidence.