On-policy distillation is quietly eating GRPO's lunch — and DPO's tax bill just got itemized

This week's post-training evidence favors distilling from a strong checkpoint over running fresh RL, and DPO's alignment tax turns out to be task-selective.