Verify Before You Distill: Prompt-Level Teacher Gating for On-Policy Distillation
A technical method for improving on-policy distillation by gating teacher prompts, with no direct safety or misuse implications.
A technical method for improving on-policy distillation by gating teacher prompts, with no direct safety or misuse implications.
Auto-scored: This is a narrow machine-learning methodology paper with no evidence of escalation, misuse, or governance impact.