{
  "id": 12060853,
  "title": "Gains and Collapse in On-Policy Distillation:A Reinforcement Learning Perspective",
  "url": "https://urgent.news/2026/10/02/gains-and-collapse-in-on-policy-distillation-a-reinforcement-learning",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-02T11:57:36.000Z",
  "source": {
    "name": "arXiv cs.AI",
    "slug": "arxiv-cs-ai",
    "url": "https://arxiv.org/abs/2610.03185v1"
  },
  "original_language": "en",
  "account": null,
  "summary": "On-policy distillation (OPD) has become an important approach to language model post-training. However, despite its performance gains, OPD can also collapse into excessively long and repetitive generation, and the mechanism underlying these divergent outcomes remains poorly understood. We explain these outcomes through a reinforcement learning perspective: the teacher implicitly rewards student…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}