{
  "id": 11346473,
  "title": "Finetuning with Sampling: SFT Learns Better Than You Think",
  "url": "https://urgent.news/2026/10/01/finetuning-with-sampling-sft-learns-better-than-you-think",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-01T17:45:07.000Z",
  "source": {
    "name": "arXiv cs.AI",
    "slug": "arxiv-cs-ai",
    "url": "https://arxiv.org/abs/2610.02140v1"
  },
  "original_language": "en",
  "account": null,
  "summary": "Introducing new capabilities to frontier models has long been the goal of posttraining, which predominantly employs supervised finetuning (SFT) and reinforcement learning (RL) to this end. Conventional wisdom dictates that RL enables strong generalization on new tasks without losing existing capabilities, while SFT is prone to weak generalization and catastrophic forgetting. At the same time, SFT…",
  "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."
}