{
  "id": 4407,
  "url": "https://arxiv.org/abs/2605.13130v1",
  "title": "GRACE: Gradient-aligned Reasoning Data Curation for Efficient Post-training",
  "summary": "Existing reasoning data curation pipelines score whole samples, treating every intermediate step as equally valuable. In reality, steps within a trace contribute very unevenly, and selecting reasoning data well requires assessing them individually. We present GRACE, a gradient-aligned curation method that views each reasoning trace as a sequence of optimization events and scores every step by two complementary signals: its alignment with the answer-oriented gradient direction, and its consistenc",
  "authors": "Junjie Li, Ziao Wang, NingXuan Ma, Jianghong Ma, Xiaofeng Zhang",
  "category": "research",
  "topics": "safety-alignment",
  "orgs": null,
  "regions": null,
  "published_at": "2026-05-13T07:55:39.000Z",
  "fetched_at": "2026-07-14T16:30:59.236Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/4407",
  "original_url": "https://arxiv.org/abs/2605.13130v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}