{
  "title": "Scaling Laws for Generative Reward Models",
  "authors": ["Angel Raychev", "Nathan Lile", "Rafael Rafailov"],
  "publication_date": "2026-10-21",
  "updated_date": "2026-10-23",
  "status": "arxiv-preprint",
  "arxiv_id": "2610.30765",
  "arxiv_version": "v1",
  "arxiv_url": "https://arxiv.org/abs/2610.30765",
  "arxiv_publication_date": "2026-10-22",
  "arxiv_pdf_sha256": "c442b081068e134ec72c9b62ef58c280e9676e99e7b6070788fd051cc42bb4a7",
  "revision": "arxiv-v1-public-analysis-2026-10-23",
  "pages": 28,
  "pdf": "scaling-laws-for-generative-reward-models.pdf",
  "sha256": "43aabe39876750e05f155592eac00652b724c5692d9897537d11f5621a4cd446",
  "source_repository": "https://github.com/RaychevAngel/generative-reward-models",
  "source_tag": "v2026.10.23-analysis.1",
  "source_commit": "03fbe84695610c65257058095bde5a56a3dfd185",
  "public_package_scope": "Analysis code, derived numerical results, figures, prompt templates and manuscript artifacts. Custom training implementation is maintained privately and is not included.",
  "original_manuscript_release": {
    "tag": "v2026.10.21.1",
    "source_commit": "0a5797a5bcbb35ee1912f84c612086b0ee6fbaef",
    "training_source_revision": "b13b2057c6c87ba754b1271513504bc985e034d8",
    "arxiv_source_sha256": "753c72c376ca83844f07577ad5782d6ebe515b0ed679ca330612409035a83bde",
    "availability": "Historical source and release provenance retained in the private archive; not a tag or training-code release in the public analysis repository."
  },
  "reader": "reader.html",
  "pdfjs_version": "5.6.205",
  "pdfjs_license": "Apache-2.0"
}