Metraipolis › archive › dc_f07528a3562f

Rubrics as Privileged Information for Open-Ended Generation — arXiv:2608.02948

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.02948
License (as archived)
CC0-1.0
Archived
2026-10-31T12:18:22.668Z
SHA-256 of the archived text
e00e43ea9f76c4cdd79c01ffdf0113d46d53277a7348ad8d81572fcc06ec1c82
Stable URL of this page
https://metraipolis.com/documento/dc_f07528a3562f

How to cite

Rubrics as Privileged Information for Open-Ended Generation — arXiv:2608.02948 — arXiv (en), https://arxiv.org/abs/2608.02948 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f07528a3562f (Metraipolis archive dc_f07528a3562f, sha256 e00e43ea9f76c4cdd79c01ffdf0113d46d53277a7348ad8d81572fcc06ec1c82)
BibTeX
@misc{dcf07528a3562f,
  title = {Rubrics as Privileged Information for Open-Ended Generation — arXiv:2608.02948},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.02948},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f07528a3562f; sha256 e00e43ea9f76c4cdd79c01ffdf0113d46d53277a7348ad8d81572fcc06ec1c82},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.02948. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

On-policy self-distillation (OPSD), where a single model acts as both student and teacher with different contexts, has shown promise in verifiable domains like math, where hard privileged information (PI) in the form of ground-truth answers structurally constrains valid continuations. We extend OPSD to open-ended generation using soft PI in the form of rubrics that guide preferences but admit many valid responses. Rubrics have served as scalar rewards for reinforcement learning (RL); we show that they provide substantially richer signal as dense PI for distillation, and contrary to intuition, soft rubric PI provides a larger and more effective training signal on student roll-outs than hard reference completion PI in this regime. A reference completion is one point in a set of valid responses, so distilling towards it over-constrains the student, while rubrics specify the preference structure shared across the set of valid responses. We show the effectiveness of using rubrics as PI for open-ended generation across Qwen and Llama model families and show that it outperforms rubric-as-reward (RaR) RL using HealthBench, a benchmark that grades open-ended health responses against physician-created rubrics, providing dense token-level supervision for open-ended tasks; RuPI beats RaR RL by up to +0.10 absolute score and, under matched recipe and KL direction, beats reference-PI by +0.034 to +0.079 absolute score across three models. We further show that these findings generalize to training on the RubricHub Science corpus and evaluating on ResearchQA: soft rubric PI outperforms both reference-PI distillation and RaR RL (66.6% vs. 64.2% and 57.6%).

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-05 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.02948 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-31T12:05:52.965Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-05?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).