Metraipolis › archive › dc_cc25385ece46
One Offset, Two Directions: Baseline Adjustment via Subgroup Offsets — arXiv:2611.00782
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.00782
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T10:31:29.986Z
- SHA-256 of the archived text
2ef71c856b9c09291780dd7248a4f7a749349ffe21bc53d9ce7cd753c651b2e8- Stable URL of this page
https://metraipolis.com/documento/dc_cc25385ece46
How to cite
One Offset, Two Directions: Baseline Adjustment via Subgroup Offsets — arXiv:2611.00782 — arXiv (en), https://arxiv.org/abs/2611.00782 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cc25385ece46 (Metraipolis archive dc_cc25385ece46, sha256 2ef71c856b9c09291780dd7248a4f7a749349ffe21bc53d9ce7cd753c651b2e8)
BibTeX
@misc{dccc25385ece46,
title = {One Offset, Two Directions: Baseline Adjustment via Subgroup Offsets — arXiv:2611.00782},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.00782},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cc25385ece46; sha256 2ef71c856b9c09291780dd7248a4f7a749349ffe21bc53d9ce7cd753c651b2e8},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.00782. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Group Relative Policy Optimization (GRPO) estimates advantages against the within-prompt mean reward, a baseline shared across the group. In practice, rollouts within a prompt are not always homogeneous, differing in their solution strategies and reasoning trajectories, along which they may group into subgroups. We further find that these heterogeneous subgroups may exhibit distinct reward distributions. The global baseline, however, collapses them onto a single credit axis, where the credit reflects neither a rollout's standing within its subgroup nor the subgroup's standing relative to the rest. We therefore study how to adjust the baseline when the rollouts of a prompt fall into heterogeneous subgroups. We define the \emph{subgroup offset} as the deviation of a subgroup mean from the global mean, and shift the baseline along it by a single signed coefficient $\lambda$. The sign of $\lambda$ yields two opposing directions: \emph{Alignment} ($\lambda{>}0$) adds the offset to the baseline, drawing credit toward a rollout's relative position within its subgroup; \emph{Contrast} ($\lambda{<}0$) subtracts it, drawing a rollout's credit toward the relative quality of the subgroup it belongs to. The adjustment acts on the baseline alone, leaving the reward signal and the optimization objective unchanged. In this work, we induce subgroups from strategies, grouping rollouts that share a strategy into the same subgroup. We evaluate both directions on Qwen3-8B-Base and Olmo3-7B-Base across four competition-level math benchmarks, against a $\lambda{=}0$ anchor that shares the same strategy-conditioned sampling and differs only in the baseline. Against this anchor each direction yields modest but consistent gains in multi-sample coverage, with Contrast the stronger of the two.
Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2611.00782 — 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 2026-11-03T10:29:48.477Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).