Metraipolis › archive › dc_61b65fbbcd95
STAGE: Controlled Objective Admission for Multi-Preference LLM Alignment — arXiv:2608.16553
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.16553
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T18:59:08.015Z
- SHA-256 of the archived text
5865c41ff3b3515ce1f4b2d8cdc8282ca3274c9be14664ffcfa2a4ed4d12ce48- Stable URL of this page
https://metraipolis.com/documento/dc_61b65fbbcd95
How to cite
STAGE: Controlled Objective Admission for Multi-Preference LLM Alignment — arXiv:2608.16553 — arXiv (en), https://arxiv.org/abs/2608.16553 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_61b65fbbcd95 (Metraipolis archive dc_61b65fbbcd95, sha256 5865c41ff3b3515ce1f4b2d8cdc8282ca3274c9be14664ffcfa2a4ed4d12ce48)
BibTeX
@misc{dc61b65fbbcd95,
title = {STAGE: Controlled Objective Admission for Multi-Preference LLM Alignment — arXiv:2608.16553},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.16553},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_61b65fbbcd95; sha256 5865c41ff3b3515ce1f4b2d8cdc8282ca3274c9be14664ffcfa2a4ed4d12ce48},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.16553. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Multi-preference alignment is often framed as scalarization: combine reward dimensions, then optimize. This leaves a temporal decision underspecified: when should each preference dimension enter policy optimization? We propose \methodname, a stability-guided active-set controller for controlled objective admission. \methodname starts from a small active set, retains admitted objectives, and expands when reward-deviation gates indicate low recent deviation or a patience budget is exhausted. A probing phase estimates a hard-to-easy order, and adaptive weighting emphasizes underperforming active dimensions. Automatic evaluations with 15 training preferences and 16 held-out benchmark columns show that \methodname obtains higher averages than simultaneous scalarization and shared-budget adapted baselines. Component ablations and expansion dynamics further support cumulative retention, gated admission, and probing-derived ordering as useful design choices in this setting. These results position objective-entry timing as a concrete control variable in reward-vector RLHF. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-18 in cs.CL). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2608.16553 — 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 2026-10-18T18:06:57.897Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).