Metraipolis › archive › dc_768a64bacc40
Using Context Is Not Enough: Test-Time Training for Personalized Reward Modeling — arXiv:2609.35109
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.35109
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T20:58:30.777Z
- SHA-256 of the archived text
856c2023bf01b473faaa93fd586bdc5c60ec8f5cab9d938bcc77aafe93526c6d- Stable URL of this page
https://metraipolis.com/documento/dc_768a64bacc40
How to cite
Using Context Is Not Enough: Test-Time Training for Personalized Reward Modeling — arXiv:2609.35109 — arXiv (en), https://arxiv.org/abs/2609.35109 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_768a64bacc40 (Metraipolis archive dc_768a64bacc40, sha256 856c2023bf01b473faaa93fd586bdc5c60ec8f5cab9d938bcc77aafe93526c6d)
BibTeX
@misc{dc768a64bacc40,
title = {Using Context Is Not Enough: Test-Time Training for Personalized Reward Modeling — arXiv:2609.35109},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.35109},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_768a64bacc40; sha256 856c2023bf01b473faaa93fd586bdc5c60ec8f5cab9d938bcc77aafe93526c6d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.35109. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Reinforcement learning from human feedback (RLHF) aligns large language models (LLMs) with human preferences, yet most pipelines learn a single reward model that overlooks individual differences in preferences. Personalized reward models (PRMs) address this by conditioning rewards on user-specific feedback, most commonly through in-context learning (ICL), where a user's historical comparisons are supplied as contextual preference pairs. However, we identify a key limitation of ICL-based PRMs: they fail to capture the preference relations conveyed by contextual pairs. To address this, we propose Preference-Aligned Test-Time Training (P-TTT), which explicitly encodes these relations into user-specific fast weights for personalized reward prediction. P-TTT introduces sequence-level update and apply operations to match the response-level granularity of preference feedback, together with a preference-aligned objective that directly uses pairwise preference relations to guide fast-weight adaptation. Notably, P-TTT is simple to implement and computationally efficient, updating fast weights within a single forward pass without inference-time backpropagation. Extensive experiments show that P-TTT more effectively captures historical preference relations and outperforms state-of-the-art methods by a large margin. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.35109 — 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 2026-10-07T20:29:01.780Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).