Metraipolis › archive › dc_2472add04032

PB-GRPO: Learning Socially Adaptive LLM Agents from Persona-Driven Simulation with Preference-Batched GRPO — arXiv:2610.04132

Source
arXiv
Original
https://arxiv.org/abs/2610.04132
License (as archived)
CC0-1.0
Archived
2026-10-06T05:25:59.941Z
SHA-256 of the archived text
c0543d642a4ec58f025dba15a21f112720d1711f97ca256edf6453cf8e178ee4
Stable URL of this page
https://metraipolis.com/documento/dc_2472add04032

How to cite

PB-GRPO: Learning Socially Adaptive LLM Agents from Persona-Driven Simulation with Preference-Batched GRPO — arXiv:2610.04132 — arXiv, https://arxiv.org/abs/2610.04132 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2472add04032 (Metraipolis archive dc_2472add04032, sha256 c0543d642a4ec58f025dba15a21f112720d1711f97ca256edf6453cf8e178ee4)
BibTeX
@misc{dc2472add04032,
  title = {PB-GRPO: Learning Socially Adaptive LLM Agents from Persona-Driven Simulation with Preference-Batched GRPO — arXiv:2610.04132},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.04132},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2472add04032; sha256 c0543d642a4ec58f025dba15a21f112720d1711f97ca256edf6453cf8e178ee4},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.04132. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Building LLMs that behave well socially, not merely correctly, requires Building LLMs that behave well socially, not merely correctly, requires more than producing locally helpful responses. A socially competent agent must infer users' unstated goals, respect their preferences, and adapt as the conversation unfolds. These behaviors are inherently multi-turn and social, making them hard to optimize: real interaction data is scarce, and user preferences are typically latent rather than directly observable. To address these challenges, we build on a persona-driven social simulation environment (consisting of a persona library, LLM-based user simulators, and a user-satisfaction scoring system ranging from [0, 1]), to introduce preference-batched GRPO (PB-GRPO), a post-training algorithm that learns socially adaptive policies from conversation-level feedback. Compared to vanilla GRPO, PB-GRPO computes advantages using a normalization estimated across a bucket of users with similar preferences, stabilizing training across a diverse social population. Empirical evidence shows that PB-GRPO improves models' social behavior over strong reinforcement learning baselines in our simulated environment.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.04132 — 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 2026-10-06T05:24:05.795Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).