Metraipolis › archive › dc_d2472d00a8fe

JHiMA-WebRL: A Joint Hierarchical Multi-Agent Reinforcement Learning Framework for Long-Horizon Web Tasks — arXiv:2611.05263

Source: arXiv (en)
Original: https://arxiv.org/abs/2611.05263
License (as archived): CC0-1.0
Archived: 2026-11-04T18:35:29.384Z
SHA-256 of the archived text: c53c60b970bf23beed60189bb43d7f165cfde2d9d83d83650945e932945324b5
Stable URL of this page: https://metraipolis.com/documento/dc_d2472d00a8fe

How to cite

JHiMA-WebRL: A Joint Hierarchical Multi-Agent Reinforcement Learning Framework for Long-Horizon Web Tasks — arXiv:2611.05263 — arXiv (en), https://arxiv.org/abs/2611.05263 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d2472d00a8fe (Metraipolis archive dc_d2472d00a8fe, sha256 c53c60b970bf23beed60189bb43d7f165cfde2d9d83d83650945e932945324b5)

BibTeX

@misc{dcd2472d00a8fe,
  title = {JHiMA-WebRL: A Joint Hierarchical Multi-Agent Reinforcement Learning Framework for Long-Horizon Web Tasks — arXiv:2611.05263},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.05263},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d2472d00a8fe; sha256 c53c60b970bf23beed60189bb43d7f165cfde2d9d83d83650945e932945324b5},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.05263. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Multimodal web agents have shown promising capabilities for interacting with websites through graphical user interfaces (GUIs), yet realistic web tasks often require long-horizon, multi-step planning that remains challenging for monolithic policies. Such agents must jointly infer high-level plans, navigate through intermediate states, track progress over extended interaction histories, and execute visually grounded actions. Although prior work has explored modular multi-agent systems with a separate planner, the planner is typically not updated jointly with the actor. We investigate whether explicitly separating and jointly training planning and execution provides a stronger inductive bias for long-horizon web navigation. We introduce JHiMA-WebRL, a joint hierarchical multi-agent reinforcement learning framework for web agents that couples a task decomposer with an actor. To train the decomposer, we use a strong teacher model to construct structured subtask supervision from multimodal GUI trajectories, and use these annotations to fine-tune both the decomposer and the subtask-conditioned actor. We then extend Group Relative Policy Optimization (GRPO) to jointly optimize the decomposer and actor in the interactive WebArena environment through hierarchical rollout grouping and role-specific advantages. On WebArena-Lite-v2, JHiMA-WebRL-14B, consisting of a 7B actor and a 7B decomposer, achieves an average success rate of 37.2%, compared with 24% for a ScaleCUA-32B agent that we fine-tune with RL on the same training tasks and evaluate under the same protocol. Ablations show that jointly updating the decomposer during RL outperforms updating the actor alone. These results suggest that jointly learned role factorization with smaller models can be a competitive alternative to scaling monolithic agents for long-horizon web tasks.

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/2611.05263 — 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 2026-11-04T18:32:37.360Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).