Metraipolis › archive › dc_0ff71bb473cb

ARISE-RL: Agentic Rubric-Grounded Iterative Self-Evolution with Reinforcement Learning — arXiv:2609.01058

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.01058
License (as archived): CC0-1.0
Archived: 2026-10-17T06:32:21.401Z
SHA-256 of the archived text: a27833b450bbd4317a1da81f285d9566ec644a7f8a7b5758ac3b764d56d987d1
Stable URL of this page: https://metraipolis.com/documento/dc_0ff71bb473cb

How to cite

ARISE-RL: Agentic Rubric-Grounded Iterative Self-Evolution with Reinforcement Learning — arXiv:2609.01058 — arXiv (en), https://arxiv.org/abs/2609.01058 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0ff71bb473cb (Metraipolis archive dc_0ff71bb473cb, sha256 a27833b450bbd4317a1da81f285d9566ec644a7f8a7b5758ac3b764d56d987d1)

BibTeX

@misc{dc0ff71bb473cb,
  title = {ARISE-RL: Agentic Rubric-Grounded Iterative Self-Evolution with Reinforcement Learning — arXiv:2609.01058},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.01058},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0ff71bb473cb; sha256 a27833b450bbd4317a1da81f285d9566ec644a7f8a7b5758ac3b764d56d987d1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.01058. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Training open-ended agents via reinforcement learning (RL) is hindered by the lack of verifiable gold answers and scalable rubrics. Moreover, even near the model's capability boundary, long-horizon open-ended agentic tasks often yield brittle and unstable rewards, resulting in weak or noisy rollout contrast that obscures fine-grained optimization signals for group-based policy learning. To address these challenges, we propose ARISE-RL, a novel full-cycle self-evolution framework that couples a task/rubric Generator and a reasoning Solver through rubric-mediated co-evolution. The Generator grounds tool-related rubric criteria in real tool observations and is rewarded for producing valid, intermediate-difficulty tasks aligned with the Solver's evolving capability boundary. The Solver, in turn, learns from fine-grained rubric satisfaction signals through multi-step reasoning and tool use. We further introduce Reward-Gated Self-Evolution Distillation (RG-SED), which selectively distills a memory-augmented variant of the same policy back into itself only when the memory yields empirical reward improvement, thereby reducing distribution mismatch and avoiding blind imitation of noisy guidance. Finally, to support rigorous evaluation, we present ECR-Bench, an expert-calibrated rubric benchmark suite covering single-tool deep research and multi-tool travel planning. Extensive experiments demonstrate that ARISE-RL consistently achieves robust and stable overall state-of-the-art performance across all evaluated benchmarks.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-02 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.01058 — 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 2026-10-17T06:15:12.674Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).