Metraipolis › archive › dc_6b1d4db74df7
RLHarness: Co-evolving Procedural Skills with Reinforcement Learning for Long-horizon Multimodal Reasoning — arXiv:2609.32326
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.32326
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T20:55:18.887Z
- SHA-256 of the archived text
536fdb98846b3a3ee077d573152489ce53a28847f9d253039b96c83d69f7d413- Stable URL of this page
https://metraipolis.com/documento/dc_6b1d4db74df7
How to cite
RLHarness: Co-evolving Procedural Skills with Reinforcement Learning for Long-horizon Multimodal Reasoning — arXiv:2609.32326 — arXiv (en), https://arxiv.org/abs/2609.32326 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_6b1d4db74df7 (Metraipolis archive dc_6b1d4db74df7, sha256 536fdb98846b3a3ee077d573152489ce53a28847f9d253039b96c83d69f7d413)
BibTeX
@misc{dc6b1d4db74df7,
title = {RLHarness: Co-evolving Procedural Skills with Reinforcement Learning for Long-horizon Multimodal Reasoning — arXiv:2609.32326},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.32326},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_6b1d4db74df7; sha256 536fdb98846b3a3ee077d573152489ce53a28847f9d253039b96c83d69f7d413},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32326. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Multimodal reasoning requires models to preserve visual evidence through long decision chains while selecting appropriate procedures across diverse scenarios and rules. When learning is guided only by terminal verifiers, reinforcement learning (RL) reveals whether a final answer is correct but not how it should be produced. The policy must therefore discover reusable reasoning procedures while learning to execute them, creating a program cold-start problem. Skills can externalize successful procedures, reduce repeated exploration, and provide inspectable guidance. However, a fixed Skill Bank assumes that this guidance remains compatible with an evolving policy, while updating Skills alone can leave their triggers, execution protocols, and demonstrations stale or mutually inconsistent. We introduce RLHARNESS, which organizes Skills, selection and execution protocols, few-shot demonstrations, and task contracts into a unified, versioned Harness and alternates Harness evolution with policy learning. An Exploration-Distillation Harness builds the initial Harness and version-aligned verified traces for SFT and DAPO I. After the first RL block, a Post-RL Reconstruction Harness rebuilds Skills, protocols, and demonstrations from fresh success-failure rollouts, and DAPO II adapts the policy to the reconstructed program. RLHARNESS improves Accuracy from 16.25%/27.50% to 62.00%/50.00% on MetroMap/TravelMap and raises F1 score from 37.13%/45.50% to 65.81%/65.51% on Fee-VL/Cancel-VL. All four tasks achieve their best results only after reconstruction and DAPO II, showing that an evolving Harness complements RL by continually updating the external program that the policy learns to execute. 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/2609.32326 — 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 2026-10-07T20:29:00.754Z 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).