Metraipolis › archive › dc_265dfbf811a6
LLMs as Adaptive Meta-Solvers: Strategy-Diverse RL for Industrial-Scale Optimization — arXiv:2609.34427
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.34427
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T21:54:12.723Z
- SHA-256 of the archived text
22c1343bc54cddacd0ae36b2720d3de1835b23fd3df8fa570ea7d0c35c589bc8- Stable URL of this page
https://metraipolis.com/documento/dc_265dfbf811a6
How to cite
LLMs as Adaptive Meta-Solvers: Strategy-Diverse RL for Industrial-Scale Optimization — arXiv:2609.34427 — arXiv (en), https://arxiv.org/abs/2609.34427 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_265dfbf811a6 (Metraipolis archive dc_265dfbf811a6, sha256 22c1343bc54cddacd0ae36b2720d3de1835b23fd3df8fa570ea7d0c35c589bc8)
BibTeX
@misc{dc265dfbf811a6,
title = {LLMs as Adaptive Meta-Solvers: Strategy-Diverse RL for Industrial-Scale Optimization — arXiv:2609.34427},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.34427},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_265dfbf811a6; sha256 22c1343bc54cddacd0ae36b2720d3de1835b23fd3df8fa570ea7d0c35c589bc8},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34427. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Scaling LLM-based optimization from textbook-scale instances to real-world, industrial tasks remains a critical open challenge. Existing approaches are predominantly evaluated on small, self-contained textual problems and often commit to a solver-integrated paradigm, limiting their ability to handle the scale and structural diversity of practical optimization workloads. In this work, we propose a practical framework for training open-source LLMs to tackle real-world, industrial-scale optimization. We first show empirically that solver-integrated reasoning, exact combinatorial algorithm, and heuristic search exhibit complementary strengths across different problem structures and scales. Motivated by this, we introduce Strategy-Diverse Reinforcement Learning (SDRL), which trains LLMs as adaptive optimization meta-solvers. SDRL leverages this complementarity through a correctness-gated hierarchical diversity reward that promotes robust exploration across varying strategies and within each strategy, effectively preventing premature strategy collapse. We further introduce a mixed-format training scheme that jointly supports both self-contained textual problems and file-grounded instances. Across comprehensive evaluations, our framework outperforms existing fine-tuned methods and frontier models including DeepSeek-V4-Pro and GPT-5.5, both on average across benchmarks and on industrial-scale optimization tasks. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.CL). Categorie arXiv: Machine Learning (cs.LG) ; Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2609.34427 — 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 2026-10-07T21:27:08.812Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).