Metraipolis › archive › dc_6d75dbb32d63

FORGE: Form-Optimal Routing of Grounded Evidence for Frozen LLM Agents — arXiv:2609.34358

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.34358
License (as archived)
CC0-1.0
Archived
2026-10-07T21:54:12.218Z
SHA-256 of the archived text
ebbf93ba029968624e282431cb4dd14e91120c2ac0d802c98d4441f7dda218d5
Stable URL of this page
https://metraipolis.com/documento/dc_6d75dbb32d63

How to cite

FORGE: Form-Optimal Routing of Grounded Evidence for Frozen LLM Agents — arXiv:2609.34358 — arXiv (en), https://arxiv.org/abs/2609.34358 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_6d75dbb32d63 (Metraipolis archive dc_6d75dbb32d63, sha256 ebbf93ba029968624e282431cb4dd14e91120c2ac0d802c98d4441f7dda218d5)
BibTeX
@misc{dc6d75dbb32d63,
  title = {FORGE: Form-Optimal Routing of Grounded Evidence for Frozen LLM Agents — arXiv:2609.34358},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.34358},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_6d75dbb32d63; sha256 ebbf93ba029968624e282431cb4dd14e91120c2ac0d802c98d4441f7dda218d5},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34358. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

In agentic AI systems, frozen foundation models are increasingly deployed as closed-weight API endpoints, making downstream adaptation possible only through the inputs and inference procedures surrounding the model. As a result, for each input query, two coupled decisions largely determine both answer quality and token cost: what evidence to provide and how much reasoning budget to allocate. Fixed defaults along these axes are often suboptimal, misallocating support form or reasoning depth on roughly 80% of queries in our analysis. To address this challenge, we propose FORGE, a unified framework for adapting frozen models through per-query routing over a joint action space that spans both support form and thinking depth. Under an entropy-regularized, cost-aware utility objective, we derive a closed-form Boltzmann routing target and instantiate the policy as a lightweight 269K-parameter factorized router. The routing policy is trained around the frozen host, without any weight access, through a three-stage pipeline: offline arm enumeration, supervised Kullback-Leibler (KL) distillation from the Boltzmann target, and Group Relative Policy Optimization (GRPO) refinement with host feedback. Across 5 knowledge-intensive benchmarks and 8 frozen backbones ranging from 7B to 671B parameters, FORGE improves accuracy at 42-45% lower token cost on both main hosts, transfers zero-shot across hosts at lower token cost, and composes with intrinsic thinking budgets where available.

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/2609.34358 — 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 2026-10-07T21:27:08.810Z 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).