Metraipolis › archive › dc_2f8e2d47a2c8

Planned Test-Time Scaling with Coordinated Reasoning Paths — arXiv:2609.27374

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.27374
License (as archived)
CC0-1.0
Archived
2026-10-08T09:12:32.530Z
SHA-256 of the archived text
31fe2840f31ade2ababe436092a547263fdd40128015560110a2d12376c373aa
Stable URL of this page
https://metraipolis.com/documento/dc_2f8e2d47a2c8

How to cite

Planned Test-Time Scaling with Coordinated Reasoning Paths — arXiv:2609.27374 — arXiv (en), https://arxiv.org/abs/2609.27374 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2f8e2d47a2c8 (Metraipolis archive dc_2f8e2d47a2c8, sha256 31fe2840f31ade2ababe436092a547263fdd40128015560110a2d12376c373aa)
BibTeX
@misc{dc2f8e2d47a2c8,
  title = {Planned Test-Time Scaling with Coordinated Reasoning Paths — arXiv:2609.27374},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.27374},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2f8e2d47a2c8; sha256 31fe2840f31ade2ababe436092a547263fdd40128015560110a2d12376c373aa},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.27374. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Test-time scaling with parallel branches is widely adopted to improve performance on challenging reasoning tasks. The predominant approach, repeated sampling, draws branches independently from a single policy, which can produce redundant attempts and thereby limit the gains from additional inference compute. To address this limitation, we propose Planned Test-Time Scaling (PTTS), which replaces independent sampling with a coordinated joint policy: a planner generates a solution outline for each branch, steering the branches toward distinct reasoning paths, and an executor produces a full solution conditioned on each outline. Formally, we show that PTTS strictly generalizes repeated sampling and, in a stylized setting, provably promotes coverage of complementary reasoning modes and yields better pass@k scaling. We instantiate PTTS on top of strong reasoning models, keeping them fixed as executors while replacing repeated sampling with PTTS inference to further enhance test-time scaling. Concretely, we develop two variants: PTTS-ZS prompts a model to jointly generate outlines for all branches in a single autoregressive pass, while PTTS-RL directly optimizes the planner against the pass@k reward using truncated execution rollouts for efficient training and a sharper reward signal. Across five mathematical reasoning benchmarks with Qwen3-1.7B and 4B, PTTS-ZS improves pass@64 over repeated sampling by up to 6.7 points, while PTTS-RL further increases the gain to up to 13.4 points. Further analysis indicates that broader coverage of distinct reasoning paths contributes to these gains. Overall, PTTS provides a general framework for improving test-time scaling by coordinating reasoning branches, with zero-shot and trainable instantiations that yield substantial performance gains.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-24 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.27374 — 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 2026-10-08T08:13:26.664Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).