Metraipolis › archive › dc_a949ecbcc852

When LLM Meets Tree Search: A Systematic View of Inference as Search in Large Language Models — arXiv:2608.30395

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.30395
License (as archived)
CC0-1.0
Archived
2026-10-17T08:07:31.362Z
SHA-256 of the archived text
840bda40d081a73a03a8fc2e3afe866456fa50e2e2cc3af54c83d84b98ac4355
Stable URL of this page
https://metraipolis.com/documento/dc_a949ecbcc852

How to cite

When LLM Meets Tree Search: A Systematic View of Inference as Search in Large Language Models — arXiv:2608.30395 — arXiv (en), https://arxiv.org/abs/2608.30395 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a949ecbcc852 (Metraipolis archive dc_a949ecbcc852, sha256 840bda40d081a73a03a8fc2e3afe866456fa50e2e2cc3af54c83d84b98ac4355)
BibTeX
@misc{dca949ecbcc852,
  title = {When LLM Meets Tree Search: A Systematic View of Inference as Search in Large Language Models — arXiv:2608.30395},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.30395},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a949ecbcc852; sha256 840bda40d081a73a03a8fc2e3afe866456fa50e2e2cc3af54c83d84b98ac4355},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.30395. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

As pretraining scaling laws approach saturation, Test-Time Scaling (TTS) has emerged as an important direction for improving reasoning by allocating inference-time compute to a fixed model prior. Viewed at a high level, TTS reframes inference as search over a space of partial reasoning states. While Chain-of-Thought (CoT) exposes intermediate steps, common instantiations rely on single-trajectory decoding, limiting recovery from early errors and exploration. This survey systematizes recent progress in tree-search-based reasoning, viewing inference as instance-specific optimization rather than decoding. We trace the evolution from uninformed search to Monte Carlo Tree Search (MCTS), highlighting how sampling-based control supports principled exploration-exploitation trade-offs. To unify a fragmented literature, we introduce a Unified Design Space spanning search topology, evaluation signals, and control dynamics, and advocate a standardized compute-reporting abstraction to make compute-accuracy trade-offs explicit and comparable.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.30395 — 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 2026-10-17T07:47:52.809Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).