Metraipolis › archive › dc_d73cdae576e1

BODHI: Do LLMs Branch Out and Discover Heterogeneous Inferences? — arXiv:2608.02867

Source: arXiv (en)
Original: https://arxiv.org/abs/2608.02867
License (as archived): CC0-1.0
Archived: 2026-10-31T12:18:21.159Z
SHA-256 of the archived text: 050799f9a7bef9711b82445e8036478e952bccf1f169509e721fa24619d43eb1
Stable URL of this page: https://metraipolis.com/documento/dc_d73cdae576e1

How to cite

BODHI: Do LLMs Branch Out and Discover Heterogeneous Inferences? — arXiv:2608.02867 — arXiv (en), https://arxiv.org/abs/2608.02867 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d73cdae576e1 (Metraipolis archive dc_d73cdae576e1, sha256 050799f9a7bef9711b82445e8036478e952bccf1f169509e721fa24619d43eb1)

BibTeX

@misc{dcd73cdae576e1,
  title = {BODHI: Do LLMs Branch Out and Discover Heterogeneous Inferences? — arXiv:2608.02867},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.02867},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d73cdae576e1; sha256 050799f9a7bef9711b82445e8036478e952bccf1f169509e721fa24619d43eb1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.02867. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Although reinforcement learning with verifiable rewards (RLVR) has improved the performance of large language models (LLMs) across a variety of reasoning tasks, there is significant debate as to whether RLVR expands the reasoning capability boundary, or just improves sampling efficiency. In this paper, we investigate the nature of test-time exploration in RLVR-trained LLMs by employing controlled maze-solving experiments and extracting a tree structure from mathematical reasoning traces (BODHI-Trees) based on semantic equivalence. This helps us delineate between entropy arising from stylistic variations and genuine inferential branching. Our findings demonstrate that the policy entropy collapse observed in RLVR models is not merely syntactic, and is accompanied by a significant reduction in semantic branching entropy. While RLVR improves adherence to environmental constraints and backtracking capabilities, it constricts the space of continuations; we provide evidence suggesting that this might be responsible for the sample efficiency gains of RLVR, albeit at the cost of genuine rollout diversity.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-05 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.02867 — 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 2026-10-31T12:05:52.962Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-05?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).