Metraipolis › archive › dc_cfa42a219e3e
Not All Reasoning Generalizes: A Case Study on Learning Generalizable Reasoning Strategies in RLVR — arXiv:2611.03627
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.03627
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T11:23:25.604Z
- SHA-256 of the archived text
0b09999584e1f54c79819c1296c15e37ad9e8736113c9242662860fa02a1e97d- Stable URL of this page
https://metraipolis.com/documento/dc_cfa42a219e3e
How to cite
Not All Reasoning Generalizes: A Case Study on Learning Generalizable Reasoning Strategies in RLVR — arXiv:2611.03627 — arXiv (en), https://arxiv.org/abs/2611.03627 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cfa42a219e3e (Metraipolis archive dc_cfa42a219e3e, sha256 0b09999584e1f54c79819c1296c15e37ad9e8736113c9242662860fa02a1e97d)
BibTeX
@misc{dccfa42a219e3e,
title = {Not All Reasoning Generalizes: A Case Study on Learning Generalizable Reasoning Strategies in RLVR — arXiv:2611.03627},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.03627},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cfa42a219e3e; sha256 0b09999584e1f54c79819c1296c15e37ad9e8736113c9242662860fa02a1e97d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.03627. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Reinforcement learning with verifiable rewards (RLVR) can enable generalizable improvement for large language model (LLM) reasoning. This paper studies how to incentivize such generalization via reasoning strategies. We study this question with a controllable equation-solving framework: finding multiple inputs x for equations f(x)=y, where f is a non-bijective string function. This setting makes reasoning strategies observable while supporting a clear definition of easy-to-hard out-of-distribution (OOD) shifts, e.g., more solutions, longer y, etc. We uncover a subtle RLVR failure: it teaches the model to solve in-distribution tasks with high accuracy and structured step-by-step derivation, yet under OOD shifts, the same model abandons derivation and collapses to ineffective brute-force enumeration, causing sharp accuracy drops. This observation reveals the deeper challenges of RLVR generalization as it is neither incapable base model nor reward hacking, since the model succeeds in-distribution with the correct reasoning strategy. The key to mitigating this failure is an indirect intervention that surprisingly helps: co-training with bijective inversion tasks. Although these bijective tasks are actually not uniformly harder and even further from the non-bijective OOD evaluation, they make derivation more certain and substantially restore the correct strategy and accuracy under OOD shifts. This effect further motivates our theoretical analysis: outcome-only RLVR can shape reasoning processes through implicitly crediting the effective reasoning strategies more, even without process supervision. To conclude, our analysis framework and findings systematically reveal the crucial role and incentivization of reasoning strategies in RLVR generalization under OOD shifts. Code is available at this https URL . Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2611.03627 — 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 2026-11-03T11:02:37.732Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).