Metraipolis › archive › dc_2bdead408e67

Incentivizing Self-Learning in Large Language Models via Reinforcement Learning — arXiv:2610.39671

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.39671
License (as archived): CC0-1.0
Archived: 2026-10-31T06:17:01.830Z
SHA-256 of the archived text: 4bc8f42fe50abb20eb4b3478fc8f57c2b50ecfdcf34d853973dfd8848924dba0
Stable URL of this page: https://metraipolis.com/documento/dc_2bdead408e67

How to cite

Incentivizing Self-Learning in Large Language Models via Reinforcement Learning — arXiv:2610.39671 — arXiv (en), https://arxiv.org/abs/2610.39671 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2bdead408e67 (Metraipolis archive dc_2bdead408e67, sha256 4bc8f42fe50abb20eb4b3478fc8f57c2b50ecfdcf34d853973dfd8848924dba0)

BibTeX

@misc{dc2bdead408e67,
  title = {Incentivizing Self-Learning in Large Language Models via Reinforcement Learning — arXiv:2610.39671},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.39671},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2bdead408e67; sha256 4bc8f42fe50abb20eb4b3478fc8f57c2b50ecfdcf34d853973dfd8848924dba0},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.39671. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models (LLMs) draw on parametric knowledge and retrieval-augmented generation (RAG) to solve knowledge-intensive tasks. Yet problem solving in unfamiliar domains calls for tightly coupled knowledge acquisition and reasoning, beyond factual retrieval and aggregation or learning solely from a preassembled context. We study self-learning -- autonomously retrieving, studying, and applying knowledge from long-form reference materials on the fly. To investigate this capability, we formulate a textbook-based mathematical problem-solving task. LLMs must navigate textbook exposition, learn and apply relevant principles, and construct proofs for exercises, with reference solutions withheld. We introduce a BookNavigator interface to support this human-inspired study process and incentivize self-learning through reinforcement learning (RL) with fine-grained rewards. Experiments show that our method improves performance across in-domain and out-of-domain textbooks and subjects, raising Intern-S2-Preview-35B's overall proof accuracy from 51.5% to 79.5% and outperforming RL-trained textbook-free and fixed-retrieval baselines. These gains extend to complex long-context benchmarks involving different task types, with improvements over the base model of 1.7 and 2.3 percentage points on BrowseComp-Plus and CL-bench, respectively. Together, these findings suggest that RL can foster a domain-agnostic self-learning strategy, highlighting the potential of self-learning from long-form corpora to strengthen broader problem-solving capabilities on complex, long-context tasks.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2610.39671 — 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 2026-10-31T06:07:23.361Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).