Metraipolis › archive › dc_7eb5cbe11829
What Drives the Zero-to-One Jump? Diagnosing and Steering Hard-Prompt Learning in RLVR — arXiv:2611.02747
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.02747
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T10:37:53.483Z
- SHA-256 of the archived text
34f36315434c1b4589026bd020b3e5505651c8bb825b4b608350c0732ada96a4- Stable URL of this page
https://metraipolis.com/documento/dc_7eb5cbe11829
How to cite
What Drives the Zero-to-One Jump? Diagnosing and Steering Hard-Prompt Learning in RLVR — arXiv:2611.02747 — arXiv (en), https://arxiv.org/abs/2611.02747 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7eb5cbe11829 (Metraipolis archive dc_7eb5cbe11829, sha256 34f36315434c1b4589026bd020b3e5505651c8bb825b4b608350c0732ada96a4)
BibTeX
@misc{dc7eb5cbe11829,
title = {What Drives the Zero-to-One Jump? Diagnosing and Steering Hard-Prompt Learning in RLVR — arXiv:2611.02747},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.02747},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7eb5cbe11829; sha256 34f36315434c1b4589026bd020b3e5505651c8bb825b4b608350c0732ada96a4},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.02747. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Reinforcement Learning with Verifiable Rewards (RLVR) has substantially improved mathematical reasoning, yet progress on hard prompts often stalls: a prompt can remain unsolved for many training steps before abruptly transitioning to solved--a discrete "zero-to-one" jump whose underlying dynamics and steerability remain poorly understood. We investigate two questions: what drives this transition, and can it be steered? Through event-based analysis measuring the time-to-first-positive (TTFP), we find that negative rollouts preceding successful transitions consistently retain higher early-stage generation entropy, distinguishing near-solvable failures from dead ends. This signal motivates NAPO (Negative-Augmented Policy Optimization), a method integrated with Group Relative Policy Optimization (GRPO) that ranks negatives in unsolved groups via an early-branching driver and assigns graded advantages, injecting informative gradients precisely where standard group-relative updates vanish. Under fixed rollout budgets, NAPO accelerates early solution discovery and improves Pass@256 by 10.0 points on AIME24 and 6.6 points on AIME2025, demonstrating that zero-to-one transitions are diagnosable and steerable through driver-guided negative learning. Our code is available at this https URL . Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2611.02747 — 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 2026-11-03T10:29:48.893Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).