R2-Write: Reflection and Revision for Open-Ended Writing with Deep Reasoning
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913001583411200 |
|---|---|
| author | Liu, Wanlong Zhang, Bo Li, Chenliang Lai, Shaopeng Wu, Yuning Lei, Xuanyu Yan, Ming |
| author_facet | Liu, Wanlong Zhang, Bo Li, Chenliang Lai, Shaopeng Wu, Yuning Lei, Xuanyu Yan, Ming |
| contents | While deep reasoning with long chain-of-thought has dramatically improved large language models in verifiable domains like mathematics, its effectiveness for open-ended tasks such as writing remains unexplored. In this paper, we conduct a systematic investigation revealing that existing mainstream reasoning models achieve limited gains on open-ended writing tasks. Our further analysis shows that these models lack deep reflection and revision patterns in open-ended writing, resulting in substantially smaller improvements compared to mathematical reasoning tasks. To address this limitation, we introduce R2-Write: an automated framework that synthesizes high-quality thinking trajectories enriched with explicit reflection and revision patterns through iterative writer-judge interaction. To prevent redundant reflections, we design a process reward mechanism that supervises reflection quality during reinforcement learning, improving both performance and token efficiency. Extensive experiments across multiple creative writing and deep-research benchmarks demonstrate significant improvements, validating that explicitly incorporating reflection and revision patterns unlocks deep reasoning capabilities for open-ended writing tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_03004 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | R2-Write: Reflection and Revision for Open-Ended Writing with Deep Reasoning Liu, Wanlong Zhang, Bo Li, Chenliang Lai, Shaopeng Wu, Yuning Lei, Xuanyu Yan, Ming Computation and Language Artificial Intelligence While deep reasoning with long chain-of-thought has dramatically improved large language models in verifiable domains like mathematics, its effectiveness for open-ended tasks such as writing remains unexplored. In this paper, we conduct a systematic investigation revealing that existing mainstream reasoning models achieve limited gains on open-ended writing tasks. Our further analysis shows that these models lack deep reflection and revision patterns in open-ended writing, resulting in substantially smaller improvements compared to mathematical reasoning tasks. To address this limitation, we introduce R2-Write: an automated framework that synthesizes high-quality thinking trajectories enriched with explicit reflection and revision patterns through iterative writer-judge interaction. To prevent redundant reflections, we design a process reward mechanism that supervises reflection quality during reinforcement learning, improving both performance and token efficiency. Extensive experiments across multiple creative writing and deep-research benchmarks demonstrate significant improvements, validating that explicitly incorporating reflection and revision patterns unlocks deep reasoning capabilities for open-ended writing tasks. |
| title | R2-Write: Reflection and Revision for Open-Ended Writing with Deep Reasoning |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2604.03004 |