跳到正文
arXiv cs.CL· Jinlong Liu, Mohammed Bahja, Mark Lee·· 3 天前AI 评分27

推理监督反而有害:基于 TTCW 的长篇文学评论生成研究

When Reasoning Supervision Hurts: TTCW-Based Long-Form Literary Review Generation

AI 导读

研究构建了含 263,911 篇长篇故事、覆盖 14 个 TTCW 维度的数据集,测试推理轨迹能否提升结构化文学评论生成。在 Qwen3-4B、Qwen3-8B 和 Nemotron-8B 上,无轨迹监督的分数模仿保真度更高且报告生成近乎完整,而轨迹监督使完成率降至 85.9%–90.9%。失败分析显示 88.6% 的失败推理生成在轨迹内出现语义重复循环。

正文 · 原文

View PDF HTML (experimental)

Abstract:Automatic evaluation of long-form literary writing remains challenging, particularly for creativity-related dimensions such as originality and flexibility. We address the lack of large-scale supervision for structured Torrance Test of Creative Writing (TTCW)-based evaluation by constructing a dataset of 263{,}911 long-form stories, each paired with scalar scores and meta-synthesised review comments across 14 TTCW-based dimensions. Using this dataset, we study whether reviewer-style reasoning traces improve structured literary review generation across Qwen3-4B, Qwen3-8B, and Nemotron-8B. Across all three models, no-trace supervision yields higher score imitation fidelity and near-complete report generation, whereas trace supervision reduces completion to $85.9$--$90.9\%$. Failure analysis shows that $88.6\%$ of failed reasoning generations contain semantic repetition loops inside the trace. These loops concentrate in particular story--criterion interactions and progressively harden after onset, while comparable repetition is rare in the supervision data. Our results suggest that, for long-form tasks requiring strictly structured outputs, explicit reasoning traces can introduce generation instability that outweighs their overall benefit.
Comments: Submit to NAACL 2027
Subjects: Computation and Language (cs.CL)
Cite as: arXiv:2605.20364 [cs.CL]
  (or arXiv:2605.20364v2 [cs.CL] for this version)
  https://doi.org/10.48550/arXiv.2605.20364

arXiv-issued DOI via DataCite

Submission history

From: Jinlong Liu [view email]
[v1] Tue, 19 May 2026 18:16:58 UTC (222 KB)
[v2] Wed, 14 Oct 2026 11:05:55 UTC (321 KB)

来源:arXiv cs.CL · arxiv.org