arXiv cs.CL· Zaynab Merimi, Noussayba El Marrakchi, Arthur Satouf·· 13 小时前AI 评分31
扩散语言模型在表格到文本生成中的忠实性诊断研究
A Diagnostic Study of Faithfulness in Diffusion Language Models for Table-to-Text Generation
AI 导读
研究在 ToTTo 表格到文本任务上对比掩码扩散模型 MDLM、LLaDA 与自回归基线 GPT-2、GPT-Neo，发现参考相似度与源忠实度会脱钩：LLaDA 增加示例后 BLEU 从 11.3 升至 22.0，但针对源表格的 NLI 蕴含分数从 0.78 降至 0.52。
正文 · 原文
Abstract:Diffusion language models generate by iteratively denoising a masked sequence under bidirectional attention, without the strict left-to-right commitment of autoregressive decoding. Whether this difference affects how reliably they remain grounded in structured inputs is largely unexplored. We study this question in table-to-text generation using ToTTo, where highlighted source cells make factual grounding directly testable. We compare the masked diffusion models MDLM and LLaDA with GPT-2 and GPT-Neo under an identical retrieval-based few-shot protocol, together with a QLoRA-adapted LLaDA, and we score reference similarity and source grounding as separate axes. The two pull apart. Within LLaDA, additional demonstrations raise BLEU from 11.3 to 22.0 while NLI entailment against the source table falls from 0.78 to 0.52; QLoRA reaches the best BLEU of the study, 24.15, and grounds no better for it. Manual annotation shows the failure modes differ in kind: the autoregressive baselines tend to wander off the table, while LLaDA-based models more often stay on topic and swap out individual entities. Reference-based metrics alone are insufficient to assess faithfulness in table-to-text generation.
| Comments: | 9 pages, including references and appendix. Diagnostic study on source faithfulness in table-to-text generation |
| Subjects: | Computation and Language (cs.CL); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.29602 [cs.CL] |
| (or arXiv:2610.29602v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.29602 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Noussayba El Marrakchi [view email]
[v1]
Wed, 21 Oct 2026 11:42:19 UTC (32 KB)
来源：arXiv cs.CL · arxiv.org