Format Inertia: A Failure Mechanism of LLMs in Medical Pre-Consultation
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866912629223587840 |
|---|---|
| author | Lim, Seungseop Kim, Gibaeg Han, Wooseok Seo, Jean Lee, Hyunkyung Yoo, Jaehyo Yang, Eunho |
| author_facet | Lim, Seungseop Kim, Gibaeg Han, Wooseok Seo, Jean Lee, Hyunkyung Yoo, Jaehyo Yang, Eunho |
| contents | Recent advances in Large Language Models (LLMs) have brought significant improvements to various service domains, including chatbots and medical pre-consultation applications. In the healthcare domain, the most common approach for adapting LLMs to multi-turn dialogue generation is Supervised Fine-Tuning (SFT). However, datasets for SFT in tasks like medical pre-consultation typically exhibit a skewed turn-count distribution. Training on such data induces a novel failure mechanism we term Format Inertia, where models tend to generate repetitive, format-correct, but diagnostically uninformative questions in long medical dialogues. To mitigate this observed failure mechanism, we adopt a simple, data-centric method that rebalances the turn-count distribution of the training dataset. Experimental results show that our approach substantially alleviates Format Inertia in medical pre-consultation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_01688 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Format Inertia: A Failure Mechanism of LLMs in Medical Pre-Consultation Lim, Seungseop Kim, Gibaeg Han, Wooseok Seo, Jean Lee, Hyunkyung Yoo, Jaehyo Yang, Eunho Computation and Language Artificial Intelligence Recent advances in Large Language Models (LLMs) have brought significant improvements to various service domains, including chatbots and medical pre-consultation applications. In the healthcare domain, the most common approach for adapting LLMs to multi-turn dialogue generation is Supervised Fine-Tuning (SFT). However, datasets for SFT in tasks like medical pre-consultation typically exhibit a skewed turn-count distribution. Training on such data induces a novel failure mechanism we term Format Inertia, where models tend to generate repetitive, format-correct, but diagnostically uninformative questions in long medical dialogues. To mitigate this observed failure mechanism, we adopt a simple, data-centric method that rebalances the turn-count distribution of the training dataset. Experimental results show that our approach substantially alleviates Format Inertia in medical pre-consultation. |
| title | Format Inertia: A Failure Mechanism of LLMs in Medical Pre-Consultation |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2510.01688 |