Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This study investigates the effectiveness of various post-training alignment methods for small language models (SLMs) in generating patient-friendly narratives from complex biomedical data, specifically focusing on medication leaflets. By comparing supervised fine-tuning (SFT), direct preference optimization (DPO), odds ratio preference optimization (ORPO), and group relative policy optimization (GRPO), the authors demonstrate that aligned SLMs can surpass proprietary models like GPT-5 in performance. Notably, GRPO achieves the best cross-dataset generalizability, indicating its potential as a superior method for biomedical data-to-text tasks.
Aligned small language models can outperform GPT-5 in generating patient-friendly medication leaflets, with GRPO showing the strongest cross-dataset performance.
Translating complex biomedical data into patient-friendly narratives is central to modern biomedical informatics. This study presents a comparative analysis of training small language models (SLMs) in specialized biomedical datato-text generation tasks. We explore widely adopted post-training methods including supervised fine-tuning (SFT), direct preference optimization (DPO), odds ratio preference optimization (ORPO), and group relative policy optimization (GRPO) with Qwen-based SLMs on a medicine package leaflets dataset. To assess cross-dataset generalizability, we also curated drug label data from openFDA. We evaluate models using both standard lexical overlap metrics like ROUGE as well as semantic similarity measures. Across our experiments, the results show that (1) the aligned SLMs outperform proprietary models like GPT-5; (2) ORPO outperforms the SFTbaselines; (3) GRPO yields the most robust cross-dataset performance among the alignment methods tested as well as GPT-5.