Research

Paper

AI LLM March 23, 2026

Parameter-Efficient Fine-Tuning for Medical Text Summarization: A Comparative Study of Lora, Prompt Tuning, and Full Fine-Tuning

Authors

Ulugbek Shernazarov, Rostislav Svitsov, Bin Shi

Abstract

Fine-tuning large language models for domain-specific tasks such as medical text summarization demands substantial computational resources. Parameter-efficient fine-tuning (PEFT) methods offer promising alternatives by updating only a small fraction of parameters. This paper compares three adaptation approaches-Low-Rank Adaptation (LoRA), Prompt Tuning, and Full Fine-Tuning-across the Flan-T5 model family on the PubMed medical summarization dataset. Through experiments with multiple random seeds, we demonstrate that LoRA consistently outperforms full fine-tuning, achieving 43.52 +/- 0.18 ROUGE-1 on Flan-T5-Large with only 0.6% trainable parameters compared to 40.67 +/- 0.21 for full fine-tuning. Sensitivity analyses examine the impact of LoRA rank and prompt token count. Our findings suggest the low-rank constraint provides beneficial regularization, challenging assumptions about the necessity of full parameter updates. Code is available at https://github.com/eracoding/llm-medical-summarization

Metadata

arXiv ID: 2603.21970
Provider: ARXIV
Primary Category: cs.CL
Published: 2026-03-23
Fetched: 2026-03-24 06:02

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.21970v1</id>\n    <title>Parameter-Efficient Fine-Tuning for Medical Text Summarization: A Comparative Study of Lora, Prompt Tuning, and Full Fine-Tuning</title>\n    <updated>2026-03-23T13:35:11Z</updated>\n    <link href='https://arxiv.org/abs/2603.21970v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.21970v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Fine-tuning large language models for domain-specific tasks such as medical text summarization demands substantial computational resources. Parameter-efficient fine-tuning (PEFT) methods offer promising alternatives by updating only a small fraction of parameters. This paper compares three adaptation approaches-Low-Rank Adaptation (LoRA), Prompt Tuning, and Full Fine-Tuning-across the Flan-T5 model family on the PubMed medical summarization dataset. Through experiments with multiple random seeds, we demonstrate that LoRA consistently outperforms full fine-tuning, achieving 43.52 +/- 0.18 ROUGE-1 on Flan-T5-Large with only 0.6% trainable parameters compared to 40.67 +/- 0.21 for full fine-tuning. Sensitivity analyses examine the impact of LoRA rank and prompt token count. Our findings suggest the low-rank constraint provides beneficial regularization, challenging assumptions about the necessity of full parameter updates. Code is available at https://github.com/eracoding/llm-medical-summarization</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CL'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.AI'/>\n    <published>2026-03-23T13:35:11Z</published>\n    <arxiv:comment>9 pages, 5 figures, presented at 6th International Conference on NLP &amp; Text Mining (NLTM 2026), March 21-22, Sydney, Australia. Published in Computer Science &amp; Information Technology (CS &amp; IT), pp. 01-09, 2026</arxiv:comment>\n    <arxiv:primary_category term='cs.CL'/>\n    <arxiv:journal_ref>Computer Science &amp; Information Technology (CS &amp; IT) 16(06), 01-09 (2026)</arxiv:journal_ref>\n    <author>\n      <name>Ulugbek Shernazarov</name>\n    </author>\n    <author>\n      <name>Rostislav Svitsov</name>\n    </author>\n    <author>\n      <name>Bin Shi</name>\n    </author>\n    <arxiv:doi>10.5121/csit.2026.160601</arxiv:doi>\n    <link href='https://doi.org/10.5121/csit.2026.160601' rel='related' title='doi'/>\n  </entry>"
}