Taiyi : a bilingual fine-tuned large language model for diverse biomedical tasks
© The Author(s) 2024. Published by Oxford University Press on behalf of the American Medical Informatics Association. All rights reserved. For permissions, please email: journals.permissionsoup.com..
OBJECTIVE: Most existing fine-tuned biomedical large language models (LLMs) focus on enhancing performance in monolingual biomedical question answering and conversation tasks. To investigate the effectiveness of the fine-tuned LLMs on diverse biomedical natural language processing (NLP) tasks in different languages, we present Taiyi, a bilingual fine-tuned LLM for diverse biomedical NLP tasks.
MATERIALS AND METHODS: We first curated a comprehensive collection of 140 existing biomedical text mining datasets (102 English and 38 Chinese datasets) across over 10 task types. Subsequently, these corpora were converted to the instruction data used to fine-tune the general LLM. During the supervised fine-tuning phase, a 2-stage strategy is proposed to optimize the model performance across various tasks.
RESULTS: Experimental results on 13 test sets, which include named entity recognition, relation extraction, text classification, and question answering tasks, demonstrate that Taiyi achieves superior performance compared to general LLMs. The case study involving additional biomedical NLP tasks further shows Taiyi's considerable potential for bilingual biomedical multitasking.
CONCLUSION: Leveraging rich high-quality biomedical corpora and developing effective fine-tuning strategies can significantly improve the performance of LLMs within the biomedical domain. Taiyi shows the bilingual multitasking capability through supervised fine-tuning. However, those tasks such as information extraction that are not generation tasks in nature remain challenging for LLM-based generative approaches, and they still underperform the conventional discriminative approaches using smaller language models.
| Medienart: |
E-Artikel |
|---|
| Erscheinungsjahr: |
2024 |
|---|---|
| Erschienen: |
2024 |
| Enthalten in: |
Zur Gesamtaufnahme - year:2024 |
|---|---|
| Enthalten in: |
Journal of the American Medical Informatics Association : JAMIA - (2024) vom: 29. Feb. |
| Sprache: |
Englisch |
|---|
| Beteiligte Personen: |
Luo, Ling [VerfasserIn] |
|---|
| Links: |
|---|
| Themen: |
Biomedical multitasking |
|---|
| Anmerkungen: |
Date Revised 29.02.2024 published: Print-Electronic Citation Status Publisher |
|---|
| doi: |
10.1093/jamia/ocae037 |
|---|
| funding: |
|
|---|---|
| Förderinstitution / Projekttitel: |
|
| PPN (Katalog-ID): |
NLM369122690 |
|---|
| LEADER | 01000naa a22002652 4500 | ||
|---|---|---|---|
| 001 | NLM369122690 | ||
| 003 | DE-627 | ||
| 005 | 20240301233107.0 | ||
| 007 | cr uuu---uuuuu | ||
| 008 | 240301s2024 xx |||||o 00| ||eng c | ||
| 024 | 7 | |a 10.1093/jamia/ocae037 |2 doi | |
| 028 | 5 | 2 | |a pubmed24n1313.xml |
| 035 | |a (DE-627)NLM369122690 | ||
| 035 | |a (NLM)38422367 | ||
| 035 | |a (PII)ocae037 | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
| 041 | |a eng | ||
| 100 | 1 | |a Luo, Ling |e verfasserin |4 aut | |
| 245 | 1 | 0 | |a Taiyi |b a bilingual fine-tuned large language model for diverse biomedical tasks |
| 264 | 1 | |c 2024 | |
| 336 | |a Text |b txt |2 rdacontent | ||
| 337 | |a ƒaComputermedien |b c |2 rdamedia | ||
| 338 | |a ƒa Online-Ressource |b cr |2 rdacarrier | ||
| 500 | |a Date Revised 29.02.2024 | ||
| 500 | |a published: Print-Electronic | ||
| 500 | |a Citation Status Publisher | ||
| 520 | |a © The Author(s) 2024. Published by Oxford University Press on behalf of the American Medical Informatics Association. All rights reserved. For permissions, please email: journals.permissionsoup.com. | ||
| 520 | |a OBJECTIVE: Most existing fine-tuned biomedical large language models (LLMs) focus on enhancing performance in monolingual biomedical question answering and conversation tasks. To investigate the effectiveness of the fine-tuned LLMs on diverse biomedical natural language processing (NLP) tasks in different languages, we present Taiyi, a bilingual fine-tuned LLM for diverse biomedical NLP tasks | ||
| 520 | |a MATERIALS AND METHODS: We first curated a comprehensive collection of 140 existing biomedical text mining datasets (102 English and 38 Chinese datasets) across over 10 task types. Subsequently, these corpora were converted to the instruction data used to fine-tune the general LLM. During the supervised fine-tuning phase, a 2-stage strategy is proposed to optimize the model performance across various tasks | ||
| 520 | |a RESULTS: Experimental results on 13 test sets, which include named entity recognition, relation extraction, text classification, and question answering tasks, demonstrate that Taiyi achieves superior performance compared to general LLMs. The case study involving additional biomedical NLP tasks further shows Taiyi's considerable potential for bilingual biomedical multitasking | ||
| 520 | |a CONCLUSION: Leveraging rich high-quality biomedical corpora and developing effective fine-tuning strategies can significantly improve the performance of LLMs within the biomedical domain. Taiyi shows the bilingual multitasking capability through supervised fine-tuning. However, those tasks such as information extraction that are not generation tasks in nature remain challenging for LLM-based generative approaches, and they still underperform the conventional discriminative approaches using smaller language models | ||
| 650 | 4 | |a Journal Article | |
| 650 | 4 | |a biomedical multitasking | |
| 650 | 4 | |a large language model | |
| 650 | 4 | |a natural language processing | |
| 650 | 4 | |a supervised fine-tuning | |
| 700 | 1 | |a Ning, Jinzhong |e verfasserin |4 aut | |
| 700 | 1 | |a Zhao, Yingwen |e verfasserin |4 aut | |
| 700 | 1 | |a Wang, Zhijun |e verfasserin |4 aut | |
| 700 | 1 | |a Ding, Zeyuan |e verfasserin |4 aut | |
| 700 | 1 | |a Chen, Peng |e verfasserin |4 aut | |
| 700 | 1 | |a Fu, Weiru |e verfasserin |4 aut | |
| 700 | 1 | |a Han, Qinyu |e verfasserin |4 aut | |
| 700 | 1 | |a Xu, Guangtao |e verfasserin |4 aut | |
| 700 | 1 | |a Qiu, Yunzhi |e verfasserin |4 aut | |
| 700 | 1 | |a Pan, Dinghao |e verfasserin |4 aut | |
| 700 | 1 | |a Li, Jiru |e verfasserin |4 aut | |
| 700 | 1 | |a Li, Hao |e verfasserin |4 aut | |
| 700 | 1 | |a Feng, Wenduo |e verfasserin |4 aut | |
| 700 | 1 | |a Tu, Senbo |e verfasserin |4 aut | |
| 700 | 1 | |a Liu, Yuqi |e verfasserin |4 aut | |
| 700 | 1 | |a Yang, Zhihao |e verfasserin |4 aut | |
| 700 | 1 | |a Wang, Jian |e verfasserin |4 aut | |
| 700 | 1 | |a Sun, Yuanyuan |e verfasserin |4 aut | |
| 700 | 1 | |a Lin, Hongfei |e verfasserin |4 aut | |
| 773 | 0 | 8 | |i Enthalten in |t Journal of the American Medical Informatics Association : JAMIA |d 1997 |g (2024) vom: 29. Feb. |w (DE-627)NLM074735535 |x 1527-974X |7 nnns |
| 773 | 1 | 8 | |g year:2024 |g day:29 |g month:02 |
| 856 | 4 | 0 | |u http://dx.doi.org/10.1093/jamia/ocae037 |3 Volltext |
| 912 | |a GBV_USEFLAG_A | ||
| 912 | |a GBV_NLM | ||
| 951 | |a AR | ||
| 952 | |j 2024 |b 29 |c 02 | ||