Slovene instruction-following dataset for large language models GaMS-Instruct-MED-Anatomy 1.0

  Avtorji

Plesnik, Emil ; et al.prikaži vse

Plesnik, Emil ; Tovornik, Robert ; Fabjan, Borut ; Korošec, Filip ; Žabkar, Ines ; Kuzman, Ema ; Rigler, Martin ; Škufca, Lara

  Identifikator vnosa

http://hdl.handle.net/11356/2085

 URL projekta

https://www.cjvt.si/povejmo/

 Datum objave

2026-02-03

 Vrsta

corpus, text

 Velikost

711805 units

 Jezik(i)

Slovenian

 Opis

GaMS-Instruct-MED-Anatomy is an instruction-following dataset containing 711,805 prompt-response units in Slovene (with English and Latin terminology). The units form a structured, machine-readable database of Slovenian anatomical terminology for training language models. The collection is based on anatomical data collected, translated and validated by medical experts. The data was processed, structured and enriched with automatic scripts and explanations generated using large language models. It includes: • Anatomical terminology in Slovene, English and Latin • SNOMED CT classification (standardized medical coding system) • Classification by body systems • Synonyms and alternative terms (original and generated) • Popular explanations of anatomical structures for the general public • Expert explanations of anatomical structures for medical experts The result is a standardized database in an instructional format that is suitable for use in computational linguistics, natural language processing (NLP), medical informatics and for training and adapting large language models. The corpus is intended for research and development of fine-tuning language models, training and adapting large language models for the medical and anatomical domains, development of medical chatbots and assistants in Slovene, support for healthcare professionals in anatomical terminology, translation of medical documentation, standardization of medical terminology in Slovene, and education in the field of anatomy. For more details on the structure of the dataset, please consult 00README.txt.

 Izdajatelj

Better, d.o.o.

 

Faculty of Computer and Information Science, University of Ljubljana

 Zahvala

ARIS (Slovenian Research and Innovation Agency) NOO "PoVeJMo research project (Adaptive Natural Language Processing with Large Language Models)"

ARRS (Slovenian Research Agency) P6-0411 "Language Resources and Technologies for Slovene"

 Ključne besede

instruction following dataset medical texts large language models anatomy

 Zbirke

CLARIN.SI data & tools

Prikaži polni zapis vnosa