Source data for manuscript: Zero-shot data citation function classification using transformer-based large language models (LLM)
收藏资源简介:
This repository contains the supplementary data files for the manuscript "Zero-shot data citation function classification using transformer-based large language models (LLM)" by Byers, et al. (currently in submission). File information: prompt_endpoint_static_XXX - contains metadata information for cited accession numbers (key_metadata.zip) XXX_pairs.csv - contains article>accession pairs for PubMed Central articles citing listed accession numbers (key_pairs.zip) XXX_labels.json - contains structured, manually-determined labels describing data use in listed articles (labels.zip) PMCXXX.txt - contains the full text of a PubMed Central articles as processed by the Data Citation Explorer. All included articles are reproduced in consideration of stated licenses and restrictions. See reference list below. (papers.zip) eval1_init_0temp_meta_full_l405.csv - Evaluation file for the 'initial' article set (Llama 3.1-405B) (results.zip) eval1_val_0temp_meta_full_l405.csv - Evaluation file for the 'evaluation' article set (Llama 3.1-405B) (results.zip) eval1_val_0temp_meta_full_COpus41.csv - Evaluation file for the 'evaluation' article set (Claude Opus 4.1) (results.zip) XXX_RESULTS.json - All machine generated results for a given batch (results.zip) XXX_PARAMS.json - All individual chat messages and accompanying parameters for a given batch (results.zip) XXX_MESSAGE.json - Final chat message exchanges for a given batch (results.zip)



