This repo contains the code for paper Attributes as Textual Genes: Leveraging LLMs as Genetic Algorithm Simulators for Conditional Synthetic Data Generation, which will appear at Findings of EMNLP 2025.
conda env create -f environment.yml
conda activate SDG
We use eight datasets covering diverse domains and tasks.
We also provide preprocessing scripts in /Data-preprocess to convert the raw data into expected format.
| Dataset | Domain | Task | Download |
|---|---|---|---|
| AG News | News | Classification (CLS) | Link |
| StackExchange | Science | Classification (CLS) | Link |
| ChemProt | Biomedicine | Relation Extraction (RE) | Link |
| DDI | Pharmacology | Relation Extraction (RE) | Link |
| SemEval | Web | Relation Extraction (RE) | Link |
| CoNLL04 | News | Relation Extraction (RE) | Link |
| SciTLDR | Science | Abstractive Summarization (ABS) | Link |
| MeQSum | Medical | Abstractive Summarization (ABS) | Link |
See /Generation and /Generation-scripts for details.
See /Downstream and /Downstream-scripts for details.
Feel free to contact ghan AT memphis DOT edu for any questions and collaboration opportunities.
If you find this repository helpful, please kindly consider citing the corresponding paper. Thanks in advance!
@inproceedings{han2025attributes,
title = "Attributes as Textual Genes: Leveraging {LLM}s as Genetic Algorithm Simulators for Conditional Synthetic Data Generation",
author = "Han, Guangzeng and
Liu, Weisi and
Huang, Xiaolei",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-emnlp.1055/",
pages = "19367--19389",
ISBN = "979-8-89176-335-7"
}
Inspired by and partly based on AttrPrompt. Thanks to the authors for open-sourcing it.