Do LLMs Really Adapt to Domains?

Code of paper: Do LLMs Really Adapt to Domains? An Ontology Learning Perspective (ISWC 2024)

Description

This repository contains the code for the experiments in the paper. The code is split into two repositories, unified by this main repository for convenience:

• wordnet-gibberish: Contains the code for creating synthetic datasets from WordNet ontologies, by replacing the lexical forms with gibberish.
• finetune-llm-ontology: Contains the code for evaluating and fine-tuning LLMs on WordNet ontologies (including the datasets created by the wordnet-gibberish package).

Please refer to the README files in each repository for more information.

Citation

If you use our software or datasets generated by it, please cite our paper:

@misc{mai2024llmsreallyadaptdomains,
      title={Do LLMs Really Adapt to Domains? An Ontology Learning Perspective}, 
      author={Huu Tan Mai and Cuong Xuan Chu and Heiko Paulheim},
      year={2024},
      eprint={2407.19998},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2407.19998}, 
}

License

This software is open-sourced under the AGPL-3.0 license. See the LICENSE file for details. Both sub-repositories also contain the same license.

The Open English WordNet (2023 Edition) is released under the Creative Commons Attribution 4.0 International License. See their LICENSE file here for details.

Contact

For any inquiries or questions, please contact the project maintainer at [email protected].