ChunLiu-cs / LLMEmbed-ACL2024

16 stars 3 forks source link

LLMEmbed: Rethinking Lightweight LLM's Genuine Function in Text Classification

This code is for the LLMEmbed paper accepted in the 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024) https://aclanthology.org/2024.acl-long.433

llama2_embedding / bert_embedding / roberta_embedding

The rep_extract.py uses language model to extract the representation of dataset and saves the representation as .pt file.

MyDataset

MyDataset.py reads the representation from .pt file.

DownstreamModel

DownstreamModel.py is for the co-occurence pooling.

📜Citation

This work has been accepted to [ACL-2024](url: https://aclanthology.org/2024.acl-long.433), please cite the paper if you use LLMEmbed or this repository in your research. Thank you very much 😉

@inproceedings{chunliu2024llmembed,
  title={LLMEmbed: Rethinking Lightweight LLM’s Genuine Function in Text Classification},
  author={Liu, Chun and Zhang, Hongguang and Zhao, Kainan and Ju, Xinghai and Yang, Lin},
  booktitle={Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  pages={7994--8004},
  year={2024}
}