metadata

language:
  - en
thumbnail: url to a thumbnail used in social sharing
tags:
  - ArguGPT
license: mit
datasets:
  - SJTU-CL/ArguGPT
metrics:
  - accuracy
pipeline_tag: text-classification

ArguGPT

RoBERTa-large finetuned on ArguGPT essays.

label 1 for machine generated essays
label 0 for human written essays

Please truncate your input essay to 512 tokens

Citation

Please cite our work arXiv:2304.07666 as

@misc{liu2023argugpt,
      title={ArguGPT: evaluating, understanding and identifying argumentative essays generated by GPT models}, 
      author={Yikang Liu and Ziyin Zhang and Wanyang Zhang and Shisen Yue and Xiaojing Zhao and Xinyuan Cheng and Yiwen Zhang and Hai Hu},
      year={2023},
      eprint={2304.07666},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}