LIFE
Collection
LIFE is a distantly supervised model for factual error correction. See "Improving Factual Error Correction by Learning to Inject Factual Errors"
•
4 items
•
Updated
This repository contains the checkpoint for the corruptor of LIFE, which is trained with retrieved evidence. Please refer to the code of LIFE at https://github.com/NLPCode/LIFE for more details.
# Load the model
from transformers import AutoConfig, AutoModelForSeq2SeqLM, AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("He-Xingwei/LIFE-Corruptor-RE")
model = AutoModelForSeq2SeqLM.from_pretrained("He-Xingwei/LIFE-Corruptor-RE")
If you want to use this model in your research, please cite our paper:
@inproceedings{he2024improving,
title={Improving Factual Error Correction by Learning to Inject Factual Errors},
author={He, Xingwei and Zhang, Qianru and Jin, A-Long and Ma, Jun and Yuan, Yuan and Yiu, Siu Ming},
booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
volume={38},
number={16},
pages={18197--18205},
year={2024}
}