/FENDER

Source code of the paper "Variety Is the Spice of Life: Detecting Fake News with Dynamic Contextual Representations"

Primary LanguagePython

FENDER

This repo is the official code of our work Variety Is the Spice of Life: Detecting Fake News with Dynamic Contextual Representations

This code acknowledges and follows to Generalizing to the Future: Mitigating Entity Bias in Fake News Detection

Requirements

cudatoolkit==11.8.0
torch==1.13.1
numpy==1.23.5

Train

  1. Download datasets from GITHUB LINK, and place them to ./FENDER_en/data and ./FENDER_ch/data, respectively;
  2. Run the scripts ./FENDER_en/run.sh or ./FENDER_ch/run.sh to rerun the experiments, or you can directly run
python main.py --para_len 32 --seed 1000

Output

The log files are saved in ./logs/param, and the checkpoints can be loaded from ./para_model.

Acknowledgement

We acknowledge the source code provided by GITHUB LINK.

Citation

NONE