This repository stores a fine-tuned RoBERTa model to classify Chinese Weibo posts as protests or non-protests posts. It is thus a binary classification for strings.
This documentation shows how to reproduce our results.
- After creating a virtual env, use
pip install -r requirements.txtto install required packages.
NB: we highly recommand installing these packages manually, as those in our requirements.txt are configurated to adapt to our the server we used. We also chose to use virtualenv because conda did not work well on our server. Feel free to use your prefered method.
- For any .py scripts, use
python script.py -hto see documentation. - For any .ipynb scripts, please refer to commentaries over each cell.