TobiasLee / FormBERT

Code implementation of FormBERT, accepted by Findings of EMNLP 2021

FormBERT

Code implementation of Findings of EMNLP 2021 Paper: Leveraging Word-Formation Knowledge for Chinese Word Sense Disambiguation

Requirements

We recommend using Anaconda for the setup of experiments environment.

The following instructions would be helpful:

git clone https://github.com/TobiasLee/FormBERT
cd FormBERT
conda create -n formbert python=3.7
conda activate formbert
conda install pytorch torchvision cudatoolkit=11.0 -c pytorch 
pip install -r requirements.txt 

Data

Our FiCLS dataset will be released soon.

Run Experiment

sh train.sh

Contact

If you have any further questions about the data and code, raise a issue at the repo or send emails to the authors:

Hua Zheng: [e-mail]

Lei Li: [e-mail]