Syntactic-MTL

This repository contains the multitask learning model proposed in Granular Syntax Processing with Multi-Task and Curriculum Learning.

Requirements

• pytorch
• flair 0.13.0
• fastnlp 0.7.0

Usage

To train and test the model on two syntactic tasks, put the non-parallel labeled dataset in the p_paths, and the parallel labeled dataset in the c_paths, e.g., WSJ dataset for p_paths and CoNLL2000 for c_paths. Then run the following example script:

python run.py --batch_size 20 --lr 0.0001 --cl_size 4 --p_paths p_dataset --c_paths c_dataset

Note that the model only works on syntax processing tasks with sequence labeling setting.

Citation

If you use this knowledge base in your work, please cite the paper - Granular Syntax Processing with Multi-Task and Curriculum Learning with the following:

@article{zhang2024granular,
  title={Granular Syntax Processing with Multi-Task and Curriculum Learning},
  author={Zhang, Xulang and Mao, Rui and Cambria, Erik},
  journal={Cognitive Computation},
  pages={1--15},
  year={2024},
  publisher={Springer}
}