cooelf/limitbert

The EMNLP 2020 Findings paper is based on Limit-BERT: Linguistic Informed Multi-task BERT . Python 3.6 or Cython 0.25.2 or Pytorch 1.0.0+. PyTorch is available for download from Google Drive . Pre-trained models (PyTorch) are available for training purposes . The training scripts are available in the original repo for the training scripts . To train LimIT-berT, simply run: run run_limitbert.sh or test it after setting model path: test_bert . To test it, run make inside the EVALB/ directory to compile an evalb executable,