Pytorch reimplement of the paper "A Novel Cascade Binary Tagging Framework for Relational Triple Extraction" ACL2020. The original code was written in keras.

• keras-bert
• tensorflow-gpu
• transformers

• CMED: CHIP-2020 中文医学文本实体关系抽取

1. Get the pre-trained Chinese BERT model

   1. Download the vocab.txt of BERT-wwm
   2. Get the pre-trained BERT cache

   from transformers import *
   model = BertModel.from_pretrained("hfl/chinese-bert-wwm")
   

   p.s. I use the chinese-bert-wwm here. You can also choose other pre-trained models like this

   p.p.s. the bert cache usually will be /home/your user name/.cache/torch/transformers

2. Train the model

   python train.py
   

3. Test the model

   python test.py
   

CasRel-keras: test F1 45.59

CasRel-pytorch: test F1 47.59