This repository includes the code for buliding a very simple character-based BiLSTM-CRF sequence labeling model for Chinese Named Entity Recognition task. Its goal is to recognize three types of Named Entity: PERSON, LOCATION and ORGANIZATION.
This code works on Python 3.7 & Pytorch 1.1.0 and the following repositories or course
https://github.com/Determined22/zh-NER-TF
https://github.com/mtreviso/linear-chain-crf
http://web.stanford.edu/class/cs224n/
gives me much help.
All of code are annotated well.
python main.py
-
After 230 epoch, the accuracy is 0.97+
-
You can test your sentence, e.g.
