Tensorflow 1.x solution of chinese NER task Using ALBERT-BiLSTM-CRF model with Google ALBERT Fine-tuning

Disclaimer, this project is for techinical learning and has not been used in commercial scenarios.

You can get the chinese model from here.

This project run in python3 and tensorflow 1.x.

本人不定期更新相关博客，需梯子。

You can upload how-to-use.ipynb to google colab, or see my blog here. Please use GPU. Just follow Edit > Notebook settings or Runtime>Change runtime type and select GPU as Hardware accelerator.

• 2020.4.19 create project

data/train.txt, and also dev.txt、test.txt dataset is like this:

海 O
钓 O
比 O
赛 O
地 O
点 O
在 O
厦 B-LOC
门 I-LOC
与 O
金 B-LOC
门 I-LOC
之 O
间 O
的 O
海 O
域 O
。 O

Each line contain a token and a token's label, each sentences divide by blank line(not show here, please check train.txt).

• https://github.com/google-research/albert

• https://github.com/macanv/BERT-BiLSTM-CRF-NER

• https://github.com/ProHiryu/albert-chinese-ner