POS Tagging for Chinese

Model : CharLSTM + LSTM + CRF

Pretrained Word Embedding : giga.100.txt

Data : CTB5

requirements

python >= 3.6.3
pytorch = 0.4.1

running

mkdir save                   # or define other path to save model and vocab
python train.py --pre_emb    # train the model with pretrained embedding
python evaluate.py           # reload the model and evaluate test file

results

dev=95.84%
test=95.65%