Skip to content

Latest commit

 

History

History
32 lines (28 loc) · 730 Bytes

README.md

File metadata and controls

32 lines (28 loc) · 730 Bytes

Environment setup

  • Requirements: python3, tensorflow1.10, sklearn, numpy, pandas and jieba
pip install -r requirements.txt

Training

  • CNN
python train_cnn.py
  • RNN with attention
python train_rnn.py

the data format is as follows: label '\t' text

Testing

  • evaluation mode:
python test.py --checkpoint_dir=./runs/textcnn/trained_result_1548144557/checkpoints --model_type=CNN
  • prediction mode:
python test.py --checkpoint_dir=./runs/textcnn/trained_result_1548144557/checkpoints --model_type=CNN --test_mode=prediction

Inference

python inference.py --checkpoint_dir=./runs/textrnn/trained_result_1548145204/checkpoints --model_type=RNN