tensorflow implementation for RMN

spacy==2.0.18
tensorflow-gpu==1.15.0
numpy==1.15.4
sklearn
python 3.6.9

• install requirements
  

pip install -r requirements.txt

• download en-core-web-sm
  

python -m spacy download en

• download 'glove.840B.300d.zip' from here and extract 'glove.840B.300d.txt' to './data'
  
• generate word2idx.pkl and glove_300.pkl matrix
  

python preprocess.py

• generate train and test.pkl

python Dataset.py

• train (can modify settings in Settings.py)
  

python train.py

• test
  

python train.py --train=False

• Tips
  for finetune.py, you can get the current best model, and finetune it with smaller learning rate or remove dropout to train new model.