sentence_classification

Requirements

• Python 2.7
• Pytorch

Data preparation

• Download Stanford CoreNLP and GloVe, then extract to data/.

  wget http://nlp.stanford.edu/software/stanford-corenlp-full-2017-06-09.zip
  unzip stanford-corenlp-full-2017-06-09.zip
  wget http://nlp.stanford.edu/data/glove.840B.300d.zip
  unzip glove.840B.300d.zip

TREC

1. Download the dataset.

   wget http://cogcomp.org/Data/QA/QC/train_5500.label
   wget http://cogcomp.org/Data/QA/QC/TREC_10.label

2. Parse raw data.

   python preprocess.py

3. Get json data.

   ./gen_json.sh

Train

• Without glove

python main.py --gpu

• With glove

python main.py --gpu --glove_path=data/glove.840B.300d.txt --lr=0.008 --lr_milestones=11

Evaluate

• Without glove

python main.py --gpu --mode=eval \
    --checkpoint_path=models/TREC_test_batchsize25_input300_hidden100_lr0.001_seed10137_epoch48.pth

• With glove

python main.py --gpu --mode=eval \
    --checkpoint_path=models/TREC_test_batchsize25_input300_hidden100_lr0.008_ms11_wc0.0001_glove_seed10137_epoch12.pth