hecongqing / TDNN

The python source used in the TDNN model in Jin et al. ACL 2018.

Following python files are included:

-embedding.py: 
  Manage word embedding file in HDF5 format.

-utils.py:
  Functions used for data preparation. 

-metrics.py: 
  Evaluation metrics. 

-essay_model.py:   
  Model implementation.

-train_part_essay_glove.py:   
  Training of TDNN(SEM)

-train_part_essay_parser.py:
  Training of TDNN(SEM+Synt)