email_coherence

ndat, basma and shafiq

How to train a model.

  1. run perm_generation.py to generate 20 permuted docs for the whole dataset
  2. run cnn_email_coherence.py with different parameters to find the best model

________________ Important _____________________

The dataset contains now a folder Entity_grid_threads which contains all the threads and their permutation

Just run the run-cnn.sh

cnn.log is the log file