A simple guide for EGN3LDG: text classification
- Simple CNN and LSTM classifiers are implemented.
- Checkgrad in cpp file can be commented out as a pratical classifier.
- add_definitions( -DUSE_FLOAT ) must be commented out for Checkgrad.
- If we add new neural modules in EGN3LDG, should check the module gradients by the clasifier.
(1) cd build;
(2) rm -rf *;
(3) cmake .. (if MKL is installed, check "set(MKL_ROOT /opt/intel/mkl)", and run cmake .. -DMKL=True);
(4) make;
If you have any problem, please send an email to mason.zms@gmail.com.