run.py file is the testing code that generates the exact simulation results submitted to
AIcrowd, with the score achieves 83.7% accuracy and 0.750 F1 score. We should first use
modelcomputing.py to train the model before running run.py. Nonetheless, a pretrained models
parameters are also ready for testing.
modelcomputing.py includes the model training code. We arranged
training code and testing code separately to avoid memory being
overloaded under which the running speed could potentially be slowed down
please see the comments in the corresponding files