train.py

[kiristern]

go over different parts and sections Ahmad put in the train.py file

[kiristern]

Line 112-116 would guide you about what needs to be done to train the network.
Need to build the network first to get the input of size (4, 1, 600) (in batches) and output a vector of size (164) (again, in batches).
once you have the output of the model, you can compare with the ground truth labels to update the model params and recode statistics of training