mnist_train.py: It is pytorch code for training the MNIST data, the program saves the pickel files with only paramete with the use of "state_dict"( method 1 ) through this we save only the parameters instead the whole model.
checkpoint_mnist_train.py : It is training the mnist data but saving the entire model it allows model to be used by someone else without the access to your code.
checkpoint_mnist_test.py : It is pytorch code of loading bianry files with parameters ( method 2 )
checkpoint_mnist_test.py : It is pytorch code of loading bianry files with entire model ( method 1 )
hook.py : shows the use of register_hook() to print certain variables gradients
Gradietn_train.py : pytorch code to train mnist model and to print the gradient just before 1st FC layer
Data: is enclosed in zipped file "train&test data.tar.gz"