mic

import pickle

train_file = open('train_dataset', 'ab')
val_file = open('val_dataset', 'ab')

pickle.dump(train_dataset, train_file)
pickle.dump(valid_dataset, val_file)

train_file.close()
val_file.close()