python train.py --train_dir 'path to desired directory for training' --arch 'model(vgg or densenet)'
--lr 'learning rate for training(float)' --hidden_units 'hidden units for training(int)'
python train.py --train_dir '/dir' --arch vgg --lr 0.005 --hidden_units 1024 --epochs 30 --gpu cuda
python predict.py --arch 'vgg or densenet (since there is only one checkpoint file that this application generates
--hidden_units 'hidden units used when training(int)' --top_k 'required no. of top K classes(int)'
--print_k '1 prints a list of top_k; 0 prints only max probability' --json_file 'complete path to category names file'
python predict.py --arch vgg --img_path /predict_img.jpg --lr 0.005 --hidden_units 1024 --top_k 3 --print_k 1