The trained model, 'QDrawModel.h5', which has been run for 25 epochs and achieved a test accuracy of just over 90%, has been saved in the repository. You have already seen this code; it is reproduced here for your convenience.
So, to reiterate, you can load this trained model with the following code:
from keras.models import load_modelmodel = load_model('./QDrawModel.h5')model.summary()
Again, the training/test data can be loaded with the following code:
import h5pyimport numpy as nphf = h5py.File('x_train.h5', 'r')x_train = np.array(hf["QuickDraw"][:])hf = h5py.File('x_test.h5', 'r')x_test = np.array(hf["QuickDraw"][:])hf = h5py.File('y_train.h5', 'r')y_train = np.array(hf["QuickDraw"][:])