def load_model(filename: str):
"""Loads the model architecture and weights as a usable Model.
Requires '<filename>_model.json' and '<filename>_weights.h5'"""
with open(filename + '_model.json', 'r') as fobj:
model = model_from_json(fobj.read())
model.load_weights(filename + '_weights.h5')
return model
def load_model():
# import model
json_model = open('model/model.json', 'r')
load_model_json = json_model.read()
json_model.close()
loaded_model = model_from_json(load_model_json)
loaded_model.load_weights('model/model.h5')
return loaded_model
print('test after loading model...')
print(model.predict(x_test[:2]))
print('test accuracy :', model.evaluate(x_test, y_test))
plt.scatter(x_test, model.predict(x_test), c='r')
plt.scatter(x_test, y_test, c='b')
plt.show()
#another way to save and load
# only save the weights of model , the structure of model is not saved
model.save_weights('my_model_weights.h5')
# to bulid a new model whose structure is same as before ....
new_model.load_weights('my_model_weights.h5')
# a method to save model structure only ,well-trained weights is not saved
from keras import model_from_json
json_string = model.to_json()
new_mode = model_from_json(json_string)
model_json = model.to_json()
with open("model.json", "w") as json_file:
json_file.write(model_json)
# serialize weights to HDF5
model.save_weights("model.h5")
# imports
from keras import model_from_json
# opening and store file in a variable
json_file = open('model.json', 'r')
loaded_model_json = json_file.read()
json_file.close()
# use Keras model_from_json to make a loaded model
loaded_model = model_from_json(loaded_model_json)
# load weights into new model
loaded_model.load_weights("model.h5")
print("Loaded Model from disk")
# compile and evaluate loaded model
loaded_model.compile(loss='categorical_crossentropy',
optimizer='adam',
metrics=['accuracy'])