运行用于构建 CNN 的 SGD 优化器时出错

Question

i get following error while running a SGD optimizer for building a CNN

--> 146 optimizer = SGD(0.01, 0.8, 0.0005, nesterov=True) 147 model.compile(optimizer=optimizer, loss='categorical_crossentropy', metrics=['accuracy']) 148 #Train the model

TypeError: init () got multiple values for argument 'nesterov'

Answer 1

keras SGD optimizer has 3 parameters and it accept kwargs so it better to give keys to your parameter.

tf.keras.optimizers.SGD(learning_rate=0.01, momentum=0.0, nesterov=False, name="SGD")

Answer 2

If the problem does not solve after changing the code. You have to upgrade Keras and use it this way

optimizer = SGD(lr=0.01, momentum=0.8, decay=0.0005, nesterov=True)
model.compile(optimizer=optimizer, loss='categorical_crossentropy', metrics=['accuracy'])