kerasd的callbacks

callbacks need a list, load fit inside. Only one parameter can be, you can also adjust some parameters of the learning rate.

checkpoint = ModelCheckpoint(filepath=filepath,
                             monitor='val_acc',
                             verbose=1,
                             save_best_only=True)

callbacks = [checkpoint]

# Training.
model.fit(x_train, y_train,
          batch_size=batch_size,
          epochs=epochs,
          verbose=1,
          validation_data=(x_test, y_test),
          callbacks=callbacks)