python - AttributeError：“密集”对象没有属性“形状”（多输入模型）

Question

考虑到这种布局，我正在尝试在辅助模型上实现迁移学习，但是，我收到以下错误：

AttributeError: 'Dense' object has no attribute 'shape'

迁移学习实施

loaded_model = keras.models.load_model('pretrained_model')

#Freeze all model layer weights
loaded_model.trainable = False

print(loaded_model.summary())

base_input1 = loaded_model.layers[0]
base_input2 = loaded_model.layers[1]
base_input3 = loaded_model.layers[2]

base_inputs = [base_input1, base_input2, base_input3]

base_output = loaded_model.layers[12]

output = Dense(1, activation = "relu")(base_output)
new_model = Model(inputs = base_inputs, outputs = output)

print(new_model.summary())

标签： pythonmachine-learningdeep-learningneural-networktransfer-learning