我有3个输入将进入我的模型,它们是: 120 X 20列表 2X[2]数组或类似[[a,b],[c,d],…]的n个数 三。[a,b]阵列
这使模型变得简单,但我不明白如何定义这个模型的输入和输出。你知道吗
模型应该有两个隐藏层,每个层32个。你知道吗
inputBoard = tf.keras.Input(shape=(20,20))
x1 = tf.keras.layers.Flatten()(inputBoard)
boardDense = tf.keras.layers.Dense(32, activation='relu')(x1)
inputAgent = tf.keras.Input(shape=(None, 2))
x2 = tf.keras.layers.Flatten()(inputAgent)
agentDense = tf.keras.layers.Dense(16, activation='relu')(x2)
inputReward = tf.keras.Input(shape=(1,1))
x3 = tf.keras.layers.Flatten()(inputReward)
rewardDense = tf.keras.layers.Dense(4, activation='relu')(x3)
concat = tf.keras.layers.concatenate([boardDense, agentDense, rewardDense])
hidden = tf.keras.layers.Dense(32, activation = 'relu')(concat)
hidden2 = tf.keras.layers.Dense(32, activation='relu')(hidden)
output = tf.keras.layers.Dense(4, activation="softmax")(hidden2)
self.model = tf.keras.Model(inputs=[inputBoard, inputAgent, inputReward], outputs = output)
self.model.build()
optimizer = tf.keras.optimizers.Adam(lr = self.learningRate)
self.model.compile(loss = 'mse', optimizer = optimizer)
self.model.summary()
目前没有回答
相关问题 更多 >
编程相关推荐