tensorflow - DKeras 缺少 1 个必需的位置参数：“输入”

Question

我正在尝试使用 DKeras，但是当我使用它时

 model = dKeras(model=brain.model, init_ray=False, wait_for_workers=True, n_workers=2)

通过这个错误

TypeError: __call__() missing 1 required positional argument: 'inputs'

该模型非常简单，并且可以在没有 dKeras 的情况下工作

    states = keras.Input(shape=(616,))
  
    a = Dense(units=128, activation='sigmoid')(states)
    a = Dropout(rate=0.1)(a)

    x = Dense(units = 64, activation = 'sigmoid')(a)
    x = Dropout(rate = 0.1)(x)

    y = Dense(units = 32, activation = 'sigmoid')(x)
    y = Dropout(rate = 0.1)(y)

    q_values = Dense(units = number_actions, activation = 'softmax')(y)
    
    self.model = Model(inputs = states, outputs = q_values)
    self.model.compile(loss = 'mse', optimizer = Adam(lr = learning_rate))

标签： tensorflowkerasartificial-intelligence