tensorflowjs_converter能否与使用函数API制作的Keras模型一起工作？

Question

我试图将Python中使用tf.keras生成的模型转换为tensorflow.js格式，以便在Node.js中使用。以下是我的软件包版本：

tensorflowjs: 1.0.1
Keras: 2.2.4
tf-nightly-2.0-preview: 2.0.0.dev20190321 (from pip install tensorflowjs)

下面是我的序列模型，也是用函数API重新制作的：

^{pr2}$

当我使用tensorflowjs_converter将序列模型转换为tfjs_layers_model时，它使用tensorflowjs进行很好的加载。当我对函数模型执行同样的操作时，我会得到格式不正确的model config错误：

Error: Improperly formatted model config for layer {"_callHook":null,"_addedWeightNames":[],"_stateful":false,"id":1,"activityRegularizer":null,"inputSpec":[{"minNDim":2}],"supportsMasking":true,"_trainableWeights":[],"_nonTrainableWeights":[],"_losses":[],"_updates":[],"_built":false,"inboundNodes":[],"outboundNodes":[],"name":"dense_38","trainable_":true,"updatable":true,"initialWeights":null,"_refCount":null,"fastWeightInitDuringBuild":true,"activation":{},"useBias":true,"kernel":null,"bias":null,"DEFAULT_KERNEL_INITIALIZER":"glorotNormal","DEFAULT_BIAS_INITIALIZER":"zeros","units":128,"kernelInitializer":{"scale":1,"mode":"fanAvg","distribution":"uniform","seed":null},"biasInitializer":{},"kernelConstraint":null,"biasConstraint":null,"kernelRegularizer":null,"biasRegularizer":null}: "input_26"

我也尝试过导出为tfjs_graph_model，但是tensorflowjs_converter不允许这样做。我希望模型最终有多个输出，这就是为什么我希望使用函数式API而不是顺序的。在