将Keras模型转换为TensorFlow Lite的问题

Question

Python：

from tensorflow.contrib import lite
converter = lite.TFLiteConverter.from_keras_model_file(filepath) # Your model's name
model = converter.convert()
file = open( 'model.tflite' , 'wb' ) 
file.write( model )

输出：

ValueError                                Traceback (most recent call last)
<ipython-input-88-4a1e951d08b6> in <module>
      1 from tensorflow.contrib import lite
----> 2 converter = lite.TFLiteConverter.from_keras_model_file(filepath)
      3 model = converter.convert()
      4 file = open( 'model.tflite' , 'wb' )
      5 file.write( model )

ValueError: Unknown metric function:top_2_accuracy

我有三个输出精度，分别是top_cat_accuracy，top_2_accuracy和top_3_accuracy。该如何解决？