Hi there,
I'm getting a "TypeError: unsupported operand type(s) for +: 'NoneType' and 'str'" while converting a Keras model using a Bidirectional(LSTM) layer.
To give you more context information the model also use an Embedding layer with pretrained GloVe weights before the bidirectional layer and a final Dense layer with Softmax activation at the end.
Moreover I'm using LSTM instead of GRU as in my original model because I got a specific CoreML export error suggesting me to use LSTM (ValueError: Keras bi-directional wrapper conversion supports only LSTM layer at this time).
Everything works fine if I just use a mono directional LSTM, or even GRU, layer in the model.
Is there anyone using Keras Bidirectional layers succesfully with CoreMLTools ?
Thanks,
-Jacopo