The compilation model of the CoreML model. An error compilation model prepared in keras with an injectable layer as the first layer.

Question

I trained a simple model in keras and its first layer is the Embedding layer as shown below:

enter image description here

I converted the model to .mlmodel, but when I try to compile the code in Xcode I get this error:

/mlkitc: compiler error: Inner product layer: dense_1: Product of input C,H,W (32,1,1) must be equal to the input channels (16000)

Is there a conversion issue?

+3

xcode swift machine-learning coreml keras

Emannuel carvalho June 20. At 4:28 am

source to share

No one has answered this question yet

Check out similar questions:

8

Convert Keras Image Classification Trainable Model to Coreml and Integration in iOS11

6

Opening Keras Model with Embedding Layer in Tensorflow in Golang

3

Keras model for predicting a sequence of numbers

2

How to format input for conv1d / lstm nn in coreml converted from keras

2

Keras Model - Functional API - adding layers to an existing model

1

Layered Layers for Keras Sequential Model

1

Embedding GloVe Embedding Layer in Keras Model

0

Seralizing keras model with embedding level

0

Channels first with Keras?