Keras: multi_gpu_model seems not to work

[emergix]

simple code :

from keras.models import Sequential
from keras.layers import Dense, Activation
model = Sequential()
model.add(Dense(32, input_dim=784))
model.add(Activation('relu'))
parralelmodel =tf.keras.utils.multi_gpu_model(model, gpus=4)

gives me

/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/base_layer.py in __init__(self, outbound_layer, inbound_layers, node_indices, tensor_indices, input_tensors, output_tensors, arguments)
1758 # For compatibility with external Keras, we use the deprecated
1759 # accessor here.
-> 1760 layer.outbound_nodes.append(self)
1761 # For compatibility with external Keras, we use the deprecated
1762 # accessor here.

AttributeError: 'InputLayer' object has no attribute 'outbound_nodes'

I use keras 2.2.4 and tensorflow 1.11.0 on cuda 9.0

[emergix]

I used tf.keras.utils.multi_gpu_model because it is mentioned there:
https://www.tensorflow.org/api_docs/python/tf/keras/utils/multi_gpu_model
but if I replace by keras.utils.multi_gpu_model , now it works.

so it is more a consistency problem with the tensorflow doc, and of course a consistency problem between 2 versions of the code

[jvishnuvardhan]

Is this still an issue with latest versions of TF and keras? Thanks.

[jvishnuvardhan]

Closing due to lack of recent activity. Please update the issue when new information becomes available, and we will open a new issue. Thanks!

[UpCoder]

In TF 1.13.1 same errors.