1

I have trained a model on a cloud computer and now i want to port it to a Jetson TX2. It seems like it cannot load the model from the saver module of tensorflow.
It is possible to create the graph manually and then only load the weights?
And if so how do I save only the weights of the model?

This is not just a save and restore as I have tried that already without luck.

Martin Bagge
  • 385
  • 1
  • 2
  • 11
  • 1
    Possible duplicate of [Tensorflow: how to save/restore a model?](https://stackoverflow.com/questions/33759623/tensorflow-how-to-save-restore-a-model) – Maxim Feb 16 '18 at 15:30
  • 1
    You can also consider [freezing the graph](https://www.tensorflow.org/extend/tool_developers/#freezing) (a simple way to do it [here](https://stackoverflow.com/questions/45466020/how-to-export-keras-h5-to-tensorflow-pb/45466355#45466355)) – jdehesa Feb 16 '18 at 17:33
  • I will try with the freeze to see if that is enough thx. – Martin Bagge Feb 16 '18 at 19:10

1 Answers1

0

As @jdehesa commented the solution was to freeze the graph. I did the freeze and removed the devices set and I was able to load the model on the Jetson board.

Martin Bagge
  • 385
  • 1
  • 2
  • 11