Hi there,
I’m trying to do the ML example with the jupyter notebook, but I’m running into an error from tensorflow. I’ve tried a few different things but I don’t know enough about keras to be able to debug it properly. My question is, is there a working example that I could take a look at to know where I went wrong.
I’ve attached my Jupyter side code, the error on the anvil side says:
InvalidArgumentError: Tensor conv2d_1_input:0, specified in either feed_devices or fetch_devices was not found in the Graph
Thanks,
Eleanor
In classic deep learning style, they make a beautiful notebook and don’t show inference methods . Do you know which line is causing the issue?
I would go through the data prep section to see how they prep the data for loading to the model. You would need to do the same to your image in order to get a score from the predict method.
Thanks Robert. Interestingly I can make it work locally on my machine so it’s not necessarily a problem with the tenser code. It seems to be related to a web based issue - I’ve seen some threads relating to “sessions” and having a graph default but any attempts I’ve done to try and make that work have failed unfortunately.
I’ve tried 3 different ways of developing a web app and it is quite difficult!
Thanks,
Eleanor
. Do you know which line is causing the issue?