How to save a graph about models HOT 6 CLOSED

Any progress on this issue? I had call to implement the same procedure today and am also having trouble identifying the layer names.

from models.

Sorry you're hitting problems, the output layer names can be hard to figure out. Typically for Inception models they're "softmax" (and you'll need them to run the model too). @shlens do you know what your training script names them in this case?

from models.

Sorry for all of the trouble. We keep a track of pointers to all Tensors of
interest in the Python endpoints dict.

For instance, the logits can be found here:
https://github.com/tensorflow/models/blob/master/inception/inception/slim/inception_model.py#L328

And the normalized predictions can be found here:
https://github.com/tensorflow/models/blob/master/inception/inception/slim/inception_model.py#L329

If you run the script and print out endpoints['predictions'].op.name, this
should provide the name of the output layer.

On Mon, Apr 11, 2016 at 8:25 AM, Pete Warden [email protected]
wrote:

Sorry you're hitting problems, the output layer names can be hard to
figure out. Typically for Inception models they're "softmax" (and you'll
need them to run the model too). @shlens https://github.com/shlens do
you know what your training script names them in this case?

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#38 (comment)

from models.

@shlens @petewarden Is this resolved?

from models.

I've been trying to do the same thing so that I can run this model, and I must say it's been confusing.

I ended up figuring out that I needed to pass in 'inception_v3/logits/predictions' as the output_node (which I only found by iterating over the graph and dumping the names).

I also ended up getting stuck for a moment since I was confused by how to deal with the fact that the tensor was of shape [32, 299, 299, 3] and I couldn't figure out what I needed to do to get it to accept a tensor of the shape [1, 299, 299, 3]; in the end it just seems to work? But I have no idea if I'm doing the right thing.

from models.

@kuza55 These questions might be better posed on StackOverflow. Can you ask them there with the 'tensorflow' tag. Closing this out...

from models.