How to get the accuracy of the bonsai_example.py in TF?

[Niranjhana Narayanan]

After training the model, in the bonsai_example.py in TF, I want to separately print the accuracy of the model. How do I do that? Should I invoke the accuracyGraph() function before sess.close() in bonsai_example.py?

I understand that calling bonsaiTrainer.train() prints the accuracy after each epoch. But I want to display the accuracy alone without calling to retrain.

[Niranjhana Narayanan]

Could anyone help with this? In ProtoNN example, it is simple to call ProtoNN.accuracy to get it, but it doesn't seem to be straightforward with the Bonsai example.

[Aditya Kusupati]

https://github.com/microsoft/EdgeML/blob/master/tf/edgeml_tf/trainer/fastTrainer.py#L480

There is an accuracy graph built for Bonsai where you can feed in the data and get the accuracy. Please refer to the example in the above-mentioned line.