How did you run the code to retrieve the confusion matrix

[julijenv]

Dear @payalbajaj , I have been trying to implement and modify your code in order to be able to get the classification working - meaning that when I run sketch_rnn_classification, I intent to retrieve an estimation of the category for an existing saved list of strokes inside an ndjson files in order to run it in a loop and retrieve the seemingly confusion matrix. From now on, looks like the code is just making the training and I have been trying to use the only function remaining unused inside your code that looks like it really makes the classification (predict_model). Could you please share your insights on how you modify the main() inside your sketch-rnn_classification.py to produce the confusion matrix, which of course will unveil the way you actually classify. let me know, thanks in advance