Saving convertDicts on data_utils.py seems necessary

[2seungeun]

I think the dictionary to convert categories in test.txt (kaggle_data) is necessary. It would be better if there is an example on how to use kaagle's test.txt after training kaggle's train.txt even though current example uses only train.txt to split train, val, and test dataset.

[hjmshi]

We have not included an implementation for loading test.txt because the test set provided by Kaggle does not provide labels. From my understanding, most of the papers written on recommendation and personalization models simply split train.txt to perform their experiments.

If you would like to load the test.txt file, then you will have to modify the data_utils.py function to read that file with the same dictionary as provided from train.txt. Note that the labels will not be provided, so one will have to remove reading that part of each line.