The predictions were overwriting the actual labels.
0.1.1
This version displays the proportion of labels.
0.1.0
This version contains the main two components. One is to annotate the dataset following the file sequence, and the second is to use an active learning approach.
0.0.25
This version has the option to perform active learning.
0.0.24
This version has a parameter to shuffle the dataset. It verifies that there are instances without labels to predict only when necessary.
0.0.23
In this version, it is possible to decide the batch size.