Repository navigation
BERT_with_positional_features
·
17 commits
to main
since this release
This is the model resulted by the script Bert_pos_features.py. The model is based on the pretrained KBBertmodel_stratifiedsampling_withweights and Positional_FFNN published in earlier releases. The model was trained on the non-merged segements of the unstratified data to classify margins vs other segments. The scores of the model are (92.80, 89.90, 76.72, 82.79) for (accuracy, precision, recall, f1-score).