GitHub

All files and models for training BERT on Java source code

multiand own_vocab were trained on 30 training files each containing 10k Java files for 2 epochs, rest on standard configuration the own_vocab has a modifed vocabulary file that contains additional ...