cl-tohoku/bert-base-japanese-char

This is a BERT model pretrained on texts in the Japanese language . The model processes input texts with word-level tokenization based on the IPA dictionary . The training corpus is 2.6GB in size, consisting of approximately 17M sentences . The code for the training is available at cl-tohoku/bert-japanese. The model is trained on Japanese Wikipedia as of September 1, 2019. The models are distributed under the terms of the Creative Commons Attribution-ShareAlike 3.0 license . The data was used to generate the training corpus, WikiExtractor is used to extract plain texts from a dump file of Wikipedia articles. The data files used for training are,