You can easily train BERT from scratch both on MLM & NSP tasks using combination of BertForPretraining TextDatasetForNextSentencePrediction ...
確定! 回上一頁