BERT models fine-tuned with the garlic-specific dataset showed better performance. BERTweet models achieved accuracy of 0.897–0.911, while BERTBASE and ...
確定! 回上一頁