重写、补充、完善这段话不要出现第一人称我们因为这是在写论文。为了验证模型的分类性能本文将BERT预训练模型在THUNCNews数据集上进行微调经过若干轮的训练得到了一个在THUNCNews数据集上表现优异的BERT分类模型。具体来说本文计算了模型在每个新闻类别上的分类精确率、召回率和F1值并将结果汇总在表2中。

In order to evaluate the classification performance of the model, this study fine-tuned the BERT pre-trained model on the THUNCNews dataset. After several rounds of training, an excellent BERT classification model on the THUNCNews dataset was obtained. Specifically, this study calculated the classification precision, recall, and F1 value of the model on each news category, and the results were summarized in Table 2.