- 分词结果的向量化 (用utils中的函数Word2Vec , **或者直接用JSON文件中的word2idx+embeddings) - 分词后句子的seq_len,  mask_matrix的计算 - 根据是否使用GPU,Variable所有参数 - 计算并返回分类结果 -Quantization of word segmentation results (using the Word2Vec function in utilities, or directly using word2idx+embeddings in JSON files) -Calculation of seq_1en and mask_matrix for segmented sentences -Variable all parameters based on whether GPU is used or not -Calculate and return classification results 问题: Question: 回答: Answer: 预测结果: predictions: