An Extended Text Combination Classification Model for Short Video Based on Albert

Figure 2

Albert model believes that [CLS] vector can represent the semantics of the whole sentence.