Нейронное текстуальное следствие

AbstractWe study the problem of building text classifiers with little or no training data, commonly known as zero and few-shot text classification. In recent years, an approach based on neural …...