1: The label smoothing technology has achieved good results in deep learning models in many fields such as picture classification, speech recognition, and machine translation.
2: The role of label smoothing:
improve the model's generalization ability, prevent the model from overfitting, improve the model's correction ability, and improve the model's cluster search ability. However, label smoothing in the teacher model will reduce the knowledge distillation effect on the student model.
