Knowledge distillation is model compression method in which a small model is trained to mimic a pre-trained, larger model (or ensemble of models).
確定! 回上一頁