soft attention/hard attention soft attention输出注意力分布的概率值,hard attention 输出onehot向量, soft的优势> hard 知识蒸馏(knowledge distill)和迁移学习 相关文章: 2021-08-03 2021-08-21 2021-05-03 2021-12-19 2022-12-23 2021-12-28 2021-07-09 2022-01-16