随笔分类 -  深度学习

Transformer代码细节
摘要：###优化措施 作者采用warmup学习率，先线性增长学习率，随后指数缓慢减少学习率 class ScheduledOptim(): '''A simple wrapper class for learning rate scheduling''' def __init__(self, optimi 阅读全文
posted @ 2020-10-09 13:16 blueattack 阅读(559) 评论(0) 推荐(0)
理解LSTM在keras API中参数return_sequences和return_state
摘要：https://blog.csdn.net/u011327333/article/details/78501054 阅读全文
posted @ 2020-05-20 23:07 blueattack 阅读(328) 评论(0) 推荐(0)