机器学习和深度学习-- 李宏毅(笔记与个人理解)Day22

Day 22 Transformer

seqence to seqence

有什么用呢?

image-20240419195409787

image-20240419200600208

image-20240419200618184

image-20240419200655869

image-20240419200714492

Encoder

image-20240419200811528

image-20240419200929890

image-20240419201816447

image-20240419201856942
how Block workimage-20240419202000619

仔细讲讲Residual 的过程?

重构

image-20240419202016784

image-20240419202055831

Decoder - AutoRegressive

image-20240419202814089

image-20240419203026310

image-20240419203044694

image-20240419203125676

Mask

image-20240419203147941

image-20240419203210758

image-20240419203323124

由于是文字接龙,所以无法考虑右边的 info

image-20240419203435098

image-20240419203445328

image-20240419203517384

image-20240419203607549

image-20240419203641245
另一种decoder

image-20240419204034186

Encoder to Decoder – Cross Attend

image-20240419204156788

image-20240419204337308

image-20240419204606629

怀疑begin那里没有做 Norm是bug

image-20240419204911428

image-20240419205003779

Training

image-20240419205022013

image-20240419205132661

很像分类的问题

image-20240419205223059

image-20240419205257761

Teacher Forcing : using the ground truth as input

Tips

image-20240419205454295

image-20240419205509670

image-20240419205543623

image-20240419205805744

how to resolve that?

image-20240419213753968
image-20240419214526024

image-20240419214840792

image-20240419215040489

image-20240419215130167

image-20240419215205131

image-20240419215259187

最近更新

  1. docker php8.1+nginx base 镜像 dockerfile 配置

    2024-04-28 21:08:03       94 阅读
  2. Could not load dynamic library ‘cudart64_100.dll‘

    2024-04-28 21:08:03       101 阅读
  3. 在Django里面运行非项目文件

    2024-04-28 21:08:03       82 阅读
  4. Python语言-面向对象

    2024-04-28 21:08:03       91 阅读

热门阅读

  1. 网络编程!

    2024-04-28 21:08:03       37 阅读
  2. git bash上传本地文件报错debug

    2024-04-28 21:08:03       33 阅读
  3. Scala Extention

    2024-04-28 21:08:03       22 阅读
  4. 使用 GORM 自定义类型:解决问题与技巧分享

    2024-04-28 21:08:03       28 阅读
  5. Mockito Mybatis-plus 单元测试

    2024-04-28 21:08:03       35 阅读
  6. Qt——置灰窗口

    2024-04-28 21:08:03       33 阅读
  7. 官网设计UI设计需要考虑哪些?

    2024-04-28 21:08:03       28 阅读
  8. RTCRTC

    2024-04-28 21:08:03       30 阅读
  9. Unity中的C#事件与回调:一个简单的实例分析

    2024-04-28 21:08:03       36 阅读
  10. Lua语法

    Lua语法

    2024-04-28 21:08:03      27 阅读
  11. Excel常用函数

    2024-04-28 21:08:03       32 阅读
  12. 发布自己的npm包

    2024-04-28 21:08:03       26 阅读