深度学习入门

1.精通Tensorflow、Pytorch等深度学习框架。

(1)安装ubuntu + anconda + jupter Notebook等工具。

https://blog.csdn.net/White__Hacker/article/details/81066971?utm_medium=distribute.pc_relevant.none-task-blog-BlogCommendFromMachineLearnPai2-1.channel_param&depth_1-utm_source=distribute.pc_relevant.none-task-blog-BlogCommendFromMachineLearnPai2-1.channel_param

https://zhuanlan.zhihu.com/p/69799707

https://www.linuxidc.com/Linux/2018-01/150457.htm

pytorch离线安装：
https://blog.csdn.net/qq_36622009/article/details/104382740?utm_medium=distribute.pc_relevant.none-task-blog-BlogCommendFromMachineLearnPai2-1.channel_param&depth_1-utm_source=distribute.pc_relevant.none-task-blog-BlogCommendFromMachineLearnPai2-1.channel_param

(2)cuda的安装配置。

https://blog.csdn.net/wf19930209/article/details/81879514

(3)熟悉Pytorch基本用法。

Pytorch官网:
autograd
GPU
backward
torch.randn与numpy.random.randn
pytorch动态计算图与tesorflow静态图比较
nn
optim
控制流+权重共享

yeild()函数：https://www.cnblogs.com/BigFishFly/p/6337081.html
tuple vs list：https://www.cnblogs.com/still-smile/p/11586452.html

Torchtext
https://zhuanlan.zhihu.com/p/65833208
https://www.jianshu.com/p/71176275fdc5

einsum()函数
https://blog.csdn.net/Eric_1993/article/details/105670381
tuple()函数
https://www.cnblogs.com/still-smile/p/11586452.html
unsqueeze()函数：
https://blog.csdn.net/flysky_jay/article/details/81607289
DataLoader()函数
https://www.cnblogs.com/ranjiewen/p/10128046.html

embeding()函数
https://www.cnblogs.com/lindaxin/p/7991436.html

模型调优：
conv用法：https://www.cnblogs.com/expttt/p/12397330.html
模型参数申明 conv = torch.nn.Conv2d(1,8,(2,3),(1,1))
模型调用 x = torch.randn(2,1,7,3)  conv(x)

loss函数选择：BCELoss vs CrossEntropyLoss
https://blog.csdn.net/rosefun96/article/details/88058708

优化函数选择：Adam
https://www.cnblogs.com/dylancao/p/9878978.html

flatten()函数：https://blog.csdn.net/GhostintheCode/article/details/102530451

Linear()函数：https://www.cnblogs.com/Archer-Fang/p/10645473.html

argmax()函数：https://blog.csdn.net/weixin_43869268/article/details/107624108

view()函数：https://www.jianshu.com/p/56402c641661

求二分类标签结果：pred.data.max(1)[1]


2.熟悉典型深度学习模型，使用场景及方法。



3.熟悉分词、命名实体识别、文本分类、NLU。