Bert查询记录和caffe相关

相关论文与实现:

https://github.com/google-research/bert

https://arxiv.org/pdf/1810.04805.pdf

http://jalammar.github.io/illustrated-transformer/

https://arxiv.org/pdf/1907.00151.pdf

附件是caffe  cpu版本：

来自https://my.oschina.net/kalnkaya/blog/1824298

测试的代码来自这篇博客:

https://christopher5106.github.io/deep/learning/2015/09/04/Deep-learning-tutorial-on-Caffe-Technology.html

在公司电脑上测试时一直报python35.dll版本不匹配冲突;---这是由于我在命令行中直接输入的python文件名，然后调用的是系统的默认打开python的程序与我bat中设置的python路径不同，需要使用python 文件形式启动;

回家试了试:

pip切个国内源:

https://blog.csdn.net/sinat_21591675/article/details/82770360

这个prebuilt版的caffe是64位的需要下载对应的64位的:

https://mirrors.huaweicloud.com/python/3.5.2/

在公司我是按照上面设置的环境变量，有个io.py冲突,用改名的方法解决了,在家没出现这个问题;

可能是我在家用的是bat预准备环境:

bat如下:

REM C:\Python27\;C:\Python27\Scripts;C:\Python27-x64;
SET PYTHONPATH=E:\dev\AI\caffe\python
REM SET PYTHONHOME=D:\Python\Python36
REM SET PYTHONHOME=E:\dev\AI\python-3.5.2\PCbuild\win32
REM SET PATH=E:\dev\AI\python-3.5.2\PCbuild\win32;{other path}
REM SET PATH=D:\Python\Python36;D:\Python\Python36\Scripts;{other path}
SET PYTHONHOME=D:\Python\Python35
SET PATH=D:\Python\Python35;D:\Python\Python35\Scripts;{other path}

REM python -m pip install numpy
REM python -m pip install --upgrade pip
REM python -m pip install matplotlib==3.0
REM python -m pip install pillow
REM python -m pip install scikit-image==0.15
REM python -m pip install numpy scipy
REM python -m pip install protobuf

tknter问题解决参考:

https://stackoverflow.com/questions/36327134/matplotlib-error-no-module-named-tkinter 

其他:

backpropagation-algorithm

Backpropagation-NeuralNetwork

一文弄懂神经网络中的反向传播法——BackPropagation

如何直观地解释 backpropagation 算法？

反向传播算法（过程及公式推导）

神经网络中误差反向传播(back propagation)算法的工作原理

caffe

Windows下VS2015编译caffe(CPU ONLY)

数据挖掘系列（10）——卷积神经网络算法的一个实现

BP神经网络计算过程详解，用笔手算一遍弄懂反向传播

神经网络--反向传播（手算）

Disable same origin policy in Chrome

wiki engine

https://blog.csdn.net/climaxwind/article/details/52930738

https://blog.csdn.net/iamzhangzhuping/article/details/49993899

/*----------------------------------分割线-------------------------------------------------*/

caffe

https://yufeigan.github.io/

https://blog.csdn.net/mounty_fsc/article/details/51085654

https://blog.csdn.net/zsf211/article/details/79690082

https://www.cnblogs.com/empty16/p/4878164.html

http://ask.ainoob.cn/article/9141

https://cloud.tencent.com/developer/article/1100814

https://blog.csdn.net/elysion122/article/details/64151247

https://github.com/ZhaofanQiu/caffe-windows-3dConvNet

https://github.com/Ewenwan/MVision/blob/master/CNN/Action_Recognition/CDC/readme.md

http://zgyxwlxzz.paperopen.com/en/OA/pdfdow.aspx%3FSid%3D201804085

http://element-ui.cn/news_show_5142.shtml

pytorch

https://github.com/yunjey/pytorch-tutorial/tree/master/tutorials/03-advanced/image_captioning

tensorflow

https://github.com/yunjey/davian-tensorflow

/*--------------------------------------------------------------------------------------------------------------*/

caffe可视化

http://ethereon.github.io/netscope/#/editor

https://blog.csdn.net/u012915263/article/details/60579429

https://juejin.im/post/5ae2a7b3f265da0b9f40011a

https://www.cnblogs.com/carle-09/p/9086793.html

https://cloud.tencent.com/developer/article/1065135

/*****************************************************************************************************************/

 读懂caffe

https://blog.csdn.net/u012526120/article/details/49471485

https://github.com/Stephenfang51/Caffe_structure

https://www.thinbug.com/q/39742486

https://stackoom.com/question/3Gmwx/Caffe-prototxt%25E7%259A%2584%25E8%25AF%25AD%25E6%25B3%2595%25E6%258C%2587%25E5%258D%2597%25E5%2592%258C%25E7%2589%2588%25E6%259C%25AC%25E6%258E%25A7%25E5%2588%25B6

https://www.cnblogs.com/xiaochouk/p/10107726.html

https://www.jiqizhixin.com/articles/2019-12-19-15