<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[qiongyugaohan的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/qiongyugaohan</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; qiongyugaohan]]></copyright><item><title><![CDATA[关于pytorch、torch_geometric安装 22.12.25]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/128438850</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/128438850</guid><author>qiongyugaohan</author><pubDate>Sun, 25 Dec 2022 22:18:42 +0800</pubDate><description><![CDATA[cuda 9.2 + torch_geometric 1.61 + pytorch 1.6.0 + python3.8，torch_geometric报错；再cuda 10.2 + torch_geometric 1.61 + pytorch 1.6.0 + python3.8，可以了。装了之后，显卡驱动自动更新了？然后显示可以装CUDA11.7版本一下；系统坏了，重装系统，一开始以为电脑只能装cuda9.2版本一下；卸载CUDA，重新安装半天，终于装上CUDA 10.2,]]></description><category></category></item><item><title><![CDATA[plot-python]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/122184228</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/122184228</guid><author>qiongyugaohan</author><pubDate>Tue, 28 Dec 2021 05:43:02 +0800</pubDate><description><![CDATA[# fig, ax2 = plt.subplots(figsize=(11,6))
    # ax2.scatter(x0[:,0],x0[:,1])
    # ax2.scatter(x1[:,0],x1[:,1])
    # ax2.plot(x_line,y_line,next(colors), ls='--', lw=1) 
    # ax2.plot(x_line_simu,y_line_simu,next(colors), ls='--...]]></description><category></category></item><item><title><![CDATA[2021-05-26]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/117279429</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/117279429</guid><author>qiongyugaohan</author><pubDate>Wed, 26 May 2021 09:11:24 +0800</pubDate><description><![CDATA[装pytorch-geometric目前需要CUDA11.1以下？我装了个11.2各种报错。于是在笔记本上的CPU上装了。
]]></description><category></category></item><item><title><![CDATA[MATLAB: save 高精度csv文件]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/114818098</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/114818098</guid><author>qiongyugaohan</author><pubDate>Mon, 15 Mar 2021 08:29:06 +0800</pubDate><description><![CDATA[fjing = [];
for i = 1: size(data_fenzu_new_x,2)-1
  fjing = [fjing, '%.8f,'];
end
fjing = [fjing, '%.8f\n'];

fid=fopen('data_fenzu_new_x.csv','w')  % 16th_data_chazhi3.csv 为固定精度的数据
% fid=fopen('16th_data_chazhi2.txt','w')
for i=1:size(data_fenzu_new_x...]]></description><category></category></item><item><title><![CDATA[保存数据]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/103676995</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/103676995</guid><author>qiongyugaohan</author><pubDate>Tue, 24 Dec 2019 09:06:51 +0800</pubDate><description><![CDATA[Python -&gt; CSV
from pandas import DataFrame
import os
xishu_all = np.array(xishu_all)
domain_result = r'E:\data'

save = DataFrame(xishu_all)
info_error = 'xishu_all.csv'    #  reshape
info = os.pat...]]></description><category></category></item><item><title><![CDATA[python代码]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/103668760</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/103668760</guid><author>qiongyugaohan</author><pubDate>Mon, 23 Dec 2019 16:59:53 +0800</pubDate><description><![CDATA[打乱数据顺序

# 打乱
index_train = [i for i in range(len(image_train))]  
np.random.shuffle(index_train)   
# 1 train
data = image_train
label = label_train    
data = data[index_train]
label = label[index_t...]]></description><category></category></item><item><title><![CDATA[笔记]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/103257740</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/103257740</guid><author>qiongyugaohan</author><pubDate>Tue, 26 Nov 2019 15:40:51 +0800</pubDate><description><![CDATA[异常点检测：https://blog.youkuaiyun.com/yangziluomu/article/details/73528080
]]></description><category></category></item><item><title><![CDATA[论文LaTeX制表]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/93850738</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/93850738</guid><author>qiongyugaohan</author><pubDate>Thu, 27 Jun 2019 10:14:35 +0800</pubDate><description><![CDATA[先加载：
\usepackage{threeparttable}

然后是代码：
\begin{center}
\begin{threeparttable}
\caption{The Skewing Angles ($\beta$) for $\fam0 Mu(H)+X_2$ and
   $\fam0 Mu(H)+HX$~\tnote{a}}
\begin{tabular}{rlcc}
\hli...]]></description><category></category></item><item><title><![CDATA[论文画图——eps格式的图]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/93708915</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/93708915</guid><author>qiongyugaohan</author><pubDate>Wed, 26 Jun 2019 08:50:34 +0800</pubDate><description><![CDATA[用Photoshop保存的图太大的，

比较好的方法是：



如果你有Adobe professional版本，那么恭喜这将是最简单的一种方法。

首先，把你想要转成eps格式的PPT示意图所在的那一页，转化成pdf格式。

ppt转pdf的方式有很多，有的powerpoint可以直接另存为ppt格式，也可以借助诸如tinyPDF的辅助工具。

然后用Adobe Pro打开它，并使用TOOLS...]]></description><category></category></item><item><title><![CDATA[代码记录]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/92407443</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/92407443</guid><author>qiongyugaohan</author><pubDate>Sun, 16 Jun 2019 16:47:19 +0800</pubDate><description><![CDATA[Python3找出List中最大/最小的N个数及索引
# -*- coding: utf-8 -*-
import heapq
nums = [1, 8, 2, 23, 7, -4, 18, 23, 24, 37, 2]
# 最大的3个数的索引
max_num_index_list = map(nums.index, heapq.nlargest(3, nums))
# 最大的3个数的索引
max...]]></description><category></category></item><item><title><![CDATA[论文画图+MATLAB保存高精度图片]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/92406362</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/92406362</guid><author>qiongyugaohan</author><pubDate>Sun, 16 Jun 2019 16:35:18 +0800</pubDate><description><![CDATA[print('-djpeg','-r1000','bon')



上面是直接保存。

也可以借助Visio保存，PPT是不行，PPT另存为后dpi很低，只有150.

而Visio可以自己设置，Visio导出jpg文件，设置为1000，然后再利用Photoshop（Alt+i+i），重新调整到需要的大小（Alt+F+A）。



有一个问题：Visio可能不太方便画图，也可以在PPT中画好之...]]></description><category></category></item><item><title><![CDATA[tensorboard的使用]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/89958359</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/89958359</guid><author>qiongyugaohan</author><pubDate>Wed, 08 May 2019 16:06:33 +0800</pubDate><description><![CDATA[activate tensorflow

tensorboard --logdir= 保存路径
]]></description><category></category></item><item><title><![CDATA[Windows10+CUDA+Tensorflow-gpu安装。装了一个星期，啥问题都见过了。（eg: ImportError: DLL load failed）]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/89516476</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/89516476</guid><author>qiongyugaohan</author><pubDate>Thu, 25 Apr 2019 15:16:21 +0800</pubDate><description><![CDATA[本次tensorflow-gpu安装历时整整一个星期，心中一万只。。。飘过。今天终于安装好了。在这期间各种问题都见过，关键是各种问题网上一搜都有解决方案？！这就让我一步步入了坑。其实出发点就是错的，我的问题就是CUDA版本的问题！要求9.1以上，我最开始就用了最新的10.1，不行，然后又试了10.0，还是不行，又试了9.0，更不行！由于10.0没有什么明显的报错，就在10.0的路上越走越远！最后，...]]></description><category></category></item><item><title><![CDATA[anaconda中如何安装tflearn?]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/89424735</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/89424735</guid><author>qiongyugaohan</author><pubDate>Sat, 20 Apr 2019 22:17:37 +0800</pubDate><description><![CDATA[1、 打开anaconda prompt

2、 activate tensorflow(环境名)

3、 pip install tflearn

4、 如何安装tflearn
]]></description><category></category></item><item><title><![CDATA[关于卷积网络的几个误解 及 参数计算]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/89005149</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/89005149</guid><author>qiongyugaohan</author><pubDate>Wed, 03 Apr 2019 21:18:09 +0800</pubDate><description><![CDATA[1、之前误解一：

多通道的卷积和我之前理解的不一样。只是对应通道相乘，最后再加？那我之前的残差网络具体卷积核的大小及深度是多少？？？

2、之前误解二：

不论输入图像大小参数个数是不会发生改变的！！！（但计算量会改变？？？）



3、卷积网络优化的是卷积核的参数？



卷积不仅限于对原始输入的卷积。蓝色方块是在原始输入上进行卷积操作，使用了6个filter得到了6个提取特征图。绿色方块还能...]]></description><category></category></item><item><title><![CDATA[训练模型的重新加载]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/89005074</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/89005074</guid><author>qiongyugaohan</author><pubDate>Wed, 03 Apr 2019 21:12:37 +0800</pubDate><description><![CDATA[为了使训练模型能够重新加载，并重新使用，在定义tensor时，一定要加name, 这样在重新调用时，输入类似‘training_flag:0’就可以调用。

刚开始定义的：



后面调用的：






...]]></description><category></category></item><item><title><![CDATA[如何安装Python包？]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/88880410</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/88880410</guid><author>qiongyugaohan</author><pubDate>Thu, 28 Mar 2019 22:44:29 +0800</pubDate><description><![CDATA[命令行中安装，和anaconda中手动安装有什么区别吗？
]]></description><category></category></item><item><title><![CDATA[anaconda 中安装 pytorch]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/84071738</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/84071738</guid><author>qiongyugaohan</author><pubDate>Wed, 14 Nov 2018 18:42:29 +0800</pubDate><description><![CDATA[https://www.cnblogs.com/hz-blog/p/8512148.html

按照这个步骤安装。

其中第2部分CUDA安装，在anaconda中直接搜索找到cudatoolkit包，9.0版本。

测试：import torch

 

其他：知乎上介绍到：安装之后，也千万要注意，要在主代码的最外层包上：if __name__ == '__main__':...]]></description><category></category></item><item><title><![CDATA[ImportError: No module named 'sklearn' ？]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/82995438</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/82995438</guid><author>qiongyugaohan</author><pubDate>Wed, 10 Oct 2018 13:01:30 +0800</pubDate><description><![CDATA[原来是需要安装：scikit-learn包！！！之前一直在找sklearn这个包。。。]]></description><category></category></item><item><title><![CDATA[软件清单]]></title><link>https://blog.youkuaiyun.com/qiongyugaohan/article/details/82733894</link><guid>https://blog.youkuaiyun.com/qiongyugaohan/article/details/82733894</guid><author>qiongyugaohan</author><pubDate>Mon, 17 Sep 2018 10:53:13 +0800</pubDate><description><![CDATA[QQ
微信
everything
驱动精灵
u_torrent
邮箱
chrome

notepad++

爱奇艺万能播放器

网易云音乐

百度网盘



搜狗输入法

office
PDF
anaconda
visual studio
MATLAB
解压
]]></description><category></category></item></channel></rss>