2020年08月_城俊BLOG

原创 pytorch从dataloader中取数据（python从enumerate或iterator对象中取数据）

报错：TypeError: 'enumerate' object is not subscriptableTypeError: 'DataLoader' object is not an iterator代码：_, (inputs2, labels2) = next(enumerate(test2))_, (inputs3, labels3) = next(enumerate(test3))iter应该同理。next(iter(test3))enumerate还可以加索引...

2020-08-31 16:08:21 9690 5

翻译 vs 2017 release无法调试

第三步:工具–>选项—>调试---->常规–>禁用"启动仅我的代码"https://blog.csdn.net/chinmusam/article/details/7879140

2020-08-31 14:50:47 1147

原创 python找不到包，ModuleNotFoundError: No module named

明明添加了 init.py文件了，还是找不到检查本地代码是否已经同步到了服务器端加入路径。此方法不需要什么 init.py文件目录结构：__project |__ dira |__a.py # a引用b |__dirb |__b.py # 定义了fun1() c.py # c引用d d.py # 定义了fun2()代码：a.py# a.py要引用dirb文件夹中的b.py中定义的的fun1()函数sys.path.append(r'../../')from dirb

2020-08-26 18:36:31 2579

翻译 python lambda函数 filter函数

老是记不住lambda 参数：表达式>>> c = lambda x:x**3>>> c(2)8

2020-08-26 17:28:45 234

翻译 linux ubuntu bash shell报错 xxx.sh: Syntax error: end of file unexpected (expecting “then“)

代码：# creat log dir if it not existslogDir="./ckpt/"$timeDirif [ ! -d $logDir ]; then mkdir $logDirelse echo $logDir existfi报错：$ sh run.sh: not foundrun.sh:run.sh: 16: run.sh: Syntax error: end of file unexpected (expecting "then")解决：需要转换脚

2020-08-26 14:56:20 1145

原创 windows win10 pycharm安装bash support插件

在pycharm软件内部商店搜不到只能手动安装下载：https://plugins.jetbrains.com/plugin/4230-bashsupport/versions/stable安装：file - settings- Plugins - 齿轮- install plugin from disk… - 选择下载的zip文件 - restart IDE

2020-08-26 13:39:42 1741

原创各种配置文件语法扫盲：YAML

不能用tab缩进，只能用空格多层字典嵌套：tasknum: # 注意要使用空格进行缩进 '1': # 多个小横杠 - 实现一个列表 - bitsPerAttr: 40实现的是：'tasknum': {'1': [{'bitsPerAttr': 40}],

2020-08-26 11:44:53 326

原创 tensorflow keras InvalidArgumentError: Default MaxPoolingOp only supports NHWC on device type CPU

tf.version ‘2.3.0’keras.version ‘2.4.3’tensoflow代码：out = model.predict(x_test)out = np.array(out)报错：---------------------------------------------------------------------------InvalidArgumentError Traceback (most recent call las

2020-08-21 15:16:32 8018 1

翻译 mac book air自动重启／自动注销

一会不用它就变得很干净了，像重启过长时间的网页连接也会断掉，很麻烦其实是自动注销了。在安全与隐私中重新设置：系统偏好设置->安全性与隐私->点击小锁->右下角高级按钮 -> 设置自动注销时间https://www.ancii.com/abvzgmub7/...

2020-08-20 15:35:18 604

原创 linux ubuntu scp Host key verification failed

一开始我(user2)在 456.xxx.xxx.xx 机器上执行命令：$ sudo scp -r user1@123.xxx.xxx.xx:/home/user1/folder1 user2@456.xxx.xxx.xx:/home/user2/folder2/想把123.xxx.xxx.xx 机器上的文件夹folder1拷贝到 456.xxx.xxx.xx 机器的folder2下面，但是报错：user1@123.xxx.xxx.xx's password:Host key verificati

2020-08-18 18:29:24 331

原创 python读取图片 PIL, matplotlib(plt.imshow) 、cv2.imread和skimage.imread 打开和显示图片的区别（shape，通道）

区别：PIL：（宽，高）cv2：（高，宽）原图：1920x1080Image.open(imgPath)：(1920, 1080)cv2.imread(imgPath,0)：(1080, 1920)

2020-08-17 22:48:55 4861

原创 mxnet sklearn ValueError: Input contains NaN, infinity or a value too large for dtype(‘float64‘).

Traceback (most recent call last): File "train_0723.py", line 455, in <module> main() File "train_0723.py", line 451, in main train_net(args) File "train_0723.py", line 445, in train_net epoch_end_callback=epoch_cb) File "/home/us

2020-08-17 11:39:41 560

原创 linux ubuntu apt-get报错E: Unable to correct problems, you have held broken packages.

报错内容：$ sudo apt-get install lm-sensors sensors-appletReading package lists... DoneBuilding dependency treeReading state information... Donelm-sensors is already the newest version (1:3.3.4-2ubuntu1).Some packages could not be installed. This may mean

2020-08-17 11:25:23 2433

原创 mxnet训练卡死重启后依然无法解决

3480 root 20 0 0 0 0 R 100.0 0.0 11:53.46 UVM GPU1 BH1724 root -51 0 0 0 0 R 100.0 0.0 11:59.19 irq/88-nvidia

2020-08-17 11:15:13 1157

原创 pytorch RuntimeError: expected backend CUDA and dtype Float but got backend CPU and dtype Float

代码：criterion = nn.BCEWithLogitsLoss(reduction='none')loss = criterion(output, target)loss.mul_(weights)报错：Traceback (most recent call last):File “/home/user1/main_cs_0708.py”, line 391, in main()File “/home/user1/main_cs_0708.py”, line 301, in mai

2020-08-14 17:46:08 2367

翻译 windows win10调节外接显示器亮度

安装显卡驱动，在显卡应用程序里面调如英特尔的显卡桌面应用程序https://www.cnblogs.com/anyefrozen/p/11275485.html

2020-08-13 15:44:25 7147

翻译 pytorch tensor操作

1，增加维度>>> a = torch.tensor([222])>>> atensor([222])>>> a.shapetorch.Size([1])# 变成1列>>> c = torch.unsqueeze(a,1) # torch.unsqueeze(a,0) 是变成一行>>> c.shapetorch.Size([1, 1])

2020-08-12 16:37:22 946

原创 pytorch网络结构可视化graphviz.backend.ExecutableNotFound: failed to execute [‘dot‘, ‘-Tpdf‘, ‘-O‘, ‘

报错：Traceback (most recent call last): File "/data/user1/pkgs/conda/envs/drc/lib/python3.7/site-packages/graphviz/backend.py", line 129, in render subprocess.check_call(args, stderr=stderr, **POPEN_KWARGS) File "/data/user1/pkgs/conda/envs/drc/lib/

2020-08-12 15:14:51 1466

原创 linux ubuntu 升级sklearn 0.19

sklearn 0.19各种报错还不能升级：报错1（使用时报错）：ImportError: cannot import name 'multilabel_confusion_matrix'报错2（升级时报错）：$ pip install git+http://github.com/scikit-learn/scikit-learn.gitLooking in indexes: https://mirrors.aliyun.com/pypi/simple/, https://pypi.tuna.t

2020-08-11 23:57:26 708

原创修改conda某个虚拟环境的python解释器interpreter

# 进入环境$ conda activate your_env# 查看特定环境的环境变量设置$ conda env config # 修改某个环境变量$ conda env config vars set CONDA_PYTHON_EXE=/home/data/pkgs/miniconda3/bin/python# 使变量生效$ conda activate your_envhttps://conda.io/projects/conda/en/latest/user-guide/tasks

2020-08-10 22:20:24 3920

翻译 linux ubuntu bash shell报错 xxx.sh: Syntax error: “(“ unexpected

原因可能是你使用了sh而不是bash去运行你的脚本，sh没有数组定义，而bash有bash脚本的开头应该为：#!/bin/bash运行命令为：$ sudo bash xxx.shsh脚本开头没有限制。运行命令为：$ sudo sh xxx.shhttps://unix.stackexchange.com/questions/253892/syntax-error-unexpected-when-creating-an-array...

2020-08-10 17:23:08 470

原创 windows win10装机软件清单

播放器：MPV图片浏览器：IrfanView

2020-08-09 20:17:33 512

原创 MxNet base.h:459: Check failed: e == cudaSuccess (30 vs. 0) : CUDA: unknown error nvidia-smi显卡ERR！

命令输出缓慢，并且报ERR1，重启机器2，重装驱动3，维修或换显卡

2020-08-09 11:33:27 1242

翻译 UserWarning: Matplotlib is currently using agg, which is a non-GUI backend, so cannot show the figur

画图出不来UserWarning: Matplotlib is currently using agg, which is a non-GUI backend, so cannot show the figure.解决：import matplotlib.pyplot as pltplt.switch_backend('TkAgg')okie@https://stackoverflow.com/questions/56656777/userwarning-matplotlib-is-curre

2020-08-07 11:37:53 3406

原创 linux ubuntu 查找文件、文件夹、包含指定内容的文件 grep / locate / find

find 命令$ sudo find /home/ -name 'v2'查找：/home/xxx/data/86/images/v2/act-5/002822.jpg所在的位置:$ locate v2/act-5

2020-08-06 11:26:23 3326

原创 pytorch 每次测试结果不同

真是见了鬼了，搞好半天没弄清楚啥原因。把每个预测值看了一遍，确实每次结果不同。原因排查：自定义的Metric写的有问题，会根据batch_size变化，是按batch_size大小按每个batch分别计算的，而不是按batch累计的被测试的网络结构中有Dropout层，比如nn.Dropout测试的数据loader加入了随机处理，比如transforms.RandomCrop()其他可能的原因，目前我还没遇到。感谢：https://blog.csdn.net/t20134297/artic

2020-08-04 16:58:43 5926

翻译代码差异比较 diff pycharm sublimetext

好用的代码比较工具：pycharmsublimetext安装：https://www.cnblogs.com/olive27/p/6038144.html用法：https://blog.csdn.net/a0405221/article/details/78769672https://www.diffchecker.com/ ？

2020-08-04 16:31:34 1311

翻译 linux ubuntu vi 编辑器按键异常

可能需要配置一下：https://blog.csdn.net/harbor1981/article/details/82428409

2020-08-03 17:18:07 368

翻译 IPython Jupyter Notebook方向键异常

方向键变成数字，试试关闭numlock锁定https://blog.csdn.net/Du_Shuang/article/details/77716611

2020-08-03 17:17:13 919

翻译 linux ubuntu终端terminal禁止在输出时自动滚动

在终端运行程序，如果程序不停地向终端输出内容，就没办法向上滚动查看之前的输出。这个可以通过以下步骤设置解决：terminal->右键->Profiles->Profile Preferences->Scrolling->取消选中 Scroll on output小技巧。感谢：https://qastack.cn/superuser/541106/how-to-scroll-up-in-terminal-window-while-new-input-is-generated

2020-08-03 14:15:04 2211

原创 python numpy二进制存取

根据数据特点，调整存储占用的空间大小。因为数组中全是0和1，所以用单字节（np.int8）存储。如果是浮点或其他格式，可以用其他类型。# 存>>> f = open(c, 'ab+')>>>> d1 = np.array([0,0,1], dtype=np.int8)>>> f.write(d1)3>>> d2 = np.array([0,1,1], dtype=np.int8)>>> f.wri

2020-08-03 14:05:04 1753

原创 mxnet mshadow/./stream_gpu-inl.h:62: Check failed: e == cudaSuccess: CUDA: an illegal memory access

报错：Traceback (most recent call last): File "train_0723.py", line 455, in <module> main() File "train_0723.py", line 451, in main train_net(args) File "train_0723.py", line 445, in train_net epoch_end_callback=epoch_cb) File "/hom

2020-08-03 09:56:50 754

翻译 linux ubuntu查看执行的命令历史记录 history

$ history设置显示的行数：$ sudo nano /etc/profule加入HISTTIMEFORMAT="%F %T `whoami` " # prefix of outputHISTSIZE=5000 # lines num in output of history commandHISTFILESIZE=5000 # lines num in .bash_history file# 生效$ source /etc/profile...

2020-08-02 12:13:11 4739

翻译 sudo dpkg 报错 E: Could not get lock /var/lib/dpkg/lock-frontend

报错：E: Could not get lock /var/lib/dpkg/lock-frontend. It is held by process 8888 (aptd) - open (11: Resource temporarily unavailable)N: Be aware that removing the lock file is not a solution and may break your system.E: Unable to acquire the dpkg fronte

2020-08-02 11:42:31 2655

原创 Linux ubuntu18.04使用ubuntu-drivers autoinstall安装nvidia驱动后紫屏黑屏卡在主板logo无法进系统

ubuntu 18.04 GTX 1080TI 双显卡主要是显卡驱动的问题，解决过程（仅供参考）：先尝试的安装450$ ubuntu-drivers devices$ sudo add-apt-repository ppa:graphics-drivers/ppa$ sudo apt-get update$ sudo apt update && sudo apt -f install && sudo apt full-upgrade注意几个问题：1，先解

2020-08-02 11:39:03 9394 1

Hello Word!