安装torch
直接运行conda/pip install torch安装不成功，参考https://www.cnblogs.com/andrew-address/p/12733669.html，pip install torch1.4.0+cpu torchvision0.5.0+cpu -f https://download.pytorch.org/whl/torch_stable.htmlpip install torch==1.4.0+cpu torchvision==0.5.0+cpu -f https://download.pytorch.org/whl/torch_stable.html

使用该命令即可安装成功。

**
安装Torch7，按照官网操作，运行./install.sh时报错：
**

参考：http://www.pianshen.com/article/5473287386/
2 errors detected in the compilation of "/tmp/tmpxft_00001b24_00000000-6_THCTensorMath.cpp1.ii".
CMake Error at THC_generated_THCTensorMath.cu.o.cmake:267 (message):
  Error generating file
  /home/itl/Torch7/distro/extra/cutorch/build/lib/THC/CMakeFiles/THC.dir//./THC_generated_THCTensorMath.cu.o


lib/THC/CMakeFiles/THC.dir/build.make:112: recipe for target 'lib/THC/CMakeFiles/THC.dir/THC_generated_THCTensorMath.cu.o' failed
make[2]: *** [lib/THC/CMakeFiles/THC.dir/THC_generated_THCTensorMath.cu.o] Error 1
CMakeFiles/Makefile2:172: recipe for target 'lib/THC/CMakeFiles/THC.dir/all' failed
make[1]: *** [lib/THC/CMakeFiles/THC.dir/all] Error 2
Makefile:127: recipe for target 'all' failed
make: *** [all] Error 2

解决方案
当前版本cuda与torch不兼容，设置一下环境变量
export TORCH_NVCC_FLAGS="-D__CUDA_NO_HALF_OPERATORS__"

然后执行
./clean.sh   #删除之前的安装文件

再重新安装
./install.sh
source ~/.bashrc  #使环境变量生效




检查
th

Python安装torch模块报错处理
前言
安装
报错
解决
查找文档
解决方案
安装 torch
等待安装完成
安装 torchvision
提示安装完成
总结
前言
因python项目上线，需要在海外服务器上安装指定torch模块
安装
根据官网https://pytorch.org/安装提示，执行pip3 install即可。
pip3 install torch torchvision
但安装过程中发生报错，尝试多次执行pip3 install torch torchvision，仍会重置下载。
报错
$ pip3 install torch
WARNING: pip is being invoked by an old script wrapper. This will fail in a future version of pip.
Please see https://github.com/pypa/pip/issues/5599 for advice on fixing the underlying issue.
To avoid this problem you can invoke Python with '-m pip' instead of running pip directly.
Looking in indexes: http://ap-south-1.mirrors.cloud.aliyuncs.com/pypi/simple/
Collecting torch
Downloading http://ap-south-1.mirrors.cloud.aliyuncs.com/pypi/packages/62/01/457b49d790b6c4b9720e6f9dbbb617692f6ce8afdaadf425c055c41a7416/torch-1.5.1-cp36-cp36m-manylinux1_x86_64.whl (753.2 MB)
|████████▊                       | 206.4 MB 48 kB/s eta 3:06:38ERROR: Exception:
Traceback (most recent call last):
File "/data/python3/lib/python3.6/site-packages/pip/_vendor/urllib3/response.py", line 425, in _error_catcher
yield
File "/data/python3/lib/python3.6/site-packages/pip/_vendor/urllib3/response.py", line 507, in read
data = self._fp.read(amt) if not fp_closed else b""
File "/data/python3/lib/python3.6/http/client.py", line 449, in read
n = self.readinto(b)
File "/data/python3/lib/python3.6/http/client.py", line 493, in readinto
n = self.fp.readinto(b)
File "/data/python3/lib/python3.6/socket.py", line 586, in readinto
return self._sock.recv_into(b)
socket.timeout: timed out
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/data/python3/lib/python3.6/site-packages/pip/_internal/cli/base_command.py", line 188, in _main
status = self.run(options, args)
File "/data/python3/lib/python3.6/site-packages/pip/_internal/cli/req_command.py", line 185, in wrapper
return func(self, options, args)
File "/data/python3/lib/python3.6/site-packages/pip/_internal/commands/install.py", line 333, in run
reqs, check_supported_wheels=not options.target_dir
File "/data/python3/lib/python3.6/site-packages/pip/_internal/resolution/legacy/resolver.py", line 179, in resolve
discovered_reqs.extend(self._resolve_one(requirement_set, req))
File "/data/python3/lib/python3.6/site-packages/pip/_internal/resolution/legacy/resolver.py", line 362, in _resolve_one
abstract_dist = self._get_abstract_dist_for(req_to_install)
File "/data/python3/lib/python3.6/site-packages/pip/_internal/resolution/legacy/resolver.py", line 314, in _get_abstract_dist_for
abstract_dist = self.preparer.prepare_linked_requirement(req)
File "/data/python3/lib/python3.6/site-packages/pip/_internal/operations/prepare.py", line 469, in prepare_linked_requirement
hashes=hashes,
File "/data/python3/lib/python3.6/site-packages/pip/_internal/operations/prepare.py", line 259, in unpack_url
hashes=hashes,
File "/data/python3/lib/python3.6/site-packages/pip/_internal/operations/prepare.py", line 130, in get_http_url
link, downloader, temp_dir.path, hashes
File "/data/python3/lib/python3.6/site-packages/pip/_internal/operations/prepare.py", line 281, in _download_http_url
for chunk in download.chunks:
File "/data/python3/lib/python3.6/site-packages/pip/_internal/cli/progress_bars.py", line 166, in iter
for x in it:
File "/data/python3/lib/python3.6/site-packages/pip/_internal/network/utils.py", line 39, in response_chunks
decode_content=False,
File "/data/python3/lib/python3.6/site-packages/pip/_vendor/urllib3/response.py", line 564, in stream
data = self.read(amt=amt, decode_content=decode_content)
File "/data/python3/lib/python3.6/site-packages/pip/_vendor/urllib3/response.py", line 529, in read
raise IncompleteRead(self._fp_bytes_read, self.length_remaining)
File "/data/python3/lib/python3.6/contextlib.py", line 100, in __exit__
self.gen.throw(type, value, traceback)
File "/data/python3/lib/python3.6/site-packages/pip/_vendor/urllib3/response.py", line 430, in _error_catcher
raise ReadTimeoutError(self._pool, None, "Read timed out.")
pip._vendor.urllib3.exceptions.ReadTimeoutError: HTTPConnectionPool(host='ap-south-1.mirrors.cloud.aliyuncs.com', port=80): Read timed out.
解决
查找文档
https://blog.csdn.net/chr1341901410/article/details/80995451
解决方案
PACKAGE修改成所需要安装的模块
pip install PACKAGE -i http://pypi.douban.com/simple --trusted-host pypi.douban.com
安装 torch
pip3 install torch -i http://pypi.douban.com/simple --trusted-host pypi.douban.com
等待安装完成
WARNING: pip is being invoked by an old script wrapper. This will fail in a future version of pip.
Please see https://github.com/pypa/pip/issues/5599 for advice on fixing the underlying issue.
To avoid this problem you can invoke Python with '-m pip' instead of running pip directly.
Looking in indexes: http://pypi.douban.com/simple
Collecting torch
Downloading http://pypi.doubanio.com/packages/62/01/457b49d790b6c4b9720e6f9dbbb617692f6ce8afdaadf425c055c41a7416/torch-1.5.1-cp36-cp36m-manylinux1_x86_64.whl (753.2 MB)
|████████████████████████████████| 753.2 MB 13.3 MB/s
Requirement already satisfied: numpy in /data/python3/lib/python3.6/site-packages (from torch) (1.19.0)
Collecting future
Downloading http://pypi.doubanio.com/packages/45/0b/38b06fd9b92dc2b68d58b75f900e97884c45bedd2ff83203d933cf5851c9/future-0.18.2.tar.gz (829 kB)
|████████████████████████████████| 829 kB 97.0 MB/s
Using legacy setup.py install for future, since package 'wheel' is not installed.
Installing collected packages: future, torch
Running setup.py install for future ... done
WARNING: The scripts convert-caffe2-to-onnx and convert-onnx-to-caffe2 are installed in '/data/python3/bin' which is not on PATH.
Consider adding this directory to PATH or, if you prefer to suppress this warning, use --no-warn-script-location.
Successfully installed future-0.18.2 torch-1.5.1
安装 torchvision
pip3 install torchvision -i http://pypi.douban.com/simple --trusted-host pypi.douban.com
提示安装完成
WARNING: pip is being invoked by an old script wrapper. This will fail in a future version of pip.
Please see https://github.com/pypa/pip/issues/5599 for advice on fixing the underlying issue.
To avoid this problem you can invoke Python with '-m pip' instead of running pip directly.
Looking in indexes: http://pypi.douban.com/simple
Collecting torchvision
Downloading http://pypi.doubanio.com/packages/9a/f1/535a407b4a265adf2dd7c2c2458217e37c5fe83ec97234e66c564592a9a0/torchvision-0.6.1-cp36-cp36m-manylinux1_x86_64.whl (6.6 MB)
|████████████████████████████████| 6.6 MB 985 kB/s
Requirement already satisfied: numpy in /data/python3/lib/python3.6/site-packages (from torchvision) (1.19.0)
Requirement already satisfied: pillow>=4.1.1 in /data/python3/lib/python3.6/site-packages (from torchvision) (7.2.0)
Requirement already satisfied: torch==1.5.1 in /data/python3/lib/python3.6/site-packages (from torchvision) (1.5.1)
Requirement already satisfied: future in /data/python3/lib/python3.6/site-packages (from torch==1.5.1->torchvision) (0.18.2)
Installing collected packages: torchvision
Successfully installed torchvision-0.6.1
总结
个人认为国内服务器和部分海外服务器安装相关模块不会出现这种情况。可以通过这种方法解决类似问题。

关于安装torch模块的一些问题
大家好！ 这是你第一次分享关于python相关的内容，有不理解的地方，可以留言告知。
PyTorch
PyTorch是一个开源的Python机器学习库，基于Torch，用于自然语言处理等应用程序。身为小白的作者前一阵子开始学习python，在网上有稍微看了一些资料，其中就包括PyTorch，用来做机器学习。
作者昨晚精疲力竭，大脑混乱，实在没有导入成功PyTorch。pip失败，pycharm下载PyTorch也失败。另外昨晚还帮助一些小伙伴答疑解惑到凌晨两点多。
所以今天早上再战
准备
事先还先整了一份简单的word，用来记录我安装PyTorch的过程



讲真，在我们pip下载失败的时候，是都会想到会不会是源的问题，但是作者添加过了，还是失败。




官方地址内的操作
那重点来了：ModuleNotFoundError: No module named ‘tools.nnwrap’

遇到这个问题呢，作者是建议直接去官方网站选择适合自己电脑配置的，https://pytorch.org/ 这是官方链接。


进入官网后，向下滑动，我们可以看到简单的配置选项，选择我们所需要的

（重点！！！建议看完，因为这个很慢！！！）



那翻译一下：



OK，就这样复制，cmd打开，粘贴。ENT啪的一声



内心激动啊，额 这网络!我得下到什么时候，行吧，放弃了。

好在聪明机智的我找到了他们的下载地址：https://download.pytorch.org/whl/torch_stable.html


进入这个官方下载稳定版网址，先使用pip试用下载 看看自己电脑所相对应的版本，然后调用迅雷下载








最后
那在最后呢，我们进入文件夹的地址栏 cmd



pip install 加上文件名，注意！后缀要加上
两个都使用pip下载好后，我们还可以使用pip list 进行查看





大功告成！
补充
这两天作者把机械磁盘整合成一个盘

今天重新安装 出现一个问题



那在这呢 也就是缺少了numpy模块，所以百度搜索一下





pip install numpy -i https://mirrors.aliyun.com/pypi/simple/


pip安装的模块包不在自己的site-packages\中？！

那这边 作者才发现自己没设置好 安装的包全不在自己定的D盘的\Lib\site-packages文件夹中，全去了\AppData\Roaming\python\python38\site-packages\
通过python -m site命令查看





我们看见这里的USER_SITE和USER_BASE其实就是用户自定义的启用Python脚本和依赖安装包的基础路径。 哎，我这边没定义,是None。

必须要改。我还以为什么情况，这次安装python环境后，老是说权限不够？！

USER_SITE 没有填选，默认调用C盘路径下的python.exe中的脚本pip文件去下载，就会将site-package的默认安装到这个C盘路径下。