2021.3.20

    1.anaconda安装
    （https://www.anaconda.com/products/individual#macos）
    installer下载以后跟着走，可以加个环境变量。然后开黑窗cmd⛵

    1. //查看版本
    2. conda --version
    3. //切换环境
    4. conda activate 环境名
    5. //创建环境
    6. conda create --name 环境名 python=3.x
    7. //安装扩展包
    8. conda install 扩展包名称

    requests测试

    1. conda install requests
    2. python
    3. >>import requests
    4. >>print(requests.get('https://ai-classes-conyrol.vercel.app/#/'))
    5. <Response [200]>

    2.来个pycharm🌞
    setting->py interpreter->点齿轮⚙->add…->python 3.7(DL)
    切换到DL的环境

    3.装个pytorch（cpu😥）
    （https://pytorch.org/get-started/locally/）
    选中None版复制command：conda install pytorch torchvision torchaudio cpuonly -c pytorch
    先按推荐尝试镜像源，记得把-c 删掉，优先级高于镜像。
    （https://mirrors.tuna.tsinghua.edu.cn/help/anaconda/）
    再贴个群里的交大源

    1. channels:
    2. • https://mirrors.sjtug.sjtu.edu.cn/anaconda/cloud/pytorch/
    3. • https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
    4. • https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
    5. • https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
    6. • https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge/
    7. ssl_verify: true
    8. show_channel_urls: true

    记录个全流程

    1. //查看目前已有的channel：
    2. conda config --show channels
    3. //添加新的channel：
    4. conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
    5. conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    6. conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/msys2/
    7. conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
    8. conda config --set show_channel_urls yes
    9. run command：conda install pytorch torchvision torchaudio cpuonly
    10. C:\Users\admin>    conda config --show channels
    11. channels:
    12. • defaults
    14. C:\Users\admin>conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
    15. C:\Users\admin>conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    16. C:\Users\admin>conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/msys2/
    17. C:\Users\admin>conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
    18. C:\Users\admin>conda config --set show_channel_urls yes
    19. C:\Users\admin> conda config --show channels
    20. channels:
    21. • https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
    22. • https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/msys2/
    23. • https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    24. • https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
    25. • defaults
    26. C:\Users\admin>conda clean -i
    27. C:\Users\admin>conda install pytorch torchvision torchaudio cpuonly
    28. Collecting package metadata (current_repodata.json): done
    29. Solving environment: done
    30. Package Plan
    31. environment location: D:\D\software\anaconda
    32. added / updated specs:
    33. - cpuonly
    34. - pytorch
    35. - torchaudio
    36. - torchvision
    37. The following packages will be downloaded:
    38. package                    |            build
    39. ---------------------------|-----------------
    40. conda-4.9.2                |   py38haa244fe_0         3.1 MB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    41. cpuonly-1.0                |                0           2 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch
    42. libuv-1.41.0               |       h8ffe710_0         341 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    43. ninja-1.10.2               |       h5362a0b_0         273 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    44. python_abi-3.8             |           1_cp38           4 KB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
    45. pytorch-1.8.0              |      py3.8_cpu_0       160.3 MB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch
    46. torchaudio-0.8.0           |             py38         2.7 MB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch
    47. torchvision-0.9.0          |         py38_cpu         6.6 MB  https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch
    48. ------------------------------------------------------------
    49.                                        Total:       173.3 MB
    50. The following NEW packages will be INSTALLED:
    51. cpuonly            anaconda/cloud/pytorch/noarch::cpuonly-1.0-0
    52. libuv              anaconda/cloud/conda-forge/win-64::libuv-1.41.0-h8ffe710_0
    53. ninja              anaconda/cloud/conda-forge/win-64::ninja-1.10.2-h5362a0b_0
    54. python_abi         anaconda/cloud/conda-forge/win-64::python_abi-3.8-1_cp38
    55. pytorch            anaconda/cloud/pytorch/win-64::pytorch-1.8.0-py3.8_cpu_0
    56. torchaudio         anaconda/cloud/pytorch/win-64::torchaudio-0.8.0-py38
    57. torchvision        anaconda/cloud/pytorch/win-64::torchvision-0.9.0-py38_cpu
    58. The following packages will be SUPERSEDED by a higher-priority channel:
    59. conda               pkgs/main::conda-4.9.2-py38haa95532_0 --> anaconda/cloud/conda-forge::conda-4.9.2-py38haa244fe_0
    60. Proceed ([y]/n)? y
    61. Downloading and Extracting Packages
    62. torchvision-0.9.0    | 6.6 MB    | ############################################################################ | 100%
    63. libuv-1.41.0         | 341 KB    | ############################################################################ | 100%
    64. conda-4.9.2          | 3.1 MB    | ############################################################################ | 100%
    65. pytorch-1.8.0        | 160.3 MB  | ############################################################################ | 100%
    66. ninja-1.10.2         | 273 KB    | ############################################################################ | 100%
    67. torchaudio-0.8.0     | 2.7 MB    | ############################################################################ | 100%
    68. cpuonly-1.0          | 2 KB      | ############################################################################ | 100%
    69. python_abi-3.8       | 4 KB      | ############################################################################ | 100%
    70. Preparing transaction: done
    71. Verifying transaction: done
    72. Executing transaction: done

    然后成功的没有切换环境以为装不好😶
    于是又重装了一次🤐
    【所以清华没有镜像pytorch😬】
    这次用原指令装了一次，其实网速还可以，没有想象中的慢。
    然后装的不是cuda版的：
    import torch
    test = torch.Tensor([1.])
    test.cuda()
    print(test)
    测试不了了，姑且看pytorch-1.8.0装好了。
    4.玩玩jupyter🙄
    装完conda就over了，打开resnet，愉快地看到这个本子
    在In[90]处修改图片地址反复测试，得到多个结果。（反复尝试人像都定位到物品上了，测个桥，饼啥的还可的样子）