configuration
channels: - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/ - http://mirrors.aliyun.com/anaconda/pkgs/main show_channel_urls: true default_channels: - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/r - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/msys2 custom_channels: conda-forge: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud msys2: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud bioconda: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud menpo: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud pytorch: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud pytorch-lts: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud simpleitk: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud auto_activate_base: false envs_dirs: -F:\ProgramData\miniconda3\envs pkgs_dirs: - F:\ProgramData\miniconda3\pkgs
Installation configuration
Download
Anaconda official website
Anaconda download mirror
Miniconda
Installation/Configuration Environment Variables
There can be no spaces in the installation directory
Anaconda
can be checked and added to the environment variable during installation.
The installation of the new version of Miniconda
will no longer prompt to add environment variables, and you need to add environment variables manually.
Take the installation in F:\ProgramData\miniconda3
as an example, add the following paths to the path variable
F:\ProgramData\miniconda3 D:\ProgramData\miniconda3\Scripts D:\ProgramData\miniconda3\Library\bin D:\ProgramData\miniconda3\Library\mingw-w64\bin
In order not to affect the original python environment, it is recommended to put the original py environment configuration variables above, and conda
configuration under the system python
Start and run the command box, enter conda, if the result is displayed as shown in the figure below, it means that the Anaconda environment variable configuration is successful
windows
computer modify PowerShell
running policy
In PowerShell
for security reasons, the default policy does not allow script execution. If you don’t modify it, you can only run the script in cmd.
Get-ExecutionPolicy
indicates the permission to execute PowerShell
scripts in the current environment, which has 4 values
Restricted
, the default setting, does not allow any scripts to runAllSigned
, only run trusted scriptsRemoteSigned
runs local scripts whether they are trusted or not;Unrestricted
, allowing all scripts to run, even untrusted ones;
In general, set the level to RemoteSigned
to facilitate script debugging and running.
Run the command in the PowerShell
window as an administrator and enter:
Set-ExecutionPolicy unrestricted
Conda command line initialization
conda init powershell
After conda is installed, when running conda activate xxx to switch environments, an error may be reported
Administrators are required to run cmd settings, otherwise some do not have permission to set
Conda needs to initialize the current command line environment before it can be used. Please initialize according to the command line tool you use. Enter commands such as conda init powershell or conda init bash. After inputting, restart the command line.
After the initialization is complete, under normal circumstances, the current conda environment will be displayed before each command
Cancel the base virtual environment automatically loaded by the terminal
After the Conda command line is initialized, the conda (base) environment will be automatically loaded every time you enter the command line, causing the system python environment to fail to run
conda config --set auto_activate_base false
After execution, a sentence auto_activate_base: false
will be generated at the end of the C:\Users\Hi\.condarc
configuration file
At the same time, the base environment will not be loaded every time the command is opened
Modify directory permissions
By default, only the administrator has permission to read and write the envs directory of miniconda3\envs
during a new installation, and the permissions need to be modified to facilitate all users to read and write
Mirror configuration
Modify the image of python pip in all environments
# set mirroring pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple/
There are two ways to modify the mirror source
Direct command number execution
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/ conda config --set show_channel_urls yes
-
When setting the source address to be displayed during installation, it is convenient to know which source the package comes from during installation
conda config --set show_channel_urls yes
Edit the C:\Users\Hi\.condarc
file, .condarc
may be hidden.
Windows users cannot directly create a file named .condarc
, they can execute conda config --set show_channel_urls yes
first to generate the file and then modify it.
- Tsinghua Mirror
channels: - defaults show_channel_urls: true default_channels: - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/r - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/msys2 custom_channels: conda-forge: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud msys2: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud bioconda: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud menpo: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud pytorch: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud pytorch-lts: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud simpleitk: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
-
Ali Cloud
channels: - defaults show_channel_urls: true default_channels: - http://mirrors.aliyun.com/anaconda/pkgs/main - http://mirrors.aliyun.com/anaconda/pkgs/r - http://mirrors.aliyun.com/anaconda/pkgs/msys2 custom_channels: conda-forge: http://mirrors.aliyun.com/anaconda/cloud msys2: http://mirrors.aliyun.com/anaconda/cloud bioconda: http://mirrors.aliyun.com/anaconda/cloud menpo: http://mirrors.aliyun.com/anaconda/cloud pytorch: http://mirrors.aliyun.com/anaconda/cloud simpleitk: http://mirrors.aliyun.com/anaconda/cloud
Modify the creation environment location
The envs path under the anaconda installation directory was not found during creation, and the newly created environment is in C:\Users\xxx.conda\envs\
Pay attention to change the dividing line of the path to double slashes
There is a .condarc file under C:\Users\username, open it, and add the following content at the end of it:
envs_dirs: -F:\ProgramData\miniconda3\envs pkgs_dirs: - F:\ProgramData\miniconda3\pkgs
conda config --show
View configuration directory
Modify the default startup environment
notepad $PROFILE # Will open (if not automatically create) Profile. conda activate notes # Add the following commands to the Profile file. Next time, open a terminal and the default Conda python environment will be notes .
This method works but it is a bit tasteless to enable immediate loading every time. Or honestly upgrade or downgrade the py version of base
View conda basic information
# View conda version information conda info # update conda version conda update -n base conda # Output conda configuration conda config --show
Virtual environment management
Activate/Exit Environment
# activate conda activate envname_xxx conda activate env_path # Activate the environment of a specific path # quit conda deactivate
envname_xxx
indicates the environment to be activated, if you forget the name, you can first use conda env list
to view all virtual environments
Exit the environment If you do not continue to use this command line window, just close the window and leave it alone. You do not need to specify the environment name after closing the environment, and the current environment is closed by default.
Modify the creation environment location
The conda environment is installed in the user directory C:\Users\username.conda\envs by default. If you choose the default path, then create a virtual environment later and install it in the user directory. If you don’t want to take up space on the C drive, you can modify the conda virtual environment path.
The envs path under the anaconda installation directory was not found during creation, and the newly created environment is in C:\Users\xxx.conda\envs\
Pay attention to change the dividing line of the path to double slashes
Before adding the following two, you need to prescribe the corresponding directory permissions, otherwise it will cause problems that cannot be written
There is a .condarc file under C:\Users\username, open it, and add the following content at the end of it:
envs_dirs: -F://ProgramData//miniconda3//envs pkgs_dirs: -F://ProgramData//miniconda3//pkgs
conda config --show
View configuration directory
Existing environment view
conda-env list
New installations will only have one base default environment
[External link picture transfer failed, the source site may have an anti-theft link mechanism, it is recommended to save the picture and upload it directly (img-cRdr4udU-1689435571658)(https://gitcode.net/qq_40597070/pic_go/raw/master/202301102332189.png )]
Create a new environment
conda create --name envname_xxx python=3.4 conda create -n envname_xxx python=3.4 # -n is the same shorthand for --name conda create -n envname_xxx python=3.7 numpy=1.16. matplotlib # You can also install some other packages while creating the environment, or specify the version of the package at the same time conda create -n envName python==3.7.3 pip --no-deps # create environment but do not install default packages
The above commands have the same effect, python34
refers to the name of the new environment to be created, python=3.4
refers to the python
version of the creation environment, which will be automatically installed by default new
It should be noted that an error is often reported without permission to operate the directory, and the command line needs to be opened in the management mode
The third way to create pip is strongly recommended to install otherwise it will directly use the built-in pip
Delete environment
Delete an existing environment and all packages in it
conda remove --name envname_xxx --all
To delete an environment, another violent way is to directly delete the envs/env_name folder under the anaconda installation path
Copy specified environment
conda create -n new_env_name --clone old_clone_env_name
Copy the environment, copy old_env_name, and name the new environment new_clone_env_name
Copy the specified directory and replace old_env_name with the existing environment
The environment cannot be copied by copying the folder under envs in the anaconda installation directory, because such an operation will not change the configuration information in this environment by itself, and the operation of the copied environment will affect the original environment
View environment history
This function is similar to the version record of git, but this is only applicable to the package pip package installed by conda will not record
# View the specified environment history backtracking conda list -n env_name -r
The environment package falls back to the specified version
conda list -n env_name -r # View the specified environment history backtracking conda install -n env_name --revision REV_NUMBER
First list all versions through the above command, and then return to a certain version through the following command
Environment renaming
# new version support method conda-script.py rename [-h] [-n ENVIRONMENT | -p PATH] [--force] [-d] destination conda rename -n current name new name # example conda rename -n test123 test321 conda rename --name test123 test321 conda rename -p path/to/test123 test321 conda rename --prefix path/to/test123 test321
conda create -n old_env_name --clone new_clone_env_name conda remove -n old_env_name --all
There is currently no direct renaming command, you need to clone and then remove the old environment
Package management
List installed packages
conda list # List all packages installed in the current environment
conda list --revisions # List the version history of the package management changes in the current environment, list the version history of the current environment changes, this version record is similar to the git record, record the environment record after each package management operation, this command can save lives when it is critical, For example, if you upgrade a certain environment and find a problem and want to roll back to the previous environment state
conda list -n env_name # List all packages in the specified environment
conda list --name env_name --revisions # List the version history of package management changes for the specified environment
Installation package
conda install package_name conda install PKGNAME==3.1.4 # Install the specified version of the package, one equal sign or two equal signs are acceptable conda install --yes PKG1 PKG2 # Install the package and output the command prompt, that is, direct installation does not require confirmation, and it will be useful when writing automatically run scripts
Update package
# After starting Anaconda Prompt as administrator, # upgrade is the same as update conda update conda # upgrade conda (required before upgrading Anaconda) conda update anaconda # upgrade anaconda conda update --all # Update all packages, including Anaconda and conda conda update --all -n env_name # Upgrade all packages in the specified environment, without -n env_name means upgrade all packages conda update <package> # update the specified <package> package conda update -h # other attempts conda upgrade pip # upgrade pip # pip package update python -m pip install --upgrade pip pip install --upgrade <package> # Upgrade the specified pip package, it is not recommended to use this method to upgrade the package installed by conda
Uninstall package
conda remove package_name conda uninstall package_name # Uninstall a package in the current environment conda uninstall -n env_name package_name # Uninstall a package in the specified environment
Clean unused caches and packages
conda clean --all
Sometimes you can easily clean up a few G, but if it is not a case of limited hard disk space, it is recommended not to clean up, because if the installation package has this package in the cache, there is no need to download it again
?
Environment package export and import
Export
Conda export needs to pay attention to the pip installed package will not be exported
# Export the activated environment package to YAML file conda list --explicit > "C:\Users\Hi\Desktop\requirements.yaml" # Export the activated environment package to a txt file conda list --explicit > "C:\Users\Hi\Desktop\requirements.txt" conda list -e > "C:\Users\Hi\Desktop\requirements.txt" pip freeze > "C:\Users\Hi\Desktop\requirements.txt" # pip export
Import
# Create environment through yaml file conda env create --file package_list.yaml # Create an environment from a YAML file conda env create # There is a package_list.yml file in the current directory, you can directly create the environment without entering the file name # Create an environment through a txt file conda create --name new_env_name --file pkgs.txt # Upgrade or install existing environment packages through yaml or txt conda env update --file package_list.yaml # Install to the global environment (without activating the existing environment) the yaml or txt file exported by conda can be used conda env update -n env_name --file package_list.yaml # Install to the specified environment, either yaml or txt files exported by conda pip install -r requirements.txt # pip is installed to the current environment, and the txt file exported by pip is required
It should be noted that the txt imported by conda has more information than the txt imported by pip, and sometimes it is not universal
Use of Packages (package) and Channels (source/channel)
# Add Channels conda config --add channels CHANNELNAME # Use conda to search for packages from the configured Channels, you can bring the version number, the author's note: it is less used, and it is more convenient to directly search on Baidu/google under the system of the graphical interface conda search numpy=1.16 # Use anaconda to search for packages from all Channels anaconda search FUZZYNAME # Install the package from the specified channel, such as conda install conda-forge::PKGNAME