在我安装 anaconda 后没有新的 pip 安装的模块可以导入

Question

I recently installed anaconda and now I can't find the new modules anymore after I've installed them in this new anaconda python environment.

this is the location of the python interpreter for my anaconda environment I get when I type 'which python ': /Users/user/opt/anaconda3/bin/python

this is my $PATH: /Users/user/opt/anaconda3/bin:/Users/user/opt/anaconda3/condabin:/Users/user/go/bin:/usr/local/go/bin:/Users/user/.pyenv/shims:/Users/user/.pyenv/bin:/usr/local/bin:/Library/Frameworks/Python.framework/Versions/3.6/bin:/usr/local/bin:/usr/local:/usr/bin:/bin:/usr/sbin:/sbin:/go/src/github.com:/Library/Frameworks/Mono.framework/Versions/Current/Commands: No such file or directory

can anyone please tell me why anaconda can't find the modules I pip install in the anaconda environment when it has configured its own path? referring to these: /Users/user/opt/anaconda3/bin:/Users/user/opt/anaconda3/condabin in $PATH that came automatically with the installation.

my bash profile:

# Setting PATH for Python 3.6
# The original version is saved in .bash_profile.pysave
PATH="/Library/Frameworks/Python.framework/Versions/3.6/bin:${PATH}"
export PATH

export PYENV_ROOT="$HOME/.pyenv"
export PATH="$PYENV_ROOT/bin:$PATH"
if command -v pyenv 1>/dev/null 2>&1; then
  eval "$(pyenv init -)"
fi

# >>> conda initialize >>>
# !! Contents within this block are managed by 'conda init' !!
__conda_setup="$('/Users/SirFalk/opt/anaconda3/bin/conda' 'shell.bash' 'hook' 2> /dev/null)"
if [ $? -eq 0 ]; then
    eval "$__conda_setup"
else
    if [ -f "/Users/SirFalk/opt/anaconda3/etc/profile.d/conda.sh" ]; then
        . "/Users/SirFalk/opt/anaconda3/etc/profile.d/conda.sh"
    else
        export PATH="/Users/SirFalk/opt/anaconda3/bin:$PATH"
    fi
fi
unset __conda_setup
# <<< conda initialize <<<

Answer 1

Anaconda relies heavily on virtual environments, which in turn have their own set of installed python packages. This is a good thing. It lets you switch between different versions of Python, as well as use different versions of packages for different projects, without any headaches caused by package dependency conflicts.

You need to reinstall packages for each virtual environment you create on anaconda, (but note that anaconda environments will inherit packages from the base anaconda environment). I recommend you don't install any extra packages in the base environment, though, to avoid dependency conflicts. I also highly recommend using the command line interface for Anaconda, as it is much faster than their GUI app.

In the terminal, cd to the directory containing your anaconda folder; (I recommend having the anaconda folder in your home directory for convenience).

Then activate conda (this enters the conda base environment):

source anaconda3/bin/activate

Create and enter a conda virtual environment

conda create -n myenv python=3.7
conda activate myenv

Note that you may have to type . anaconda3/bin/activate . anaconda3/bin/activate instead of using source , depending on your type of shell.

Now that you're in your virtual environment, you can install packages like so:

conda install jupyter
conda install -c pytorch -c fastai fastai pytorch torchvision cuda92

To leave your virtual environment:

conda deactivate

This returns to the base anaconda environment. To leave the base environment, type conda deactivate a second time.

Other useful stuff:

To enter an already created environment:

conda activate myenv

To remove an environment package:

conda remove -n myenv package_name