It's possible to add specific terminal to VS Code on Windows machine when you have Anaconda package installed. You will have Python interpreter under your fingertips. Just add to settings.json the following configuration
"terminal.integrated.profiles.windows": {
"miniconda": {
"path": [
"${env:windir}\\System32\\cmd.exe"
],
"args": ["/k C:\\Users\\neupo\\miniconda3\\Scripts\\activate.bat"]
},
"MinGW": {
"path": [
"C:\\msys64\\msys2_shell.cmd"
],
"args": ["-defterm", "-here", "-no-start", "-mingw64"],
"icon": "terminal-bash"
},
}
And then select the profile in local workspace settings .vscode/settings.json
"terminal.integrated.defaultProfile.windows": "MinGW",
For a long time I was against any virtual environment for Python. You only need it for the clean experiment, right? The package manager must resolve all conflicts and it's not a problem to jump a few versions ahead. That's what I thought. But more and more I started see conflicts and new versions got uninstalled and another random version installed in my system.
I got tired of it. And when the time has come to update Python 3.8 to the new version, I decided to go with Conda. Just at the same time Arch rolled another confusing for me politics - pip
refuses to install any package even with the --user
flag and insists to use pypenv
.
Since that I started using in my day after day experiments
- Transformers and PyTorch
- Also PyTorch, but in WSL2 and this version is craving for
CUDA_HOME
- Diffusers from Hugging Face
- TensorFlow 2.12 on Windows can only be installed in WSL2
Also, can be helpfull Anaconda cheatsheet