Just create a directory and put the .cmd files inside it:
-
Execute launchLMS.cmd
- It will create some directories and will ask you to download "LM-Studio-x.y.z-Setup.exe" from his website
- Download it and put it inside <currentPath>\Installers path
-
Execute launchLMS.cmd again
-
LMStudio will install.
- In case you don't skip onboarding, you'll be prompted to "Enable local LLM service on login" in the "Download your first local LLM" page:
⚠️ Disable it⚠️
- In case you don't skip onboarding, you'll be prompted to "Enable local LLM service on login" in the "Download your first local LLM" page:
-
Download a model, check configuration, etc...
-
Close LM Studio
-
-
Execute launchLMS.cmd again
- Everything should be as it was before.
After normal installation, you can start launchLMS_service.cmd to run LMS in headless mode.
LMS will populate <currentPath>\UserProfile\.cache\lm-studio with some (empty) dirs and config-presets (json).
Those config-presets aren't actually being downloaded from the internet, but recreated upon launch.
I've set up the script to grab those and sync them with the portable dirs:
The path is being used by electron upon launch and the app will crash if its forcefully deleted while it's running (content is not)
You can either...
- Execute setFolderIcon.cmd
or
- Open a CMD/terminal window inside the portable app dir and execute three comands:
attrib +R .
attrib +H +S desktop.ini
attrib +H +S icon.ico
GNU General Public License v3.0