Would it be possible to use a locally run LLM instead of Codeium, e.g. via Ollama? #784
Replies: 1 comment 5 replies
-
|
Interesting! The codeium integration is currently achieved using: Monaco editor: Monaco Codeium Provider, which is a fork of the codeium official codeium-react-code-editor. It allows you to configure the languageServer URL. CodeMirror: codemirror-codeium, which although it does not allow to configure the URL out-of-the-box, this can be easily achieved. So to use another backend server (including a local one), you will have to provide the same API and change the URL to it. Thank you. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you. Unfortunately, I am not a developer myself and I have no experience with nodejs. So please excuse my naive follow up questions: Where can I find this? I looked in the livecodes/src folder but could not locate those files you referenced. They seem to be in different places in the original projects on Github and I don't understand how you integrated them. |
Beta Was this translation helpful? Give feedback.
-
|
These are indeed external projects that are used for the Codeium AI integration. These are loaded in LiveCodes when the AI assistant is activated: |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for your patience in explaning this stuff. But if the code is loaded "live", how can I modify it? |
Beta Was this translation helpful? Give feedback.
-
|
If you need to modify these, you will have to fork the projects, make your modifications and then re-publish them as new packages which can then be used instead of the original ones. The real problem you are trying to solve are not modifying these. The actual problem is providing the code suggestions in the same format that the codeium server does. The actual codeium backend server running the AI models is not open-source. |
Beta Was this translation helpful? Give feedback.
-
|
Okay, so this is not easily done. :( Thank you for the explanations! |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
I would rather not have my code uploaded to any 3rd party. And I am already running various LLMs locally.
Beta Was this translation helpful? Give feedback.
All reactions