Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Cannot specify inline completion model for provider registries #889

Closed
3coins opened this issue Jul 10, 2024 · 11 comments
Closed

Cannot specify inline completion model for provider registries #889

3coins opened this issue Jul 10, 2024 · 11 comments
Labels
bug Something isn't working

Comments

@3coins
Copy link
Collaborator

3coins commented Jul 10, 2024

Description

There is no way to specify an inline completion model in settings for providers such as HuggingFace and Ollama, because they don't have the models listed within their provider configuration.

Screenshot 2024-07-10 at 3 21 15 PM
@3coins 3coins added the bug Something isn't working label Jul 10, 2024
@krassowski
Copy link
Member

krassowski commented Jul 10, 2024

This does not look right to me. I was able to use huggingface models before. Do you have inline completer enabled? It looks greyed out on the screenshot.

Edit: Of course it could be a recent regression.

@3coins
Copy link
Collaborator Author

3coins commented Jul 10, 2024

You might be right, but I cannot get the inline completion list to be enabled. I have the "enabled" checks on within the settings panel.

inline-completion-settings

@krassowski
Copy link
Member

krassowski commented Jul 10, 2024

Can you upgrade JupyterLab to the latest version? There was a related big which was patched in 4.2.3 ( jupyterlab/jupyterlab#16508 ).

@3coins
Copy link
Collaborator Author

3coins commented Jul 10, 2024

Yes, that seemed like the issue, I opened a document, and then used the bottom button to enable, which worked. Will check with JLab 4.2.3 shortly.

Will close this issue, as it doesn't seem to be the problem.

@3coins 3coins closed this as completed Jul 10, 2024
@3coins
Copy link
Collaborator Author

3coins commented Jul 10, 2024

@krassowski
Can confirm with JLab 4.2.3, it works perfectly, I also see the state sync between the bottom button and settings, which was great. Thanks for clarifying.

@cboettig
Copy link

Maybe I'm missing something obvious but I also cannot specify an inline completions model (it is greyed out). I am running Jlab 4.2.3 . Using ollama as the language model in the chat window works wonderfully though, huge thanks to the team for supporting that!

@krassowski
Copy link
Member

@cboettig with what version of jupyter-ai? Can you post a screenshot?

@cboettig
Copy link

2.19.0

image

image

@krassowski
Copy link
Member

From your screenshot I see that you have the history provider enabled, but do you also have the jupyter-ai provider enabled? You need to scroll down in the settings panel to check.

@cboettig
Copy link

argh, thanks! I totally didn't spot that next check box hiding in there!

@sjdemartini
Copy link

sjdemartini commented Aug 1, 2024

For what it's worth, in case others are struggling with this: when using Notebook v7 to edit the Jupyter-AI settings (via "Left sidebar -> Show Jupyter AI Chat"), the "Inline completion model" select box remains grayed out for me regardless. You can instead go to the JupyterLab-based interface at /lab to update the AI settings, and inline completions will then work in both the JupyterLab and Notebook v7 cells.

(And somewhat similarly, when clicking the gear next to "Inline completions model" while in a Notebook v7 notebook, it leads to a JS error Uncaught (in promise) Error: Command 'settingeditor:open' not registered. visible in the devtools console, since I guess the settings editor can only be opened from the top-level /tree view for Notebook v7 and not while on an individual notebook.)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working
Projects
None yet
Development

No branches or pull requests

4 participants