OK, so Ollama is running, now it’s time to test the Open WebUI

For my specific setup, I visit http://framework:3000 and I get the WebUI that can talk to Ollama. I set it up with one account for me as the admin. This lets me pull models, switch models, configure the incredible number of settings for each model. There are probably lots of videos on how to use Open WebUI.

I won’t watch any of them. It’s way more fun to click on things and see what breaks. This is why docker containers are great. I can blow everything away and start over anytime I want. I’ve already done this about 5 times to be honest.

So anyway. Time for a test!

NVIDIA just released Nemotron which is an awesome name. I’m sad I never thought of this word before. I figured this would be a great model to test first.

I gave it this prompt Create a html web page that draws a christmas tree in the middle of the page which I am 100% aware is a terrible prompt. Part of learning is failing on purpose sometimes. I wanted to see what would happen.

alt

It clearly nailed it!

You can’t see on the image, but the thing on the bottom, I think it’s a drum maybe, would move. I’m too lazy to create an animation of this. I’m also impressed by the LLM taking the initiative to add some Christmas themed messages without being asked.

There are a lot of models to explore, some are text only, some can process images. I have no idea what the limits of any of this is.