I made a (so far) simplified version that runs entirely in a chrome browser using the built in Gemini Nano model that they've been pushing to people.
The code is here https://github.com/differentprogramming/ChromeHalupedia/blob/main/index.html
I took a little bit of your prompt, and even though I'm a programmer, I let Gemini write the site. The first version was (believe it or not) written by the version that inside the google search bar.
I did a bit of improvement with Gemini 3.0 Pro inside of github copilot.
It has to throw 12 random words into the model before it starts because otherwise the model will more or less generate the same article over and over. It needs semantic prodding.
It keeps the last two articles in the buffer when it generates the next one in order to have some consistency, or at least that's what it's supposed to do (and it seems to work). But I need to look at the code to make sure it makes sense.
You have to open links in the same window to maintain the context.
I may later make it use persistent storage in the browser to make it more like the full site.
It could use some tweaking, it's not as whimsical and fun as the original so far.
Anyway I'd like permission to use the word you made up "Halupedia". I'll change it to link to the original before I put it up.
update: This little model may not be good enough at abstraction to worth it... Needs more work. I'll keep you updated.
I made a (so far) simplified version that runs entirely in a chrome browser using the built in Gemini Nano model that they've been pushing to people.
The code is here https://github.com/differentprogramming/ChromeHalupedia/blob/main/index.html
I took a little bit of your prompt, and even though I'm a programmer, I let Gemini write the site. The first version was (believe it or not) written by the version that inside the google search bar.
I did a bit of improvement with Gemini 3.0 Pro inside of github copilot.
It has to throw 12 random words into the model before it starts because otherwise the model will more or less generate the same article over and over. It needs semantic prodding.
It keeps the last two articles in the buffer when it generates the next one in order to have some consistency, or at least that's what it's supposed to do (and it seems to work). But I need to look at the code to make sure it makes sense.
You have to open links in the same window to maintain the context.
I may later make it use persistent storage in the browser to make it more like the full site.
It could use some tweaking, it's not as whimsical and fun as the original so far.
Anyway I'd like permission to use the word you made up "Halupedia". I'll change it to link to the original before I put it up.
update: This little model may not be good enough at abstraction to worth it... Needs more work. I'll keep you updated.