Working, but cant pick better ( as in bigger LLM ) what is that I do wrong?

[jaspweowaale]

So it's running and working, but I have a bit more power on my 24GB GPU and would like to pick the biggest model to test. I'm also trying this out with some historical photos and would like to include people who are all publicly well-known, with their names in the keywords.

And looks like I cant select any how to add to your list

[jabberjabberjabber]

Generally larger models don't work much better for the speed decrease, but you can run any model you like. Just go into the resources directory and run the koboldcpp executable and load the model in the there, then open the llmii launcher and select option 3.

[jaspweowaale]

I give up speed for better results. I'm not in front of my setup right now, but does KoboldCpp pop up a web GUI that hooks into the same port your GUI uses? I'll have a look at that — it's getting a bit late here in Asia.

Thanks for the fast feedback! Do you have one of those Buy Me a Coffee links or PayPal? I'd be happy to drop something in.

[jabberjabberjabber]

It will open a web browser if you have that option checked, but there is no conflict unless you try to chat with it while running the indexer, in which case you have to wait for the chat response to finish before it can work on an image.

No need to worry about coffee or paypal, but I appreciate the offer.

[jaspweowaale]

Offer is real, I always support projects I use, that been the case from back in the days when I was running BBS and all Windows Shareware into Eur from the US, many small rivers a ocean make, , I play around a bit tomorrow,