Wake me up when someone creates an LLM that at least delivers deterministic, reproducible results.
reproducible results
isn't this already the case? use the same model with same temperature locally and it should produce the same output
they already can, it's just undesirable so they have random numbers added to make them unpredictable
Man, at first glace at the picture without reading the model, I thought that was the Samsung Galaxy Note 2. I thought I was gonna be able to pull that old puppy outta storage and get Nura running on it.
I'm not a fan of AI, but this use case of getting older dead hardware some life is interesting. Hopefully the process you used could allow it to one day bring other old hardware back into use.
LLM code is not workable for widespread adoption, but for individual use cases it can do a lot of interesting things!
It depends on how much human care and attention was used. Vibe coding gets all the media attention but real work is being done by experienced engineers that know how to guide this tool and wrangle the code into shape.
But the docs/code can definitely be used to support manual labour.
I’m hoping this trend will enable some bootloader hacks/bypasses for older Apple devices, as well.
I was curious what AI model / product you used to do this? I'd like to help the efforts for some of my own old phones.
Claude Opus 4.8 - 5.5.
@sv1sjp congrats, very encouraging
Linux
A community for everything relating to the GNU/Linux operating system (except the memes!)
Also, check out:
Original icon base courtesy of lewing@isc.tamu.edu and The GIMP