@tacobelmin@sierracatalina yeah, excuse me for a brief response to you excellent writeup, my bronchitis has gotten the best of me. this song embodies a lot of this - https://t.co/S0IOak4h4Y
im not sure anybody could write a LLM back then, maybe a concept - but please provide a proof
Large Language models were technicaly possible only recenlty, I guess try writing it now in basic on a modern computer and youll have a cult following soon :)
maybe, now with a hindsight one could use 1 bit LLMs from microsoft to run a small version on underpowered computres
https://t.co/x7qsUCAbG6
but this is much more interesting now: https://t.co/uj2R8Y4Jrf
i guess iphone side is getting fed up with diminishing upgrades, but mac side is happy with the M series CPUs
everything changed for me when I bought my first Macbook Air M1
it would be heaven if they never left ObjectPascal and if something like Delphi for Mac was reality instead of Xcode
free tier ChatGPT works well mostly, its all similar and depends on the prompt
o1 is the best for now but of anybody has an M1 or better mac with 16 or more GB of RAM ollama works great with OpenWeb UI for free after you download the models
it can even work with a cheap AMD PC but the process is a bit more involved
yeah i feel OpenAI did a rug pull, and LLMs in general are screwed on iOS devices because of low RAM
local Siri voice recognition works amazing on WatchOS
ollama works amazing with OpenWebUI, latest M4 machines and open source LLMs
i used to be scared for Intel
now Im a bit scared for AMD
long term Chinese RiscV designs will save us, unless we invent replicant brains sooner
Dillo browser
it can show a basic popular rather objective unbloated news site from my country
it can show LWN
it can show HN
it can show old popular local forum
it can show rss feeds
it automatically blocks (does not work on) all the popular social networks
what else would one need?