• 1 Post
  • 3 Comments
Joined 10 months ago
cake
Cake day: August 28th, 2023

help-circle


  • I installed Haiku on this Laptop from 1999 once since it was actually the only non-windows OS I could get to run for some reason. Video driver was bugged tho so the screen was visually offset by ~100 pixels which made it too hard to use. Otherwise though it ran at a bearable normal speed which is a huge feat for the something like 500MHz processor and 500MB/1GB of RAM (I forget the exact specs).


  • Sorry but has anyone in this thread actually tried running local LLMs on CPU? You can easily run a 7B model at varying levels of quantization (ie. 5 bit quantization) and get a generalized prompt-able LLM. Yeah, of course it’s going to take ~4GB of RAM (which is mem-mapped and paged into memory), but you can easily fine tune smaller more specific models (like the translation one mentioned above) and have surprising intelligence at a fraction of the resources.

    Take, for example, phi-2 which performs as well as 13B param models but with 2.7B params. Yeah, that’s still going to take 1.5GB RAM which Firefox wouldn’t reasonably ship, but many lighter weight specialized tasks could easily use something like a fine tuned 0.3B model with quantization.