150
Simple local ai app
(flathub.org)
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 17 Sep 2024
150 points (100.0% liked)
Linux
48324 readers
558 users here now
From Wikipedia, the free encyclopedia
Linux is a family of open source Unix-like operating systems based on the Linux kernel, an operating system kernel first released on September 17, 1991 by Linus Torvalds. Linux is typically packaged in a Linux distribution (or distro for short).
Distributions include the Linux kernel and supporting system software and libraries, many of which are provided by the GNU Project. Many Linux distributions use the word "Linux" in their name, but the Free Software Foundation uses the name GNU/Linux to emphasize the importance of GNU software, causing some controversy.
Rules
- Posts must be relevant to operating systems running the Linux kernel. GNU/Linux or otherwise.
- No misinformation
- No NSFW content
- No hate speech, bigotry, etc
Related Communities
Community icon by Alpár-Etele Méder, licensed under CC BY 3.0
founded 5 years ago
MODERATORS
something tells me its not local
local AI models are a thing you know, just run this on a computer with no internet access and you'll swiftly see whether it relies on servers or not.
I think something is lying to you.
so like, turn your network connection off and try it
Why would that be the case?
Computation and storage limitations
You overestimate the hardware required to run AI.
I can run Llama3 on my desktop with a 3060, answers are near instant.
model training is often exorbitantly high-energy and storage, but the model output is not: GPT3 cost $7 million dollars just in electricity costs to train, but the model is "only" 500 GB, and can be run locally on just CPU processing
cool
You can run a basic model on pretty mid-range hardware, the smaller ones are only 1-2GB in size.
To be fair, the first time I tried running local AI (and it actually worked), I was so surprised that I actually unplugged my Ethernet and tried again. I'm still surprised, but it's possible for the massive amounts of training data to be compressed to a model under only 10 or 20 GB.
Username checks out
Just use firejail to sandbox it and find out.