exactly. The people who say this clearly have literally no idea how it actually works (I’ve seen people unironically saying it compresses all the images and stores them) and just jump off the next twitter post as reference
Description: A picture of a simple box trap. The bait consists of a Blahaj, a plushie of Bridget (from Guilty Gear), a Monster energy drink and a copy of the video game “Celeste”. A caption above the trap says “pspsps”.
Hi, I’m currently starting to learn how LLM works in depth, so I started using nanoGPT to understand how to train a model and I’d like to play around with the code a little more. So I set myself a goal to train a model that can write basic French, it doesn’t to be coherent or deep in its writing, just French with correct...
Sure! You’ll probably want to look at train-text-from-scratch in the llama.cpp project, it runs on pure CPU. The (admittedly little docs) should help, otherwise ChatGPT is a good help if you show it the code. NanoGPT is fine too.
For dataset, maybe you could train on French Wikipedia, or scrape from a French story site or fan fiction or whatever. Wikipedia is probably easiest, since they provide downloadable offline versions that are only a couple gigs.
So this isn’t a major thing, but I’ve blocked some bots (e.g. ones that provide an alternative video link on every YouTube post), and whenever I see a post like this, it still shows 1 comment. So I go in, and there’s nothing there. Obviously it’s still counting the blocked account commenting within this total....
A simple way would be to load the comments themselves, and then check for blocked users. But this would basically ddos the instance servers and would be extremely janky lol
The technology of compression a diffusion model would have to achieve to realistically (not too lossily) store “the training data” would be more valuable than the entirety of the machine learning field right now.
Rap names (i.postimg.cc)
The human sized thumb is getting angry. (sh.itjust.works)
hypothetical quick time rule (lemmy.dbzer0.com)
rule
Pick your poison. Dystopian style (lemmy.world)
REFERENCE rule (sh.itjust.works)
let’s see how low effort posts can be rule...
What are your tips for saving money and reducing waste?
Here’s a couple examples from my life:...
rule (lemmy.blahaj.zone)
Why do Lemmy users rarely use emoticons?
Reading comments in different communities, I noticed that users hardly leave smilies. Why is that?...
egg🪤irl (lemmy.blahaj.zone)
Description: A picture of a simple box trap. The bait consists of a Blahaj, a plushie of Bridget (from Guilty Gear), a Monster energy drink and a copy of the video game “Celeste”. A caption above the trap says “pspsps”.
Good enough title (lemmy.world)
Rule (i.imgur.com)
uncomfortable (sh.itjust.works)
lol
ornament (sh.itjust.works)
I hope this isn’t a repost
For Whom The Fridge Rules (lemmy.world)
Training a model without a GPU
Hi, I’m currently starting to learn how LLM works in depth, so I started using nanoGPT to understand how to train a model and I’d like to play around with the code a little more. So I set myself a goal to train a model that can write basic French, it doesn’t to be coherent or deep in its writing, just French with correct...
Rule (lemmy.blahaj.zone)
animefeminist.com/love-yall-lonely-catgirl-seeks-…
Voyager 1.31.0: Change comment account, happy holidays!
Hey, happy holidays everyone!~...
Feature request - hide blocked accounts from comment total
So this isn’t a major thing, but I’ve blocked some bots (e.g. ones that provide an alternative video link on every YouTube post), and whenever I see a post like this, it still shows 1 comment. So I go in, and there’s nothing there. Obviously it’s still counting the blocked account commenting within this total....
egg_irl (lemmy.blahaj.zone)
⚠️ Trade offer ⚠️...
Firefox is on the brink of being dropped by the US Government (www.brycewray.com)
usb formatting (sh.itjust.works)
shamelessly stolen from nixCraft on mastodon
Asking ChatGPT to Repeat Words ‘Forever’ Is Now a Terms of Service Violation (www.404media.co)
Wake up, it's time for rule (lemmy.blahaj.zone)