I think they’re open sourcing it because they actually make all their money from content recommendation and ad placement models
LLMs are just a side gig for them

I think they’re open sourcing it because they actually make all their money from content recommendation and ad placement models
LLMs are just a side gig for them
Oh yeah, thats it. I was mistaken
It’s classified as an “assault rifle” in my state. But yeah, I suppose it varies from state to state.
Replying to @the_crotch@sh.itjust.works
Depends on whether you consider an AR-15 to be an assault rifle or not
Replying to @return2ozma@lemmy.world
Oh good, all my assault rifles are still okay though
Replying to @Enkrod@feddit.org
Partially sarcastic, but I did convince 3 people to switch to Linux this year
Replying to @FoxtrotDeltaTango@sh.itjust.works
Ironically, their stock is dropping because they started treating their users badly in order to improve stocks.
Replying to @friend_of_satan@lemmy.world
Lemmy stocks to the moon haha
Replying to @eicker@lemmy.world
I’m a big fan of local AI and I think it has to be the future.
It’s ridiculous l, like Bonsia AI’s Q1 models are like 3.5GB easily doing basic tasks that most people are asking 600GB flagship models.
Who on earth would pay for something that needs a basically terabyte or RAM that only performs 10% better
Replying to @fizzle@quokk.au
I think it’s important to realize that a lot of these data centers aren’t just doing LLM for chat bots
I’m guessing a lot of Meta’s and xAI’s data centers are mostly training and running content recognition and recommendation models.
Just speculating, but like HUGE embedding and CLIP models for high resolution databases of all the trash on Instagram/X/tiktok/YouTube paired with large multimodal Recurrent Nueral Nets that look at user’s activity to recommend more content and ads
Replying to @Patnou@lemmy.world
Now all of a sudden the US AI companies care about intellectual property theft.
If they can steal the entire worlds work, I think other people can distill their models into more efficient models