Microblog Memes

[–] Bebopalouie@lemmy.ca 9 points 2 months ago (1 child)

We opted for downloading as many books as possible from a number of pirates sites and stashing them on a usb drive away plan instead. Wife and son have been very busy.

https://www.kavitareader.com/

hideshow 1 child comment

[–] ILikeBoobies@lemmy.ca 2 points 2 months ago

parent

[–] Skullgrid@lemmy.world 12 points 2 months ago (10 children)

that last one doesn't add up

hideshow 10 child comments

[–] reddfugee@lemmy.world 25 points 2 months ago (9 children)

I think it's directed at the "I don't use commercial LLMs, I just run small local models" crowd

hideshow 9 child comments

parent

[–] Skullgrid@lemmy.world 6 points 2 months ago (8 children)

yes and what books are they destroying?

hideshow 8 child comments

parent

[–] tyler@programming.dev 16 points 2 months ago (7 children)

Those smaller models are trained on the larger models. They take just as many resources to train and then you add on more training, so they’re literally worse than the big models.

hideshow 7 child comments

parent

[–] Skullgrid@lemmy.world -1 points 2 months ago (6 children)

What if your power comes from renewables? And if you're training private data?

hideshow 6 child comments

parent

[–] Voixe2@lemmy.blahaj.zone 14 points 2 months ago

What if a unicorn gave me the first ethical local model, now all my sins are absolved

parent

[–] ayyy@sh.itjust.works 4 points 2 months ago* (3 children)

And what if tech oligarchs weren’t evil and used this new technology to make lives better instead of concentrating capital? We don’t have that either so there’s no use speculating.

Snark aside, I think you’re underestimating just how much data it takes to make these models work. There is no such thing as enough “private” data for training.

hideshow 3 child comments

parent

[–] Skullgrid@lemmy.world 3 points 2 months ago (2 children)

There is no such thing as enough “private” data for training.

what the fuck are you on about? You can build local datasets and train additional LORAs locally to have more relevant outputs, and it doesn't go back to the creators of the original weights.

hideshow 2 child comments

parent

[–] ayyy@sh.itjust.works 1 point 2 months ago (1 child)

You are describing fine tuning, no? Which still requires a base model that was trained on a massive amount of other data.

parent

hideshow 1 child comment

[–] Skullgrid@lemmy.world 1 point 2 months ago

Yes and those models already exist, for free.

It's not the unit 731 data ffs

parent

[–] tyler@programming.dev 3 points 2 months ago

Then they still used just as many resources? All these local models are distillations of the big models.

parent

[–] minorkeys@sh.itjust.works 3 points 2 months ago

Your only access to information will be through AI and what you get to access will be based on how poor you are. We can stop this trend, we can stop this future, you'd all have to actually do something though, is that beyond you? Doing a bit of a thing to save the future?

[–] csolisr@hub.azkware.net 1 point 2 months ago

In related news, I genuinely have no idea of why did that account block me before even interacting with it