You don't have to sanitize the weights, you have to sanitize the data you use to get the weights. Two very different things, and while I agree that sanitizing a LLM after training is close to impossible, sanitizing the data you give it is much, much easier.
Oh no, it's very difficult, especially on the scale of LLMs.
That said, we others (those of us who have any amount of respect towards ourselves, our craft, and our fellow human) have been sourcing our data carefully since way before NNs, such as asking the relevant authority for it (ex. asking the post house for images of handwritten destinations).
Is this slow and cumbersome? Oh yes. But it delays the need for over-restrictive laws, just like with RC crafts before drones. And by extension, it allows those who could not source the material they needed through conventional means, or those small new startups with no idea what they were doing, to skim the gray border and still get a small and hopefully usable dataset.
And now, someone had the grand idea to not only scour and scavenge the whole internet with no abandon, but also boast about it. So now everyone gets punished.
At last: don't get me wrong, laws are good (duh), but less restrictive or incomplete laws can be nice as long as everyone respects each other. I'm excited to see what the future brings in this regard, but I hate the idea that those who facilitated this change likely are the only ones to go free.
So now LLM makers actually have to sanitize their datasets? The horror....
There has to be a law against such heretical actions somewhere! Even if it's .00, this computer is an affront to order! I propose we burn it alongside those frivolous computers who think they can simply name themselves .0 or .255!
I wish Denmark never gets an active presence of these groups. Oil, coal, and natural gas is irrefutably fatal and one has loads of reasons to despise those who lobby for it, but I despise those who cause terrorism (doesn't matter if it's destruction of cultural artifacts nor environmental disasters or indirect support and growth of oil, such as what Greenpeace is known for) with about equal amount.
Combating oil lobbies is extremely hard, albeit absolutely necessary, but I don't think having the common populace hate you even more than they hate the lobbyists is a good first step.
Also, Denmark appears to already be trying to work towards solutions for a better, more green future using windmills and by reclaiming farmland for more natural zones (hopefully they don't end up selling it to international farmers instead), but do indeed lack quite a lot behind on actual usefulness of all their hard work.
Regarding this, a lot of technology is not fully mature yet, we have windmills, yet are not sure how to properly maintain and decommission them in a proper manner. We have the tech, yes, but, as far as I'm aware, were currently facing much the same way as plastic. We CAN recycle plastic, but how much is really recycled? Also, do we commit to lithium polymer batteries which are very dirty, or do we attempt to use non-mature technologies such as hydrogen-pastes?
Huh, is Windows screwing over GRUB and Linux not a bi-monthly experience anymore? Sad that it happened, but glad that it's become novelty enough to write about.
Just detach the blades. You can always re-attach them when you've landed.
We get to choose the genes when genetically modifying, and it usually takes a few years (plus health metrics and research once complete).
Contrary, when selectively breeding we can breed for traits which we are not guaranteed to actually get, and it takes a few decades (plus health metrics and research once complete).
Huh, I'm not sure they are comparable.
Didn't USB A and USB B use a master-slave relationship in which the male would (generally) always be the slave, whereas USB C uses agreement and discussion to decide the master and slave roles regardless of connector gender.
Please do correct me if I'm wrong. Also, do we say "agent" now instead of "slave", or what is the new term?
Should have seen him before!
The absolute ridicule! I'm sorry, but I might not survive this! How could this come to be?!
Well, almost (49/51%).
Needless to say, the steam deck has definitely found its place playing Monster Hunter, Graveyard Keeper, and sometimes even Guild Wars 2 and factorio.
It does run Deep Rock Galactic and Vermintide 2 too, but I feel those are better played on the rig.
Who else is excited for Rootkit "anticheat/DRM" requirements for web browsers? We all already give games full system access, so why not do the same for cookies?