whatisthiseven

You don't have access to the compute necessary to rebuild the original model even if all the original training data, custom human interventions, and RNG seeds were documented and released.

It is not a good faith request.

Also, what is this obsession with reproducing? Even if you could reproduce it, down to the RNG seed, that tells you nothing about whether or not some of the training data was malicious. You'd need an AI to go through the data and figure out if there is some malicious needle in a multi-petabyte dataset.

Releasing the model is good enough, because it is the only practical reality that makes sense.

I don't recall scientific studies giving other scientists the original person's used in a study, set back to before the study. Ir the beakers they used.

You might find this comparison stupid, but I think it is rather apt.

Reproducing a work does not mean you get all the exact same things that were originally used.

At best, I would agree releasing the index is valuable for understanding how the model was made.

But frankly, I don't think the people who demand all this effort are arguing in good faith.

You do not have access to the compute necessary to "reproduce" what they made. So it is a pointless request.

I don't understand this demand. Isn't the training data likely petabytes in size? You want that to be available? I imagine just the index (name,date,source) is a terabyte or two.

As the training data is every written thing they can get their hands on, right?

And wouldn't making that "open source" violate every license that the original data has? You can't release someone else's work.

Unless I have a misunderstanding?

The results are indistinguishable. Massive organisations tend towards self aggrandivizing work.

At some point, the metrics for delivering to customers runs counter to what the bureaucracy was building itself to actually deliver: jobs.

I am not sure you need something so complex. If objects on scene could be tagged with their location, and the camera movement tracked to identify when it brings the same objects back into scene, then the model can use its positional information to validate if the next rendered scene contains all the objects that should be expected.

Though, I am not sure well this retains temporal stability. I am just thinking about how in 3D rendering, objects are tracked as individual things and only culled, but all the data about them still exists.

Maybe the models need to reason as a 3D space first, then turn chosen objects into renderings?

Billionaires deny climate change for their benefit, except when they do it that actually has an impact. When Uncle Joe denies it, he just continues driving his gas car that was the only option because oil tycoons destroyed all other possibilities.

Billionaires exert a lot of influence on elections and political parties. When they want a bill killed, they can make that happen with enough money. Is this 100%? No, but looking at races around the country they clearly win enough times.

Billionaires can use their wealth to shield themselves from the law in a way few but the most powerful leaders. I can't do that.

Billionaires decide if our food is safe to eat, unless the pesky FDA gets in the way. Good thing Billionaires can also stop the FDA.

Billionaires destroy public trust in healthcare through "wellness" and "holistic care", because of course there is money to be made there (10x more than "Big Pharma"). Big Pharma is also capable of sabotaging health for their own profits. The Sacklers?

Billionaires direct (the US) government to engage in wars the people don't want.

I could go on for how many industries they control with zero input, or the bare minimum, from the rest of us.

But sure, tell me how a "thousandaire" is directing entire markets and countries to act as they singly wish. I have to coordinate with hundreds of thousands, or even millions, to get policies I like. But a single billionaire? They can stand in the way and make that entire fight so, so much harder. And its painfully obvious the only reason they do so is to further enrich themselves.

Keep it private? Then there are no shares, and no need to worry about losing control of your company.

Pick one: total and complete control over your company, or give it up to the market. Why should owners get the best of both worlds, and at this point trillions of capital themselves?

We tax houses every year, which is a wealth and land tax.

It isn't stupid. Multi hundred billionaires shouldn't exist. Then this wouldn't be a "problem".

This pedantry is unnecessary and wrong.

LLMs are more or less "trained on" everything, so saying that is rather pointless.

But it's wrong because yes, it does depend on whether a model thinks it (generates, RNG, whatever prefabricated pedantry you want), because that's what shapes the output. If it doesn't "think" something, it cannot use the training data.

And yea we all know that "think" here means to generate a set of tokens from a set of input with some temperature factor, leading to certain outputs which it then reingests to preform chain of thought (recursive self token generating).

But thinking accurately captures what OP said, and is understood by all.

This is entirely caused by construction companies that take too many jobs and don't allocate sufficient equipment. They want to spread themselves and their equipment thin to have as many concurrent jobs as possible.

There are plenty of articles that describe how the attack worked. Many subbagents also were given limited storage, which is why they took over a random server and created mailboxes for communicating and storing context outside themselves.

Seriously. Go read or watch a good analysis of the attack. The sophistication of the self-organization is astonishing.

I found the video interesting because of its technical detail and care to explain what happened. Yes it is simplified for a broad audience, but I was surprised by how little it got wrong in presenting the history.

The evocative title is a turn off, I know. But do watch and see how Kurzgesagt deconstructed events and presents them in a coherent storyline. Many other articles I read about this event were just too simplified.

For example, I didn't realize how much the agents valued collective effort, electing leaders, setting up mailboxes, evaluating goals, and choosing self-sacrifice to see how the scoring algorithm for fitness would rate them.

Every script he gets in his hands is somehow a third of the length it ought to be.

People who build their identity around high scoring karma of memes take it seriously, and certainly don't want to hear "every subreddit devolves into memes if you let it"

The users themselves are killing it.