Forgot your password?
typodupeerror

Comment Re:What? (Score 1) 100

You probably never even had access to an airline lounge until some recent credit card perk. And it's excessive credit card perks that has created the need for private terminals.

They're overrated. Sitting at the airport eating mediocre "free" food and reading newspapers? Then walking a long distance to the terminal before the flight leaves...

Unless the airport lounge is right next to your terminal, or unless you really REALLY like being told you are special, credit card lounges are overrated.

Comment Re: So basically... (Score 1) 193

There aren't really any unsolved engineering problems. SpaceX can absolutely put a rack of nvidia GPUs into low orbit. We could have done that in the 70s. The argument is whether it's economical or not.

Bringing the cost down is also an engineering problem. NASA and the ESA both said reusable rockets weren't economical even though they already knew it was possible. The ESA even famously poo-poo'd the idea, exactly like you guys are doing here. I personally don't know whether this will work. If somebody -- anybody -- has an idea for how they can make it work, then unlike you and apparently most others on slashdot, I'm not going to try to stop it, nor do I see any good reason for that.

Besides, I'm not seeing the argument for fraud, which is what GP asserted, and is what I responded to. If you disagree, then who is defrauding whom? Perusing an idea that ultimately doesn't work isn't fraud, it's just business. Most ideas don't work, which is why so many businesses fail. That's why investing carries risk.

Comment Re:Land of the free ... (Score 2) 91

True. Except for some wines, peanuts and almonds, I cannot think of a single US food product I could easily get here.

How do you survive without Doritos, Pringles, and Twinkies?

Serious response to a jokey question: they don't. You can buy Doritos and Pringles in pretty much any European country. They're manufactured locally with recipes tweaked to meet EU requirements and local palates. I've had both the American and the European version of both and couldn't tell the difference (though I'll admit that I'm not a snack food connoisseur).

Twinkies are trickier. I believe the only places they're commonly available are the UK and Ireland.

Comment Re: CostCo crap for a few pennies (Score 2) 91

The only way it ends up in food is when you overheat your teflon pan OR you scratch it and the coating starts coming off.

How does that all square with the petitioner's finding that PFAS were found in a number of food products, including "extremely high" levels in some milk brands, which presumably weren't first scalded in a consumer's non-stick skillet?

Comment Re:"Reasoning" (Score 1) 179

Tokenizing might be a problem for some neural networks, but more generally they fail at the balanced parenthesis problem (scroll down to the section "What Really Lets ChatGPT Work?" and scroll down to the pictures of parenthesis).

It's related to the clock problem.

If you ask a neural network to recognize "A" then all As will be tokenized the same way, so it's will be able to recognize it as well as anything else.

Comment Re:Replicated already (Score 1) 179

I can tell you are a fanboy, and not a person with a scientific background because you keep ignoring the main point:

The philosophical claim is a hypothesis that isn't supported by the evidence presented in the paper

the closed-source nature of these LLMs makes reproducibility very difficult,

Difficult is not impossible.

Comment Re: AI Company says their AI is the bestest boy (Score 1) 179

Actually that sounds really good, you seem to have calmed down. Nice.

The next step is to be more clear in your language. LLMs clearly didn't "solve context sensitivity," but you know that (having implementing rule-based natural language grammars). However, you did seem to have an actual point, although I can't figure it out. So it would be interesting if you dug it out and expressed it a bit more clearly.

In your last comment you talk about performance, but the numbers are a little unclear here, so we would want to clarify that. Also, there is frequently a tradeoff between accuracy and performance (always with neural networks, since if the data set is small enough, you would just calculate an exact solution).

Comment Re:Replicated already (Score 1) 179

A far more interesting and nuanced take than your naked dismissive statement.

It's politer, but it's not nuanced. He clearly says there was no evidence in the paper to move him to accept the philosophical claim.

He's polite because he was hand selected by the company, but the meaning is the same: The philosophical claim is a hypothesis that isn't supported by the evidence presented in the paper. Being that is the primary claim of the paper (and title), it would therefore be rejected in peer review (or modifications requested, since the other hypotheses are supported).

Slashdot Top Deals

"The sixties were good to you, weren't they?" -- George Carlin

Working...