stu2b50

I don’t see why there’s outrage. Facebook released both the raw models and a few fine tuned on chat prompts for a reason. In many commercial cases, safer is better.

But you don’t want that? No problem. That’s why the raw model weights are there. It’s easy to fine tune it to your needs, like the blogpost shows.

It’s hard to blame him for that, tbh

That's on the things (vbulletin forums, etc) that Discord and Reddit have replaced for being so bad that people would prefer using Discord over them. It's hard to blame Discord for making a good product. As to users, a UX principle is that users aren't wrong, the UX is wrong. There's a reason people use Discord for these purposes, and you need to resolve that reason, not make punitive actions towards users.

They can be nice as small servers. OSX is not as suited to server usage as a BSD or Linux and doesn’t have expandable storage - NUCs have additional m2 or sata drive ports. The NUCs also cost less.

I don't think it's a scandal, it's a natural thing that happens when iterating on models. OP doesn't mean they literally train on those tests, but that as a meta-consequence of using those tests as benchmarks, you will adjust the model and hyperparameters in ways that perform better on those tests.

For a particular model you try to minimally do this by separating a test and validation set, but on a meta-meta level, it's easy to see it happening.

I think more Apple. It's not like Google or Microsoft would want to use LLaMA when they have fully capable models themselves. I wouldn't be surprised if Amazon does as well.

Apple is the big laggard in terms of big tech and complex neural network models.

It's not like they ditched them for x86 devices. If people said that 8GB M1s were more capable than their x86 machines, it can be logically consistent with them going for even more power as more M* series chips released.

PPP was vacuously legal since it was introduced in… a law.

Then it should be explicitly a penalty to white people, no? This is the same country that passed a law called the "Chinese Exclusion Act".

I'm not sure I'd lump UNC in there. It's a fairly elite public school, but still a public school, with 20k students and a 20% admission rate.

Sure, but the only qualification to be on IBR is that your IBR payment isn't higher than your standard repayment payment, so it's less a qualification and more a guardrail. Filling out a single form every year on a website is not the worse fate.

The debt is of course on your credit report, but if you are making payments, more likely than not it's going to be beneficial, as young people are unlikely to have other debts that are considered "old". My own student loans carried the "age of credit" criteria on my credit report for a while.

They are a percentage of income, but that is no different than the UK version. And to be fair to IBR, it's more nuanced that that - the income cutoff for the percentage varies state-by-state (so, someone in New York has a much higher cutoff than someone in Arkansas).

I’d also note that this is how it works in the US as well for loans provided by the federal government. In addition, as long as you make payments, excess interest will not accrue and your principal won’t increase. They’re also written off in 20 years.

It’s mostly with private student loans that people go completely underwater in the US

The starting premise is that some students have families in different areas than their schools?

Because it looks ugly? It's like when you shove all of the mess in your room into the closet before a guest comes in. The prompt is an ugly implementation detail you don't want users unintentionally coming across.

It’s a sliding scale in the end. You can have a fusion system that barely breaks even, or only slightly breaks even, and then is still more expensive than traditional energy sources after factoring in fixed costs and maintenance.

That's not really a correlation that exists anywhere else. A white collar office worker on average makes significantly more than a blue collar worker but they don't particularly take more risk or anything.

People are paid based on opportunity cost. Executives are paid very highly because the cost of them making a poor decision, or not making a good decision, is very high. Meanwhile, the cost of hiring me and me not performing to expectations is low. It's not because they're God's or anything. The supply of people with leadership experience is also definitionally low.

Of course there's a lot of old boys club in executive roles in reality but that's the boards and shareholders problem when they do a bad job.

Making a mistake is not in-it-of-itself a problem. Mistakes are inevitable, if nothing else because people cannot be oracles. If you bet into the river because you have a flush draw, you'll lose around 75% of the time - but that doesn't mean it was a mistake in those 3/4s of a time.

If as a board you unduly punish executives when bets they make don't pan out, you risk causing a culture of fear that prevents innovation and big bets.

Obviously there's a limit, and the reasoning is important behind the bets.

That being said it's not like all of the tech ceos were perfectly competent - that's not what I'm saying. But the knee-jerk "well, fire them too" reaction from the internet is mostly an emotional response.

Hindsight is 20/20, though. The narrative from the company isn’t that “we purposefully overhired and are now laying the excess off for fun”, it’s that they believed they would have use for the excess given the trajectory of the company and the overall economy but they were mistaken. Which is not particularly strange - the future is famously hard to predict.

It’s not necessarily clear, but it’s also definitely not clear that the layoffs will be detrimental to the company either. The title of these types of articles implies that the dividend and the layoff are at odds with each, which they aren’t.

The placebo effect is strong, especially for things like alertness. There’s been studies purposefully showing that participants who think they’ve been given a mental stimulant perform better in cognitive tasks in a statistically significant way, for instance.

Liquidity?

Only if the product has completely inelastic demand? Also, that'd depend on the goal of the tax in question?

That seems less efficient than just searching through the contents without training an LLM to adopt the persona of the suspect. You'd have to double check everything they say, because obviously any of it can be false. So at best it's pointing you to the right direction, but realistically not.

Might be a good lead generation if the FBI is completely stumped I guess? Or they're just bored at work?

The hype is because Intel and AMD have an enforced duopoly of x86, because only they own the licenses to x86-64 and you'll take it from their cold dead hands.

So, if you want to compete, you'll have to go ARM. Amazon graviton and Apples M series have only made it more appealing.

The main reason to switch is cost.

The subject of the article are internships which pay very well. Big tech internships are usually between $45-50/hr. Some people I know make more money at internships than their entire annual PhD stipend.

Sometimes things just don’t have good solutions in one space. We solved in another space, as SSD and ram manufacturers made memory exponentially cheaper and more available over the last few decades.

So we make the trade off of software complexity for hardware complexity. Such is how life goes sometimes.

Hardware cost money? The rate at which AWS buys more hardware is consistent with their income from their cloud rental business. If they're not going to buy the hardware otherwise, then evidently at cloud market rates the marginal income isn't worth it.

Hardware, rackspace are all resources. If something is using up hardware and rackspace, there is opportunity cost.

That doesn’t really change anything. The opportunity cost would be the same, which is what matters for companies. Those resources could be sold for market value rates to a customer if they weren’t being allocated to DPReview.

The point I got from the article is to bias against medical interventions. That doesn't mean never consider them, but they really need to have a serious reason to occur.

It doesn't have to that doctors are bad at their jobs, but rather that they're humans, and humans are hardwired to try to make other humans happy, and patients like being actively given treatment as opposed to "have you considered sleeping well, eating well, and just letting your body do its thing?"

The sources that the article has, an article in the BMJ[0] and Cochrane[1], are actually targeted at primary care physicians (after all, random laypeople don't read medical journals!), to basically tell them to chill, and tell their patients to just let themselves be.

[0] https://www.bmj.com/content/348/bmj.g3680 [1] https://www.cochrane.org/news/featured-review-are-general-he...

You also have to remember that this study was done by Danish researchers and involved a meta-analysis on not just US healthcare, but also European healthcare systems.