Not an expert, but this paper explores double descent with simple models. The interpretation there: when you extend into the overparameterized regime, that permits optimization towards small-norm weights, which generalize well again. Does that explain DD generally? Does it apply to other models (e.g. DNNs)?

https://arxiv.org/pdf/2303.14151.pdf

I usually have a good experience, but just in the last few weeks it’s been bad:

- Driver started watching TikTok videos on his dashboard-mounted phone while driving

- Driver going way over the speed limit in stormy weather

- A driver was clearly sick, coughing and sniffling and sneezing the whole ride

- Got in a car and it reeked of marijuana

- Got in a car and it reeked of BO

At this point I’d probably prefer a robot driver.

Some of these will still happen, even in a self-driving car, because of prior passengers. In fact they might be more of a problem because passengers won't have another human in the car inhibiting their impolite or gross behavior.

I look forward to Snow Crash's self-cleaning public restrooms.

Pretty sure you get filmed while you are in the car so smoking is going to be a challenge.

While some may have doubted the iPhone, Apple sold 1.9 million iPhones in the year it launched, and cell phones had already been established as an enormous and growing market.

VR seems very different from that. Most people simply don’t want what companies are trying to sell them, and I think multiple major technical breakthroughs would be necessary to change that.

While there were absolutely people who thought the iPhone would fail, the idea that most people were skeptical about the iPhone is simply ahistorical. The iPhone had a huge amount of hype from the moment it was announced, and it was pretty clear from the outset that this was the direction cell phones were headed.

The Vision Pro just doesn't have that. Apple was able to clearly articulate what the iPhone allowed you to do that you couldn't do before. But the use cases presented for the Vision Pro just are a lot more niche, and a lot less compelling. Some people will certainly find uses for it, but right now the Vision Pro feels like a solution in search of a problem.

It hasn't been a year of AVP sales yet.

And as you say, the VR market is small compared to the phone market of 2007.

It wasn’t a full year of iPhone sales either: it launched in June and sold 1.9 million units in just six months.

The VR market is small for a reason.

> You should see Google's turnover numbers from 4 years ago, much less now.

High turnover was industry-wide a few years back because pay went through the roof and job hopping was the best way to capture that.

I suspect it’s lower now, following mass layoffs and substantially fewer openings.

It’s easy to dismiss everyone who doesn’t like being surrounded by fentanyl and meth addiction and crime as “right wing,” but it just makes them sound pretty sensible.

All my friends moved away over a span of about three years. I held on as long as I could but I finally threw in the towel too.

Easy way to tell which longtime SF residents have Stockholm Syndrome: they blame everything that is completely broken in the city on "right wing media."

ImageNet was very influential, but this just shows he was eighth author on a twelve author paper from almost a decade ago. Is there better evidence of sustained contributions to the field?

Hm, well, I see on his resume that he was a founder of OpenAI, recruited to be Tesla's head of AI, went back to OpenAI, and also has the most viewed educational videos in this space.

So, he has made theoretical contributions to the space, contributions to prominent private organizations in the space, and broadly educated others about the space. What more are you looking for?

Tesla fumbled big on AI, and as for his work at OpenAI, he just left, had he been good enough they would have made him a financial offer that would have made him continue. But, I'll give him that, he seems to be a really good teacher.

Not everyone is purely motivated by money so. I know that the moment I decided to quit or switch jobs, no, and I mean litterally no, amount of money would change my mind.

Me changing can never be used as an appraisal of my old organisation so.

Disclaimer: regarding money, if I get enough in max a year to rezire forever after that, I might be tempted. Which won't happen, because a) I'd just leave a year later anyway and b) nobody would pay me high 7 figures just to not quit.

I doubt he left because he wasn't being compensated fairly.

People just get bored and go do something else for a while sometimes. Or he's got some beef.

many SOTA papers for multi-camera deep fusion/birds eye view perception in autonomous driving were based on copying teslas homework after their AI day 2022 talk

the commenter is probably a junior boot camp web dev...

You left out a few prepositions like off the space, toward the space, from the space, and down the space.

He’s first author on a ton of those papers. That’s a tenure worthy CV almost anywhere. Gimme a break.

OP shared a link to just the Imagenet paper.

Agree he had a decent overall track record at Stanford, but that’s not how tenure works — it might have got his foot in the door as an assistant professor somewhere. He chose a much more lucrative path.

Offering worse content while continually raising prices makes it difficult to compete

There’s more about the effort to publish his complete works here: https://en.m.wikipedia.org/wiki/Opera_Omnia_Leonhard_Euler

It should be submitted at least annually

Same here. No matter who you are, if you conspire to steal from so many employees and workers, you deserve prison.

Especially since it's acutely pertinent to Masimo v Apple that is currently ongoing.

A few years back, 1-bedrooms in dire need of renovation were going for $3-$4k. Sounds like things have really fallen off. I still wouldn’t move back, unless I were 15 years younger and needed to jumpstart my career