site-packages1

He did breach basic ethics and standards of professional conduct by his actions, for sure. I would lean against considering what he did illegal, but I think there is an argument to be made that it would be illegal under the CFAA.

I think this is the public one that's generally posted in the html for the client side stripe portion, not a secret.

This is really cool!

I used to have these same pains. My trick with spot instances has been to set my maximum price to the price of a regular instance of that class or higher and to sync weights to S3 in the background on every save. The former is a parameter when starting the instance in the console or terraform, the latter is basically a do...while loop. I've noticed that often one gets booted from a spot instance causing interruption because the market price increases a few cents above the "70% savings price." Increasing the maximum to on demand is basically free money because you don't get booted often, and your max price is the regular on demand price.

This has seemed to mitigate about all of the spot downsides (like interruptions or losing data) because you don't easily get kicked out of the spot unless there's a run on machines and the prices rarely fluctuate that much (at least for the higher end p3 instances). This has seemed to prevent data loss and protected the downside risk by setting the instance to a knowable max price. There are instances where the spot price goes higher than the on demand price, so you still get booted once in a while, but it's very infrequent as you can imagine most people don't want to spend more for spot than on demand.

Anecdotally I still average out to getting the vast majority of the spot savings with this method with very few interruptions. Looking at the SpotML it seems to be a lot of tooling that achieves these same goals (assuming one would be interrupted when a spot dies and moves to full-freight on-demand with SpotML), which makes SpotML's solution feel very over-engineered to me given that the majority of what SpotML provides can be had with a simple maximum cost parameter change when spinning up a spot instance.

I would be very interested in using anything that doesn't have great overhead and saves money. Our bill seems "big" to me (but I realize it may be small to many others), so even these small savings add up. Would you compare the potential benefits of SpotML to the method I described above?

In my experience, may of these companies (Riot, et. al) are not particularly consistent over time or in the application of even current rules. And it's also hard to divine what "would be considered cheating" by these companies as one support person may say something is cheating, and another might consider it cheating and ban your account.

It seems very fraught and risky to use for gamers who desire that their accounts don't get banned.

CopperheadOS is an open source OS that builds on Android and can be used on the Pixel devices. I've found it to be quite secure.

Not a scientist or otherwise expert, but this seems like a silver lining in the climate change that will eventually lead to humans not being able to inhabit the planet and the impact people are having on the planet: plants get more food and grow more. Am I wrong in that thought?

One reason is that they now have the highest stock price and that's a cool selling point for BK. Not their only one, clearly, but a neat point to make.

Jobs were good, but lots of people move somewhere for work. I know people who moved to the miserable (according to them) tundra in Canada for super well paying oil jobs, but they didn't want to move there because of the tundra, but because of the jobs. Calculated decision. Many people "want" a high paying, stable job and thus move somewhere because of that.

Anecdotally, I spent two decades in Houston and have a pretty good feel for it. It has many redeeming qualities and will always have a place in my heart, but most people I talk to find the flatness, lack of seasons, and oppressive heat to be too much of downsides to "want" to live there. I don't think I have ever met anyone ever who moved to Houston because they wanted to move to Houston in the same way people move to Florida or San Diego or Denver because they want to move to those places.

i.e. not necessarily less trustworthy, but less able to enforce their will upon you

Respectfully, I was quoting you in my post, and don't think I was taking it out of context.

You said

"It's because the government isn't siphoning everybody's information"

which is a blanket statement, and now you're limiting that statement to being via PRISM, which is a tightening of the requirements you were talking about. The link you posted states that the government is siphoning Americans' data via NSL letters and the like, which is true. I do not know enough about PRISM to say one way or another, but I can see in the last few comments you're thrashing on this.

This isn't correct information. I thought you were referring to something more meta, and not a reverse-conspiracy theory that the government _isn't_ doing what we all have seen it is. Additionally: Your link doesn't support the general claim that the gov isn't siphoning data, but provides some evidence that it doesn't have a _direct_ link to data creators. It simply states that it's using mechanisms like NSLs to obtain such information, which is basically the same.

"He seemed absolutely convinced that PRISM was a big story, which anybody with a fifth grade reading ability could tell was a non-story."

What do you mean by this? I read the article, or at least skimmed it. Is it because everyone siphons everyone's data and the fact that the gov is doing it is unremarkable?

It definitely seems to be. I just opened the app right now and the room titles I see are:

"Breakfast w/ Champions the Millionaire's Breakfast Club"

"Money vs Passion: Big Entrepreneurs Open Up"

"Welcome to Clubhouse! Get +1000 followers, likes, and comments"

"NFTs for Dummies"

"SuperDuper Coin takes over CRYPTO!! Less than $0.01!"

"Big Instagram Room: Marketing 101"

"Rising Affirmations"

"NFTs tips 2 Month Anniversary get your POAP"

"Marketing and Selling NFTs"

"BOSS MOMS: Branding Like a Boss"

"DAILY WORKSHOP: Reclaim your passion"

That's just a snippet, surprised to not see more around NFTs, and you can imagine there's rocket and star emojis strewn throughout all those titles. I basically stopped logging into clubhouse. There are some good conversations here and there in my areas of interest (none of the above) but even in those people are just preening to get "on stage" and when they talk say nothing of substance "let's take a step back here and think about X" or "where do we draw the line?" seem to be the go-to phrases of someone who has nothing tangible to say about the topic.

I would welcome a Clubhouse type startup with actually vetted people that eschewed topics/influencers or which didn't attempt to expand. Like, cap Clubhouse new members 6 months ago and that would have been an amazing app with lots of interesting and successful people talking about interesting topics without all the noise.

I have thought about this a bit. My take is that almost all of these blog posts are from one-off contributors and not one you'll see over and over. I kind of see it as them taking their shot as making you a recurring visitor or recurring reader of their content. I've personally never heard of Statista and don't think I ever will again, if past experience is an indicator. I don't fault them for that, although it is slightly annoying.

Though if it were me, I'd personally wait until someone was partway through the article at least before presenting a newsletter sign up. There's no way I'd sign up for a newsletter before I saw the site's content.

But multiple headlines can both be accurate with one being more engaging. These aren't mutually exclusive.

Article about a bank robbery:

"First National Bank Robbed"

"Gunmen Rob First National Bank in Daring Robbery"

~I think you have these two backwards: "1. Implicit distribution learning - VAE or autoregressive based techniques 2. Explicit learning of the distribution - GAN based models. The way these two model the distribution is fundamentally different."~

edit, someone else pointed this out hours ago and provided a much more detailed answer.

StyleGAN was trained on something like 3 million images, and it's very difficult to fine tune GANs on a small subset of images past that, unless perhaps your images are very similar to the ones StyleGAN used. It's likely you'd need at least thousands or hundreds of thousands of images to get the results you probably want. I've been successful with thousands, but not less than that.

I would pay so much more than this for help with random Pytorch/Tensorflow questions throughout the day.

I don't think I would use a service called "leetresumes" but having your own site is good advice as a place with high SEO value when people DDG your name. However, this doesn't replace LinkedIn, unfortunately. Also to echo someone else, hiring managers love LI for good reason, it's easy to source a bunch of candidates with potentially niche experience.

Or, presumably, holding the spacebar down and letting the repeat function do its thing

I think the joke is that the automated scanners weeding out resumes as in the article posted look for keywords like "React" or "Machine Learning" but without context, so when they see "Machine Learning" on a CV, they are more inclined to accept it, not knowing the context. As an interviewer you can simply play dumb on that aspect.

I think the title on HN is different than the title in the actual article, and has different implications. The HN title implies that getting "demoted" wasn't the goal but a bad effect of following dreams. The title from the actual article, "I followed my dreams to get demoted to software developer," is much better: clearly the "demotion" was the goal of following their dreams.

This is neat, but it's less "How to Train a YOLO4" and "How to put your data into [our product that abstracts the training of YOLO4 behind the scenes so you don't even have to know what you're doing]." That's totally fine, to be clear, but the title makes it seem like one will learn about training YOLO4 models, when instead one is really learning how to use the roboflow dashboard to put images into their system, the fact that it happens to be YOLO4 under the hood is incidental.

Really interesting. There's an empty spot in my building that's a coffee shop storefront. When we moved in, I saw that it was vacant (beginning of 2020) and asked about rent. It was $9k/mo. I offered them $5k and they declined. Still vacant. That's a lot of lost income, but perhaps it makes sense in the sense of commercial real estate.

This is spam of a landing page for some concept, not any sort of tech article explaining anything, let along what a "MrBeast Burger" is.

I played with a really nicely Oculus set up the other day, it was totally amazing and immersive. In particular this was a racing sim, and when I was headed toward a wall I got the feeling like in a real car when you're about to hit something. It was very cool.

Unfortunately, I lost my lunch after 5 minutes or so and had to lay in a dark room for a couple hours to get over a bout of nausea. Excited for the future though if they can solve that problem for people like me (I also get seasick easily).

I don't car for Uber because of their business practices, but wow is this article biased against them!

The basic idea is that there are different information systems governed by different agencies under NARA regulations.

Preservation of materials regardless of admin is paramount to democracy.