I build art using an M1 Ultra with 20 cores. Many times a day, I have all cores running for 5-20 seconds at a time. I have also never heard a fan running in my Mac Studio. Having a single core that runs at a crazy speed but is cooled by some crazy cooling fluid will not make much difference to me. I can see a single-thread game being helped by these processors, however. I can't see anyone being silly enough to run these in a datacenter all day long unless you build it in space.
I think AI-generated images are worse for training AI generative models than LLMs, since there are so many now on the internet (see Instagram art related hashtags if you want to see nothing but AI art) compared to the quantity of images downloaded prior to 2021 (for those AI that did that). Text will always be more varied than seeing 10m versions of the same ideas that people make for fun. AI text can also be partial (like AI-assisted writing) but the images will all be essentially 100% generated.
I found out I was barely in the diabetic range a couple of months ago; hopefully, with diet changes and Metformin, I will see my A1C at a better number in a month or so. Otherwise, I would like to try a CGM to see what I am missing.
I had a Lisa on my desk in late 1983, along with an Apple II and Apple III and later a PC/XT (not sure if that appeared before they took the Lisa away). The Lisa was for evaluation, so I wrote some reports with it but didn't get to write any code. It was ridiculously expensive, but I could tell that it represented the future of personal computing. Explaining to my friends how it worked was difficult; even the concept of a mouse, much less fonts and a bit-mapped display, was foreign to them. None of my friends even owned any kind PC so they had little to compare my description to.
But they do own their own kitchen. I thought of moving to Berlin (from US) at one time, but the idea of an apartment not having a built-in kitchen is insane. Apparently, it is very common for the apartment to have an empty kitchen with nothing but hookups, and when you leave, you either take the kitchen you put in with you or sell it to the next renter.
As a musician, I found the pieces unremarkable. Of course, a lot of contemporary music is forgettable as well, as people try to create songs that all sound like hits but, in doing so, create uninteresting songs. I wonder what music the model is based on. I suppose for game music/sounds, perhaps its good enough?
It's not like the history of working for game companies is unknown, major studios have always done this. Imagining it won't happen to you is wishful thinking. I worked for a small, poor game company once with a FPS MMO and it was probably my funnest job ever, but it paid very little, and I eventually had to leave and make real money again. The best thing was that the player base appreciated every one of us, and constantly thanked us for making things better. You can have that dream job building games, but it's probably not at a big company with big salaries.
No one tests their apps on declared unsupported hardware. How would you identify potential unsupported-but-might-work hardware? It's not worth the cost.
Still, it sounds like someone simply screwed up a compiler setting.
In this modern age where inventions are generally built on those who came before, it's hard to imagine how inventive this design was with so little prior history (and how crazy it was to fly at all). I wonder if the Wright brothers would have been as successful as they were without Lilienthal thinking about it first.
I once worked for a company that stored all of its passwords for production servers and databases in a text file in the code repository, all because the chief architect didn't want to have to remember any passwords. Pointing out how stupid that was to the CTO, got me a statement "we trust our employees" and "we passed our security audit". My face is still hurting from the facepalm.
OK, let's see it make full-sized videos first; making tiny demo videos is a long way from showing it at 4K. Also, let's see the entire paper and note how many computing resources were required to build the models. Until everyone can try it for themselves, we have no idea how cherry-picked the examples were.
Hudson Yards in NYC makes sense on paper for that size/density of a city, yet it's a financial failure. Something like this building in OKC makes zero sense to even start.
I have never flown Spirit due to being 6'5". 30 in seat pitch is anatomically the limit, and that hurts. JetBlue having 33 in makes it a good choice for me.
As a newly diagnosed Type 2, but not overweight, I take metformin and have a carefully altered my diet to slow down my digestion and have few easily digestible carbs (i.e. such as the whites, flour, sugar, potatoes and rice). But it takes 3 months to see if my A1C goes down (a measure of the past 3 months of blood sugar), so I won't know for a couple more months. Going keto to me seems like using a shotgun to kill a fly on your foot, it is too hard to keep up with, and causes other issues, while simply understanding how carbs affect your blood sugar is much easier to do.
Part of the problem is that all the tables of carbs assume you are only eating the one thing (which is how they make the tables), but the actual affect on blood sugar spikes is the sum of what you ate. For example, eating a boiled potato by itself will spike a Type 2 steeply, but eating it with lots of butter will not.
I prefer to understand my health issues by reading a wide variety of information from people who specialize in them, not just read one guy's opinion.
A designer complained to me that her Photoshop blur requirements did not match my iOS implementation. I had to manually adjust my code until it met precisely, which was a pain. A few years later, she became a developer and admitted she was wrong to complain about exact matching between platforms.
Thanks for the explanation. Reading the paper, it seems it still takes 2-4 hours to train (one scene?). I imagine that's still faster than any manual method.
Could someone explain what splatting is suitable for? I see lots of recreations of photographs. Is this some a kind of compression technique, or is there some other usage?
Visiting the Tank Museum is something I do when I go to England. I wish I could be there on Tiger Day and see it driving around. It's an amazing machine to see in person. Having four actual WW2 Shermans plus the Tiger in Fury made the movie so much more amazing; it's not a great movie, but the actual vehicles made it special.
I wonder if holding stolen digital data acquired by illegal means in cloud storage is akin to holding stolen real-world goods, like by a pawn shop or fence. Of course, the hackers may have sold the contents already making deleting it moot.
As a generative artist, I experimented with flow fields for a while but found them rather limited, especially given how many people use them. It's hard to distinguish yourself if you do what everyone else is doing.
Search was the biggest feature of the web in the early '00s. Now it's such a mess. I can't imagine Search will ever be amazing again, given all the complexity of providing quality while still avoiding all the crap.
Given that the game is always the same (layout and wind/gravity/physics don't change), it's not surprising the "robot" succeeded. This is less about reasoning and more about trial and error and memorization. I doubt you need much AI to build such a system.
Why does Midjourney continue to insist on using Discord as a UI? I'd much rather they release a command line tool I can run anywhere if they insist on a text interface.
FYI, that API requires entitlements to be used, which are only available if you request them from Apple and justify their use. It's not a general-purpose API any app can use.