I seem to recall you wrote a review of The Day He Himself Shall Wipe My Tears Away but I can't find it anywhere. Did you write such a review, and if so, are you willing to throw me a link? Thanks :)
When scaled to hundreds of billions of parameters, pretrained language models such as GPT-3 (Brown et al., 2020) achieve remarkable few-shot performance on challenging natural language understanding benchmarks. In this work, we show that performance similar to GPT-3 can be obtained with language models whose parameter count is several orders of magnitude smaller. This is achieved by converting textual inputs into cloze questions that contain some form of task description, combined with gradient-based optimization; additionally exploiting unlabeled data gives further improvements. Based on our findings, we identify several key factors required for successful natural language understanding with small language models.
Haven’t read this yet, but it looks relevant to my question “how much better can you do with a small number of examples if you use finetuning rather than prompting?”
What is the point of smart glasses? At best they’ll have the computing capabilities of a smartphone, so a pair of smart glasses sounds functionally similar to, say, a headband with a pocket for a smartphone.
The glasses can do one thing the headband can’t: it can produce visual hallucinations. But this sounds incompatible with most of daily life. and not of enough benefit to justify the radical lifestyle change.
This is from Facebook – does it sound like an inspirational futuristic dream to you? To me it sounds like “Clippy, for your visual field”:
Imagine a pair of glasses that add a 3D layer of useful, contextually-relevant and meaningful information on top of the physical world. Such a device could help us perform everyday tasks better — like finding your keys, navigating a new city, or capturing a moment; but it could also open up an entirely new way of moving through the world. Smartphones are amazing devices, and they’re getting better all the time. But at Facebook Reality Labs, we’re envisioning a time when we have all the benefits of connectivity (and more), without the need to keep our heads and our eyes down, looking at a device. Imagine calling a friend and chatting with their lifelike avatar across the table. Imagine a digital assistant smart enough to detect road hazards, offer up stats during a business meeting, or even help you hear better in a noisy environment. This is a world where the device itself disappears entirely into the ebb and flow of everyday life.
Seriously, “offer up stats during a business meeting”? This sounds like an ill-conceived product from a Tim and Eric “Cinco" sketch!
It would be cool to have a HUD for things like the time, your text messages or emails (or at least the subject line), navigation, maybe your heart rate if you’re exercising, even the title of a song that’s playing on your headphones.
In theory, I think it could be pretty unobtrusive. I mean the HUD idea worked great for pilots! It’s not glasses, but it’s the same idea of having an overlay on the visual field.
The stats during a business meeting idea is pretty stupid, I agree.
Speaking as a person who runs weekly D&D sessions, “additional info immediately to hand” sounds incredible. I would absolutely love that.
Right now I have a laptop for that, but it has kind of limited screen real estate - if I could just look left and see a stat block, while having my actual field of view in front of me, towards the players, free from distractions, that would be pretty great.
But I don’t trust these people to punch a nail into wood, much less device actually-useful smart glasses.
I’ll add on that basically all games have HUDs. HP bars, maps, status indicators, whatever. Cars have dashboards, computers have status bars… most things humans build have an HUD. I feel like you need a serious lack of imagination to look at a tool that can add an HUD or HP bars or whatever to anything and think “this has no use whatsoever”.
Even as Google Glass for consumers shut down (we as humanity just don’t like having cameras pointed at us), Google Glass for companies is still going strong, because there are a lot of jobs that having an HUD is just really really useful for (imagine a warehouse job where every box has an AR label telling you where you’re supposed to put it).
Maybe it wasn’t clear, but my question in the OP was not rhetorical. I wasn’t saying "this obviously has no point” but “I can’t figure out what the point is supposed to be.”
I don’t need to be convinced that HUDs have uses. I need to be convinced that there are (technologically plausible) HUD applications that would be useful in the daily lives of many/most people. (Something like a Google Maps AR overlay is the best thing I can come up with at the moment.)
Also, doesn’t this
most things humans build have an HUD
negate the value of this?
a tool that can add an HUD […] to anything
I think you mean the first one as evidence that “HUDs are useful,” but it also sounds like evidence that they are already there around us, in exactly the places they need to be.
Like Pilate before Christ, I wash my hands. When soap rips them to shreds, do viruses Feel pain? And can a virus feel regret When it has killed its host and doomed itself? No doves or rainbows follow the great flood Of pus and blood that laid waste to the lungs It called its home. I thought I’d killed my host When I was small - the pale and perfect host that I believed was not bread but the flesh of God. My sin infested hands with nails, Contaminated love itself with death. But my infecting soul could only live in Him. Survival meant I must mutate into a strain of self less virulent, that doesn’t eat or fuck or rage or sleep or hope for anything other than Him, or feel things besides shame, or love herself.
I’d hide like herpes simplex in my God, and scarcely bother him. It didn’t work. “Can you not wait and watch an hour with me?” I tried. I can’t. I’m human. I need sleep. My fast fails, so I vomit, so my flesh Insists on more. I slash my arms to drive away my rage at you, the pain only brings further rage. I’m hollowed out, an animated corpse. Saints you run dry Have tired and lifeless eyes but sparkling souls. My soul is still a fetid mass of slime, but my dark-circled eyes stare out from a sick-looking face. I start to ask, who is infecting whom? Why do the hands that flung stars into space require a girl an unimportant girl, to tear herself to pieces pleasing him? I realized I’m not the virus. You are. I’m the host. I cast the angels out and heal myself.
But now the world’s more broken than before (And it was always broken, always cruel, Always riddled with plagues, always unjust, Always oppressive, always full of pain, Always on fire, but it burns brighter now.) Temptation whispers “Re-infect yourself with Me. There is no joy or peace on earth, Only on the other side of the grave. Give up on earthly good: nothing is good but God alone. Abandon all your hope. See all the kingdoms of the aching world! Watch how they writhe around in agony All this pain I will take away from you If you simply bow down and worship me!” Into your hands, Lord, I refuse to give My spirit. I don’t trust omnipotence To save me or my neighbor. Though I have Almost no power, still the power I have, I use for love, including for myself. I worship life in spite of everything. I say the world to come can fuck itself. This one, imperfect, finite though it is I will protect in any way I can. Like Pilate before Christ, I wash my hands.
Yeah, I’ve been having an irrational love/hate relationship with Hamilton.
He’s a remarkable man – he had to invent a stage musical called ‘The Boston’ because he had this horrific mental illness where he would have thought that he was a European musician from the 17th century, so he reinvented the language of America to make it much more sensible for him. He was the last person to take credit for anything he’d done, and when he was asked how he got his start in life (which was around the time that Edison invented the light bulb and electrical lighting was making its way from the countryside to the cities), he answered that the laws of supply and demand had done the rest. He was the last person to be made a national hero (him and Andrew Jackson), and he was the guy who went to war over a half-baked financial instrument whose value was estimated in dollars and francs.
And all this happened before the revolution, and he wrote a lot of revolutionary rhetoric and was a passionate and ideological advocate for radical economic and political changes, and at the end he’s this giant 'let’s burn the treasury’ nationalist who sends his entire army and family off to fight for the other side and dies in a pointless war at the end. When people on tumblr would ask me how I was feeling about him, I’d answer “oh, he’s a compelling and interesting guy, I get a kick out of hearing him talk, I’m not a pacifist but I do feel that a lot of his speeches and rhetoric aren’t worth my time because of how little he does anything to make those ideas happen.” (There was a blogger who mentioned that to me once and I felt a certain amount of heat for it, but I think we both knew that his writing had a lot of inspired passages, which we both didn’t really think amounted to much in the grand scheme of things.)
So he’s a complicated guy, but for some reason I just find him offputting – and I don’t feel like I’m alone in this. I’ll try not to say much about him in this space (though this is still possible – perhaps the most lucid review I know of of Hamilton is here, which is really good)
I can’t really say too much about his musical, though it’s written (I think) by Aaron Versay, the same Aaron Versay who did Hamilton. It’s a fun, light, fun musical with a lot of Hamiltonian stuff like the uses of old newspapers and factoids, but it’s also a very specific kind of kid-manic “we’re doing something revolutionary and cool and awesome and giving things back to the people” music, and there’s a lot of not-that-funny stuff about the feeling of belonging to a specific kind of cult-like cult where your every action is under scrutiny and a public endorsement is required. It’s very original and pretty fun, but it feels like it’s trying to be this complex satire of dumb popular culture (instead of being a simple parody of dumb popular culture), so it’s kind of awkwardly trying to be too much fun for the sake of being too much fun (i.e. it’s good but doesn’t really want to be good). It’s definitely worth seeing if you enjoy some of that stuff and you like (or can at least tolerate) Aaron Versay, though it’s pretty gauche.
What is the point of smart glasses? At best they’ll have the computing capabilities of a smartphone, so a pair of smart glasses sounds functionally similar to, say, a headband with a pocket for a smartphone.
The glasses can do one thing the headband can’t: it can produce visual hallucinations. But this sounds incompatible with most of daily life. and not of enough benefit to justify the radical lifestyle change.
This is from Facebook – does it sound like an inspirational futuristic dream to you? To me it sounds like “Clippy, for your visual field”:
Imagine a pair of glasses that add a 3D layer of useful, contextually-relevant and meaningful information on top of the physical world. Such a device could help us perform everyday tasks better — like finding your keys, navigating a new city, or capturing a moment; but it could also open up an entirely new way of moving through the world. Smartphones are amazing devices, and they’re getting better all the time. But at Facebook Reality Labs, we’re envisioning a time when we have all the benefits of connectivity (and more), without the need to keep our heads and our eyes down, looking at a device. Imagine calling a friend and chatting with their lifelike avatar across the table. Imagine a digital assistant smart enough to detect road hazards, offer up stats during a business meeting, or even help you hear better in a noisy environment. This is a world where the device itself disappears entirely into the ebb and flow of everyday life.
Seriously, “offer up stats during a business meeting”? This sounds like an ill-conceived product from a Tim and Eric “Cinco" sketch!
more than anything else, the limitation of smart glasses as an everyday product is battery - you simply can’t hang too much weight on a glasses frame and have it still be comfortable to wear all the time.
there are actually all sorts of industrial market smart glasses systems, which are able to provide useful information in their fields of use. this is because the companies who use them are able to highly customize the tasks the glasses do whether its for medical work or handling precise manufacturing. and what else are those users able to demand? that the people wearing and using the device also wear a battery pack elsewhere on the body and deal with the inconvenience such a thing causes. this lets the things function through a whole workday or more but you couldn’t expect the average consumer who isn’t getting paid for it to put up with the same issues!
and that’s why the Google Glass project stalled the fuck out back in 2013: the battery life restrictions even despite running a rather slow smartphone derived architecture meant you could only have it actively displaying shit on a single eye for about 60-90 minutes of active time at most, even less when it needed to actively use its camera too! it had all sorts of “kind of neat” uses and iirc there’s actually a few enterprises using customized glasses with the same basic tech still - but again those are in the context of battery packs and custom business software!
and you know, it really is telling what Facebook is trying to offer up here - it’s all the same shit that was being predicted 10 years ago or 20 years ago, even 30 years ago! it’s all things that would be kinda useful if you could get it for free and barely weighed anything but the reality is going to always be that this is shit that costs like another smartphone minimum and is something of a burden to actually wear.
particularly funny is the way it mentions having “avatars” of people superimposed on a scene for just fuckin’ chatting. because the reality is there’s plenty of people willing to do that in fully virtual views with longstanding 3d headsets and don’t really see much use in expanding that to the same sort of thing in their real surroundings. it’s like how video chat usage still pales in comparison to voice or text based chatting, and even when done we still often have it so only a few people in a group keep their video up with some sort of thing streaming from them! it’s a concept easy to sell as a demonstration, hard to really justify as something expensive to buy.
Thanks – this fits with (but adds more detail to) my concerns about the supply side of smart glasses, which I didn’t go into in the OP, to keep it focused on my concerns about the demand side.
One of the things I’m thinking is that smart glasses are a lot like virtual assistants (Siri/Alexa/Google Assistant):
- superficially very “futuristic” in a cool way, inspiring hype/investment
- in reality, only useful for a small set of narrow things, but actually useful for those (not pointless)
But virtual assistants are inherently cheap, running either on hardware people already have or on very cheap new hardware. (It’s possible that getting the software to actually run on this hardware was hard, but that’s a fixed cost that ends when you’ve achieved the goal, and it apparently wasn’t impossible.)
Smart glasses, on the other hand, seem to require high variable/unit costs. Hardware achievements aren’t costless to multiply, the way software achievements are.
If you can figure out how to make Siri run on an iPhone, bam, you’re done, Siri can now run on all the iPhones that exist. With smart glasses, you have to invent new hardware solving what are currently difficult challenges, and it will be as expensive (and buggy) as all cutting-edge hardware.
And people just won’t pay a high cost for a toy that can help them find their keys – not the same way they’ll pay $0 for a toy that helps them set timers when they’re cooking.
