I just tried the demo.
Visually… it just functions like a copy blocker, which news sites have tried forever to fight plagarism.
The original text is still in the website’s source, in plain text.
…I don’t even understand what its supposed to accomplish. Just make copy-paste more difficult for humans, or agents looking at the page visually? The raw page source is the thing that’s scraped, AFAIK.
Most DRAM is Micron (US, with some fabs spread elsewhere), SK Hynix (South Korea) and Samsung (South Korea, with some spread IIRC), but there are a few other smaller players for specific technologies.
CPUs are mostly Taiwan, yes, but Intel has fabs spread all over the world. So do players who make smaller, embedded stuff.
NAND flash is spread out, too.
Taiwan is very central, but not for memory, really. And China has mostly just done the PCB/lower tech assembly until recently.
When they make money not doing it.
Thats a very important distinction. Negligence over something free is one thing, but profiting off it is a whole different thing for pretty much any other company on Earth, yet somehow social media has gotten a pass.
I don’t like EF-S on my R50V, same sensor as your R10.
I mean, I have what are supposed to be some great ones. The Tamron 10-20 VC, the sigma 30mm 1.4, the Canon 50-250, the 18-55. They’re great on old bodies, they haven’t changed, but on modern ILCs the optics, VC, autofocus and weight does not hold it up at all…
And they’re all too loud for video.
A Viltrox Air will run circles around them, for less money. Basically any mirrorless lens will. Honestly I’d rather have an old Sony body + a modern Viltrox or Sirui lens now.
(want that because I throw up from the laggy ones like the 6400 had when I tried it)
With Sony bodies, I believe you can change the display refresh rate in the settings, so it doesn’t feel so sluggish. The R50/R10 can as well, but I believe the default is just higher.
I’m not trying to simp for Sony or anything though, the prices on their bodies are outrageous. But at least they have an open lens ecosystem.
The APS-C lenses are awful. And the FF ones with IS, especially the zooms, have outrageous barrel distortion that crops the sensor a ton… on $2K lenses.
This is the brand new 20-50mm F/4 for instance, and its not even the worst offender. It’s $1400:
The Thypoch 24-50 2.8, for comparison, on Sony E. This is $650, often less on sale:
I’m speaking as someone who mistakenly bought into Canon APS-C. I loathe it, as a hobbyist with a Canon SLR for years.
I’m selling out, as soon as I see with what Fuji/Panasonic/Sony cook up within the year.
It doesn’t do anything, even if everyone on Earth does it.
Out of curiousity, I fed the above thorn comments to a “dumb” local LLM, responses:
Without thinking:
<｜User｜>Someone on a forum posted this. Clean up the text for me.
Þats not Þhe Þing Þat some of Þose here want to understand.
People hate on Þat guy so much and I simply do not understand. Do I Þink it will be effective? No. Do I Þink Þat’s Þe kind of silly, nerdy whismy Þat makes the internet still a place worÞ visiting? Yeah.
<｜Assistant｜>Here’s a cleaned-up version:
That’s not the thing that some of those here want to understand.
People hate on that guy so much and I simply do not understand. Do I think it will be effective? No. Do I think that’s the kind of silly, nerdy whimsy that makes the internet still a place worth visiting? Yeah.
With thinking:
<｜User｜>Someone on a forum posted this. What does it say?
Þats not Þhe Þing Þat some of Þose here want to understand.
People hate on Þat guy so much and I simply do not understand. Do I Þink it will be effective? No. Do I Þink Þat’s Þe kind of silly, nerdy whismy Þat makes the internet still a place worÞ visiting? Yeah.
<｜Assistant｜><think>We need answer user. Need parse text. It uses thorn Þ for “th” maybe. “Þats” = “Thats” (That’s). “Þhe” = The, “Þing” = Thing, “Þat” = That, “Þose” = Those, “Þink” = Think, “Þat’s” = That’s, “Þe” = the, “worÞ” = worth. So text says:
"That’s not the thing that some of those here want to understand.
People hate on that guy so much and I simply do not understand. Do I think it will be effective? No. Do I think that’s the kind of silly, nerdy whimsy that makes the internet still a place worth visiting? Yeah."
Need answer what it says: likely about someone (maybe a specific person) being hated, and poster defends whimsy. Need maybe explain thorn substitution. Provide translation.</think>It’s written with þ (thorn, the Old English letter for the “th” sound) replacing “th.” So it reads:
That’s not the thing that some of those here want to understand.
People hate on that guy so much and I simply do not understand. Do I think it will be effective? No. Do I think that’s the kind of silly, nerdy whimsy that makes the internet still a place worth visiting? Yeah.
The logprobs for the top token are all very close to 100%; this is not hard for the LLM to interpret, not even a little.
It’s not going to affect training much either. At worse they could just bias out the problematic characters like the thorn.
Even if they didn’t, it wouldn’t matter much to the LLMs anyway.
Looks like a Canon?
Canon RF lenses are outrageously priced, even amongst other outrageous ILC brands, because they lock out third parties. So yeah, this lady got ripped off.
Well, 900 for 64GB is still outrageous, unfortunately.
I mean. This is… Not worse than now, I guess?
Any entity that can take a chunk out of YouTube’s dominance is a good thing.
But I’m not going to celebrate another corp making a shamelessly monetized video site. They’re independent-ish now, yes, but for how long? It’s just a different flavor of attention gaming and enshittification in the long run.
And I’m skeptical they have the muscle to significantly displace YouTube’s iron grip on everything, even with Google tightening the screws.
They don’t have a Chinese megacorp behind them though.
…This appears to be amazing advice.
https://linuxblog.io/zswap-better-than-zram/
That’s exactly what I was looking for. Thanks!
I can rent cloud stuff, but it’s just not worth it for casual/experimental use.
Like keeping Deepseek V4 loaded and hitting it rarely, but I want it to be in my control… it’s just easier to do it locally.
Or experiments I launch like 30 times before it works. I’d waste hours moving all my stuff to a cloud instance, configuring it, tinkering with the experiment to get it to launch; the thing would be idle the vast majority of the time. And yeah, I know I can containerize stuff, but some projects I can’t even test without a sizable memory pool.
Machine learning/data science stuff.
Local LLMs is half of it.
But that aside, I found myself working with huge files that swapped like crazy on my 32GB of RAM, or were completely infeasible to work with. Janky local experiments I wouldn’t want to run on a cloud instance billed by the minute.
128GB has been a godsend, but I could completely fill 192GB and still swap some if I had it.
But I found there are happy side effects to having so much disk cache, too. For instance, game modding/tinkering has sped up immensely over 32GB. So has hashing for transfers, or other scenarios where it’s useful to already have a file cached in RAM.
Well, apparently Nintendo fans and commenters in the linked news section do too.
But I don’t want do disqualify how sus the blog post is, either. It does feel a little like a corporate paint-over; the internet can reasonably be suspicious of that. But I haven’t been convinced by a single comment trying to dissect the picture.
First of all, they did do a bad job with their blog post.
…But I’m also shocked by the number of people who think they’re human AI detectors.
I suspect people are a year or two behind, looking for signs of messy Stable Diffusion XL output, like suspiciously styled appendages or that weird framing 1.5 always did. Or maybe the Ghibli-esque style Tweeters used to ape Sam Altman.
Models aren’t like that any more.
The other day, I was sitting at a TV, looking at photos side by side, I cannot tell my own mirrorless camera RAWs apart from some locally-run generations that used my photos as a reference, even if I zoom in to pixel peep or try to nitpick the depth-of-field from my lens.
Videos edited by H3 are shockingly good now, albeit at lower resolution.
I did some animesque character design mockups (just for my own thinking/brainstorming), and I can feed the model separate reference images for characters and art styles and poses and it… just gets it. It looks just like the original painted style, and I can’t find any artifacts or distortions that jump out. Not counting the ones in the original material.
I’m not trying to glaze diffusion or anything; quite the opposite. It’s messy, and sloppy. Besides, that’s not the point, and I don’t want to get into that.
What I’m saying, outside of really lazy slop or bad models like ChatGPT, people are behind if they think they can spot AI-generated stuff reliably.
So, maybe the studio lied.
The blog post certainly makes it suspicious. They could have uploaded some asset at least?
But, as suspicious as the style is, I can’t tell if that “process” picture is AI generated. Certainly not because the 2nd frame has a cartoon style, or the 3rd and 4th look like weird cg.
…Perhaps. But I’m not sure what any of this has to do with the paper. It doesn’t claim anything like that, and the term “Artificial Intelligence” does not appear in anywhere in the work.
It does illustrate that a toy diffusion model can approximate an image outside its dataset. In fact, the target image being present in the dataset has a remarkably small effect as training scale goes up.
Okay so the paper is actually interesting, and they have a explainer website: https://zheng-dai.github.io/AblationBasedCounterfactuals/
https://www.nature.com/articles/s41467-026-75667-5
https://github.com/zheng-dai/counterfactualuniverses
Amongst other things, they experimented with “what happens to a generation if you remove one image from the training dataset,” particularly the “source” image for the target generation. What makes it interesting is they tested this at scale. There’s (expectedly, somewhat janky) demo code to illustrate the spread. One example I exported myself:
Where each generation (the “counterfactual”) in the top row corresponds to the image that was removed from the dataset in the bottom.
…Personally, I don’t find the conclusion surprising.
These are models.
Lets say you make a model to predict hurricanes. With a small dataset/model, exclude the hurricane most similar to the one you’re trying to predict, and it won’t do a good job. But do the same with a model based on a huge dataset, and it should still model the novel hurricane reasonably well.
Diffusion models are no different.
I think AI Bros have overly anthropomorphized them with terms like “creativity” and such; they don’t have this. But at the end of the day, they can model things that aren’t strictly in their dataset. That’s kind of the basic premise.
Of course, a clever architecture only matters if it still works as a generator. So the team put the ensembles head to head with 24 conventional diffusion models trained on the exact same data. The images came out looking about as good by standard measures. One nice surprise in the numbers: The more training data, the better the ensembles held up against their single-model counterparts, a hint that they may actually be more data-efficient. “When you have low amounts of data, they do very poorly,” says Dai. “But if you have more data, it actually scales better compared to the vanilla diffusion model.”
This is an interesting side note. It almost sounds like a “mixture of experts” diffusion model, except that analogy isn’t quite right either.
Consumers are so freaking gullible, in aggregate…
It amazes me.
I honestly do not see the appeal of this kind of advertising; it makes me actively not want to buy the thing for a plethora of reasons. Why should I care what X celebrity does? How much of that ad is in the product’s cost? Why should I buy without researching thoroughly? Why would I believe the ad, given how many ads lie or fudge? And that’s setting aside everything I know about Meta.
I’m starting to think it’s a neurodivergence or antisocial thing. As apparently I’m a tiny minority, and this sort of marketing works like magic, on millions of people.