• 1 Post
  • 36 Comments
Joined 9 months ago
cake
Cake day: July 1st, 2024

help-circle


  • I think so? It’s section 3.2 of the paper: https://www.tandfonline.com/doi/full/10.1080/17452759.2024.2404157#d1e604

    They don’t talk that much about current in particular; figure 5 only shows the resistance in the regulated output path as a function of the voltage on the control path, which isn’t enough to actually say whether there’s specifically current amplification. (Also, the gain would be negative; does that matter?) But they do discuss the fact that the output channel is much wider, which strongly suggests it’s able to pass more current (since they mention the resistivity drops as the cross-sectional area of the trace increases). The wider trace is one that wouldn’t have the fuse behavior on its own, because the resistivity is too low for it to heat up enough to trigger that at the voltages they’re using, but the close proximity of the very thin fusing wire of the control signal is enough to cause a nonlinear resistivity change in the output path as well. I think that means they’re using a single voltage for both kinds of path, and that the control current is thus lower than the output current because the resistance on the control path is higher, but I’m not certain. I am not an electrical engineer, just an enthusiastic amateur.


  • They did do the thing that HewlettHackard is describing. Check out the AND gate in the linked article. The input paths are short and use small wires, but also cross the larger paths that normally link the output to ground. If both are active, the paths to ground are interrupted, and the resistor to VCC pulls up the output. So they did make logic gates. In the paper they also demonstrate NOT and OR.

    I gather there’s a technical sense of “active” that’s used in electrical engineering that might not apply here, but to someone like me, with only a tinkerer’s knowledge of components, logic gates seem like enough to justify the term in the headline.



  • Citizens United would be a decent candidate. Once it was established that donations were protected political speech, it effectively legalized bribery, and made oligarchy essentially inevitable. Most of the missteps since then have been motivated by folks trying to simultaneously play to populist talking points but also placate billionaire donors. The left needed an actual positive message, like the kind Bernie Sanders was pushing, that would energize folks and unite the overeducated with the working class, but that was never going to be acceptable to the donor class, and so candidates like him always had to be shoved aside for someone who would clearly cater to corporate needs. And someone who would clearly cater to corporate needs was always going to be a really tough sell and not really a solution to the needs of the moment.

    That doesn’t really account for the rise of the tech bro fascist accelerationists like Mencius Moldbug and the Dark Enlightenment, which is a big part of the current moment and accounts for how the far right was able to hoodwink some billionaires into voting for a social collapse that seems very likely to hurt them also. But Citizens United still seems like a fair candidate for a point at which some of the last paths away from this outcome were foreclosed.



  • I think the issue is with it being just a single wall. The slicer tends to have some trouble with that.

    My first thought was just to model the return paths as well, making it a loop, print that in vase mode, then just cut away the unwanted parts. My second thought was to wonder whether there would be a way to use post-processing to set the extrusion to zero for those particular paths, maybe using the multicolor features, and thereby avoid printing the extra area.



  • I think it’s reasonably likely. There was a research paper about how to do basically that a couple years ago. If you need a basic LLM trained on a specialized form of input and output, getting the expensive existing LLMs to generate that text for you is pretty efficient/inexpensive, so it’s a reasonable way to get a baseline model. Then you can add stuff like chain of reasoning and mixture of experts to improve the performance back up to where you need it. It’s not going to be a way to push the state of the art forward, but it’s sure a cheap way to catch up to models that have done that pushing.


  • I do love games, but most of what I do at my computer is maker projects. CAD, 3d printing, electronics design, coding. Lately I’ve been building a puzzle box for my niece’s birthday.

    Interestingly, I did upgrade my GPU a year and a half or so ago (to a used 3070, I’m not made of money) and since then the main thing I’ve used that GPU for is actually AI experiments rather than games. E.g. for the puzzle box, I got Stable Diffusion to generate images for a puzzle for me. It’s four images, and when you combine them in the right way they reveal a fifth image. I don’t think I could have done the same puzzle without AI.

    I do still play games, though. I’m just kind of off the big budget stuff these days.









  • The “idiot” part comes in where I encountered this problem, and didn’t even stop to consider whether this might be specific to this model, or even try something as basic as turn the model on the print bed, which wouldn’t have fixed the slicing, but would have told me my assumption about how the “bridging angle” setting worked was wrong. Instead, I leapt straight from “huh, this model sliced in a weird way” to “this basic slicer feature is designed in a bizarrely poor way and I’m the first one to ever notice,” and posted about it on social media.

    So I appreciate the sentiment, and I’ll leave the post up as it I agree it’s a mildly interesting and counterintuitive result, but I still maintain I acted kinda dumb. :)


  • A bit of a weird one because it never actually came out, but I was really excited about the news that Michel Gondry was set to direct a film adaptation of Rudy Rucker’s novel The Master of Space and Time. I really like both Gondry and Rucker, and their sensibilities would have worked really well together. The story is kind of a sci-fi three wishes fable. It’s not my favorite of his–that’s gotta be White Light–but it’s really light-hearted and fun. And Gondry is terrific at getting bizarre, dream-like ideas onto film. I’m still bummed out that they cancelled that.

    For something that actually did come out but then got cancelled, I enjoyed Disney’s adaptation of The Mysterious Benedict Society. Then they threw that one down the memory hole for tax reasons, so there’s no way to watch it anymore short of piracy. I feel like that shouldn’t be allowed. Seems like they should add least have to provide the Library of Congress with a copy. It’s weird that our cultural history can just be yanked away like that now.