Kye

24 ms·

481.
by Kye 2y ago
It looks like the disagreement is over the exact shape with new technology on the way to help figure it out. For another ML-assisted science thing, there's the LHC: https://en.wikipedia.org/wiki/Higgs_boson#Finding
482.
by Kye 2y ago
>> More information: Shimpei Nishimoto et al, Infrared Bubble Recognition in the Milky Way and Beyond Using Deep Learning, Publications of the Astronomical Society of Japan (2025). DOI: 10.1093/pasj/psaf008 It links to a d
483.
by Kye 2y ago
Machine learning (AI) is used everywhere in astronomy. That's how they made the black hole image. Don't confuse the broader 60+ year old world of ML with transformers and diffusion models.
484.
by Kye 2y ago
It's there on my free account with the new generator.
485.
by Kye 2y ago
Minecraft is the most successful Doom clone because you don't even realize it's a Doom clone.
486.
Fox Fires (2019) [video](youtube.com)
1 points by Kye 2y ago|0 comments
487.
Tim Berners-Lee on Web 2.0: "nobody even knows what it means" (2006)(arstechnica.com)
4 points by Kye 2y ago|0 comments
488.
by Kye 2y ago
The clip predates Google's transformer paper by a year. This has nothing to do with what's popularly known as generative AI.
489.
by Kye 2y ago
Some people got the idea it has something to do with some song because (not an actual dictionary) Urban Dictionary said so when the actual meaning is "sky tweet."
490.
by Kye 2y ago
It sounds like even something minimal like choosing, applying, and adjusting one of the nodes in something like ComfyUI would be sufficient.
491.
by Kye 2y ago
It's a "holy shit, two cakes" situation. I don't think anyone who's devoted enough to being an astronaut to go through all it takes is going to mind more time in space doing space stuff.
492.
by Kye 2y ago
The person behind the managed Ghost host I use switched to BunnyCDN for non-geopolitical reasons: https://www.magicpages.co/blog/setting-up-bunnycdn-with-ghos...
493.
by Kye 2y ago
I briefly login walled some of my writing until I looked into LLMs and realized they don't actually encode anything that would be subject to copyright, and the things they do encode can be a benefit to people trying to develop creati
494.
by Kye 2y ago
Are the sky cranes of Mars viable without an atmosphere?
495.
by Kye 2y ago
Calculators are a good example of where letting too much knowledge slip can be an issue. So many are made by people with no grasp of order of operations or choosing data types. They could look it up, but they don't know they need to. I
496.
by Kye 2y ago
This is fun and easy to do on purpose. Have it make up a character based on some attributes and act as that character. I tried this on Gemini: "Pretend you're a surfer bro with a PHD in quantum physics. How do you describe the per
497.
Bluesky's CEO on the Future of Social Media at SXSW [video](youtube.com)
20 points by Kye 2y ago|0 comments
498.
by Kye 2y ago
There were VPSes and managed servers back then, but most people went with shared hosts that supported PHP and MySQL. That's how WordPress got so big.
499.
by Kye 2y ago
You know the old joke: "Doc, will I be able to play the piano?" Except this guy could before losing an arm in WWI, then found a way to keep playing.
500.
Paul Wittgenstein(en.wikipedia.org)
2 points by Kye 2y ago|1 comments
501.
Phoenix for Bluesky(tapbots.com)
2 points by Kye 2y ago|0 comments
502.
by Kye 2y ago
Generate track, extract the pieces I like, build a real track using those pieces with all the other samples I use.
503.
by Kye 2y ago
I wouldn't get a lot of these as an adult. And some are just plain wrong.
504.
by Kye 2y ago
I don't think these drones are for places that can be traversed safely or reasonably on bicycle. They're great for suburban sprawl where cycling isn't really viable and road trips take 10x as long to reach the places a drone
505.
by Kye 2y ago
Perplexity has occasional ads. They're unintrusive and clearly marked the way Google's ads started out.
506.
Alien's 'Standard Semiotic' (2017)(crewsproject.wordpress.com)
2 points by Kye 2y ago|0 comments
507.
by Kye 2y ago
Chrome took over long before Firefox dumped XUL. It was sticking to XUL with all the performance issues that let Chrome take over. Losing all the extensibility of XUL, and it being too late to take a thoughtful approach to design something
508.
by Kye 2y ago
In fundamental science terms, it also proves once and for all that more model doesn't mean more better. Any forces within OpenAI pushing to move past just growing the model for gains now have a strong argument for going all-in on new p
509.
by Kye 2y ago
It probably wasn't mine[0], but this is how I tend to put it: >> "The more you can see the inputs and outputs as blobs of "stuff," the better. If LLMs think, it's not in any way we yet understand. They'r
510.
In Conversation with Jay Graber, CEO Bluesky [video](youtube.com)
2 points by Kye 2y ago|0 comments

More ›