KeplerBoy

70 ms·

1441.
by KeplerBoy 4y ago
Doesn't seem to be true. Surely the stocks are held by all kinds of people and institutions and only a very small fraction of the shares are up for sale at any given time. Am i missing something? Also the stockprice listed is the price
1442.
by KeplerBoy 4y ago
ChatGPT's and Bing's German is flawless.
1443.
by KeplerBoy 4y ago
It's crazy, but it seems to be happening already. Granted, that's probably still a far-cry from Chat-GPT, but it seems inevitable a few years down the line. https://news.ycombinator.com/item?id=35100086
1444.
by KeplerBoy 4y ago
Can't you just link Octave against mkl?
1445.
by KeplerBoy 4y ago
A lot of that classical engineering stuff is moving to Python. At least that's what I'm observing at uni. The last real stronghold of Matlab is controls engineering, because dynamic systems and simulink are just a good match, but
1446.
by KeplerBoy 4y ago
Apparently not all penguins are equal.
1447.
by KeplerBoy 4y ago
Python 2 was an odd choice 4 years ago.
1448.
by KeplerBoy 4y ago
was it really 8 GB? Seems like a lot for 2006-ish.
1449.
by KeplerBoy 4y ago
which would make traditional distributed training prohibitively slow. You would have to sync the weights with residential internet speeds (10s of MB/s) instead of PCI-e Speeds (10s of GB/s).
1450.
by KeplerBoy 4y ago
quintillion bytes of social media probably have very limited value.
1451.
by KeplerBoy 4y ago
You can attach gdb to any process. Debug information is not needed unless you want nice things like knowing which lines of code you are stepping through.
1452.
by KeplerBoy 4y ago
This is a neat little reference to pig iron, probably the only kind of iron one could produce with a settlers style smelting operation. https://en.wikipedia.org/wiki/Pig_iron
1453.
by KeplerBoy 4y ago
You might get the correct answer on the second try, but that would be down to random seeds.
1454.
by KeplerBoy 4y ago
Because it has no idea if it's spouting bullshit at the moment or not. And that doesn't seem to be addressable as long as the model is just stringing together the next most likely words as it learned them from some huge dataset. T
1455.
by KeplerBoy 4y ago
Stuff like PyTorch? They're building large parts of the infrastructure everyone uses.
1456.
by KeplerBoy 4y ago
The best way to benchmark CUDA code is probably using nvtx annotations and nsight systems.
1457.
by KeplerBoy 4y ago
With the small restriction that conversion from PDF is one of the few things pandoc does not do.
1458.
by KeplerBoy 4y ago
pandoc doesn't ingest PDFs, it can only output them. Getting PDFs into the pandoc intermediate representation would probably work on such a small subset of PDFs, pandoc does not even bother trying.
1459.
by KeplerBoy 5y ago
Those teams still wouldn't switch because they wouldn't gain any performance from switching. After all CPython is just calling highly optimized, already compiled C, Fortran or Accelerator specific (CUDA, ROCm, TPU) code as far as
1460.
by KeplerBoy 5y ago
He has a valid point, generating the image on a separate device (heck, you could do that on AWS Lambda for free) and only grabbing the resulting image could vastly improve battery life. Then you could also switch from a power hungry Pi to a
1461.
by KeplerBoy 5y ago
weird how that works. For me eggs are now a no-go.
1462.
by KeplerBoy 5y ago
What solution did you end up switching to? I have a feeling that most python GUIs are lacking in some area and for full creative freedom someone would have to go for flask or fastapi, build the UI with modern js-frontend frameworks and conn
1463.
by KeplerBoy 5y ago
It's a flaw, but also a feature. If the enhanced comfort of a wider tire reduces the perceived extertion, even though the actual power needed increases, it's a net win in my view.
1464.
by KeplerBoy 5y ago
Unfortunately i'm rooting for the Bozo, the current AI Revolution won't lead us anywhere and will ebb down eventually.
1465.
by KeplerBoy 5y ago
Probably the mentioned 25 Gbit/s. At least now nobody needs that at home. I'm not even sure how one would hook up a computer to such a fast connection. certainly not with conventional ethernet.
1466.
by KeplerBoy 5y ago
The web just got worse in a lot of ways, because everything needs to generate money.