Skip Navigation

Posts
53
Comments
879
Joined
2 yr. ago

New account since lemmyrs.org went down, other @Deebsters are available.

  • I went with ntfy as well - you can set the different levels to alert in different ways and my max priority is set to always ring even if the phone is on silent. Mostly I use max prio as a find-my-phone tool, but there are real alerts that would use it.

  • Ebooks: I use Calibre locally and Calibre-web on the server (read-only metadata db, I overwrite with the Calibre version as tagging, etc is far easier on desktop).

    You can connect Koreader to Calibre-web and until maybe a fortnight ago you could jailbreak a Kindle and use Koreader instead of the default software. Now you'll need to manually move files over, or use the email-to-Kindle option (probably a bad idea, but I expect Amazon can tell if you've side loaded pirated content anyway). Nowadays I buy from not-Amazon sources, strip any DRM and send it over.

    Manga/comics/graphic novels: I use Kavita on the server and I use comictagger on desktop to fix the metadata.

    I'm happy to use different set ups for the different types as they're quite different experiences and specialist tools work better.

  • Look at bottom centre

    Edit: I'm getting upvotes but I'm not technically in the right here...

  • Deleted

    Permanently Deleted

    Jump
  • It uses a neutral net that he designed and trained, so it is AI. The public's view of "AI" seems mostly the generation stuff like chatbots and image gen, but deep learning is perfect for science and medical fields.

  • Deleted

    Permanently Deleted

    Jump
  • I found his paper: https://iopscience.iop.org/article/10.3847/1538-3881/ad7fe6 (no paywall 😃)

    From the intro:

    VARnet leverages a one-dimensional wavelet decomposition in order to minimize the impact of spurious data on the analysis, and a novel modification to the discrete Fourier transform (DFT) to quickly detect periodicity and extract features of the time series. VARnet integrates these analyses into a type prediction for the source by leveraging machine learning, primarily CNN.

    They start with some good old fashioned signal processing, before feeding the result into a neutral net. The NN was trained on synthetic data.

    FC = Fully Connected layer, so they're mixing FC with mostly convolutional layers in their NN. I haven't read the whole paper, I'm happy to be corrected.

  • I think I'll wait for a better source than the Daily Mail before I consider it dead.

  • lemsha.re - Lemmy shareable links

    Jump
  • lemsha.re - Lemmy shareable links

    Jump
  • You missed a great opportunity to use lemshare/lemmyverse here

  • Yup, same as they could clone git repos in one shot, but they instead crawl every single page.

  • This is brilliant, I hope we see loads more of these edits.

  • We're all getting invited.

  • As an IDE

    Jump
  • I do if I can raise the laptop up so that the screen is where it should be for good ergonomics.

  • It's the meme version of a laugh track

  • I just watched Hit Man, which claims there aren't any such thing as hit men in real life - clearly that's not the case!

  • "Commentator posts hot take, demonstrating a massive lack of empathy to people doing a stressful and important job."

  • I guess some people might go with f-s-tayb, but I wouldn't necessary recognise what they were saying.

  • Yes, I still have it showing up in Windows/Android, and phone numbers show their cost per minute.

  • I have a load of credit on there still (got tricked by them deactivating my credit and topped up unnecessarily). I still use it for international calls at least once a month, I hope this news story is overblown.

  • This is one of my favourites, despite the lack of Hobbes.