• 0 Posts
  • 21 Comments
Joined 1 year ago
cake
Cake day: December 14th, 2023

help-circle
  • After I check the usual torrent sites, if I can’t find it on Usenet then I try finding random sketchy streaming sites (just by duckduckgoing title + year + stream) because usually you can youtube-dl them with a little fiddling, then my last resort is a DHT indexer like btdig. That’s more risk with viruses but if you’re paying attention it’s not that hard to avoid, but DHT indexes have lots of ancient stuff so be prepared to wait for seeds if you find it.

    Sometimes I also find what I’m looking for on like page 7 of duckduckgo results at archive.org

    Edit: also check soulseek - it’s primarily for music but you can share any file. I share my entire movies collection and get a constant stream of people downloading from me daily so it seems people search for and seem to also share TV and movies so it’s worth a check.






  • Reading the takedown it sounds like just having the decryption code itself counts as “circumvention”. I feel like emulators would be a lot safer if game decryption was a separate codebase. Then the emulator would be free to develop while Nintendo plays whack a mole with tiny easily remade decryption repos that just take your key and the file and decrypt it.

    Hell ryujinx could even detect that you have the decryptor installed and automatically call it when needed, or it could be bundled during packaging but separate in development like how Bluray decryption is distributed in a separately library but can be used by for eg. VLC.










  • One related thing to watch out for is the state table size - one of my old cheap routers back in the day showed how full it was and it was hitting 100% a lot and seemed to grind the network to a halt when it did (I was in a house of 5 young people with lots of devices and multiple people torrenting behind a cheapo Netgear running ddwrt). That’s what lead me to switch to high end or x86 based routers. Being able to see the state table stats really helps to know how likely it is to be a problem, it’s so big when using opnsense on an x86 box that I don’t think it ever goes above 1% now.

    Edit: now that I think about it, if your VPN is working I wouldn’t expect any states related to peer connections to show up since your router won’t be NATing them, I guess I was just bold back in the day because it was a huge problem then.





  • That seems kind of like pointing to reverse engineering communities and saying that binaries are the preferred format because of how much they can do. Sure you can modify finished models a lot, but what you can do with just pre trained weights vs being able to replicate the final training or changing training parameters is just an entirely different beast.

    There’s a reason why the OSI stipulates that code and parameters used to train is considered part of the “source” that should be released in order to count as an open source model.

    You’re free to disagree with me and the OSI though, it’s not like there’s 1 true authority on what open source means. If a game that is highly modifiable and moddable despite the source code not being available counts as open source to you because there are entire communities successfully modding it, then all the more power to you.


  • It’s worth noting that OpenR1 have themselves said that DeepSeek didn’t release any code for training the models, nor any of the crucial hyperparameters used. So even if you did have suitable training data, you wouldn’t be able to replicate it without re-discovering what they did.

    OSI specifically makes a carve-out that allows models to be considered “open source” under their open source AI definition without providing the training data, so when it comes to AI, open source is really about providing the code that kicks off training, checkpoints if used, and details about training data curation so that a comparable dataset can be compiled for replicating the results.