• 0 Posts
  • 88 Comments
Joined 2 years ago
cake
Cake day: June 17th, 2023

help-circle
  • I would look into Tailscale based on your responses here. I don’t know what your use case is exactly but you set TS up on your server and then again on your phone/laptop and you can connect them through the vpn directly. No extra exposed ports or making a domain or whatnot.

    If you want other people to access the server they will need to make a TS account and you can authorize them.



  • I mean that is already a problem, if you ask a question you have to be ready for the answer to be a mismatch of version conflicts.

    But that is ok. ChatGPT is a tool that can either help you or hurt you. I like to think of it like a power hammer. If you are doing a roofing job, it can help you get things done faster compared to a manual hammer, but you still need to know how to build a roof to get started.

    ChatGPT is great at helping you organize your thoughts or finding an answer to some error message buried in some log file, but you still need to know what questions to ask and you need to be ready for it to give you a stupid answer and how to get around that.









  • I’m not sure what you were trying, but this works for me:

    Never use hardware encoding. That is intended for real time transcoding. There are not many settings that work since it is just sending the file to the video card and letting it do its thing.

    Slower is better. If you set the software encoder to very slow it will produce an output that is very high quality per megabyte. I generally don’t care if it takes twice as long to encode it as to watch it. I queue it up and let it run over night.

    Choose the right codec. I like 10 bit HEVC, because I know it will work on the clients I play it from. When you rip a DVD using MakeMKV, the video will be MPEG-2, it was designed in the 1990’s and converting the file to a modern codec will save a lot of space. I don’t reencode 4K UHD rips much since I don’t want to mess with losing the hdr or other color features that I like in watching those files.

    Audio tracks: I will rip out audio for languages I don’t speak, or desctiptive audio track, but go out of my way to label things like director commentaries. I don’t reencode the audio tracks at all, you won’t save much disk space by messing with them compared to the video tracks.


  • To be honest, that seems like it should be the one thing they are reliably good at. It requires just looking up info on their database, with no manipulation.

    That’s not how they are designed at all. LLMs are just text predictors. If the user inputs something like “A B C D E F” then the next most likely word would be “G”.

    Companies like OpenAI will try to add context to make things seem smarter, like prime it with the current date so it won’t just respond with some date it was trained on, or look for info on specific people or whatnot, but at its core, they are just really big auto fill text predictors.



  • The network effect is real. You can have the best, most awesomely-designed social media platform ever and it will be useless if you are the only person on it.

    You can try to convince all your contacts to switch away from whatever app is causing the most evil today, but you also have to convince all of your contacts’ contacts and all of theirs as well.