• 0 Posts
  • 63 Comments
Joined 1 year ago
cake
Cake day: September 7th, 2023

help-circle



  • Yes, thankfully the reasonable tech companies offering these services have decided to stop the training process after it was done once. The insane increase in energy consumption and hardware manufacturing for datacenter components and accelerators is purely coincidental and has nothing to do with demand for gimmicky generative AI services. Let’s also conveniently ignore the increasing inference cost of more complex models, while we’re at it.




  • XML aims to be both human-readable and machine-readable, but manages neither. It’s only really worth it if you actually need the complexity or extensibility, otherwise it’s just a major pain to map XML structures to any sensible type representation. I’ve been forced to work with some of the protocols that people like to present as examples of good XML usage and I hate every single one of them.

    Fuck YAML though. That spec is longer and more complex than any other markup language I know of and it doesn’t have a single fully compliant implementation.


  • C has not aged well, despite its popularity in many applications. I’m grateful for the incredible body of work that kernel developers have assembled over the decades, but there are some very useful aspects of rust that might help alleviate some of the hurdles that aspiring contributors face. This was not a push by rust evangelists, but an attempt to enable modernization efforts at least for new driver development. If it doesn’t work out, that’s fair enough but I’m grateful for the willingness - especially of Linus - to try something new.


  • The whole premise of this discussion was about technological progress and growth going by your initial comment. That means refining existing models and training new ones, which is going to cost a lot of energy. The way this industry is going, even privacy conscious usage of open source models will contribute to the insane energy usage by creating demand and popularizing the technology.




  • anyhow2503@lemmy.worldtoCyanide and Happiness@lemm.ee30 June 2024
    link
    fedilink
    arrow-up
    1
    arrow-down
    1
    ·
    6 months ago

    Training your dangerous dog isn’t a requirement to owning it and kids naturally do stupid shit. That’s not a good reason to handwave kids getting mauled or killed. Pitbulls especially will just randomly decide to bite something on occasion, not because something genuinely threatened them. A kid getting overly excited (or scared) of a dog isn’t some reasonable excuse for the kid getting hurt. Looking at the statistics by breed, it’s not even close. Sure there are other dangerous breeds out there, but Pitbulls are relatively popular, owned by idiots who think they are just the cutest little angels (until they bite something they shouldn’t) and were literally bred for their dangerous bite. It should not be a surprise to anyone when that leads to 66% of the all time deaths to dog bites. There are multiple independent studies, showing that they are responsible for about half of all dog bite related injuries and then they have one of the highest rates of severe injuries, because their bite is really powerful and they tend to latch on for longer. There are plenty of breeds who don’t even make the statistic, despite their popularity.


  • anyhow2503@lemmy.worldtoCyanide and Happiness@lemm.ee30 June 2024
    link
    fedilink
    arrow-up
    4
    arrow-down
    3
    ·
    6 months ago

    That’s just the shootings with injuries or deaths. The total number of school shootings is way more impressive. Thanks for putting things into perspective though. A few dozen kids are an acceptable sacrifice, to keep a breed that is more likely to bite without warning or apparent reason and not let go. Also, check out this statistic to feel even more safe about the dog breed that’s apparently responsible for 50% of reconstructive surgery cases for children.