I’m just some guy, you know.

  • 3 Posts
  • 236 Comments
Joined 6 months ago
cake
Cake day: May 7th, 2024

help-circle
  • “Open Source” is mostly the right term. AI isn’t code, so there’s no source code to open up. If you provide the dataset you trained off of, and open up the code used to train the model, that’s pretty close.

    Otherwise, we need to consider “open weights” and “free use” to be more accurate terms.

    For example, ChatGPT 3+ in undeniably closed/proprietary. You can’t download the model and run it on your own hardware. The dataset used to train it is a trade secret. You have to agree to all of OpenAI’s terms to use it.

    LLaMa is way more open. The dataset is largely known (though no public master copy exists). The code used to train is open source. You can download the model for local use, and train new models based off of the weights of the base model. The license allows all of this.

    It’s just not a 1:1 equivalent to open source software. It’s basically the equivalent of royalty free media, but with big collections of conceptual weights.












  • There is no CPU that is ever going to be supported for 10 years for a consumer application. ARM CPUs today are 20x faster than they were 10 years ago, and the ARM/RISC-V chips a decade from now will likely be 10-20x faster than today.

    Regardless, the Kryo 670 CPU in the Fairphone 5 is already 3.5 years old, and it’s not super special, it’s just a semi-custom Snapdragon SoC. Consider that 4G LTE launched 13 years ago in the USA, and in 10 years that Kryo chip in the FP5 will be older than that. Could you handle the performance of your last 3G phone today?





  • Lemmy.ml is such a weird situation. Personally, I worry about the future of Lemmy when I consider who runs the circus.

    It started off as the first and only Lemmy instance run by the maintainers. It was classified as general purpose, but leaned hard to the left. The maintainers are mostly Tankies, but the users are not, and Lemmygrad is created for the purpose of being focused on far-left politics. Lemmy.ml was my first instance, and I was an early user. Early on it was nice there.

    As Lemmy grows, new instances appear, and eventually ML is not the largest, just the oldest. As that happens, the original intent of the ML TLD choice starts to become visible, as the instance shifts from general purpose to an instance for “Marxist Leninists” just like Lemmygrad. Early users have now mostly migrated elsewhere because it is no longer the “flagship” instance.

    Now we’re in a position where the majority of Lemmy users exist on non-ML instances, and think that the ML instances are run by morons, yet those morons maintain the codebase that runs every Lemmy instance.

    It’s a ticking timebomb. I expect we’ll all be on Kbin by 2030 after something boils over and the Lemmy maintainers start fucking around with the code to get their way.