• 3 Posts
  • 170 Comments
Joined 1 year ago
cake
Cake day: July 1st, 2023

help-circle






  • You don’t need to wonder, Apple has said as much that their AI is built on LLMs, just like everybody else. While hallucinations are still a major unsolved problem, that doesn’t mean they aren’t able to be reduced in frequency and severity. A ChatGPT like chatbot is going to hallucinate because you’re asking it to give extremely open ended responses to literally any query. The more data you feed it in the prompt, and the more you constrain its output, the less likely it is to hallucinate. It’ll likely be extremely rare that using the grammar check or rephrasing tools in Apple AI will be affected by hallucinations for that reason. Siri is more comparable to ChatGPT with regards to open ended questions, but it’s likely that they will integrate LLMs primarily for transforming inputs and outputs rather than the whole process. For example, the LLM could be prompted to call a function based on the user’s query. Then, that function finds a reliable result, either using existing APIs for real time information like weather, or using another LLM with a search engine. The output from this truth-finding process is then fed back into an LLM to generate the final output. The role of the LLM is heavily constrained at every step of the way, which is known to minimize hallucinations.

    You arguing that this is an unsolvable problem is defeatist and not helpful to actually mitigating the real issue.




  • The part you’re missing is that while C++ does have newer safer ways of doing memory management, all the old ways are still present, in wide use, and are easier. Basically, C++ makes it easy to do the wrong thing and hard to do the right thing, and most codebases are built around the wrong things. It’s often easier to just rewrite it in rust than it is to refactor an existing code base, so if you’re going to expend that effort why not do it in a language that has stronger safety guarantees, a better dependency and build management system, and a growing community?


  • bamboo@lemm.eetoSelfhosted@lemmy.world*Permanently Deleted*
    link
    fedilink
    English
    arrow-up
    11
    ·
    3 months ago

    As a person who has been managing Linux servers for about a decade now, trust me that a few hours or days of learning docker now will save you weeks if not months in the future. Docker makes managing servers and dealing with updates trivial and predictable. Setting everything up in docker compose makes it easy to recover if something fails, it’s it’s self documenting because you can quickly see exactly how your applications are configured and running.





  • Tor browser is something else, I don’t group it in with stuff like Librewolf.

    For librewolf, I just took a look to try and figure out what binary blobs are being talked about. This is the repository I was looking at, I think its the right place: https://codeberg.org/librewolf/source/src/branch/main. There isn’t much documentation on the patches besides the file names for the most part, but do you have any idea which of these relates to binary blobs? Or is it in the settings file? Really nothing I see here convinces me that this project is worth anybody’s time over regular firefox, it just changes some defaults, disables pocket (they patch it out, but there’s already a setting), and changes the branding. I don’t disagree with most of their changes, I just don’t see the point of maintaining and marketing an entire derivative browser for what could just be a settings hardening guide on a wiki somewhere.


  • I do genuinely believe that these Firefox forks are mostly pointless rebrands of Firefox to satisfy a small crowd of people who are fine with Firefox but don’t want Firefox or Mozilla branding. Other than branding, they tweak the default config, pre-install ublock origin, and that’s about it. I guess this one exposes some already existing about:config flags in the settings UI. The best part is they are managed by small teams that run a few versions behind Firefox persistently, leaving 0-days unpatched and thus their users vulnerable. Their small userbase also opens their users up to tracking that wouldn’t be possible with larger browsers.



  • I also haven’t wanted an Intel processor in a while . They used to be best in class for laptops prior to the M1, but they’re basically last now behind Apple, AMD, Qualcomm. They might win in a few specific benchmarks that matter very little to people, and are still the default option in most gaming laptops. For desktop use the Ryzen family is much more compelling. For servers they still seem to have an advantage but it’s also an industry which requires longer term contracts that Intel has the infrastructure for more so than it’s competitors, but ARM is also gaining ground there with exceptional performance per watt.