Live data from Hacker News

Viewing profile — SwellJoe

SwellJoe

HN member
Joined
Mon, Feb 19, 2007, 11:23 PM UTC
HN karma
35,458
Public activity
7,518 items

About SwellJoe

YC W07

I work on Virtualmin. You can find it at http://www.virtualmin.com

I blog at https://swelljoe.com

I toot at https://mas.to/@swelljoe

Recent public activity

  1. comment
    Comment #49253408

    I've noticed most fine-tunes, whether "heretic" models or something else, tend to be over-fitting, or something, at least some of the time, and get kind of chaotic. I want to belie…

  2. comment
    Comment #49251165

    If you need a GUI, Unsloth Studio and LM Studio are both great frontends for llama.cpp. If you don't need a GUI, llama.cpp is the business for single-user deployments. Easy to use,…

  3. comment
    Comment #49251102

    I saw that one in the "Popular models" sort at Hugging Face and tried it on some tasks I do frequently to compare models, and it feels damaged by the fine-tune, to me. It wrote sec…

  4. comment
    Comment #49245517

    Yeah, Podman would be a better basis for that kind of use case. I'd built an early implementation of `flar` with Podman first, but it was more annoying than simply having my regula…

  5. comment
    Comment #49241337

    Holler if you run into any problems. I use it every day, but I have a pretty limited set of use cases.

  6. comment
    Comment #49241323

    Yes, bubblewrap is superior to Docker for this. I wrote a tool to use bubblewrap for the purpose. It needs a tool to start it, or is at least much more convenient with a tool, beca…

  7. comment
    Comment #49241278

    I think Linux has a better solution than Docker. I wrote a tool to use `bubblewrap` to containerize any agent (at least all the agents I've used a couple of times), and bind mount …

  8. comment
    Comment #49239442

    I made a tool to bubblewrap any agent (well, any agent I've used more than once), so I can run them in whatever YOLO mode they have with a pretty reasonable level of safety (it pro…

  9. comment
    Comment #49225897

    It is not slower than DS4. That's crazy talk. It's much smaller, with smaller active parameters, and thus runs faster.

  10. comment
    Comment #49215782

    LLMs are not neutral, either. They don't have "opinions", per se, as they don't want anything, but they do have biases baked into their training.

  11. comment
    Comment #49215769

    I truly doubt it will be timeless. There are already a few excellent English translations of the Odyssey.

  12. comment
    Comment #49215403

    I'll grant it's very thorough when assigned a troubleshooting task. I'm not as confident of its code review though it is very good at security vulnerability auditing, and isn't hob…

  13. comment
    Comment #49215022

    DeepSeek is my cheap and cheerful Chinese model of choice for API use. Has been for a while, but now it's Flash instead of Pro. Even cheaper, and now better then Pro. I feel like m…

  14. comment
    Comment #49202471

    If anybody has, it's DeepSeek. But, with the promised price hikes, I'm sure that'll change. I'm guessing they're raising prices not because they're not making a profit at those pri…

  15. comment
    Comment #49201052

    I find that surprising. I've been trying it on several projects and have found it's pretty sloppy. It leaves stuff broken, doesn't reliably write tests to check its own work unless…

  16. comment
    Comment #49200005

    I dread the day when my employer's Slack includes chatbots rambling all over it.

  17. comment
    Comment #49198776

    The question is: How soon? If you make a specialist model and harness that takes six months to train and launch, will you have enough time to recoup your costs and make a profit be…

  18. comment
    Comment #49191012

    This is one reason I've been trying to figure out tasks (and products based on those tasks) that can be pushed to the edge, either via small specialized models or small general pur…

  19. comment
    Comment #49190693

    I've said before that in one small way, having access to the sycophantic models is like being a billionaire: You never have to hear "no" or "that's not a good idea". And, it has th…

  20. comment
    Comment #49189899

    There are, but only for using IP of large corporations. If you put 12 seconds of a major label song in your video review of an album, that video belongs to the major label now.

  21. comment
    Comment #49187976

    And, everyone will go to Chinese/open models. Which, I guess, is why the US providers are angling for regulatory capture to lock in their monopoly for a few years and guarantee US …

  22. comment
    Comment #49187305

    Gemini CLI didn't refuse to perform security audits of code. That makes it better than Antigravity, IMHO. Security audits are the one thing everyone needs to be doing with AI, now,…

  23. comment
    Comment #49187265

    I did a review of all of them recently (OK, there are way too many to review all...but, all the official and semi-official ones), and my favorites were Reasonix and MiMo Code, part…

  24. comment
    Comment #49187149

    I don't see a 2.0 of the CLI, thus far. When they killed Gemini CLI (a better product), they said it was to share effort across the GUI and TUI, but I guess that doesn't mean what …

  25. comment
    Comment #49186028

    I'm not convinced it's "wrapped up". This release sounds prickly and dismissive of the new Commodore company, asserting multiple times that the only thing that doesn't belong to Am…