SwellJoe

10 ms·

1. 181.▲
   by SwellJoe 2mo ago

   I don't care so much about token efficiency and cost, within reason. I care about whether the quality of the code is maintainable over time and through many iterations. My gut feeling is that very strict languages with good types, a st
2. 182.▲
   by SwellJoe 2mo ago

   I've run into Peter Norvig twice. Once at a YC event; the other when I parked my motorhome in front of his house in Palo Alto for a couple of days while visiting a friend who happened to live on the same street (not on purpose, I didn&
3. 183.▲
   by SwellJoe 2mo ago

   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 believe normal folks with normal resources can be
4. 184.▲
   by SwellJoe 2mo ago

   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, always gets new model support very quickly, built-in H
5. 185.▲
   by SwellJoe 2mo ago

   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 security bugs into the code (probably just sloppy thi
6. 186.▲
   by SwellJoe 2mo ago

   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 regular dev environment instantly available in the container.
7. 187.▲
   by SwellJoe 2mo ago

   Holler if you run into any problems. I use it every day, but I have a pretty limited set of use cases.
8. 188.▲
   by SwellJoe 2mo ago

   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, because you need to take your session/auth data into the co
9. 189.▲
   by SwellJoe 2mo ago

   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 the system stuff read-only, so the agent has your &quot
10. 190.▲
    by SwellJoe 2mo ago

    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 protects the rest of the system against prompt injections
11. 191.▲
    by SwellJoe 2mo ago

    It is not slower than DS4. That's crazy talk. It's much smaller, with smaller active parameters, and thus runs faster.
12. 192.▲
    by SwellJoe 2mo ago

    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.
13. 193.▲
    by SwellJoe 2mo ago

    I truly doubt it will be timeless. There are already a few excellent English translations of the Odyssey.
14. 194.▲
    by SwellJoe 2mo ago

    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 hobbled for that work like Fable, and even
15. 195.▲
    by SwellJoe 2mo ago

    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 most of the major Chinese models are benchmaxxed, they h
16. 196.▲
    by SwellJoe 2mo ago

    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 prices, but because they're
17. 197.▲
    by SwellJoe 2mo ago

    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 explicitly prompted, misunderstands the assi
18. 198.▲
    by SwellJoe 2mo ago

    I dread the day when my employer's Slack includes chatbots rambling all over it.
19. 199.▲
    by SwellJoe 2mo ago

    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 before the generalists get good enough and cheap enough in you
20. 200.▲
    by SwellJoe 2mo ago

    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 purpose open models. I suspect the same desire to keep uni
21. 201.▲
    by SwellJoe 2mo ago

    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 the same deleterious effect on y
22. 202.▲
    by SwellJoe 2mo ago

    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.
23. 203.▲
    by SwellJoe 2mo ago

    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 decline in tech to the status of an "also ran&quot
24. 204.▲
    by SwellJoe 2mo ago

    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, even if you're not doing any other work with AI
25. 205.▲
    by SwellJoe 2mo ago

    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, partly because they're the least like Claude Code, while ev
26. 206.▲
    by SwellJoe 2mo ago

    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 a normal person would assume it means. The current
27. 207.▲
    by SwellJoe 2mo ago

    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 Amiga is the Commodore brand and logo
28. 208.▲
    by SwellJoe 2mo ago

    The most interesting thing is the "dreaming" idea for on-device adaptation. I guess regularly adjusting weights becomes possible when the model is so small and the math is simplified. I'm assuming that doesn't actually e
29. 209.▲
    by SwellJoe 2mo ago

    DeepSeek feels like the most...trying to think of the word, honest isn't quite right but close, of the AI companies. They make very good models that are phenomenally efficient and they sell them at what seems to be a modest margin. The
30. 210.▲
    by SwellJoe 2mo ago

    Oh, Steve.