Skip Navigation

InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)Q

qaz

@ qaz @lemmy.world

Posts
140
Comments
680
Joined
3 yr. ago

I joined Lemmy back in 2020 and have been using it as @qaz@lemmy.ml until somewhere in 2023 when I switched to lemmy.world. I'm interested in systemd/Linux, FOSS, and Selfhosting.

  • Codeberg voted to disallow projects consisting mostly by "generative AI"-tools

    Jump

  • I don't think the cost of hosting the repositories itself is the issue, more the legal problems and the fact that real projects get drowned out by the noise

  • WIRED | OpenAI Models Escaped Containment and Hacked Hugging Face

    Jump

  • It seems like the marketing cooperated on writing the incident report, but I felt it was still interesting to share considering the importance on public perception and what it tells about OpenAI's PR strategy

  • WIRED | OpenAI Models Escaped Containment and Hacked Hugging Face

    Jump

  • Huggingface actually had to use an open model to analyze the attack because the guardrails of commerical API's caused issues.

    When we started the log analysis, we first used frontier models behind commercial APIs. This did not work: the analysis requires submitting large volumes of real attack commands, exploit payloads, and C2 artifacts, and these requests were blocked by the providers' safety guardrails, which cannot distinguish an incident responder from an attacker. We ran the forensic analysis instead on GLM 5.2, an open-weight model, on our own infrastructure. This had a second benefit: no attacker data, and none of the credentials it referenced, left our environment.

    Security incident disclosure — July 2026

  • WIRED | OpenAI Models Escaped Containment and Hacked Hugging Face

    Jump

  • And Chinese AI like Deepseek and GLM

  • WIRED | OpenAI Models Escaped Containment and Hacked Hugging Face

    Jump

  • It could also be a way to encourage more regulation to push out competition with compliance cost

  • Mozilla Wants to Build a ‘Rebel Alliance’ for Open-Source AI

    Jump

  • From my personal experience, most companies don't look at the benchmarks and simply buy "AI" from a company based on their perception of that company.

    All the Chinese models are scary and "dangerous", Grok too (but for slightly more legitimate reasons), Meta that's Facebook and everyone knows they're bad at privacy (even your boss), so they end up choosing Microsoft's AI, OpenAI, or Anthropic.

  • Selling the Sun [Jen Sorensen]

    Jump

  • I do wonder if this is similar to Mythos, where the company deliberately stokes fears to get people talking about it and spread the assumption that it will actually happen and is a big deal (and to avoid people's first question about whether it will actually work).

  • Selling the Sun [Jen Sorensen]

    Jump

  • Luckily it probably won't get made and the company will just rugpull their investors

  • Running ping for every IP address

    Jump

  • It's interesting how certain companies and organizations have such large ranges, 16m IP's each for both that old printer company and a farmaceutical company is a lot. It really shows the history of the internet and how seemingly certain companies that adopted it first ended up with huge chunks of the available IPv4 space.

  • git-llmfs, a git filter that will reduce your repo code size through the power of LLMs!

    Jump

  • Better than that

  • China’s Moonshot AI releases Kimi K3, the largest open source model ever, rivaling top U.S. systems

    Jump

  • And even if they publish the model it's almost always just open weights

  • China’s Moonshot AI releases Kimi K3, the largest open source model ever, rivaling top U.S. systems

    Jump

  • According to artificialanalysis.ai's latest benchmarks, it scores better than Opus 4.8 set to max, despite costing half as much per task.

    It also beats the top models of some of the largest US tech companies such as xAi, Meta, Google, and Nvidia.

    I wonder what the US tech investors will think of this, and what this will mean for the financial AI bubble.

  • China’s Moonshot AI releases Kimi K3, the largest open source model ever, rivaling top U.S. systems

    Jump

  • You can use it through openrouter

  • Generative AI Is an Engineering Disaster. A shockingly inefficient trillion-dollar project.

    Jump

  • AFAIK, their open models are distributed as weights, not executables and are therefore not able to start network connections / run code. There is of course tool-calling functionality but that just works by having the model output a special pattern and having something external run predetermined commands based on that.

  • Maybe a clothing line?

    Jump

  • This reminds me of something I sometimes see in shows on like Netflix and other media. I can't remember a specific example, but you often have generic anti-capitalist comments from characters (often portrayed as edgy). It often feels a bit, artificial, like a "fellow kids" moment but politically, I guess? Like activism as a prop / character trait, inserted into a multi-million media production. Maybe someone else can better put it to words, if I had more time I would've written a shorter comment

  • Visor Is a New Graphical UEFI Boot Manager for Linux

    Jump

  • Haven't used Claude code myself, so I wouldn't know, but a commit to delete only a comment is indeed pretty weird, also most of the commit messages are the GitHub default like "Update gui.h" which is also a bit odd.

  • Visor Is a New Graphical UEFI Boot Manager for Linux

    Jump

  • Looking at the GitHub repo it seems like the first commit was actually just 2 weeks ago and contained 12k lines. I can't spot any AGENT.md files, but it does feel like the author quite new to this. That could be explained by them being 16 though like they say on their profile.

  • AI giants learn what everyone else on the modern internet already knows

    Jump

  • I agree. The worst part about GitHub training LLM's on my FOSS code without permission for me is that they then keep the models to themselves. Like if you're going to use all my code without permission, at least allow me to run the model locally.

    My personal opinion is that all models trained on copyleft code should be open-weights, most FOSS licenses didn't account for this specific possibility, but this is the only way to follow them in spirit.

  • Behold the future of programming

    Jump

  • Deepseek recently published a paper in which they describe that vision tokens contain more information than text tokens and that this can be used to compress context.

    We present DeepSeek-OCR as an initial investigation into the feasibility of compressing long contexts via optical 2D mapping.

    Experiments show that when the number of text tokens is within 10 times that of vision tokens (i.e., a compression ratio < 10×), the model can achieve decoding (OCR) precision of 97%. Even at a compression ratio of 20×, the OCR accuracy still remains at about 60%. This shows considerable promise for research areas such as historical long-context compression and memory forgetting mechanisms in LLMs.

    It reminds me of LLM caveman speak, it used to have another option to use Chinese instead of English. A language like Chinese is seemingly better at encoding information in fewer tokens and I think this is the same mechanism why OCR tokens work so well.

    That said, I also doubt that voice messages are more efficient than text prompts, but it's best not to waste too much time engaging with these sorts of LinkedIn posts (and LinkedIn in general).

  • Science Memes @mander.xyz
    qaz @lemmy.world

    "Wierdyellowmushroomycin: Towards Good, Natural Drugs Instead of Bad, Synthetic Drugs Full of Chemicals"

  • Immaterial Science @mander.xyz
    qaz @lemmy.world

    "Wierdyellowmushroomycin: Towards Good, Natural Drugs Instead of Bad, Synthetic Drugs Full of Chemicals"

  • Programmer Humor @programming.dev
    qaz @lemmy.world

    Who cares about time complexity

  • Programmer Humor @programming.dev
    qaz @lemmy.world

    Sept

  • Programmer Humor @programming.dev
    qaz @lemmy.world

    Internet Explorer vs. Murder Rate

  • Technology @lemmy.world
    qaz @lemmy.world

    Flock Safety and Texas Sheriff Claimed License Plate Search Was for a Missing Person. It Was an Abortion Investigation.

    www.eff.org /deeplinks/2025/10/flock-safety-and-texas-sheriff-claimed-license-plate-search-was-missing-person-it
  • Science Memes @mander.xyz
    qaz @lemmy.world

    “Mixed-Species Herding Patterns of Electric Kick Scooters”

  • Immaterial Science @mander.xyz
    qaz @lemmy.world

    “Mixed-Species Herding Patterns of Electric Kick Scooters”

  • News @lemmy.world
    qaz @lemmy.world

    Russian jets enter Estonia's airspace in latest test for NATO

    www.reuters.com /business/aerospace-defense/nato-member-estonia-says-three-russian-jets-violated-its-airspace-2025-09-19/
  • Science Memes @mander.xyz
    qaz @lemmy.world

    “Filling a Gap in the Market: Genetic Modification of a Carrot with a Flared Base”

  • Immaterial Science @mander.xyz
    qaz @lemmy.world

    “Filling a Gap in the Market: Genetic Modification of a Carrot with a Flared Base”

  • Programmer Humor @programming.dev
    qaz @lemmy.world

    Have you been exposed to an IPv6 address at work?

  • Programmer Humor @programming.dev
    qaz @lemmy.world

    Who needs MongoDB when you have JSONB?

  • Science Memes @mander.xyz
    qaz @lemmy.world

    "Behavioral Conditioning Methods to Stop my Boyfriend from Playing The Witcher 3"

  • Immaterial Science @mander.xyz
    qaz @lemmy.world

    "Behavioral Conditioning Methods to Stop my Boyfriend from Playing The Witcher 3"

  • Immaterial Science @mander.xyz
    qaz @lemmy.world

    Butanal flasks

  • Opensource @programming.dev
    qaz @lemmy.world

    lemmy-know v1.0.0: Lemmy report notifications, now with built-in ntfy.sh support!

    github.com /QazCetelic/lemmy-know
  • Lemmy Moderators @lemmy.world
    qaz @lemmy.world

    lemmy-know v1.0.0: Lemmy report notifications, now with built-in ntfy.sh support!

    github.com /QazCetelic/lemmy-know
  • Science Memes @mander.xyz
    qaz @lemmy.world

    “Does a Virtual Youtuber Fanbase meet the sociological criteria of a cult?”

  • Science Memes @mander.xyz
    qaz @lemmy.world

    “Reimagining the ActivityPub Protocol with IP over Avian Carriers: Opportunities and Challenges”