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/)S

SuspiciousCarrot78

@ SuspiciousCarrot78 @aussie.zone

Posts
19
Comments
326
Joined
3 mo. ago

  • Removed Locked

    Most slopcode projects are abandoned and deleted within months of release

    Jump

  • I agree with that...but I wonder if we only ever hear from the so called parasitic ones?

    Suppose a kid in Africa uses Claude to code an app that tracks the spread of a disease in their community and predict the next outbreak site based on x,y,z.

    That's technically slop code too. Do they get a pass because the cause is virtuous or not crowed about? By the letter of the law...no. But by the spirit of the law, probably yes.

    I guess the difference is, how much leeway do we have for genuine enthusiasm vs parasitism. It's hard to tell sometimes on social media where too many people are doing preening displays in public - and we've probably all been guilty of that.

    As a rule, if a thing interests me, I'll read the post, hit the repo, and dig around the files. If I see obvious use of llm in the code (like the stupidly verbose comments that LLMs like to pepper throughout), that usually means that the person either didn't look, didn't know to look or doesn't care. That's bad.

    Bad intro post + bad readme.md + weird commit history + weird AI comments = I'm out.

    Honestly, I'm usually out after the first one or two these days.

  • Removed Locked

    Most slopcode projects are abandoned and deleted within months of release

    Jump

  • That assumes prices actually come down.

    On purely dollar costs, $5K for a coding / AI only PC doesn't amortise vs subscription or API access. Not for occasional shenanigans anyways.

    I've done the calcs, because I wanted to justify the indulgence and the numbers dont stack up. Yet.

    Of course, I based that on current rates. If basic subs go to $50/month, exclude coding agents, and API skyrockets...well fuck, I'll go back to modding and retro gaming. I still remember how to code in ARexx, BlitzBasic etc and I keep threatening to make an Amiga or C64 stuff...there are other rabbit holes :)

    PS: Lemmy spying on me; YouTube just recommended this

    https://www.youtube.com/watch?v=ox1mW2N9Z_Y

  • Removed Locked

    Most slopcode projects are abandoned and deleted within months of release

    Jump

  • Maybe but right now, according to the Steam hardware survey, the most commonly owned VRAM tier is 8GB - sitting at around 27% of surveyed systems, with 16GB closing the gap fast.

    Even with MoE and llama.cpp tricks, you're not running frontier anything at decent context length without significant fiddling on that - it's possible on 8GB but you're operating with almost no headroom, 16GB might let you scrape by with a Q4 quant MoE.

    The very best local coding models (arguably the Qwen3-Coder-30B-A3B MoE or Qwen 3.6-27B) need 16-17GB at Q4 quantisation, and building a system from scratch to run one is probably a $3.5K proposition. With the cost of living crisis hitting everywhere, that means the table ante gear is beyond the reach of many. Even a decent GPU is north of $1K in many local markets.

    I adore small LLMs, and know a lot of tricks to leverage them, but 14B is the bare minimum for what I would start to consider competent.

    We're boned until 2030ish, when gear gets cheaper (allegedly).

    Part of me thinks there's a dark conspiracy at play here. Give people affordable access to frontier LLMs, make self hosting hardware cost prohibitive, then jack up subscription prices.

    I think there's a way out of that mess, but it needs people to stop chasing "bigger, better" and start chasing "actually, how can I use what I have to do X instead of needing bigger and better?" but that needs talented devs and a mind shift.

    ICBW and YMMV.

  • Removed Locked

    Most slopcode projects are abandoned and deleted within months of release

    Jump

  • Just to nit pick (though maybe not) is there a social contract? That usually implies in exchange for X, you get Y.

    If the thing is provided as FOSS, what does the dev get, contract wise?

    Isn't that how we end up with devs walking away entirely due to "I downloaded your project, you owe me xyz, you fuck"? I've seen that happen more than once and it's a real factor in projects being abandoned, even before slopcode.

    Speaking for myself only: when I share something, it's usually something I made for myself that I think others might enjoy or find useful.

    As the dev, I'm happy to look at suggestions or reports, with no guarantee that your idea will be implemented. If it is, I credit it and you.

    I also refuse PRs, because if I am developing for me and sharing, then I'm not developing a product for sale to spec or running a democracy. I don't know you, you don't know me and you likely don't know what the long term road map or invariant constraints are, so I'd rather just not. I realise that's not a commonly held position but it's in the same "limited time" category.

    I'm happy for you to fork it, ask questions and spin up your own tho - that's why I like AGPL-3.

    Between all that, I've been able to avoid the excesses of both sides but YMMV.

  • Removed Locked

    Most slopcode projects are abandoned and deleted within months of release

    Jump

  • How's it made?

  • Removed Locked

    Most slopcode projects are abandoned and deleted within months of release

    Jump

  • Still a problem with GLMs etc in that very, very few of us have the rigs capable of running it at true frontier equivalence.

    Gate keeping AI is here - now. I don't expect China et al to be the bastion of future alternatives. Qwen is already on record of going closed source. Subscriptions and API costs will go up.

    It's gonna be a mess. Plan accordingly.

  • Removed Locked

    Most slopcode projects are abandoned and deleted within months of release

    Jump

  • Yeah, but OTOH, do we cheer for centralisation of technology for thems that can affords to pay?

  • Removed Locked

    Most slopcode projects are abandoned and deleted within months of release

    Jump

  • Then you're not actually against AI, you're against hustle culture, influencers and unsustainable practices.

    I'm 300% against that myself.

    What I don't like is when nuance gets compressed for sake of point scoring, tribalism or as an end run for due diligence. Not saying you did that BTW.

  • Removed Locked

    Most slopcode projects are abandoned and deleted within months of release

    Jump

  • Dammit...did you have to pivot this into a pitch?

  • Removed Locked

    Most slopcode projects are abandoned and deleted within months of release

    Jump

  • I think that compresses the nuance too much. I think of vibe coding like this:

    Back in the day, the C64 bedroom coder wasn't trying to "disrupt" anything. They were making something that worked for them, modifying example code (usually...poorly) and then sharing it. That's how the demo scene started, bugs and all.

    I was there Gandalf, 3000 years ago etc

    The pride was in the craft, not the pitch deck.

    The difference between that and the "we shipped" crowd isn't that the C64 coders didn't want recognition - it's that they weren't substituting the performance of shipping for the thing itself.

    Or using a woodworking analogy; you want people to appreciate the dovetail joint, not the Instagram reel of you cutting it.

    If you made the dovetail joint, it's obvious. Does which tool you used particularly matter? Shouldn't I independently assess the quality of the joint regardless?

    It's very trendy right now to be anti-ai everything, and I get that, but we're at risk of missing the forest for the trees here.

    AI is a force magnifier and it's not going anywhere. Take the same due diligence with projects you see here you do with other areas of your life - if it sounds too good to be true etc etc.

    I hate being the wowser police on this, but from where I sit both the "FuckAI" and "we just shipped... curious to hear" crowds have the same issue.

  • Removed Locked

    Most slopcode projects are abandoned and deleted within months of release

    Jump

  • Look, I think there's probably a charitable and uncharitable read for this.

    The charitable read is that people are excited to be able to bring their projects to fruition and want to share it, but don't realise how much structure is required around that.

    There's probably also a selection pressure that forces most projects into that same shape of "we built", "we shipped", "curious to hear your thoughts".

    Reddit is rife with that lingo, but Reddit is also probably 3/4 of the way to Moltbook now.

    The other reading is to just assume that this is a symptom of the same influencer hustle culture / LinkedIn brain that's everywhere in mainstream.

    I also think - with enough engineering discipline, project management skills and QC audits, someone competent (but who otherwise lacks programming syntax and fluency) could use AI to create something worthwhile, safe and sane. To my understanding, that's what PMs do IRL.

    Obviously, I'm not saying that most people who vibe code do that or there won't be bugs or issues. OTOH, you can also say that for hand-coded projects.

    Adjacently, there's also been a trend towards harassment of FOSS developers that's almost the mirror image of this - the same lack of accountability, just aimed in the opposite direction.

    If you didn't pay for it (and I don't mean throwing $10 into a ko-fi link, once) don't expect 24hr Zendesk ticket support. As Bluey is fond of reminding us: you get what you get and you don't get upset.

    Slop code isn't an AI problem. It's a people problem.

  • Xbox One X to Linux Questions

    Jump

  • Sell it and get a uFF or sFF PC. $200 at most - and even cheaper if you scope it right. It should be cost neutral swap.

  • Xbox One X to Linux Questions

    Jump

  • I always get mixed up with Xbox naming conventions (heaven forbid Microsoft went something simple like Xbox 1,2,3 etc) but are you sure that Xbox can run Linux?

    If the answer to that is no, all else is moot. Better to get a tiny/micro/mini for $100 and go from there.

    Also... I have to imagine that Xbox sucks down power from the wall, prodigiously, compared to the kind of Linux / server compute it gives you.

    Sell it. Get a PC.

  • Gamers push back: 71% say they’re not ready to let physical games die as PlayStation and Xbox move toward a digital-only future”

    Jump

  • Nah, not all of us. When switch 2 came out (and because I wanted a hand held) I looked at it, poked at in the shop and decided to buy a DS instead. $80 vs $650. If I'd been feeling more generous, I would have gotten a Retroid pocket 5 or something. I didn't need it.

    The switch 2 is shiny, but it didn't fill any niche for me that one of my other rigs couldn't do (Wii, Wii u, $100 emulation box I built).

    "Bigger, better, faster" is a mug's game. It cannot be won and it's the lynchpin of capitalism. The way to win is to not play.

    Gaming is a luxury good. If the big players are going to go full retard, let them. Let it all burn while you find greener pastures.

    This goes for all tech (and luxury goods), IMHO. Do more with less.

    Reject the status ladder → identify the actual function → satisfy it with the smallest tool that works = you win.

    Then enjoy the absurdity of $650 Switch 2 (and marvel at the cost of Switch 1 increasing in price. Lol no).

  • What's the worst part about getting older that young people don't know about?

    Jump

  • Some general questions on AI setup

    Jump

  • Ooh...hang on. Doesn't a headless server in Linux require a dummy HDMI plug if you have an Igpu + GPU? You might need to confirm that.

  • Some general questions on AI setup

    Jump

  • On the container stuff - can't help much. I'm a bare metal sorta guy. Ask BruceTheMoose from above, who also posts on !selfhosted. One thing I will tell you - shunting around a GPU between containers in Linux sounds like a pain in the ass. I would be tempted to keep Comfyi and llama.cpp in one container so you don't have to so pass thru / rebinding bullshit.

    Ask claude for advice on LXC and CUDA pass thru here to reduce pain.

    Re: second NVIDIA card - probably would yes, unless you get something like a Quadro P1000...which wouldn't do much for you.

    Re: LocalAI. I've never used it so can't comment. I either use OWUI (heavy but feature rich) or the webui that's inbuilt with llama.cpp (light, fast, but somewhat cut down).

    If you're willing to use ComfyUI, then chat (while tts stays elsewhere) then probably 1 gpu could do it. Try the Qwen 3.6 35B model I suggested - it should get you 25+ tok/s on that GPU (show Claude the YouTube video and tell it to pull the settings from the video description for you).

  • Some general questions on AI setup

    Jump

  • Depends. Two GPUs can make sense if you want the chat/radio stack and ComfyUI to be able to run independently without fighting over VRAM. But it depends on how big the underlying models are and if they run at same time.

    Using 1 card tho, logically, there are 3 things that can happen -

    • If there's enough headroom: everything coexists (if models are small enough or well orchestrated)
    • ComfyUI throws an OOM and either crashes or falls back to CPU offload (slow, but it won't usually fail silently)
    • LocalAI typically won't auto-evict a loaded model just because another process wants VRAM (it'll just sit there blocking)

    In other words, on a single card, you're either manually managing load/unload cycles, eating CPU offload penalties on ComfyUI, or playing VRAM Tetris.

    None of those are fun if you want both services available simultaneously....but that depends a lot on how big / greedy the models are. Do you want / need everything at same time?

    Eg: there are small, medium and large versions of this, https://github.com/ace-step/ACE-Step-1.5 and there are small, medium and larger versions of image generators.

    So I would say: yes, two cards are worth considering if you want both workloads available at the same time. The practical way to do this is pin each app to its own GPU via CUDA_VISIBLE_DEVICES so they never see each other: LocalAI on CUDA_VISIBLE_DEVICES=0, ComfyUI on CUDA_VISIBLE_DEVICES=1 and HDMI output via your iGPU / CPU for desktop etc.

    BTW, the cards don't need to match either - a cheaper smaller card can handle the chat/TTS stack (or even CPU like we said above) while the bigger one handles image gen. If you are happy to manually switch between AI chat/TTS and ComfyUI, then two cards may not change much besides convenience.

    PS: Worth considering a three-way split too, if you want everything all at once / separated.

    • TTS offloaded to CPU,
    • chat/music stack on one GPU,
    • image gen on the other GPU.

    Small TTS models like Piper or Kokoro run fine on CPU, and for a radio context where you have even a few seconds of buffer, the latency is hidden. That frees up VRAM on your chat GPU.

    Actually, that's how I would leverage 1 GPU for everything / mix and match you CPU / GPU but ICBW

    PS: Might be worth chatting to your LLM (lol) about this too (or using a cloud one). These are not general AI question :)

  • Some general questions on AI setup

    Jump

  • That's not hard to find. Let me help you out.

    https://huggingface.co/bartowski/Qwen_Qwen3.6-35B-A3B-GGUF

    On the second thing, I wouldn't worry too much about it. Speech to text and text to speech is really efficient on CPU. In fact, I am dictating this directly into the CPU with a microphone using the program I'm about to share with you below. It works perfectly fine and is instantaneous. You might like to try it because we're both running on the same CPU.

    https://handy.computer/

    https://github.com/cjpais/handy

    I recommend Parakeet V2 by the way.

    I'm not sure how you are thinking of using TTS. If you switch over to something like Open Web UI as your front end, you can have various plugins for that sort of thing.