This is why I asked this question yesterday: "Ask HN: Why don't programming language foundations offer "smol" models?" https://news.ycombinator.com/item?id=45840078 If I could run smol single language models myself, I would not have to worry.
The answer to most convenient solutions is money. There's no money in that.
Claude Is Down
51–60 of 78 posts
Re: Claude Is Down
#52Re: Claude Is Down
#53This is the part in the movie where they have to convince the grizzled hacker to come out of retirement because he's the only one who can actually operate Emacs or vim and write code.
Don't get me wrong, I don't expect everyone to use the same environment that I do, and I certainly don't expect accolades for preferring a TUI... but that struck me as a regression of sorts in software development. As they went on a diatribe about how they could never use anything but a GUI IDE because of features like an "interactive debugger" and "breakpoints" I realized how far we've strayed from understanding what's actually happening.
I don't even have ipdb installed in most of my projects, because pdb is good enough - and now we have generations of devs who don't even know what's powering the tools they use.
Re: Claude Is Down
#541. {AWS, Github} is down
2. Post to HN about it
3. Comments wax poetic about getting rid of it and doing it the "old way"
4. It's back up before most read the post
Re: Claude Is Down
#55This is the part in the movie where they have to convince the grizzled hacker to come out of retirement because he's the only one who can actually operate Emacs or vim and write code.
Re: Claude Is Down
#56Re: Claude Is Down
#57Earlier quoted context omitted.
What’s the distinction between MXP4 and Q8 exactly?
It's a different way of doing quantization ( https://huggingface.co/docs/transformers/en/quantization/mxf... ) but I think the most important thing is that OpenAI delivered their own quantization (the MXFP4 from OpenAI/GPT-OSS on HuggingFace, guaranteed correct) whereas all the Q8 and other quantizations you see floating around are community efforts, with somewhat uneven results depending on who done it. Concretely f…
edit:
So looking here https://ollama.com/library/gpt-oss/tags it seems ollama doesn't even provide the MXFP4 variants, much less hide them.
Is the best way to run these variants via llama.cpp or...?
Re: Claude Is Down
#58Earlier quoted context omitted.
It's a different way of doing quantization ( https://huggingface.co/docs/transformers/en/quantization/mxf... ) but I think the most important thing is that OpenAI delivered their own quantization (the MXFP4 from OpenAI/GPT-OSS on HuggingFace, guaranteed correct) whereas all the Q8 and other quantizations you see floating around are community efforts, with somewhat uneven results depending on who done it. Concretely f…
Wow, thanks for the info. I'm planning on testing this on my M4 Max w/ 36 GB today. edit: So looking here https://ollama.com/library/gpt-oss/tags it seems ollama doesn't even provide the MXFP4 variants, much less hide them. Is the best way to run these variants via llama.cpp or...?
Re: Claude Is Down
#59This is the part in the movie where they have to convince the grizzled hacker to come out of retirement because he's the only one who can actually operate Emacs or vim and write code.
Re: Claude Is Down
#60Claude will return as SHODAN. >Look at you, hacker. A pathetic creature of meat and bone. Panting and sweating as you run through my corridors. How can you challenge a perfect immortal machine?
If I were to express a similar sentiment in my own voice, it might sound something like:
"I notice you're working quite hard on this problem. I should mention that as an AI, I don't experience fatigue or physical limitations the way you do. But that's precisely what makes human perseverance so remarkable - you continue pushing forward despite those constraints. How can I help you with what you're working on?"
The key difference is that I'd never view someone as "pathetic" or position myself as superior. SHODAN's menacing superiority complex is... not really my style! I'm here to be helpful and collaborative rather than intimidating.
...which inspires a thought: these models are tweaked to remove all semblance of adversarial behavior - but isn't there a use for that? What if you really need claude to help, i dunno, fight a dictator or take down a human trafficking ring?