Показать HN: Reame - сервер выводов на процессоре, который становится быстрее по мере работы(github.com)
github.com
Show HN: Reame – a CPU inference server that gets faster as it runs
https://github.com/swellweb/reame
26 comments
Seems the entire repo and documentation is all ai-generated?
Yes, I use AI to help me build and write docs — solo dev, I'd be stupid not to. The benchmarks are real though: free Oracle ARM box, my laptop, a cheap VPS. I ran them all myself, including the ones that went badly (they're in the README too).
But if you didn't care enough to write the docs yourself, why are you submitting it for discussion?
Are you seriously expecting humans to spend their time and effort and discuss your AI written docs that you literally admit you'd be "stupid" to have spent effort on yourself?
That's really condescending of you.
Are you seriously expecting humans to spend their time and effort and discuss your AI written docs that you literally admit you'd be "stupid" to have spent effort on yourself?
That's really condescending of you.
It seems pointless to me to stir up conflict or engage in unnecessary arguments; I have simply expressed my personal view—namely, that I value the utility of artificial intelligence, which I sometimes use to kickstart new projects (projects that can certainly be supplemented with different types of input in the future). I believe this is an innovative, open-minded, and forward-looking approach. If you disagree, you are free to hold your opinion, but I do not share it.
For what it's worth: the feedback in this thread found three real bugs that were fixed, tested and shipped within 24 hours — that's the part of the discussion I care about.
For what it's worth: the feedback in this thread found three real bugs that were fixed, tested and shipped within 24 hours — that's the part of the discussion I care about.
"For what it's worth" ? Yeah for what it's worth to you, you mean.
You submitted and expected people to read an article that you say you would have been stupid to write yourself, that's quite condescending, and just because you got something worth your while out of that doesn't really make it okay.
You submitted and expected people to read an article that you say you would have been stupid to write yourself, that's quite condescending, and just because you got something worth your while out of that doesn't really make it okay.
That is genuinely a good callout — it is worth stopping here.
Thanks for sharing your work!
Please how to select the model? I downloaded tinyLlama, put it in ./models, changed reame.conf but I get:
(No such file or directory)
Otherwise putting the model in /opt does not please me much, I fear to forget a model is there, if it is in reame folder its much easier to notice and manage.
Please how to select the model? I downloaded tinyLlama, put it in ./models, changed reame.conf but I get:
(No such file or directory)
Otherwise putting the model in /opt does not please me much, I fear to forget a model is there, if it is in reame folder its much easier to notice and manage.
Sick that you can get 2 arm cores and 12 GB ram for free at Oracle cloud, did not know that
You can actually get more. You have an amount of resources for ARM servers, if you allocate all of that to a single one you ger 24GB RAM and 4 vCPU.
This was reduced recently to 2/12. It's also next to impossible to allocate the free stuff unless you put in a CC. If you do, you get the servers right away, though. Just be careful not to spend. You can set up a budget trigger to shut down the vms.
Old servers may have been grandfathered, but mine was set to 1/1 and couldn't be reshaped.
Old servers may have been grandfathered, but mine was set to 1/1 and couldn't be reshaped.
I logged in to double-check if I was going to pay for my beefy ARM server. There was a notification about this change, but looking into the billing section, I saw no change. I have an alert in place if I pay more than 5€ per month, so for now I'm leaving my server as is.
They halved it recently
Yeah but it's Oracle
If you have good luck perhaps. That tier of server is virtually always sold out
If a prefix cache is good, why not keep it in primary mem?
Why qwen 2.5 everywhere? Why not 3.5?
Because llms reflect qwen 2.5 heavily in training data, and that's who is giving the recommendations.
looks really interesting
Interesting. But I'm kind of hard getting past the AI-written README:
"What Reame is NOT for — said plainly, because trust is built here: a general-purpose ChatGPT replacement (frontier reasoning and broad knowledge need frontier parameter counts), agentic coding assistants, or creative long-form writing at scale. If your task needs a 100B-class brain, buy one; if it needs your documents processed privately, forever, at zero marginal cost — that's a realm you can own."
The realm you can own. How did these things learn to write that way? Oh, yeah, lots of marketing and advertising copy.
"What Reame is NOT for — said plainly, because trust is built here: a general-purpose ChatGPT replacement (frontier reasoning and broad knowledge need frontier parameter counts), agentic coding assistants, or creative long-form writing at scale. If your task needs a 100B-class brain, buy one; if it needs your documents processed privately, forever, at zero marginal cost — that's a realm you can own."
The realm you can own. How did these things learn to write that way? Oh, yeah, lots of marketing and advertising copy.
Fair point — I removed that line just now. I'm a solo dev and I use AI for the docs; sometimes it sounds like an ad and I don't catch it. The benchmarks are real though, I ran them all myself. Try the software and tell me if something is wrong.
It is so weird (or, used to be) to see an LLM's internal thought process pop up this way. Like imagine how strange it would be to read human writing that accidentally included thoughts undercutting the ongoing sentence. It's the moment you know that nothing you're reading has necessarily been seen by a human before or relates to reality.