segmondy

11 ms·

271.
by segmondy 6mo ago
They have had the best math models for about a year most folks just didn't know about it. You can't find inference on APIs, but I run these at home, this is also the advantage of open models. https://huggingface.co&#x2
272.
by segmondy 6mo ago
Which Monopoly? Are all large enterprises in USA? There are tons of them outside and they will run the open ones and cheapest ones to infer and those are Chinese. I run Chinese models at home and don't bother with cloud. If I co
273.
by segmondy 6mo ago
It's not about 2 cultures, but 2 timelines. China has seen the game and adapted, they will not respond with prior losing responses. Meanwhile, America is playing the same moves because it worked in the past.
274.
by segmondy 6mo ago
Pick people at random from countries around the world. Ask them what bad things have happened to them or their country because of China or USA. What do you think the result is going to be?
275.
by segmondy 6mo ago
My guess is Chinese govt is going to mandate that labs switch all future training and inference to Huawei. DeepSeek has shown it's possible. Once they are done, the rest of the world is going to be buying Huawei! I for one can&#x2
276.
by segmondy 6mo ago
Seems there should be a "hobbyist kernel" with all that kernel code, you run at your own risk but get all the toys for your obscure use cases.
277.
by segmondy 6mo ago
what do you mean nobody? there are a few of us using it, and we are completely broken when support is taken away.
278.
by segmondy 6mo ago
It's ping pong.
279.
by segmondy 6mo ago
pi does, it can talk to any OpenAI API
280.
by segmondy 6mo ago
It should crush Qwen3CoderNext especially since it beats 397B
281.
by segmondy 6mo ago
96gb of ddr5 ram is about $800.
282.
by segmondy 6mo ago
Obviously he's telling you their spec incase you wish to reproduce his results. Why don't you try it and tell us how your result compares.
283.
by segmondy 6mo ago
You can use AI, tell AI to read the page first and if it's AI anything to block it. Vibecode an extension.
284.
by segmondy 6mo ago
I like it, it's a good start.
285.
by segmondy 6mo ago
They see that the new KimiK2.6 will eat their lunch. They don't care about you, they just care about your money and will take away your options if they don't believe you have a solid alternative.
286.
by segmondy 6mo ago
The great thing bout the net is also it's biggest problem. Anyone can write a blog and if it looks nice, sounds polished they could sway a large group. I roll my eyes so strong at folks that reject SOLID principles and design patter
287.
by segmondy 6mo ago
You can't buy 16 64gb dimms for $3000. Go shop memory prices again. But yes an old epyc can run this with no GPU at reasonable speed and if you throw a few GPUs you can get very manageable speed. I run this at home on an old system
288.
by segmondy 6mo ago
It has long been Claude level since 2.5
289.
by segmondy 6mo ago
I don't think so, it's just how things played out. Thanks to Meta, after llama leak and meta followed up with llama2 and llama3 that caused everyone else to follow up with open models, Stablediffusion, Mistral, Cohere, Microsoft
290.
by segmondy 6mo ago
Tim saw the ram shortage and said, "WTF am I suppose to do with this? I'm out of here!" Better leave a hero ...
291.
by segmondy 6mo ago
I never said that, my point is that paying 20 people at $35 for 24/7 is about $6 million. You can replace that with a $10k system running 20 parallel requests for a year and save lots of money.
292.
by segmondy 6mo ago
Humans are not cheaper than AI models. Let's go with $35 an hour. 24 365 = 8760 8760 $35 = $306,600 Yeah, a human working non stop will run $300k. Now you said, the "best" models. I personally reckon that 80-90% of most w
293.
by segmondy 6mo ago
This is obviously a continuation training of 3.5, it's not a new model architecture but an incremental improvement.
294.
by segmondy 6mo ago
I can't believe you're such a party pooper. It's exciting times, the silly things do matter!
295.
by segmondy 6mo ago
You can do that with llama-server
296.
by segmondy 6mo ago
This is why you should have local models. The local models are good enough for private chats, they might not be as good as the cloud models for precise technical work, but for general sensitive chat you definitely should stick to local.
297.
by segmondy 6mo ago
I moved my DO server to a pi that was gathering dust. I agree, folks need to get off the cloud, find an old laptop or an old $40 mac mini, they are usually low power enough.
298.
by segmondy 6mo ago
So what? It will be back up.
299.
by segmondy 6mo ago
They are mass copying any idea they see out there. They are not happy enough being a platform, they want to be everything. Their dream is to be the one AI that eats it all. It's so stupid that folks are using their system. I will
300.
by segmondy 6mo ago
I don't think most people realize that. Quality of tokens beats quantity of token. I always tell folks to go as high a quant as you can only go lower if you just don't have the memory capacity.

More ›