buyucu

22 ms·

601.
by buyucu 2y ago
Seeing what china is doing to the car market, I give it 5 years for China to do to the AI/GPU market to do the same. This will be good. Nvidia/OpenAI monopoly is bad for everyone. More competition will be welcome.
602.
by buyucu 2y ago
and why should we believe anything an openai employee says. openai tried to hide their financial relationship with frontier. why do that if you're honest?
603.
by buyucu 2y ago
have you tried asking chatgpt something even slightly controversial? chatgpt censors much more than deepseek does. also deepseek is open-weights. there is nothing preventing you from doing a finetune that removes the censorship. they did
604.
by buyucu 2y ago
%90 of the time all the LLM does is to point me to the website where the solution is.
605.
by buyucu 2y ago
they were caught hiding a financial relationship with a benchmark owner, while making that benchmark a centerpiece of their advertisements. epoch is also very vague about how much access to the dataset openai had, but it's clear that t
606.
by buyucu 2y ago
I look forward to the showdown between big tech and big copyright.
607.
by buyucu 2y ago
I expected this number to be higher. I guess more leave between postdoc and phd.
608.
by buyucu 2y ago
Ollama is so close to greatness. But their refusal to support Vulkan is hurting them really bad.
609.
by buyucu 2y ago
OpenAI funded the frontier dataset, and hid this relationship. Looks very fishy.
610.
by buyucu 2y ago
OpenAI was caught gaming benchmarks recently with FrontierMath. Just (yet another) sign that benchmarks are very flawed and everyone is training on them. So I would not put too much weight on how the models are doing on benchmarks.
611.
by buyucu 2y ago
Huawei already has A100-equivalent hardware that they are selling in China. I give them 5 years to do to GPUs what BYD has done to cars.
612.
by buyucu 2y ago
Chinese censorship is less than American censorship. Have you tried asking anything even slightly controversial to ChatGPT?
613.
by buyucu 2y ago
I don't think asking coding problems to a model by itself is fair. Almost all commercial models are combining a RAG and web-search. I find that most correct answers come from that, not from the actual model.
614.
by buyucu 2y ago
thanks. I'm running the qwen distillation right now and it's amazing.
615.
by buyucu 2y ago
running the 7b distilled version with ollama right now.
616.
by buyucu 2y ago
good
617.
by buyucu 2y ago
When you try to sanction 1.3 billion people, what you are really doing is to sanction yourself.
618.
by buyucu 2y ago
the 7b distilled version works great on my laptop cpu and iGPU with vulkan. You can use llama.cpp (for iGPU with vulkan) or ollama (for cpu). My laptop is a cheap laptop from 5 years ago. Not cutting edge hardware.
619.
by buyucu 2y ago
You can use it locally on ollama: https://ollama.com/library/deepseek-r1 Also all providers are training on your prompts. Even those that they say they aren't.
620.
by buyucu 2y ago
If you crawl the internet and train a model on it, I'm pretty sure that model will say that it's ChatGPT.
621.
by buyucu 2y ago
It's not surprising. Large organizations are plagued with bureaucracy, paperwork and inertia. It's much more easier to innovate in a smaller setting.
622.
by buyucu 2y ago
H200 has less memory than MI300
623.
by buyucu 2y ago
I'm confused why there is an 7b and an 8b version: https://ollama.com/library/deepseek-r1/tags
624.
by buyucu 2y ago
I would love my country to ban Instagram and Twitter. No need for American propaganda.
625.
by buyucu 2y ago
because politicians are honest and don't lie, right?
626.
by buyucu 2y ago
CUDA lock-in is not what it once was. I do a lot of stable diffusion and I was pleasantly suprised that I could just run the same code on AMD with no changes.
627.
by buyucu 2y ago
MI300 is an insanely good GPU. There is nothing that Nvidia sells that even comes close. The H100 only has 80GB of memory, whereas MI300 has 192GB. If you are training large models, AMD is the way to go.
628.
by buyucu 2y ago
How will you react when other countries start banning Meta/Twitter with the same argument?
629.
by buyucu 2y ago
The Tiktok ban is an attempt to herd the Tiktok customers to Meta products. Nothing more.
630.
by buyucu 2y ago
where did I pick anything? USA is the greater evil, with a much higher bodycount. that is something we agree on.

More ›