osti

7 ms·

91.
by osti 11mo ago
I wouldn't say that, but just that qwen 3 max thinking definitely underperforms relative to its size.
92.
by osti 11mo ago
Qwen 3 max has been getting rather bad reviews around the web (both on reddit and chinese social media), and from my own experience with it. So I wouldn't expect this to be worse.
93.
by osti 1y ago
Apple chips are fastest in single core in most benchmarks, not just passmark.
94.
by osti 1y ago
What do you mean by imaginary patterns from Codeforces?
95.
by osti 1y ago
This kinda tracks with the latest estimate of power usage of llm inference published by google https://news.ycombinator.com/item?id=44972808 . If inference isnt that power hungry like people thought, they must be able to mak
96.
by osti 1y ago
In here https://blog.google/products/gemini/gemini-2-5-deep-think/ , the professor google worked with also claimed proving some previously unproven conjecture.
97.
by osti 1y ago
It kinda is, at least I'd say a rich person is on average more intelligent than a poor person.
98.
by osti 1y ago
Looks like no more open source models :( "We believe the benefits of superintelligence should be shared with the world as broadly as possible. That said, superintelligence will raise novel safety concerns. We'll need to be rigorou
99.
by osti 1y ago
Looks like no more open source :( We believe the benefits of superintelligence should be shared with the world as broadly as possible. That said, superintelligence will raise novel safety concerns. We'll need to be rigorous about mitig
100.
by osti 1y ago
For the coding benchmarks, does anyone know what are OJBench and CFEval?
101.
by osti 1y ago
Is it time to send Matt Damon?
102.
by osti 1y ago
Yes, but with a 400B parameter model, at fp16 it's 800GB right? So with 800GB/s memory bandwidth, you'd still only be able to bring them in once per second. Edit: actually forgot the MoE part, so that makes sense.
103.
by osti 1y ago
Interesting, so with enough memory bandwidth, even the server CPU has enough compute to do inference on a rather large model? Enough to compete against M4 gpu? Edit: I just aked chatgpt and it says with no memory bandwidth bottleneck, i can
104.
by osti 1y ago
How? These don't even have GPU's right?
105.
by osti 1y ago
True, but openai definitely isn't trying to do public research on science, they are all about money now.
106.
by osti 1y ago
The proofs also seem more human reable than openai's?
107.
by osti 1y ago
I think this is them not being confident enough before the event, so they don't wanna be shown a worse result than competitors. By being private they can obviously not publish anything if it didn't work out.
108.
by osti 1y ago
"Note that I'm also an immigrant." This sentence doesn't really help your statement, it just makes you a gatekeeper.
109.
by osti 1y ago
If you are seriously equating these two with AI, then you have horrible judgements and should learn to think critically, but unfortunately for you, I don't think critical thinking can be learned despite what people say. Note that I&#x2
110.
by osti 1y ago
What I don't get is, education is one of US's big "exports", and basically easy money; so why try to kneecap that revenue stream?
111.
by osti 1y ago
I bet they did at one point in time, then they stopped doing that, but still not bug free.
112.
by osti 1y ago
Funny how as a Chinese, I told my Canadian friend that I'd bet on one of the Chinese companies catching up to Nvidia (at least partly) before AMD or Intel.
113.
by osti 1y ago
As opposed to bow down to Apple and oh, still can't make an iOS phone?
114.
by osti 2y ago
Pretentious.
115.
by osti 2y ago
I think this is the one where they train LLM without NVIDIA GPU's.
116.
by osti 2y ago
No, he said software engineering is just programming over time, so the two are not distinct at all.
117.
by osti 2y ago
I get what you mean but calculus and algebra are quite distinct branches of mathematics..
118.
by osti 2y ago
Are 5090's able to run 32B models?
119.
by osti 2y ago
Looking forward to its release. But hopefully it will show its "thinking" process.
120.
by osti 2y ago
If you think ppl on HN or elsewhere aren't racist against the actual Chinese people, you are just incredibly naive. People have been talking about the Chinese like automotons of the government with no agency of their own for a long tim

More ›