p1esk

21 ms·

481.
by p1esk 2y ago
In 2025? Probably a consulting business to help companies replace their software engineers with AI. In 2024 it would be too early, and in 2026 it would probably be too late.
482.
by p1esk 2y ago
Apple Vision Pro 3
483.
by p1esk 2y ago
I hope this will go into AVP3
484.
by p1esk 2y ago
Interesting how it will compete with 4o mini.
485.
by p1esk 2y ago
Not necessarily. As others said, could be caused by increasing number of applications per applicant, or foreign students. Or maybe increasingly important reasons to graduate from top school that have nothing to do with actual education. If
486.
by p1esk 2y ago
Is confidence in good schools rising?
487.
by p1esk 2y ago
And yet acceptance rates at top schools have been consistently going down.
488.
by p1esk 2y ago
The lazy one gets exactly the same *opportunity* to have a good education as the bright one. Do you see the difference?
489.
by p1esk 2y ago
What draconian reason could you possibly have to make the argument that we shouldn't try to give every child an equal opportunity for a high quality education? Opportunity, sure. But an opportunity to get a good education is not the
490.
by p1esk 2y ago
His point is an AI teacher cannot force someone to learn, while a human teacher can (maybe).
491.
by p1esk 2y ago
Or they simply kill everyone who complains.
492.
by p1esk 2y ago
Are you using gpt4?
493.
by p1esk 2y ago
Is this another wrapper of chatgpt?
494.
by p1esk 2y ago
OK, I just looked at the first blog post: “ROCm is nowhere near where it needs to be to truly compete with CUDA.” That’s all I need to know as an AI/ML customer.
495.
by p1esk 2y ago
I'm building this business in a way to de-risk things as much as possible Then why did you go with AMD and not Nvidia? Are you not interested in AI/ML customers?
496.
by p1esk 2y ago
How do you choose the companies?
497.
by p1esk 2y ago
the business knows the right path better than you do. If its trying to go a direction that you didn't have in mind, you're smarter to go with that direction than to try to force the path to the one you had planned. I don’t get i
498.
by p1esk 2y ago
You do realize they are just having fun in that paper, right?
499.
by p1esk 2y ago
Is this your website? If so, can you tell us who you are and what exactly are you trying to do? The website is light on information.
500.
by p1esk 2y ago
What are you going to say in an interview when they ask you to describe your Java experience?
501.
by p1esk 2y ago
No idea, I just summarized TFA.
502.
by p1esk 2y ago
OK, would you say poetry is the best for ML engineers?
503.
by p1esk 2y ago
Looks like Conda is still the best package/env manager for ML engineers.
504.
by p1esk 2y ago
LLMs are trained to predict the next word in a sequence. As a result of this training they developed reasoning abilities. Currently these reasoning abilities are roughly at human level, but next gen models (gpt5) should be superior to human
505.
by p1esk 2y ago
It doesn’t matter what is the most commonly used model. If you want to make claims about LLM capabilities you better be using the best model available.
506.
by p1esk 2y ago
Probably because they used GPT 3.5 to make general claims about LLMs.
507.
by p1esk 2y ago
There’s a huge difference between 3.5 and 4. 3.5 is pretty much useless. 4 is very decent, it saved me countless hours at work.
508.
by p1esk 2y ago
Who cares what is default? If you want to make claims about LLMs you use the best model.
509.
by p1esk 2y ago
They used GPT 3.5 so these findings are irrelevant.
510.
by p1esk 2y ago
4 months is plenty of time to rerun experiments using GPT4.

More ›