_zoltan_

4 ms·

1.
by _zoltan_ 5d ago
billions? who is running BILLIONS of agents? tens, hundreds, maybe a couple thousand at a time? absolutely.
2.
by _zoltan_ 7d ago
I optimize a lot of CUDA and it got really, really good at it.
3.
by _zoltan_ 10d ago
you might be OK with calling everybody using LLMs a bunch of noobs. I disagree. this has nothing to do with my own usage.
4.
by _zoltan_ 10d ago
Anti LLM is not justified, it's crazy. They are digging their own grave.
5.
by _zoltan_ 11d ago
I don't need to go that far, living in Switzerland. I don't know why you'd assume I'm in the US, or somehow that I regard Kraft as great cheese.
6.
by _zoltan_ 11d ago
for your point a: writing code by hand puts you ahead of the people who don't write code by hand, eg, the people using LLMs to generate code? that doesn't make much sense assuming both of those personas are senior SWEs.
7.
by _zoltan_ 11d ago
buy another sub.
8.
by _zoltan_ 11d ago
Yes, it is a waste of time right now not to use AI to write code. It's the same category of "artisan code" like "artisan cheese": you can do it if you have a ton of free time as a hobby but other than a hobby? Not w
9.
by _zoltan_ 13d ago
Eh, take your fav painter then.
10.
by _zoltan_ 13d ago
that's a dumb comparison. are you equating writing code to painting like Michaelangelo?
11.
by _zoltan_ 13d ago
coding for the sake of coding is, well, dumb.
12.
by _zoltan_ 13d ago
> Focusing on data structures, architecture and algorithms is what competency in programming has looked like since forever yes, being able to bang out leetcode hard from head on an interview on paper matters so much... oh wait. it does n
13.
by _zoltan_ 13d ago
a lot of people don't understand that the joy of making something is not coding but actually being able to build anything I want. AI just made it 50x better, as exactly as you said, we don't need to deal with the useless junk (w
14.
by _zoltan_ 15d ago
Welcome to the new world. I, for one, welcome our new math overlords. :)
15.
by _zoltan_ 15d ago
> The benchmark was very simple. A single-shard point select, one row fetched per-query by primary key. No writes, joins, or cross-shard queries. The workload that each shard receives is isolated, in that there are no single queries that
16.
by _zoltan_ 15d ago
https://www.youtube.com/watch?v=n0U8RdRdFDU do this. thank me later. (yes, it takes multiple days!). as good as when we hop over to France.
17.
by _zoltan_ 16d ago
hahah :) sorry, typo... in hungarian both v and w are the same sound spoken so if I don't concentrate, easy to mix up.
18.
by _zoltan_ 16d ago
what competitor? the x20 Max plan for Claude gives you way lover limits.
19.
by _zoltan_ 16d ago
I almost never pay cash.
20.
by _zoltan_ 17d ago
Thank you!
21.
by _zoltan_ 17d ago
> same approach of casually spending a few million on agentic computation is going to solve end to end materials design or drug development." you're not actually spending that money. it's sunk cost, as you already bought t
22.
by _zoltan_ 17d ago
ZDR is a thing. extremely large companies with a lot of IP rights use GPT/Claude without issues.
23.
by _zoltan_ 17d ago
They could have used an enterprise or team subscription with ZDR.
24.
by _zoltan_ 18d ago
What exact enclosure are you using?
25.
by _zoltan_ 20d ago
Does USB work already or is that a 2027 item?
26.
by _zoltan_ 21d ago
that's clear, of course. but I can't just use claude/codex cli, as it needs to see the platform emulator (android/iOS), does it not? so it's cowork?
27.
by _zoltan_ 22d ago
Never heard of Granite? https://huggingface.co/ibm-granite (I'm also an IBMer.)
28.
by _zoltan_ 22d ago
I wish you study the Wikipedia page, please. it's a very old company and we have invented really cool stuff. signed: a proud IBMer
29.
by _zoltan_ 22d ago
why is your actual benchmark question so secret?
30.
by _zoltan_ 22d ago
I'd love to learn more about this as I have some toy app ideas as well :)

More ›