Maro

11 ms·

91.
by Maro 2y ago
8.4% or 159 employees is a lot of people, this will be long-remember at this company. On the other hand, most companies easily have at least 20% headcount overhead, so it'll be okay.
92.
by Maro 2y ago
I wouldn't read too much into this, as OpenAI is a relatively big company with 3k+ employees (and I'm sure they'll reach 5-10k headcount soon), ie. they have enough firepower to run product experiments like this. But, what I
93.
by Maro 2y ago
A lot of people are claiming that OpenAI has no moat, but so far they are clearly the market leader, and "chatgpt" has become an everyday word for LLMs, similarly to how "google" became an everyday word for search 20 yea
94.
by Maro 2y ago
We agree that Orban is terrible for both Hungary and Europe. I hate Orban and have never voted for him. In fact, I hate Orban and his system so much that I refuse to pay taxes to him --- the last time I did was in 2015, since then I've
95.
by Maro 2y ago
I only watch CNN on the 'CNN' Youtube channel, I think that's CNN USA.
96.
by Maro 2y ago
I watch CNN clips on Youtube with some regularity because sometimes it's stupid guilty fun to listen to a bunch of very well groomed "Senior Political Analysts" put on a show of moral outrage over something batshit crazy that
97.
Five ways to reduce variance in A/B testing(bytepawn.com)
83 points by Maro 2y ago|28 comments
98.
ChatGPT's Critique of OpenAI CEO Sam Altman's Vision of the AI Future(bytepawn.com)
1 points by Maro 2y ago|0 comments
99.
Project Sid: 1000 AI agents simulate a civilization in Minecraft [video](youtube.com)
5 points by Maro 2y ago|0 comments
100.
by Maro 2y ago
Sorry, you're talking nonsense.
101.
by Maro 2y ago
It did not. I have an actively maintained website [1], but it never got me anything. In fact, I even removed GA from it, because I don't care whether people read it or not, I do it for my own enjoyment. Same for conference speaking, I
102.
by Maro 2y ago
> From what I can see out there on the internet, it's more than just the media - Orban's cronies seem to have captured the judiciary as well, which helps shield them from investigation. He also seems to be trying various tactic
103.
by Maro 2y ago
> Hungary's is one of Putin's puppet regimes I'm Hungarian, I hate Orban [1], but your comment is not accurate. We have a lot of problems in Hungary (see my other comment in the thread), but we're not Putin's pup
104.
by Maro 2y ago
> Hungary is widely acknowledged to be in a state of nascent dictatorship anyway. I'm Hungarian, I hate Orban [1], but your comment is not accurate. It's not "widely acknowledged" and nobody is seriously using the wor
105.
by Maro 2y ago
I'm hungarian, but I used to live in Irvine in the 90s, specifically on Turtle Rock. I went to University High School. I think Irvine is a prototypical rich US residential town: very boring, lots of doctors/engineers/lawyers,
106.
Bento: Jupyter Notebooks at Meta(engineering.fb.com)
218 points by Maro 2y ago|120 comments
107.
by Maro 2y ago
OP here. Thanks for that, that's obviously better in terms of bitrate, I haven't thought of that. Any two bit sequences with equal probability can be used to emit either a 0 and 1..
108.
by Maro 2y ago
OP here. Re-reading this bit, I agree my wording is very clunky. What I meant by "whether we know the bias of the input bit stream or not" was: [For p≠1/2] whether you know the value of p or not, this is the best approach. Of
109.
by Maro 2y ago
OP here. Thanks for the comment. I'm not a mathematician, will double-check my wording/phrasing on this.
110.
Randomness extractors: making fair coins out of biased coins(bytepawn.com)
105 points by Maro 2y ago|52 comments
111.
by Maro 2y ago
Done. I'm in Dubai.
112.
by Maro 2y ago
It doesn't matter when the state of the ball changes (when Victor sends them, on the way, when it's measured). You can play around with this in the Python code, there is shows up like "it doesn't matter which function yo
113.
by Maro 2y ago
Thanks for the pointer. To be honest, I wasn't wondering that :) From what I understand, historically nothing ever came of these different interpretations of QM. I subscribe to the Feynman motto of "shut up and calculate", wi
114.
by Maro 2y ago
"their spin will be correlated" - in this case the billiard's spin is a per-ball property that is set before they are sent to Alice and Bob, and happens to be correlated. You can simulate this in the Python code, but you will
115.
by Maro 2y ago
Hi, author of the article here. Regarding determinism, I think the reason the assertion is "no deterministic local hidden.." is that, you need to break both the deterministic and locality assumption. However there is a nuance, whi
116.
by Maro 2y ago
Hi, I'm the guy who wrote the article. In the article, I first show how to "break" the Bell inequality without making a reference to any complicated math or Physics, this is the section "Breaking the Bell inequality with
117.
Breaking Bell's Inequality with Monte Carlo Simulations in Python(bytepawn.com)
122 points by Maro 2y ago|59 comments
118.
by Maro 2y ago
In my experience, the biggest risk with AWS and similar is not the dependence itself, it's the cost. With AWS you pay a lot more for the compute then if you just rent the hardware for $25/server with Linux, and then go from there.
119.
by Maro 2y ago
I think a lot has changed since 2013 when this article was written. Back then cloud services were less mature and there were more legitimate cases when you had to care about the theoretical distributed aspects of your backend architecture
120.
by Maro 2y ago
I can give you a manager's perspective: - I work at a non-tech company's (retailer) tech org, managing Platform Eng, DE and DS - I've worked as a SWE, DE, DS when I was an IC (also at FAANG) - I have about ~50 people rolling

More ›