hazard

7 ms·

1.
by hazard 3d ago
I worked very briefly at Baseten, and I can say that it was a perpetual annoyance (from an engineering perspective) that customers would complain about issues with their models but we couldn't actually see the inputs/outputs. I do
2.
by hazard 5d ago
It's a true belief, but that doesn't mean it's valid or correct. Many doomsday cults had true believers, and were able to retcon their beliefs when the predicted day came and went. AI doomers have the luxury of not having com
3.
by hazard 21d ago
In a moment of weakness, I joined a high-IQ group (Triple Nine Society). The whole mailing list was just people having the same political arguments that were on the rest of social media, just with a bigger vocabulary.
4.
by hazard 1mo ago
Put an airtag into literally every bag you check. Helps you figure out not just whether they made it on the flight, but also if they ended up at some baggage claim on the other end of the airport instead of the one they were supposed to be
5.
by hazard 1mo ago
your LLM's temperature is set too high
6.
by hazard 1mo ago
I recommend reading some foundational materials on federated learning and distributed inference. The fundamental issues are (1) compute (2) latency and (3) security. On the compute side, you can't load most individual experts into even
7.
by hazard 2mo ago
Didn't sfcompute launch with a vision of something like this and then cancel it because CFTC decided it was running an unregulated futures exchange? How are you getting around regulatory issues like this?
8.
by hazard 2mo ago
>if they’re going to also impose a slight delay between when posts show up in the API vs when they’re posted on the website. help Yes, they say it will be a few milliseconds. This service is squarely targeted at algo trading firms.
9.
by hazard 3mo ago
After being in the workforce for decades, this whole issue is just so incomprehensible to me. I went an ungrad school that was top-5 in engineering. But my experience - and in the experience of other people I've talked to - formal unde
10.
by hazard 3mo ago
This looks great, but I'd really like to see associated exercises (and solutions) to make it useful for self-study
11.
by hazard 4mo ago
> Some engineers, given a fixed token budget, generate exponentially more (and better) output. Other engineers waste their tokens. The variance is enormous, and unlike most performance variance, it is now directly measurable. HR has neve
12.
by hazard 4mo ago
The 100-point buckets are fine-tuned on blitz games from users at that Lichess rating. Lichess ratings tend to be a bit high compared to FIDE/USDF/chess.com ratings. There's a good post at https://chessgoals.com&#x
13.
Show HN: I trained a chess engine to play like humans
14 points by hazard 4mo ago|3 comments
14.
by hazard 5mo ago
> The reason that the rich were so rich, Vimes reasoned, was because they managed to spend less money. The premise is just false. The parable might be true when comparing, say, lower class vs lower-middle-class, or lower-middle-class to
15.
by hazard 5mo ago
I'm always a bit confused by "written in X" as a feature. Rust devs are the worst about this, but not the only offenders. The language is not a feature! I'm a C++ programmer, and even I don't care what language the
16.
by hazard 8mo ago
And of course equity futures immediately dropped on the news
17.
by hazard 8mo ago
https://www.1e4.ai/ A transformer-based (but not LLM) chess model that plays like a human. The site right now is very rudimentary - no saving games, reviewing games, etc., just playing. It uses three models: * A move mod
18.
by hazard 9mo ago
tldr appears to be that if you work to fatigue it doesn't matter if you fatigue out with high weights vs low weights
19.
by hazard 1y ago
"We believe that continuous exposure to transportation support systems like cars may lead to the natural human tendency to over-rely on their engines, leading to travelers becoming less motivated, less focused, and less responsible whe
20.
by hazard 2y ago
Kbit | Data Visualization Engineer + 2 other roles | Full time | 100% Remote, Hybrid available in certain locations if you prefer Strong base salary plus quarterly cash bonuses depending on firm performance. We're a digital asset hedge
21.
by hazard 3y ago
A Twitter filter to take back control of your social media feed from recommendation engines. Put in natural language instructions like "Only show tweets about machine learning, artificial intelligence, and large language models. Hide e
22.
Show HN: AI-Powered Twitter Filter
1 points by hazard 3y ago|0 comments
23.
by hazard 3y ago
The best course by NVidia looks like "Fundamentals of Accelerated Computing with CUDA C/C++" which I think used to be publicly available, but is now offered "By invitation only"
24.
Ask HN: Best way to learn GPU programming?
55 points by hazard 3y ago|10 comments
25.
Ask HN: Do you homeschool?
59 points by hazard 3y ago|56 comments
26.
by hazard 3y ago
> Do you really think it's fine blowing 400 watts because you can't be arsed to think or do not have the creative intelligence to get over the blank page syndrome and have to lean on a crutch? Yes, I think it is absolutely 100%
27.
by hazard 4y ago
This seems pretty interesting, do you have links that provide more details on this? Is it all functionality that's available by default on pihole or did you use some mods/custom blocklists/etc?
28.
Ask HN: Parents: Best screen-time limiting software?
27 points by hazard 4y ago|24 comments
29.
by hazard 4y ago
Want to bet? Literally, I mean. I'll bet you $100 that they will not hire all Rocky contributors in under 3 years and turn it into a rolling distribution.
30.
by hazard 4y ago
If only the authors had considered this... > Noncausal mechanisms, including socioeconomic status and reverse causation bias, have been proposed as contributors to the KIHD findings (Kivimaki et al., 2015). Although differences in socioe

More ›