Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
RC_ITR
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
45 ms
·
601.
▲
by
RC_ITR
4y ago
Dallas does the same thing with 75 and it's so wild because it seems like you're in an intensely urban area for miles, but really it's a wall of 10-15 story buildings surrounded by single family homes.
602.
▲
by
RC_ITR
4y ago
I think the implication is that you do noise abatement on the AC units (i.e. earthen berms, better site planning, etc.).
603.
▲
by
RC_ITR
4y ago
> Humans had thousands of years experience having fun w/o wasting non-renewable resources. Genuinely curious, are you speaking of pre-Bronze Age civilizations?
604.
▲
by
RC_ITR
4y ago
The current known helium reserves are estimated to last (at current consumption rates) for 100-200 more years. Fwiw, all this non-renewable talk is only true in a world without space exploration or fusion (two things that will hopefully be
605.
▲
by
RC_ITR
4y ago
First some background: About 85% of payment fees are taken by the bank that issues the credit cards (this portion is called interchange). The remaining 15% is split between the card network and the merchant acquirer (in this case Stripe). S
606.
▲
by
RC_ITR
4y ago
It’s just very theoretical and therefore hard to care about. It boils down to the question “do we get richer by making more or demanding more” and there are plenty of paradoxes depending on which side you choose (If it’s because we make mor
607.
▲
by
RC_ITR
4y ago
Again, reading these papers, Grokking can happen in very limited circumstances for non-algorithmic datasets. > They verify this observation in a student teacher setup, and show that it can arise in non-algorithmic datasets if initializ
608.
▲
by
RC_ITR
4y ago
>Not so. Actually, (for example) the phenomenon of "grokking" is when with enough training a NN eventually experiences a phase-change from memorising data to learning the general rules underlying it. Reading the paper, what the
609.
▲
by
RC_ITR
4y ago
It's important to remember the first principle of what GPT does. It looks at the pattern of a bunch of unique tokens in a dataset (in this case words online) and riffs on those patterns to make outputs. It will never learn math this wa
610.
▲
by
RC_ITR
4y ago
A lot of people don't really get that our brains are a bunch of specialized subcomponents that work in concert (Your pre-frontal cortex just cannot beat your heart, not matter how optimized it gets). This is unsurprising, as our bra
611.
▲
by
RC_ITR
4y ago
>Will receive Turing Award This is the weird thing - hopefully not! Hopefully there's even better NN models coming out every 5-10 years and we look back on transformers as 'just a phase' sort of like how we look back at R
612.
▲
by
RC_ITR
4y ago
>side from both incidents happening in countries not particularly well known for long-term follow-up on the consequences of failures by the state Journalists followed the other people. We know their names. We have evidence of them dyi
613.
▲
by
RC_ITR
4y ago
>Just have a "committed to the cause" person just carry something like this around as they tour the city. Please don't fear-monger; the impact of radiation on the human body is cumulative and the inverse square law makes
614.
▲
by
RC_ITR
4y ago
>visiting my friend in Chinatown is either a 10-15 minute drive, or a 40-50 minute bus ride + walk You're describing a 4 mile journey. A comparable journey in NYC is SoHo to Central Park. At best, that's a 30 minute commute in
615.
▲
by
RC_ITR
4y ago
Because some people don’t make their life choices by “out of sight out of mind”.
616.
▲
by
RC_ITR
4y ago
I think a lot of people who live in SF realize the solution isn’t hiding/shunning homeless people, so seeing it affects them less than people who live places where they think hiding/shunning homelessness is a solution. Just becaus
617.
▲
by
RC_ITR
4y ago
Stocks are priced on future events
618.
▲
by
RC_ITR
4y ago
Say what you will about Dang and their moderation style, but whatever is causing this thread to stay up for 7+ hours (and climb the front page) is very intriguing. I wonder what the logic is for this 'flamewar bait' being unmodera
619.
▲
by
RC_ITR
4y ago
The reason it’s not implemented yet is because none of these products are being monetized yet, so there’s no incentive. Google can legally return a copyrighted image in Image search (it’s not selling the image, it’s selling ads against sear
620.
▲
by
RC_ITR
4y ago
I mean feel free to have a different opinion than me, but I don’t really understand the point you’re making? Twitch is the US’s dominant streaming platform and it predates (as Justin.tv) all the China players. It’s had less of an impact bec
621.
▲
by
RC_ITR
4y ago
>a great proven model that the lagging west took ages to adopt. I’d be careful with this type of thinking. Sometimes things can just be different. Take KOLs/live-streaming e-commerce for example. Works in China for a lot of reasons
622.
▲
by
RC_ITR
4y ago
Just to add context for anyone who doesn't know the history - Bytedance's first product was Toutiao, which means 'headlines' [0]. Douyin/Tik Tok grew out of that. The concept works well one a relatively homogeneous
623.
▲
by
RC_ITR
4y ago
It's interesting that you bring up Google, a company founded to 'organize the world's information and make it universally accessible and useful.' Isn't that exactly what Artifact is doing (but instead of rules-based
624.
▲
by
RC_ITR
4y ago
>Well the AI has to infer what math is with many examples. You give it enough examples it will be able to calculate equations it's never seen before. This is categorically not true and an important distinction to understand. All Op
625.
▲
by
RC_ITR
4y ago
If you look through their releases, the growth appears to be coming from doing a better job of getting a free-trial included in almost every new car and like half of used cars (I assume this is done through dealer networks). Given that, if
626.
▲
by
RC_ITR
4y ago
There’s various kinds of learning (and to be clear transformer-based LLMs do sort of “learn”). ChatGPT is a really finely tuned probability model (the tuning came during the training and that’s what I would equate to learning), but it lacks
627.
▲
by
RC_ITR
4y ago
You’re not supposed to ask people if they read the article on here, because usually the answer is yes (and you should expect it to be). I did read it. The point of neural nets is that they are really good at probabilisticly linking tokens
628.
▲
by
RC_ITR
4y ago
Understanding language and understanding symbols are two very different things. I don’t disagree that we may be able to build symbolic neural nets 15 years from now, but they will look almost nothing like LLMs.
629.
▲
by
RC_ITR
4y ago
> So, ChatGPT seems to understand how filesystems work, how files are stored and can be retrieved later. It understands that linux machines are stateful, and correctly retrieves this information and displays it. No. ChatGPT has seen mill
630.
▲
by
RC_ITR
4y ago
>And replaced it with natural gas... Yeah. >Coal is nowhere near dead. Europe is increasing production right now. People have greatly exaggerated the death of fossil fuels. We are decades away from moving away from fossil fuels. Again
More ›