SergeAx

40 ms·

601.
by SergeAx 4y ago
> Passing a driver's test was already possible in 2015 or so I think we can talk about 2005. Check out the DARPA Grand Challenge, it was way harder: https://en.wikipedia.org/wiki/DARPA_Grand_Challenge_(2005)
602.
by SergeAx 4y ago
> you will not kill someone if you ride after a night drinking; It actually depends, but chances are better. > do not have to worry about car maintenance, parking tickets or theft; Isn't bicycle theft wider than car theft in Euro
603.
by SergeAx 4y ago
Aside question: what if I write an aptX decoder from scratch just looking at the source text of encoder? Will it be an independent or derivative work?
604.
by SergeAx 4y ago
aptX HD is one of the greatest things happened with Android ecosystem. One day I just bought a cheap BT-receiver with aptX HD support and optical output and connected it to my old-fashioned music system (receiver + amp + floorstanding speak
605.
by SergeAx 4y ago
I am a bit lost here, and I am by no means an expert in US corporate law. Is there a law that company's CEO should always record all their conversations? What about phone calls? In-person meetings?
606.
by SergeAx 4y ago
This is literally the hacker news. I wish HN have more content like this instead endless topics swirling around money.
607.
by SergeAx 4y ago
My opinion is that you one hundred percent should not learn programming from YouTube videos. It is tremendously ineffective.
608.
by SergeAx 4y ago
I am trying to avoid videos and podcasts too. But it is an uphill battle. Text, while easier to produce, would give the platform only 10 minutes of engagement at best instead of 30, and won't generate comparable number of subscriptions
609.
by SergeAx 4y ago
I will try another analogy. What if we have a parrot with exceptional memory, which can not only repeat things it heard some time ago, but to continue words it hearing now. I come to the cage and say: "Cogito?" and parrot continue
610.
by SergeAx 4y ago
No single effort will overthrow Putin, but a combination of many efforts will.
611.
by SergeAx 4y ago
What is the point of these "exploits"? What interesting is in making language model (e.g. Markov chain on steroids) to produce arbitrary text? What is proved by it? I am genuinely interested.
612.
by SergeAx 4y ago
> Just write > I just finished listening to a podcast Ironic.
613.
by SergeAx 4y ago
Finding most probable words from the internet for a given prompt have nothing to do with reasoning.
614.
by SergeAx 4y ago
Just spin up three of these babies.
615.
by SergeAx 4y ago
We actually made a wide swing from reasoning to intelligence. So I propose to ditch ants and get back on track.
616.
Nasdaq tells Yandex, other Russian firms of plan to delist stocks(reuters.com)
13 points by SergeAx 4y ago|0 comments
617.
by SergeAx 4y ago
But... ant hive does not posess any intelligence, right? Despite colonies of ants are able to perform quite complex tasks.
618.
by SergeAx 4y ago
Our organization currently caching all and every external dependencies we are using: Go, Python, npm and .NET packages, Docker images, Linux deb packages, so everything is contained inside our perimeter. We did that after one day our self-h
619.
by SergeAx 4y ago
I am not talking about the concrete realization, I am talking about the principle. You are right, LLMs are just Markov's chains on steroids, thus they cannot "reason". For reasoning you need a knowledge model, a corpus of fac
620.
by SergeAx 4y ago
Wow. 16 threads, 64 GB RAM and 2 TB NVME SSD will make a fantastic Postgres/Mongo node for a fraction of managed solutions cost.
621.
by SergeAx 4y ago
You take the prompt and caclulate what next word after the prompt is most probable. Like T9 with letters, but bigger.
622.
by SergeAx 4y ago
Please. Language model cannot "reason", it can just show next most probable word based on text corpus downloaded from the internet.
623.
by SergeAx 4y ago
Onboarding email should be just a template. Or, even better, a small template with a link to bigger template. As a manager, you should demonstrate your empathy in a welcome 1:1 meeting.
624.
by SergeAx 4y ago
Interestingly, some local "crypto" conference just rebranded as "future tech". Seems like "crypto" became a red flag deterring customers.
625.
by SergeAx 4y ago
The same way as flat-earthers. Also, showing off. Also, self-suggestion. Also, monetary incentive for vendors to keep their buyers spending money on snake oil.
626.
by SergeAx 4y ago
Of course we did. This is why otherwise ordinary "Top Gun 2" was so massively successfull: lots of live action and practical effects. I think we will lose it at all eventually. Today's audience is not making a big difference
627.
by SergeAx 4y ago
It is still quite obvious, except maybe when ML model used to enhance an original artwork.
628.
by SergeAx 4y ago
That makes sense. I thought that "technology cannot be stopped" is quite obvious and didn't even think about someone trying to do the opposite.
629.
by SergeAx 4y ago
Yes, this is true and it is hard to ignore. However, I was talking about practical applications, not just general interest. There were several hype-generating stories about diffusion model winning art contest and using LLM to write grading
630.
by SergeAx 4y ago
Yes, I meant about the same: I percept uninteresting garbage content as malicious too.

More ›