kolinko

35 ms·

721.
by kolinko 3y ago
A high chance that lead can be replaced with other metal, from what I heard. Also, it’s not extremely toxic - romans ate from lead dishes and didn’t even notice the toxicity. You don’t want to consume it, and you don’t want to breathe it, b
722.
by kolinko 3y ago
As for crypto, tgere is nothing to believe in - it’s doing well after 14 years.
723.
by kolinko 3y ago
One of the purposes of ISS, and other space programs (Apollo-Soyuz for example) was showcasing how nations can work together, and bringing hope for peace during the cold war. May sound naive, but that's the reason why nobody should be
724.
by kolinko 3y ago
With cheap fusion power we can convert CO2 back into kerosene/fuel, and use it as energy storage. It would be lovely ironic if in the future we all drove ICE cars, but just fuelled by clean and carbon-neutral gasoline.
725.
by kolinko 3y ago
It is an option, but there are two downsides: - such a current generates a huge electromagnetic field. So it won't work for a car battery, but may work for grid storage. - price - there is a limit to how much current you can store, and
726.
by kolinko 3y ago
As far as I understand, speed of light might as well be changing, but as long as it's tied to the speed of other forces, we would have no way of noticing. By the way, an amazing story of Ted Chiang, on the subject - Exhalation - about
727.
by kolinko 3y ago
GDP is not productivity. If I manage to produce/sell a hundred gizmos for $1 each, displacing my competition who were producing/selling 50 gizmos for a price of $5 each, I just halvened gdp produced even though both me and people
728.
Show HN: Answering GDPR questions using GPT(askgdpr.com)
3 points by kolinko 3y ago|3 comments
729.
by kolinko 3y ago
It looks like it's calling OpenAI's gpt-4 with function apis.
730.
by kolinko 3y ago
It's been already done with LLangchain / Plugins / Function calling by LLMs.
731.
by kolinko 3y ago
Example task that you cannot automate with a basic script - for each e-mail you receive, convert it into json of a given format for a given e-mail type (e.g. for delivery notifications, newsletters, meeting requests and so on). Also, tag it
732.
by kolinko 3y ago
With LLAMA at least, not just past tokens, but also past states for each layer and each token. So LLama has information not just on the word "Berlin", but also on why it outputted "Berlin".
733.
by kolinko 3y ago
I noticed that many people who propose this are not aware of a plethora of health issues that people have that make eating plant-based foods extremely difficult. For many people eating plants only is extremely difficult - e.g. for me person
734.
by kolinko 3y ago
x 2 or x y? x 2 is one neuron and probably a dosen iterations max. x y may be a bit more complicated - like the stackoverflow explains 13B parameters - yes, it’s essentially 13B multiplications/additions (the most basic neuron is y=act
735.
by kolinko 3y ago
Logitech has even more experience than Microsoft in doing controllers
736.
by kolinko 3y ago
„But like the car, people convince themselves they don’t want it” If you live in Europe, you often seriously don’t want or need a car. I didn’t have a car for 8 years - even though I could afford it easilt, and it took pandemic and a shutdo
737.
by kolinko 3y ago
y = x * 2 is what you get on the most basic neural network. You show it a bunch of points for that function and it will train to predict the rest or, in other words - nns and llms do exactly that - predict/invent functions that meet th
738.
by kolinko 3y ago
true, of course, can't edit reply now :)
739.
by kolinko 3y ago
There is not a word on how the AI system was designed in the study, so no idea how well it worked and if it was frustrating or easy to use. I can totally see someone using chatbots from 2021 and 2022 daily getting depressed :D
740.
by kolinko 3y ago
The study was performed in Spring 2021 and 2022. The kind of AI used in the study for testing is not specified, but back then it was very likely a poorly built chatbot - nothing like ChatGPT which was released at the end of 2023.
741.
by kolinko 3y ago
In this case you can use exactly the same method as in the paper to verify. Hallucinations are easy to verify in many cases.
742.
by kolinko 3y ago
I'm not sure binares were too big of an issue - most server simply didn't carry groups that allowed binaries. The spam issue was the real deal breaker imho - you either had a fully moderated group, or you had a fully unmoderated o
743.
by kolinko 3y ago
They tested six different strategies and only two-three of the strategies returned profit on this year. Even the ones that turned out profitable were only profitable within the first half of the year and provided no return during the second
744.
by kolinko 3y ago
With vegan you can’t estimate it 100% from menu alone - because the sauce and other minor ingredients can be animal based. If you want to do it, using “plant based” is probably better than “vegan”, and it’s always good to make sure your use
745.
by kolinko 3y ago
I had it answer properly when using the regular "thought->observation" pattern. https://chat.openai.com/share/06cd9838-a46f-4477-a8c8-95f324...
746.
by kolinko 3y ago
In car mechanics you may not see it, but some of your clients may have diagnosed - and fixed - car issues by themselves using ChatGPT. That was my case once or twice already - didn’t get to a mechanic whereas before I would need to at least
747.
by kolinko 3y ago
As a professional with >20 years of experience, I used ChatGPT for the same purpose recently, just with way more difficult programming concepts. I wanted to learn how transformers and attention mechanisms work in details. After reading a
748.
by kolinko 3y ago
LLMs can navigate sources and generate both documentation and potential answers directly from sources.
749.
by kolinko 3y ago
Posting again, this time with a direct link :) Feedback/questions welcome!
750.
Hacker Herald – a tabloid auto-generated from Hacker News posts(kolinko.github.io)
1 points by kolinko 3y ago|1 comments

More ›