jononor

46 ms·

931.
by jononor 4y ago
My point was that it is not a two-day hack job, as claimed above. It is in no way a unique challenge, but neither is it trivial. Probably take multiple man-months for a small team that has built similar systems before.
932.
by jononor 4y ago
Twitter has over 200 million Daily Active Users across the globe, with an average of 6000 tweets per second. Peak tweets per second is said to be over 100k. Each user gets a personalized feed which is regularly updated (I could not find num
933.
by jononor 4y ago
This is behind a paywall?
934.
by jononor 4y ago
The combination can be very useful sometimes, for example for transfer learning for working with low resource datasets/problems. Use a deep neural network to go from high dimensionality data to a compact fixed length vector. Basically
935.
by jononor 4y ago
WINE is implemented using Unix libraries underneath. And since it essentially supports and entire modern OS, there is quite a bit of depdencies. The wine package available in Arch Linux for example depends on 72 lib32-* packages. The majori
936.
by jononor 4y ago
For those curious about AI safety I heavily recommend Robert Miles channel on YouTube. He has lot of good content describing out the challenges we face when trying to create safe AI. Here is one potential starting point: https://
937.
by jononor 4y ago
Invest in others, and your relationship to the. Become a better human being as a consequence. Don't neglect yourself though. But on your own mask before helping others.
938.
by jononor 4y ago
A difference between ChatGPT and Google in this scenario is that Google cannot really synthesise or compose facts/knowledge together. These giga language models are decent at this, if anything they do it a bit too well/much, putti
939.
by jononor 4y ago
What you describe is it what the models can do today, but what they might be able to do with more work. It is possible conceptually now, but getting it to work on a product or MVP level - still risk there.
940.
by jononor 4y ago
Sure, it is aspirational - but at least doable. Others in sibling threads are claiming that CNC machining will never be safe, affordable or approachable due to fundamental reasons. Which I do not agree with.
941.
by jononor 4y ago
What has happened is that services offering medium scales production has become quite a lot more accessible. One can get a quote by uploading to online form, no need to enter a long term supplier agreement, few restrictions on minimum order
942.
by jononor 4y ago
These are all things that I believe can be improved by better software and machines that are designed to teach people how to do it. Current software is still mostly designed for the experts, assuming a whole lot of knowledge exists from bef
943.
by jononor 4y ago
Whats your goal/motivation? How to proceed would depend a lot on that. Tinkering as a hobby? Making something physical? Understand how things are made? Prototyping an idea for commercialisation? Career change?
944.
by jononor 4y ago
Why try to keep things in your head? Make notes on a piece of paper, or in a text file as you go. I find that this helps tremendously with non-trivial tasks.
945.
by jononor 4y ago
The Xtensa based ESP32 chips have upstream support in GCC, do they not?
946.
by jononor 4y ago
To be fair, this kind of thing even happens in JavaScript. Some use functional programming with immutable datastructures, higher order functions, super generic code, using some niche FP library with its own idiosyncrasies - when simple and
947.
by jononor 4y ago
Small crates is not a benefit wrt. The wrong 3 lines of code in crate is all it takes to be owned, regardless of whether the crate is 100 lines or 100k lines. Large number of crates and developers is a negative, because one needs to trust o
948.
by jononor 4y ago
They might not judge you more in a large city, or be any more interested in you per see - they might just be a little bit more eager to check if you might be dangerous. As in places with many encounters with many people, the risk is higher.
949.
by jononor 4y ago
One does not need every single copyrighted object in the world, just a large, diverse and representative set. One would have to skip the cases where rights cannot be secured. This is possible, see for example Spotify.
950.
by jononor 4y ago
One could procure the necessary rights from the copyright owners / artists for all the training data. Sure, it would be quite costly and take a long time - but it is possible.
951.
by jononor 4y ago
RapidMiner is one. In the physics and electronics/physics domain there is also Simulink. It can often hard to integrate these into production code though, as they are designed primarily as interactive desktop applications, with per sea
952.
by jononor 4y ago
The base serialization is stable and compatible across Python versions. However pickle files can serialize entire Python objects, including from custom and third party code. Compatibility of serialization then depends on the versions of the
953.
by jononor 4y ago
Text2img will be a next level in "stickers", like the photos/gifs one can attach today in Messenger etc. And for getting stock content, like in Canva or Adobe suite. Possibly with some curation in between initially. Just impr
954.
by jononor 4y ago
Realize that a perfect solution is not feasible (it does not exist). Not even an optimal solution for your particular problem is feasible (too many unknowns about both the problem and potential solutions, impossible to uncover and analyze t
955.
by jononor 4y ago
One could use it to test (embedded) code targetting RISC-V, without needing to use hardware. Unit tests is the easiest, since they preferably do no I/O. But also integration test by providing an I/O implementation (using a Hardwar
956.
by jononor 4y ago
I also had that idea. I came at it from the side of thinking about using computer vision for (semi) automated CAD file production, and gaskets was one of the best usecases I could come up with. One the fulfilment side, gaskets can be laser
957.
by jononor 4y ago
O(n) for every new samples is O(n*2) when seen over time/samples?
958.
by jononor 4y ago
Haha no, the other Flowhub :p http://flowhub.io
959.
by jononor 4y ago
That is something the apps and App stores can solve in collaboration. And if they are required to in order to make money, they very likely will.
960.
by jononor 4y ago
Very glad to see this get some focus! I wanted the same back when I worked on Flowhub.

More ›