two_in_one

8 ms·

31.
by two_in_one 3y ago
It doesn't matter how dangerous. He had years before and after to die this way. The fact that it happened right after has very low probability to be by chance. Which means it very likely wasn't.
32.
by two_in_one 3y ago
> Companies following the lead of other companies (CEOs not thinking for themselves, just looking at what others are doing) Layoff usually negatively affects company's image. It makes sense to do it when everybody's attention i
33.
by two_in_one 3y ago
> The bar to create animation is going to be so incredibly low in just a few years time, that the skills to have will be writing, imagination, O yes, and those 'creative writers' wouldn't be humans. At least at low level i
34.
by two_in_one 3y ago
5880 with 48GB has been announced. but I suspect it will be not in consumers range, more like $3k++.
35.
by two_in_one 3y ago
Theoretically you can use as many as GPUs you want in parallel. LLMs are easy to split and run in model parallel configuration (for big models which don't fit on one card). or data parallel for performance, when the same model runs dif
36.
by two_in_one 3y ago
that means that openai cannot claim copyright on images produced by dalle generator. no can other online services and offline software owners/producers. this is enough to use them without copyright violation for example for other ai mo
37.
by two_in_one 3y ago
It's a gray area. But if someone generates AI images at scale without human in the loop it's likely not copyrightable.
38.
by two_in_one 3y ago
> AI-generated images are not currently eligible for copyright. There is another bright side of it. This images can be used for AI training without copyright violation. Does this apply to texts as well?
39.
by two_in_one 3y ago
this is more recent and optimistic https://www.nextplatform.com/2023/11/27/groq-says-it-can-dep... But I don't see any selling numbers or independent performance conformations.
40.
by two_in_one 3y ago
Because it's not just a chip, it's everything else too. There were 2 startups in Boston, may be 500 meters apart, doing optical computing. 4 years and there is still no product. "everything else" is a lot. It's hard
41.
by two_in_one 3y ago
If they drop the price 2x that will make it very attractive.
42.
by two_in_one 3y ago
nearest object can be dynamic thing
43.
by two_in_one 3y ago
It's hard to call comprehensive. Transformers - one page. A picture would be nice. No "prompt engineering", no "double deep". In fact words "prompt" and "double" aren't used at all. "Re
44.
by two_in_one 3y ago
Probably can be solved by including a positive personality overseeing the actions, but not getting the direct instructions from the client. Just generic initial prompt like "you are an inspector investigating possible law breaking&quot
45.
by two_in_one 3y ago
They have the whole team working on 'alignment' ;) Just wandering if in ten years we'll have to tip robots in banks, hospitals.
46.
by two_in_one 3y ago
Your system, including subsystems, must have a filter for prohibited words, like n-word, f-word, etc. It's very easy to add another one. That could be another subsystem which very few people have access to. I suspect there are differen
47.
by two_in_one 3y ago
> people now refrain from using the word "Palestine" to avoid metas algirithms It wasn't algorithm, or glitch, or bug. It was a decision made by humans, including you know who. It's a form of crowd control. That'
48.
by two_in_one 3y ago
There is nothing truly new in this world. It goes in spiral and that will be the next loop on upper level. Think of it, you can't train in the latest news in LLM. Once you finally understand it you need a solution. ChatGPT is using Bin
49.
by two_in_one 3y ago
Once data and logic get separated this will become less important. I mean there is no need to train data into the model, like it's done today. Ideally it should be a small generic model, or several of them working together, with quick
50.
by two_in_one 3y ago
That was a joke: "Verizon made even Comcast look good". Not sure if it's still around.
51.
by two_in_one 3y ago
people don't realize it. but small high capacity memory changes a lot. now even small robot cam remember it's whole 'life' in low resolution. and this makes possible to create, say, powerful assistants. when you can ask
52.
by two_in_one 3y ago
1TB micro sd is < $100. this is more then one can listen to and still remember something from each track. And you can easily fit 10 of them in small pocket..
53.
by two_in_one 3y ago
isn't that how translation works in tranformers? encoder produces embedding then decoder generates the sentence. do it again with reverse translation and you get your text back in original language. combining two sentences can be done
54.
by two_in_one 3y ago
I think I've seen somewhere years ago an article claiming size is needed for training. Which means probably that after training model can be optimized to minimize the size. Purging? However, smaller model cannot be trained that well. F
55.
by two_in_one 3y ago
it may be not that big: "the lowest total in four years". could be just back to normal.
56.
by two_in_one 3y ago
welcome to the real world, it's cruel and unjust, but with some bright spots. those agreements are as good as they are enforceable. it should be hard to enforce when the source is a derivative from stolen data.
57.
by two_in_one 3y ago
cool, tried it with chatgpt-4. it's more verbose, and (!) you can ask for illustrations along the way.
58.
by two_in_one 3y ago
I meant different thing. ISS is big enough to cast a small shadow on Earth. It moves fast and covers a star for some fraction of second (roughly size/orbital velocity). Still every day several times.
59.
by two_in_one 3y ago
Am I right, with ISS on the orbit it should happen several times every day. Not for 7 seconds, but..
60.
by two_in_one 3y ago
this technology is already exploited in the wild for scams. probably for years. it's too late to worry or try stop. the question is what can we do about it. and the first is to make people aware of it.

More ›