junipertea

I tried many of them. All of them are innovative in some ways and scratch certain itches well, but none of them are perfect (i.e. good for my set of use cases with no compromises) yet, so I am still looking and regularly trying new ones. (As long as it is easy to mass import data from TiddlyWiki/jsons)

I found training multiple models on same GPU hit other bottlenecks (mainly memory capacity/bandwidth) fast. I tend to train one model per GPU and just scale the number of computers. Also, if nothing else, we tend to push the models to fit the GPU memory.

From my experience with Teslas, there is a huge space for improved user experience. Teslas are super fun to drive and and way ahead of the competition. BMW now allows using iphones to open the car. There can be deeper carplay integration and to be frank it can just be a really expensive way to flex on your friends, like all apple devices.

It's not like you can use safari on Windows.

Feeling too full to eat while not maxing out on your calories is a desirable goal for many people.

It's a good thing as long as there is a viable alternative to it.

Discrete nvidia GPU with CUDA. While there is Tensorflow compiled for radeon GPUs, majority of libraries are based on CUDA

I'm not sure about the combination of covid + vitamin D. But you can also overdose on vitamin D since it's fat soluble.

https://www.healthline.com/nutrition/vitamin-d-side-effects#...

Dashboard cameras are a thing already.

Why?

I wouldn't expect "this phone can run for 10 hours and train a model", even though it clearly is also possible.

in that case the government would just profit from the forced labor via additional tax income

You are, if you manage to blame the other party for sabotaging you.

Plenty of people like apple stores.

Please don't give them any ideas.

Based on usage of DDG, bing and wolfram alpha, I would think Apple can improve a great deal. I'm not saying it can necessarily achieve that, but I do think there is a lot of space to improve (indexing, ranking, UX)

It can't do long division because it literally can't read the number inputs, due to the way text is encoded (BPE). That it manages to learn any sort of arithmetic despite that is pretty impressive to me.

I think taxi apps with the exact same user experience will just crop up. You will just have to pay more.

I don’t know if they will release the models, but are you sure you can train a 170 billion parameter model? Last I heard it’s around 500GB, which would require serious infrastructure.

You are always free not to install the latest update and have the android experience of using an old version OS on a 2 year old phone. On a serious note, I don't think they reduce your battery on purpose, the class action lawsuits Apple lost handled that. It might be just optimising runtime and letting your phone handle new apps? So CPU runs faster and squeezes out more performance?

I guess research community would not use commercial products? The number of classes and datapoints is quite extreme for a “common” use case, usually you would want to train it by yourself.

Can you name those fundamental technical problems? From my understanding, the problems are adoption and states being slow in implementing the apps. I don’t have any exact numbers, but I would say even with some false positives it will be much more directed unlike testing everyone daily.

It was definitely terrible for many other countries.

In Table 2, we observe that TransCoder significantly outperforms both baselines in terms of computational accuracy, with 74.8% and 68.7% in the C++ → Java and Java → Python directions, compared to 61% and 38.3%

It beats existing commercial products that try to do the same, so to me it seems it has some value right now.

Maybe they meant chaining if statements

I thought all of that is pretty accurate summary without much bias. Which statement is misleading?

No one is happy in this comment section about there being downsides to a treatment. The majority are dissatisfied that treatment is being used because of hearsay and not scientifically tested, which puts huge doubts on our healthcare system. Furthermore, no one mentioned Trump.

Google made a report using geospatial data and tracking of GPS and even anonymized it. https://www.google.com/covid19/mobility/

There is extensive research in using data and statistical modeling ("AI") to help with the coronavirus efforts and there has been a ton of work invested. [1] You have to note covid-19 has been affecting western countries since february and it's april now. Not that much you can manage in 2 months. We probably could have and should have done more when we saw the situation in november, but that is a different topic.

[1] https://arxiv.org/pdf/2003.11336.pdf

I guess the simpler proof that they are the same thing would be: Do they work the same?

Apple was dead last in the last self driving report in California. https://www.bloomberg.com/news/articles/2019-02-13/apple-s-a...