ma2rten

I don't understand why buying music on iTunes was so popular in the US for a while, when in Europe everyone was already listening to Spotify.

Your query worked for me on first try: https://imgur.com/a/iTKwZ3v

The text-to-image thing is cool as well, as long as you can a) figure out how it works b) only enter landscape terms.

This comment explains how to use it: https://news.ycombinator.com/item?id=29338213

Thanks! This UI is something else...

Actually when I wanted to change teams I cold emailed managers of other teams at my company. I found the same thing: a short 3 line emails have a great response rate.

If you send a message that is too long, you just increase the barrier of responding.

Google has by far the best team in computer vision.

I googled it and it said $60-$100 per hour, which is not that great compared to Facebook pay or other contracting gigs.

What does this actually mean? Why does it need to be sent to you (presumably referring to the client) when it's generated client side.

How does rr work with multi-threading? I would imagine threads can mess with the order that syscalls are executed.

The OP writes: "International treaties explicitly and unambiguously support PGLAF's legal guidance as described above: that the copyright status in one country is not impacted or enforceable or otherwise relevant in other countries. Plaintiff managed to find a German Court, and some precedents from Germany (and, after the lawsuit was filed, from the EU), which were willing to flaunt international treaties by developing a theory that PGLAF is under jurisdiction of the German Court system."

This was already known since at least January.

Actually, in a closed bucket hash table why would you use a linked list and not some kind of tree structure in the leaves?

It could just be "reveal answers".

This might be somewhat surprising to some people because a lot of the traffic to news sites comes from facebook.

According to Zillow the median house price in Austin is $596k. In Fremont, CA it's $1,4m.

I don't think it's a big mystery. Bandwidth and ad revenue scale together. Sometimes the image will be embedded, hot linked or the request is otherwise not monetizable, but you can assume that those are a fixed fraction. Every image clicked on otherwise will generate some ad revenue which is multiples of the bandwidth cost of serving it.

The paper has some images: https://www.sciencedirect.com/science/article/pii/S147403462...

You are asking the wrong question. They expect it because it's the norm in Germany. Human is psychology is such that a resume with a photo is just more memorable, so it became the norm because applicants do it. Bias is less of a concern in a country that is historically more homogeneous.

This comment reeks of Americentrism. Germans are perfectly capable of coming up with their own insanities.

Are you sure those are the same people?

It's not that easy to get 50 million people to download your app and if you pulled that off there are more profitable things you could do.

Yes, but these attacks are actually easy to detect if they all come from the same IP. Sending packages from a lot of different IPs actually is expensive.

I could imagine that the wording regarding corporate control could have erroneously triggered a filter because that phrase appears for example in anti-wax and conspiracy theory posts.

It would be interesting to do some experiments here: post the same text to see if it gets removed again, and then repost it and remove sentences to see which one triggered the filter.

Those two are tied together because many of those few-shot capabilities only emerge at scale. If OpenAI had trained a large model and not analyzed it rigorously it would have very little scientific value. But it would have been impossible to have the scientific value without the engineering effort.

As someone in the deep learning community I disagree with your assessment that GPT-3 is a not scientific breakthrough. The GPT-3 paper won a best paper award at the most prestigious machine learning conferences after all. GPT-3 didn't make any modeling advances, but it introduced a completely new paradigm with few-shot learning.

I misspoke. I meant I can copy & paste on the web. On iOS I can copy paste from Google maps, but not Youtube. Maybe there just isn't a good enough use case for it, that's why it's not supported everywhere?

I just tried it, I seem to be able to copy text just fine from both apps.

I think you are reading something into it that is not necessarily there. The slides give examples of things that are appropriate to say and are not appropriate to say. It doesn't necessarily mean that one thing has to changed into the other.

Did people who clicked "yes" eventually report it to IT later?