tcpekin

Fig. 9 of this paper is really interesting if you zoom in. It looks like if the model was not trained on the appropriate class label, it just goes completely off the rails. As previous commenters have noted, I would be very hesitant to use this for anything analytical, or where you are looking for something unexpected. For faces though, this is amazing.

This happens in numpy as well, and was confusing the first time I saw it:

    b = np.ones(1)
    c = np.ones(1)
    c = c/(2**73)
    b += c

    >>>  ufunc 'add' output (typecode 'O') could not be coerced to provided output parameter (typecode 'd') according to the casting rule ''same_kind''

but `b = b + c` does work.

[1] https://stackoverflow.com/questions/12588986/typeerror-gener...

The real problem with these masks is that unless you are quite good at equalizing with your jaw, it will be hard to relieve pressure if you dive down more than 6-8 feet or so.

I think it's just very poorly worded. Sailboats can go faster than the true wind speed, like in the America's Cup, and typically do it do it going upwind. The addition of tacking is not wrong, it is zigzagging upwind, but just very oddly worded.

$36 a year for 1Password is perhaps the best value/$ I get out of any software I pay for.

I'm not sure that is Germany as a whole though, Berlin yes, but in Munich there are not nearly as many spatis, and from the people that I've interacted with, in smaller towns the only place to get a beer in the evening/night is the gas station.

Does anyone know a way to disable tab scrolling? The main thing that prevents me from switching is the fact that I can't see all my tabs at the same time. They don't continually shrink so I can't use any spatial awareness of where things are to know where certain pages are. I looked into trying to disable the tab scrolling on the top bar about two months ago but couldn't figure it out. Any tips would be very appreciated.

My favorite acid mixes have been Dr. Rubenstein's Quarantine series: https://www.youtube.com/watch?v=0XYcwN4wXhI

I can confirm acid techno is alive and well in Berlin. Lots of good music coming out and DJs playing it.

I recently read a very good book that was not so much a broad overview, but rather a closer look at the American ambassador and his family in Germany in the 1930s. I can wholeheartedly strongly recommend it.

https://en.wikipedia.org/wiki/In_the_Garden_of_Beasts

I did this with a pair of Focal Alpha 65s (albeit ~2.5 as expensive as the JBLs, and larger too), but I am not sure I can ever go back to smaller computer speakers again. The sound is so so so much better.

What a cool problem to work on - would have never have thought of it. What is the approx. hardware cost?

Could you explain how this works, or point me in the right direction on what to read? I'm interested in not using Transferwise as much if this is better.

I can also recommend looking into voroni clusters for amorphous glasses, as a level 3 type analysis. These can be combined into higher order "medium range order" clusters, which might be what call level 4, but these are all very interesting areas of research.

Currently watching it using Streamlink [0] to watch in VLC. This is so exciting! Wishing them, and the rover, all the best in the landing!

[0] https://streamlink.github.io/

> What is the common denominator between all metals?

I think most people who study this would say the common denominator is that metals lack a band gap and have a very mobile sea of electrons, and therefore good conductivity. Metals can be amorphous, or crystalline, or anywhere along the spectrum.

The "new school" certainly hasn't forgotten, or ignored, that crystals are patterns that tile all of space. These are known as space groups, and they enumerate every way to tile 3D space symmetrically (https://en.wikipedia.org/wiki/Space_group).

Anyways, I can guarantee that people who really study the subject absolutely take courses in metallurgy, crystallography, 3D math, and much more.

Also, if you're interested in simulations, I don't do that much myself, but you can always run some open source simulation software. http://www.opencalphad.com/

Happened to me in the exact same way with eBay. User (relatively inactive) for 13+ years, go to sell something (boots lol) with original pics, full description, everything set up correctly, and banned within 5 minutes. No way to appeal, and last email after they said they'd refund my $1 ad post charge, was that my account was being deleted.

In a very similar situation, I posted a pair of "Waxed Flesh" boots on eBay from my 13 year (albeit rarely used) account. They had plenty of pictures, waxed flesh is a very well known term with regards to boot leathers, but within 2 minutes, my long term eBay account was permanently terminated with no chance for recourse.

Besides it being exceedingly difficult to find someone to talk to about this, when I finally did in the chat service they offer, as soon as I brought up my account they terminated the chat with some copy/paste message about how they cannot talk to me about it as it might compromise their fraud detection. People I've talked to have mentioned that it might be due to the fact that there was "Flesh" in the title, but that is a valid description of the good I was trying to sell.

I guess what I'm trying to say, is that it felt entirely arbitrary, I had no recourse with regards to the discussion, and eBay is similarly shitty with regards to service.

Someone once posted a website they made in the HN comments that filtered Craigslist/Ebay for the exact old style laptops to reuse... does anyone remember what that was, or what it was called? Was a very nice way to keep on top of the offerings in a region without having to keep track of all the keywords.

We use them for computational imaging reconstruction in electron microscopy.

I happened to use Google Takeout to download everything I had last weekend (~200 Gb)... at first I thought my account was banned and that was the only solace I had, but this definitely is killing my productivity.

Unfortunately for me, I can't access simple analysis since the site can't figure out what country I'm in (no permissions were even asked for), and when I click on the intermediate analysis, the link doesn't seem to be valid. Only the advanced seems available to me. It would be great to not have to make an account as well.

I had the exact same experience there about 20 years ago as well with my family... was fantastic.

Here is the actual article, with much more analysis:

https://science.sciencemag.org/content/369/6504/689

As mentioned in the article, Nat Comm has been doing this for a few years now. I have found it immensely interesting and helpful for my own research to have access to the peer review files (usually in the Supplementary Information [1]). It is a great way to show younger students the process and formatting behind peer review and also have examples of what constitutes a positive review process. As an author, it does feel like a positive as well to have the peer review file out, so that readers who are interested can see how the work developed, and what points or weaknesses the reviewers focused on.

[1] randomly chosen from homepage - https://www.nature.com/articles/s41467-020-14530-7#Sec11

On the other hand, almost everyone I know who does scientific python in grad school loves miniconda...

Having moved continents last year, I think this is very apparent. European morning stores are the day/evenings of the US, and then like you say, between 2 and 4pm the active stories are more highly variable, as the US wakes up.

This looks to be a setting you can change in their "Other Settings" page.

I ended up using google keep as my most stripped down notetaking app that can can sync between devices.