unhammer

I really hope you check with native Cree speakers before you publish a system like this. Smaller language communities are already being hurt by low-quality generated text. See also https://news.ycombinator.com/item?id=35300603 for an experience of GPT with the language Northern Saami.

Also, back-translation was never a good evaluation method. Say I don't know English; and I make a new system that calls Google Translate word for word. I back-translate the Norwegian input "den sykkelen kunne jeg gjerne hatt" and my word-for-word system gives "it the bike could I happily hat". Now I evaluate that whole sentence by putting it into Google Translate and presto I get "den sykkelen kunne jeg gjerne hatt" – can I conclude that word-for-word Google Translate is perfect? Or even worse, if you use the same system in both directions, anything that's looks like a bijection if you squint will do (reverse then reverse, transliteration into Canadian syllabics then back, encrypt/decrypt, map each word to a number, etc.).

If you have a bilingual corpus of Cree text, then you can get some automated measures (some popular ones listed at https://wiki.apertium.org/wiki/Evaluation ), but it would have to be a corpus you didn't find online since otherwise GPT-4 has it in its training data and it the numbers will not be valid for sentences that are not in that corpus.

No-squint outdoor programming

huh, Intel making GIMP plugins https://github.com/intel/openvino-ai-plugins-gimp

Reading about programming makes you good at reading about programming. If you want a "guided project" to learn from instead, I can recommend https://lhbg-book.link/

I wish there was an f-droid for iOS; not for installing from, just as a way of searching for apps (including paid apps) that don't have ads or in-app purchases.

There is https://nobsgames.stavros.io/ but it's just for games ;-)

(I tried finding an iPad version of workrave/safeeyes, but all the apps are full of nags to pay for extra features or show ads in their break-time etc.)

also https://slate.com/human-interest/2012/05/how-to-cook-onions-... (onions take more than 10 minutes to caramelize, but people filter for short cooking time, so the lie propagates)

For magit, I like having it display full screen (deleting other windows), and then on quitting restore to the windows/splits I had before:

     (setq magit-display-buffer-function 'magit-display-buffer-fullframe-status-topleft-v1)
     (setq magit-bury-buffer-function 'magit-restore-window-configuration)

For more general display-buffer tweaks, OP has a whole article on it https://www.masteringemacs.org/article/demystifying-emacs-wi... (though for display-buffer I just stick to the defaults myself, never tried CIDER)

Also highly recommended to anyone who wants to nerd out on ecology: Rachel Carson's Silent Spring https://www.fadedpage.com/showbook.php?pid=20151002 , which among other things lead to the ban on DDT in agriculture in the US. Still relevant, and beautifully written.

Though if we keep extracting fossil fuels at the same rate, while electric cars lead to less demand for them, the price of fossil fuels will go down. Then people (corporations) will buy cheap fossil fuel for other stuff (now cruise tickets got even cheaper because electric cars lead to less demand for oil! When it gets really cheap you can heat your outdoor pool in January off a diesel generator.)

Of course, when the price of oil starts dropping, they'll have to extract at a faster rate to keep profits up.

Not that one should have to, but for Firefox you can

    cd .mozilla/firefox/$profile
    git init
    git add extensions
    git commit -am init
    echo '*.xpi diff=zip' >>.gitattributes
    echo '[diff "zip"]' >>.git/config
    echo ' textconv = unzip -c -a' >>.git/config

which at least lets you take a peek at what kind of nefarious updates you're getting.

Apropos, https://github.com/two-twelve/fernery a "tool for generating images of ferns and other Iterated Function Systems" was recently posted somewhere

what mic do you use?

Not just conservation, but banning some of the worst pesticides in the 60's

Or even a decade earlier:

> Opposition to DDT was focused by the 1962 publication of Rachel Carson's book Silent Spring. It talked about environmental impacts that correlated with the widespread use of DDT in agriculture in the United States, and it questioned the logic of broadcasting potentially dangerous chemicals into the environment with little prior investigation of their environmental and health effects. The book cited claims that DDT and other pesticides caused cancer and that their agricultural use was a threat to wildlife, particularly birds. Although Carson never directly called for an outright ban on the use of DDT, its publication was a seminal event for the environmental movement and resulted in a large public outcry that eventually led, in 1972, to a ban on DDT's agricultural use in the United States. https://en.wikipedia.org/wiki/DDT#History

By the way, you can get Silent Spring for free on https://www.fadedpage.com/showbook.php?pid=20151002 – and it is still very much worth reading; eloquent and poetic while still being factual, informative and balanced. The producers of DDT and other poisons started a smear-campaign after the publication, but it only lead to a Streisand-effect. See also this little clip from her single TV interview: https://www.youtube.com/watch?v=8nriVjC0H8I

Still not available in Norway :(

Weird signup captcha, it requires you to open devtools and remove `onpaste="return false;"` from their email field (either that or enter your email manually, typos and all)

Numbers like that don't even make sense. With lots of morphology, one "word" has lots of forms and derivations. With high frequency, one "word" takes up a huge chunk of any given text. Maybe it would give a better idea if they counted the frequency of Latin-contributed (vs other) tokens in a corpus.

Only every time they claim a major mathematical breakthrough using regular expressions.

https://en.wikipedia.org/wiki/A_language_is_a_dialect_with_a... but it seems that in the US, even dialects need to be armed to stand their ground.

Here in Norway, we just call any regional language difference dialectal (unless of course you get to call it a different language, cf. above link). That takes away the need to argue about how much linguistic difference is needed for two dialects to be considered separate or why the difference occurred or whether it's "correct grammar" – if people speak their language differently because they're from different places, that's a dialectal difference. (Anyone who speaks a language thus speaks a dialect of that language.) Simple, but rigorous. There's no value judgement, it's just a factual description.

… amberbeeches … Neat, Icelandic uses raf as in "amber" just like ēlektron "amber".

what a strangely fitting typo though :-)

There's some kind of parallel here to the need for including </sarcasm> in internet comments. The lack of shared context makes it harder to assume something that "obviously" everyone knows.

"finish school" as in graduate (at age ~19), see https://en.wikipedia.org/wiki/Russefeiring

Writing flash cards and doing the exercises takes time. How do you decide/prioritise what you need to remember (in the Age of Search Engines)?

Not yet. If anyone manages to read through all of https://github.com/Microsoft/vscode/issues/50140 they might find out if it's planned or not; Mickey's article is shorter =P

Wow. For me, Gnus starts in <10s. But I guess it depends a lot on what backends you use (I use dovecot+mbsync for mail, so Gnus doesn't really index anything)

It does handle HTML, but it'll look different from what people reading email in Firefox etc. see. So just like with eww.

(Also, Outlook has MS-specific niceties like replying directly to Teams messages from within an email without having to start up Teams, which I assume they made as a workaround for how long it takes to start up Teams and pick the right organisation and wait for it to tell you to log out (since your account is still connected to that one organisation that long ago invited you for a single meeting and only org admins can remove users) and then hit reload since that sometimes works and then pick the right organisation again and then reply "OK" and then quickly exit Teams before your computer runs out of memory and grinds to a halt.)

Open in Emacs, hit `control+h` then `c` with your cursor on the character, it'll show everything you want and more: https://i.imgur.com/X8xp1WS.gif (also shows the decomposition of composed characters so you can discern ä vs ä and emoji combinations)

Huh, I've always really liked the Thunderbird search; it's really fast, and the clickable date filter is so much easier than any other such UI I've tried. But maybe I search more by date-ranges than most people :)