fenazego

Without having read the article, I would say it demonstrates that benchmarks alone, without knowledge of how a model was trained (c.q. constructed), cannot reliably determine whether or not a model has certain capabilities.

This is not a case of Simpson's Parodox, at least the analogy about accuracy vs speed from the article isn't. You're not comparing global vs subgroup correlations. On the one hand you're measuring how speed and accuracy are correlated across the population when you ask subjects to solve a problem. On the other hand, (rather than measuring subgroup correlations) you're measuring how accuracy is affected when you ask an individual to speed up or slow down.

I haven't used icicles either. For most of the time I used ido, and at some point I switched to ivy.

I missed your original VOMPECCC submission. Thanks for the write-up.

Although I've been a daily Emacs user for +25 years, I've only occasionally invested time in tweaking ICR (messing with the basic editor stuff you use hundreds of times a day can really get in the way of getting things done), and the sheer number of packages and sub-packages in the ICR space meant I only had a hazy idea of how they all related and (are they complementary, alternatives, successors?). Your VOMPECCC blogpost does a terrific job clarifying that!

I use byobu for managing tmux windows and sessions https://www.byobu.org/

Once you have btrfs you don't really need rsync anymore, its snapshot + send/receive functionality are all you need for convenient and efficient backups, using a tool like btrbk.

I'm running the same, but for me it's definitely not a smooth experience compared to plasma + X. Indeed I resorted to clicking on tray icon because invocation from custom shortcut doesn't allow me to Ctrl-c (copy to clipboard). On the other hand, clicking from tray breaks Ctrl-s (save to file). Oh, well.

I have a PhD, ~2000 citations, I earn a decent amount of money as a contractor doing ML research and development at startups and a large company, and I also feel subpar regularly. Although people don't usually talk about it, it's probably very common.

Remind yourself often of the achievements you are proud of.

My first and only experience with Perl was like this: in 1997, just for fun, I tried to write a program in Perl to turn my Mozilla bookmarks into a website. After a week of not succeeding, in frustration I decided to try Python. In two days I had what I wanted, and programming it was a joy. That sealed my judgement that Perl (and all of its culture) was not for me, so I'm not surprised at all that others might feel the same. (To be fair, there's a single oneliner that does make life a lot easier: ... | perl -pe 's{...}{...}')

I have been using btrfs in raid1 setups over the past years, and have not had any issues. Quite happy with the features (snapshotting, send/receive, adding drives to grow an fs)

It’s not obvious that training a model on GPL code would constitute a breach of the license.

I like the Marathons for the same reasons. Only in cold weather it‘s like the rubber of the tyres gets quite a bit stiffer (more so than other tyres), and provides less grip, just when you need it.