That's a very interesting suggestion. Though I think this particular result was much more statistically significant than 1 in 12 (check out the top left panel of Figure 5 in the paper)

🥉

In an age when are people are increasingly reliant on machine learning, it's important to consider biases and flaws in the humans who built the training data

rough! I know the feeling, last year i had 2 x 15 hr observations on subsequent days... twice (Feb and Sept) 🫠

@manishacaleb.bsky.social should I tell him about the "happy trombone effect"?

the tmux/screen commands can be exceptionally useful for grad students with long-running processes (e.g. data acquisition, analysis, simulation).
and with that the htop command to make sure they know how much memory these processes are using 😉

It sorted itself out the moment I posted this... Just like the previous time.

Thanks anyway 🙃

must be quantum mechanics - by whinging about it I have changed it

This is stunning (both visually and scientifically)!

amazing how, once again, the issue disappears the moment I post about it here....