Sign in

Carlo Debernardi

@dbrcrl.bsky.social
270 followers 1.6K following 16 posts

Post-doc. Former philosopher, sociology atm, tomorrow who knows. Science of science. Trying to be bayesian. Python enjoyer.

PostsRepliesMedia
Reposted by Carlo Debernardi
Jordan Nafa @ajordannafa.com · 03/10/2026
The AI people co-opting and bastardizing "Bayesianism" is an unforgivable blasphemy and we ought to burn them at the stake like witches for their sins.
092
Carlo Debernardi @dbrcrl.bsky.social · 22/09/2026
I guess I just have a way slower rate of hardware replacement (and no work computers). Never used matlab, but I remember learning python in the old 2.7 days
010
Carlo Debernardi @dbrcrl.bsky.social · 22/09/2026
I know, I know. It was a very old install and it was somehow still working fine. I was just too lazy to update anything. But at some point things break and you have to move on 😅
110
Carlo Debernardi @dbrcrl.bsky.social · 22/09/2026
Finally bricked my old conda environment and made the switch to uv. It really does look amazing as people say. Just troubled a little to make cmdstanpy work but it seems to be solved now! Never going back to taking an afternoon to solve the f**ing environment 😌
120
Reposted by Carlo Debernardi
Cameron Patrick @cameronpat.bsky.social · 28/08/2026
do you have a moment to talk about the word of our lord and saviour? ↓
042
Carlo Debernardi @dbrcrl.bsky.social · 25/08/2026
Yep, it's bad
000
Carlo Debernardi @dbrcrl.bsky.social · 18/06/2026
Oh, yess! My precious!
My new copy of "Bayesian Workflow" by Gelman, Vehtari and McElreath, with contributions from others
062
Carlo Debernardi @dbrcrl.bsky.social · 15/04/2026
Yes, JS is pain but cytospace is nice. The one time I needed it I ended up writing a python function to generate the data structure to feed into it rather than working directly in JS
030
Carlo Debernardi @dbrcrl.bsky.social · 08/04/2026
Yeah, it's crazy, I didn't anticipate it either!
020
Carlo Debernardi @dbrcrl.bsky.social · 08/04/2026
She sure runs more than me, 200km in less than a month!
130
Carlo Debernardi @dbrcrl.bsky.social · 08/04/2026
I installed a counter on my hamster's wheel and I record distance and activity time every morning
110
Carlo Debernardi @dbrcrl.bsky.social · 24/03/2026
I had some fun familiarizing with count models in Stan. Thinning a Poisson distribution is pretty nice! carlodebernardi.github.io/writings/occ...
carlodebernardi.github.io
Occupational injuries and Poisson thinning - Carlo Debernardi
Just another academic personal webpage
011
Carlo Debernardi @dbrcrl.bsky.social · 06/12/2025
Fully agree. Still, my grandparents experienced this dramatic improvements during their lifetime. I was born in the '90s and, except for some areas, my experience is at best one of stagnation. Might be that the delta, on top of the absolute value, plays a role in shaping our views...
020
Carlo Debernardi @dbrcrl.bsky.social · 20/09/2025
So true though. I taught programming stuff a bunch of times and students are mostly ok, usual typos etc. Enter absolute and relative paths to external files and they just crash
020
Carlo Debernardi @dbrcrl.bsky.social · 20/09/2025
Also, yeah, the usage doesn't make a ton of sense in the case you are describing
110
Carlo Debernardi @dbrcrl.bsky.social · 20/09/2025
Definitely not straightforward if you don't know about it and it's not explained. But allows you to iterate over unique values (keys) and it's even better since default dicts are order-preserving (used to not be the case). Choosing the name of the iterating variable to be descriptive might help btw
110
Carlo Debernardi @dbrcrl.bsky.social · 20/09/2025
Super practical once you wrap your head around it though!
100
Carlo Debernardi @dbrcrl.bsky.social · 05/07/2025
Dropped some random thoughts about crowdsourced peer-review after reading a recent great paper by @lastpositivist.bsky.social carlodebernardi.github.io/writings/pee...
carlodebernardi.github.io
Carlo Debernardi
Just another academic personal webpage
010