Because this is such an important milestone for the SAGA Survey and for myself personally, I am posting here again!

The SAGA Survey characterizes satellite galaxy systems around Milky Way-mass galaxies, and we just posted our major results from our Data Release 3! Check arXiv!

Fantastic work out of the SAGA team!

Check out the tranche of papers now on arXiv: Papers III, IV, and V, which present the full satellite samples, star formation and gas properties, and galaxy-halo empirical models, respectively.

See sagasurvey.org for an overview!

Breaking my long silence to share results from a project I have been working on for more than a decade. Marla Geha and I were discussing how frustrating it was that there were so many comparisons of models to just ONE Milky Way. Why not 100 MWs, we thought? 🤔
sagasurvey.org

CNNs: ah yes those are reasonable hyperparams, you'll get pretty good convergence if you train a bit longer, keep at it dear ☺️

GNNs: oof, nowhere near convergence again? seems like a skill issue

So... Hugging Face can you not use the android DateTime widget for entering the DOB for Meta's Llama 3 license? Maybe I'm dumb and can't figure out how to jump back 30 yrs instantly, but clicking the 'back one month' arrow 400 times does not seem like good UI/UX.

After we passed a group of guys on motorcycles wearing leather jackets with skeletons, my 3 year old asked, 'Where did the X-ray shirt guys go?'

🚨 Exciting news: archival metadata viz tool release 🚨

tinyurl.com/archive-query
tinyurl.com/crimson-light

Ever start an observing proposal wondering whether or not there's ancillary ALMA Observatory📡 or NASA Webb Telescope or Gemini Observatory data for your targets? 1/

Question for anyone who has used Aaron Defazio's schedulefree optimizer... what the intuition for weight_decay? Same as usual?

I just had to decline an invited talk at a place that I'd be absolutely thrilled to speak at because of <personal reasons>.

It's a painful reminder that lots of folks in academia are regularly inhibited in their career opportunities because of factors outside their control.

This is so interesting.

I studied physics too. In 2011 I did an ML internship, and was shocked that the state of the art was basically just PCA (eigenfaces) and key point detection.

That sort of turned me off ML for 5+ years, until I realized deep learning was the real deal.

holy moly

There was this one break in the low cloud that blanketed us T-5min to T+2min of totality!

2026 Spain is a day after my ten year wedding anniversary! Guess we gotta start planning now...

Sadly we had a low cloud sit on top of us for about 6 minutes, including through all of totality 😭

Ah yes, and a cancelled United Airlines flight om the way out. My rate of bad flights is like 50% over the past 15 years.

Most current and future professors (including myself) were not Hubble fellows. It always stings when you don't get a national fellowship (or fellowships in general) though and it is okay to be sad. I was sad for a week or so when I got rejected each time.

(Just for fun, here is a deciphered version of that diagram by Claire Lamman.) 6/10

I have been entrained in the study of galactic winds! Awesome JWST data with more exciting papers on M82 and NGC 253 to come!

Ok this is next level April fools...