Tweets
2022-11-04T12:59:23+00:00 | 🔗
RT @OlorenAI: Heyo! We're trying to learn more about how computational chemists spend their time, what software they use, and how to best i…
2022-11-01T22:53:28+00:00 | 🔗
@andrewwhite01 My favorite kind of artificial intelligence
2022-10-31T18:39:19+00:00 | 🔗
Twitter didn't auto-show preview... the title is "Minimal data requirements for accurate compound activity prediction using machine learning methods of different complexity" from Friederike Maite Siemers, Christian Feldmann, and Jürgen Bajorath
2022-10-31T18:38:05+00:00 | 🔗
On my reading list: https://t.co/WOx0EGRwKD I'd be curious how they approach the idea of activity cliffs, or even just the distribution of activity cliffs. RFs at semi-low data = good, definitely inline w experiences
2022-10-31T15:06:43+00:00 | 🔗
There's some information out there for cross-validation with time series data (https://t.co/v2eFT6u2fP). What are the methods people have found success with? Particularly with activity cliffs in drug discovery data?
2022-10-31T14:31:37+00:00 | 🔗
@tunguz I think Copilot for data processing… I feel so much faster at it!
2022-10-23T02:54:30+00:00 | 🔗
The main test we run is test...slf, where slf stands for "Save Load File", where we train, predict, save, load, predict a model and compare assert predictions are close before save and after load.
2022-10-23T02:52:50+00:00 | 🔗
Spent the weekend cleaning up tests on @OlorenChemEng ! It's pretty satisfying seeing the field of green... The next step, getting code coverage up from 44% to 100%. The hardest part of this all is visualization generation--how do you test these? https://t.co/vGccQXAzK0
2022-10-21T21:01:37+00:00 | 🔗
@OlorenChemEng This few-shot learning paper! https://t.co/pzDmdlIiC8 Megan Stanley, John Bronskill, @MaziarzKris, @hubert_misztela, @marwinsegler, @foo_fighterin, Nadine Schneider, @mmjb86