“We got the chemical library... 130 million molecules... after 2000 teams competed, the winner was a random notebook; no one did better than chance on the most challenging eval”
Latent Space: The AI Engineer Podcast · Lab-in-the-Loop at Compute Speed - Ian Quigley, Leash Bio
Play from 30:05
30:05
Audio streams from the publisher’s own feed. Verbatim does not host or copy it.
What was described
designed a Kaggle competition with three evaluation splits (shared building block, no shared building block, completely new library) using a 130 M-molecule library
130 million molecules; 2000 teams; winner was a random notebook
Said nearby on the same subject
“If you want to try trigger, expiration, evaluation, a purchase, I would start building yourself a library. AI is really good at this as well, where you just start pulling different pieces of content.”
“They're using Pascal for the graph… we give them the entire infrastructure plumbing, and their internal tech team can build specific applications on top or use our application library.”
“Every single student… has a designated what I need block… they go to their scheduled high impact tutoring session… with two or three other students with similar need.”