I scored 4/5 on the Chemist Style Quiz. Can you spot the outlier?

check it out here:

We created a leaderboard ranking chemists on ChEMBL by how distinct and recognizable their molecules are.

blog:

preprint:

A very simple ML model can look at a molecule and identify the chemist that made it in ChEMBL. 60% top-5 accuracy on a 1815 class classification problem. ML practitioners should be wary of data leaks from models memorizing the intentions of the chemists making the molecules.

https://t.co/6noTcch8Xn

@EvolvedTechInc @nebiusai We will be sending Maxwell Kleinsasser to NVIDIA headquarters to present our findings about using mutagenesis data to keep protein-ligand binding models from memorizing. Read more here: Taking away the protein cheat sheet

Thanks so much to the @EvolvedTechInc team and to @nebiusai for providing H100s; your hardware is a pleasure to work with.

We won 2nd place at the Evolved 25 AIxBio Hackathon!

Tiny Recursion Model (TRM) results on ARC-AGI - ARC-AGI-1: 40%, $1.76/task - ARC-AGI-2: 6.2%, $2.10/task Thank you to @jm_alexia for contributing TRM, a well written, open source, and thorough research to the community based on the HRM from @makingAGI

https://t.co/09WEBhFyE7

selected notes from the field journal of alexei m. vishnevsky imperial ethnographic survey, volhynia oblast, april 1887 archival classification: restricted / special folklore submitted to the department of national character and rural traditions, under the protection of his

I would rather have noisy data than biased data.

Weeeeeeeeelllll.....

@LemmySmackett This is what a patch against the memetic equivalent of a SQL injection looks like

A real John Henry moment. Maybe the last

Humanity has prevailed (for now!) I'm completely exhausted. I figured, I had 10h of sleep in the last 3 days and I'm barely alive. I'll post more about the contest when I get some rest. (To be clear, those are provisional results, but my lead should be big enough)

you didn't listen

This is built on tens of billions of measurements our team performed on hundreds of proteins vs millions of molecules.