I called for “call for contributions” recently, but it didn’t end well. People were too obsessed with keeping their secrets and know little outside of ML.

So I searched myself for challenging problems in science, with high meaningful impact, potential for ML to make breakthrough, ready dataset and benchmark, and I found this ProteinNet for protein folding. These scientists seem to think for the sake of science as a whole, and want to see how ML can help advance their field. You are welcome to use it for your side project if you are already tired of old time CV or NLP tutorials.