Discover ways to transform this post
POST /speech/from-post
New MLIP model on the leaderboards! Currently #2 with an F1 score of 0.884. Congrats to the team. They provide a few pre-trained models as well as a ASE calculator for MD. Great stuff. It's a graph neural network (GNN) interatomic potential package based on NequIP. Published 2025-03-13 with 25.7M params.
https://github.com/MDIL-SNU/SevenNet
I'm most curious about its magnetic understanding and if we can get a material latent representation out easily. Will update what I find there shortly.
How this post is connected to other assets
In this post I'll share some of the work I've been doing on a Curie temperature prediction model. I finally found a decent dataset to work with. More on that here:
Discover other posts like this one
Sharing some notes as I read this paper. I uploaded it here for reference. I came across it looking for a Curie temperature dataset and so far this has been the best I've found so far.
Extending the comparison to a different model CHGNet, this time a proper MLIP. Similar to the Orb model, this model predicts energy, force, and stress, but with the addition of the magnetic moment for
M3GNet seems like a pretty popular MLIP model. Depending on the pipeline we build out, we may want to increase throughput with a model that can help us with MD and electronics predictions.