janosh/matbench-discovery
An evaluation framework for machine learning models simulating high-throughput materials discovery.
This project helps materials scientists and researchers evaluate how well machine learning models predict properties of new inorganic crystals, such as thermodynamic stability and thermal conductivity, and their atomic positions. It takes various ML models as input and produces an interactive leaderboard ranking their performance against high-throughput DFT calculations. This is useful for anyone working to build large-scale materials databases or to streamline materials discovery.
214 stars. Available on PyPI.
Use this if you are a materials scientist or researcher who needs to assess and compare machine learning models for predicting crystal properties and accelerating materials discovery workflows.
Not ideal if you are looking for a tool to generate new material structures or run DFT calculations directly, as it focuses on evaluating existing ML model performance.
Stars
214
Forks
54
Language
Python
License
MIT
Category
Last pushed
Mar 04, 2026
Commits (30d)
0
Dependencies
12
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/janosh/matbench-discovery"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related frameworks
deepmodeling/deepmd-kit
A deep learning package for many-body potential energy representation and molecular dynamics
chemprop/chemprop
Message Passing Neural Networks for Molecule Property Prediction
mir-group/nequip
NequIP is a code for building E(3)-equivariant interatomic potentials
Acellera/moleculekit
MoleculeKit: Your favorite molecule manipulation kit
CederGroupHub/chgnet
Pretrained universal neural network potential for charge-informed atomistic modeling...