Table S5 from the MatterSim paper: Comparison of property prediction performance for M3GNet and Graphormer models.
Good read. Well written, very detailed and thorough. Great contribution.