Skip to main content
. 2023 Apr 5;127(15):3484–3489. doi: 10.1021/acs.jpca.2c09030

Table 2. Performance of 2D and 3D GNNs for the Reorganization Energy Prediction Task Using RDKit Geometries as Inputs.

Model MAE (eV) RMSE (eV) R2
ChIRo 0.111 (0.002) 0.174 (0.003) 0.786 (0.008)
SchNet 0.152 (0.002) 0.214 (0.003) 0.674 (0.008)
MPNN 0.128 (0.004) 0.194 (0.004) 0.686 (0.022)