TABLE 3.
Performance of the models summarized in Table 2 on the CSAR-NRC HiQ scoring benchmark. We only report evaluation results from the original reference. RMSEs are expressed in pK units.
| Model | References | Set 1 r | Set 1 RMSE | Set 2 r | Set 2 RMSE |
|---|---|---|---|---|---|
| Kdeep | Jiménez et al. (2018) | 0.72 | 2.08 | 0.65 | 1.91 |
| RosENet | Hassan-Harrirou et al. (2020) | 0.83 | 1.78 | 0.80 | 1.44 |
| OnionNet-2 | Wang et al. (2021d) | 0.89 | 1.50 | 0.87 | 1.21 |
| graphDelta | Karlov et al. (2020) | 0.74 | 1.59 | 0.71 | 1.52 |
| GraphBAR | Son and Kim (2021) | 0.75 | 1.59 | 0.65 | 1.56 |
| PIGNet | Moon et al. (2022) | 0.77 | — | 0.80 | — |
| BAPA | Seo et al. (2021) | 0.83 | 1.06 | 0.75 | 0.98 |