Table 4 Results of the Suzuki-Miyaura dataset under ligand-based out-of-sample conditions
From: Prediction of chemical reaction yields with large-scale multi-view pre-training
Split type | Measure | YieldBERT | YieldBERT-DA | UA-GNN | ReaMVP |
|---|---|---|---|---|---|
Test 1 | MAE | \(19.357\pm 0.174\) | \(19.813\pm 0.177\) | \(16.328\pm 0.588\) | \(\mathbf {15.186\pm 0.492}\) |
| Â | RMSE | \(25.000\pm 0.095\) | \(24.975\pm 0.210\) | \(21.996\pm 0.818\) | \(\mathbf {19.564\pm 0.742}\) |
| Â | R\(^2\) | \(0.306\pm 0.005\) | \(0.307\pm 0.012\) | \(0.462\pm 0.400\) | \(\mathbf {0.574\pm 0.033}\) |
Test 2 | MAE | \(14.845\pm 0.364\) | \(15.777\pm 0.239\) | \(15.587\pm 0.356\) | \(\mathbf {13.905\pm 0.286}\) |
| Â | RMSE | \(19.592\pm 0.386\) | \(19.639\pm 0.264\) | \(20.485\pm 0.391\) | \(\mathbf {18.357\pm 0.349}\) |
| Â | R\(^2\) | \(0.469\pm 0.021\) | \(0.467\pm 0.014\) | \(0.420\pm 0.022\) | \(\mathbf {0.534\pm 0.018}\) |
Test 3 | MAE | \(15.438\pm 0.286\) | \(15.235\pm 0.492\) | \(13.624\pm 0.119\) | \(\mathbf {13.518\pm 0.284}\) |
| Â | RMSE | \(20.051\pm 0.371\) | \(19.455\pm 0.389\) | \(19.090\pm 0.342\) | \(\mathbf {18.236\pm 0.294}\) |
| Â | R\(^2\) | \(0.357\pm 0.024\) | \(0.395\pm 0.025\) | \(0.417\pm 0.021\) | \(\mathbf {0.468\pm 0.017}\) |
Test 4 | MAE | \(18.862\pm 0.095\) | \(18.644\pm 0.082\) | \(\mathbf {15.613\pm 0.382}\) | \(15.985\pm 0.615\) |
| Â | RMSE | \(23.114\pm 0.119\) | \(23.726\pm 0.141\) | \(22.176\pm 0.270\) | \(\mathbf {21.796\pm 0.700}\) |
| Â | R\(^2\) | \(0.239\pm 0.008\) | \(0.229\pm 0.010\) | \(0.299\pm 0.017\) | \(\mathbf {0.323\pm 0.043}\) |