Molecular Property Prediction On Freesolv
評価指標
R2
RMSE
評価結果
このベンチマークにおける各モデルのパフォーマンス結果
比較表
| モデル名 | R2 | RMSE |
|---|---|---|
| how-powerful-are-graph-neural-networks | 0.964±0.008 | 0.744±0.083 |
| grover-self-supervised-message-passing | - | 2.272 |
| pre-training-graph-neural-networks-on | - | 1.648±0.074 |
| n-gram-graph-a-novel-molecule-representation | - | 2.688 |
| uni-mol-a-universal-3d-molecular | - | 1.620 |
| n-gram-graph-a-novel-molecule-representation | - | 5.061 |
| chemrl-gem-geometry-enhanced-molecular | - | 1.877 |
| dropgnn-random-dropouts-increase-the | 0.972±0.005 | 0.657±0.059 |
| self-guided-masked-autoencoders-for-domain | - | 1.09 |
| principal-neighbourhood-aggregation-for-graph | 0.951±0.009 | 0.870±0.081 |
| a-bayesian-flow-network-framework-for | - | 1.418 |
| pure-transformers-are-powerful-graph-learners | 0.930±0.018 | 1.038±0.125 |
| masked-attention-is-all-you-need-for-graphs | 0.977±0.001 | 0.595±0.013 |
| pre-training-graph-neural-networks | - | 2.764 |
| graph-attention-networks | 0.959±0.011 | 0.791±0.101 |
| do-transformers-really-perform-bad-for-graph | 0.927±0.005 | 1.065±0.039 |
| are-learned-molecular-representations-ready | - | 2.082 |
| recipe-for-a-general-powerful-scalable-graph | 0.861±0.037 | 1.462±0.188 |
| semi-supervised-classification-with-graph | 0.957±0.009 | 0.815±0.086 |
| how-attentive-are-graph-attention-networks | 0.970±0.007 | 0.676±0.081 |
| grover-self-supervised-message-passing | - | 2.176 |
| molecular-structure-property-co-trained | - | 1.859 |