Table 30 Distillation quality scores-F1 score as the performance metric.
From: Distilling knowledge from graph neural networks trained on cell graphs to non-neural student models
Model | Number_of_Parameters | Best_Performance | DQ_Score |
|---|---|---|---|
ExtraTrees trained on logits | 409.975 | 0.89539 | 0.0084 |
XGBoost trained on logits | 462.97 | 0.8712 | 0.022 |
Random trained on logits | 1331.635 | 0.8806 | 0.0248 |
HistGrad trained on logits | 679.8 | 0.8666 | 0.02676 |
LightGBM trained on logits | 692.15 | 0.8794 | 0.0197 |
