File size: 865 Bytes
9a67fbe |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 |
Experiment: gin+ecfp on bindingdb
FOLD 1/5
GNN Train Loss: 0.0171
RMSE: 1.7724, MAE: 1.3869, R2: -0.2273
FOLD 2/5
GNN Train Loss: 0.0217
RMSE: 1.7963, MAE: 1.4253, R2: -0.4145
FOLD 3/5
GNN Train Loss: 0.0179
RMSE: 1.7046, MAE: 1.3570, R2: -0.3235
FOLD 4/5
GNN Train Loss: 0.0196
RMSE: 1.8985, MAE: 1.5419, R2: -0.4530
FOLD 5/5
GNN Train Loss: 0.0135
RMSE: 1.7420, MAE: 1.4135, R2: -0.3332
====== K-FOLD SUMMARY ======
RMSE: 1.7838 [1.7318, 1.8468]
MAE: 1.4238 [1.3796, 1.4852]
R2: -0.3489 [-0.4146, -0.2697]
★ Using fold 3 model for test inference
RMSE: 2.0158, MAE: 1.4913, R2: -0.2389
====== HELD-OUT TEST METRICS ======
{'rmse': 2.0158109357372025, 'mae': 1.4913249015808105, 'r2': -0.23889243602752686}
Test RMSE: 2.0116 (95 % CI: 1.7335–2.3415)
Test MAE : 1.4942 (95 % CI: 1.3729–1.6222)
Test R² : -0.2441 (95 % CI: -0.3634–-0.1405)
|