File size: 854 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+selfies on bindingdb

FOLD 1/5
GNN Train Loss: 0.9757
RMSE: 1.5584, MAE: 1.2829, R2: 0.0512

FOLD 2/5
GNN Train Loss: 0.9506
RMSE: 1.4919, MAE: 1.2230, R2: 0.0243

FOLD 3/5
GNN Train Loss: 0.9521
RMSE: 1.4578, MAE: 1.2127, R2: 0.0321

FOLD 4/5
GNN Train Loss: 0.9585
RMSE: 1.5511, MAE: 1.2722, R2: 0.0301

FOLD 5/5
GNN Train Loss: 0.9451
RMSE: 1.4885, MAE: 1.2289, R2: 0.0266

====== K-FOLD SUMMARY ======
RMSE: 1.5105  [1.4769, 1.5429]
MAE: 1.2441  [1.2201, 1.2699]
R2: 0.0331  [0.0263, 0.0428]

★ Using fold 3 model for test inference
RMSE: 1.7509, MAE: 1.3484, R2: 0.0653

====== HELD-OUT TEST METRICS ======
{'rmse': 1.7508875094999254, 'mae': 1.348360538482666, 'r2': 0.06534677743911743}
Test RMSE: 1.7480  (95 % CI: 1.5285–2.0136)

Test MAE : 1.3502  (95 % CI: 1.2410–1.4656)

Test R²  : 0.0639  (95 % CI: 0.0251–0.1043)