Dataset: energy_efficiency (regression)
Penalty: 0.0
Seed: 3
Best fitness: -2.051768832621367
Final val loss: 2.70543570
Final penalty: 0.00000000
Model saved to: logs/regression/energy_efficiency/models/best_model_penalty_0.0_seed_3.pth

Final architecture & hyperparameters:
  num_layers: 4
  layer_sizes: [8, 8, 10, 11]
  activations: [3, 3, 4, 4]
  dropout_rates: [0.014, 0.38, 0.157, 0.039]
  batch_norms: [0, 1, 0, 1]
  learning_rate: 0.0243
  batch_size: 32
  patience: 11
  optimizer_type: 2
  init_type: 1
  l2_penalty: 0.0001

Validation metrics (final):
  mae: 1.3632549047470093
  mse: 2.7054357528686523
  rmse: 1.6448208877773447
  r2_score: 0.9743955731391907
  mape: 8.06334763765335
  residual_std: 1.5726217031478882
  prediction_std: 9.48298454284668
  target_std: 10.279247283935547
  normalized_mae: 0.037952532690427955
  normalized_rmse: 0.04579122972226032
  num_samples: 115
  target_range: 35.91999816894531
  prediction_range: 31.06237030029297

Test metrics (final):
  mae: 1.2821044921875
  mse: 2.477389097213745
  rmse: 1.5739723940443635
  r2_score: 0.9766921997070312
  mape: 6.6939689218997955
  residual_std: 1.5521456003189087
  prediction_std: 9.489593505859375
  target_std: 10.309714317321777
  normalized_mae: 0.03682092198231918
  normalized_rmse: 0.04520311337849681
  num_samples: 116
  target_range: 34.81999969482422
  prediction_range: 29.400711059570312
