Dataset: energy_efficiency (regression)
Penalty: 0.2
Seed: 3
Best fitness: -0.7454253835065501
Final val loss: 3.95080880
Final penalty: 0.19082483
Model saved to: logs/regression/energy_efficiency/models/best_model_penalty_0.2_seed_3.pth

Final architecture & hyperparameters:
  num_layers: 4
  layer_sizes: [8, 9, 6, 9]
  activations: [3, 4, 3, 4]
  dropout_rates: [0.002, 0.496, 0.287, 0.035]
  batch_norms: [0, 1, 0, 1]
  learning_rate: 0.029
  batch_size: 32
  patience: 20
  optimizer_type: 1
  init_type: 1
  l2_penalty: 0.0002

Validation metrics (final):
  mae: 1.5520977973937988
  mse: 3.9508087635040283
  rmse: 1.987664147562165
  r2_score: 0.9626092910766602
  mape: 10.08928120136261
  residual_std: 1.8838540315628052
  prediction_std: 9.874682426452637
  target_std: 10.279247283935547
  normalized_mae: 0.04320985179603008
  normalized_rmse: 0.05533586438989863
  num_samples: 115
  target_range: 35.91999816894531
  prediction_range: 27.0233154296875

Test metrics (final):
  mae: 1.3854480981826782
  mse: 3.176548480987549
  rmse: 1.7822874293972757
  r2_score: 0.9701144099235535
  mape: 6.917751580476761
  residual_std: 1.6891441345214844
  prediction_std: 10.064854621887207
  target_std: 10.309714317321777
  normalized_mae: 0.03978886014719341
  normalized_rmse: 0.05118573937443779
  num_samples: 116
  target_range: 34.81999969482422
  prediction_range: 27.228561401367188
