Dataset: energy_efficiency (regression)
Penalty: 0.1
Seed: 73
Best fitness: -1.00704259829487
Final val loss: 1.68845748
Final penalty: 0.04906216
Model saved to: logs/regression/energy_efficiency/models/best_model_penalty_0.1_seed_73.pth

Final architecture & hyperparameters:
  num_layers: 3
  layer_sizes: [11, 2, 6]
  activations: [3, 4, 4]
  dropout_rates: [0.015, 0.107, 0.037]
  batch_norms: [0, 1, 1]
  learning_rate: 0.0238
  batch_size: 32
  patience: 16
  optimizer_type: 2
  init_type: 0
  l2_penalty: 0.0005

Validation metrics (final):
  mae: 1.0516178607940674
  mse: 1.6884574890136719
  rmse: 1.2994065911075223
  r2_score: 0.9813995957374573
  mape: 5.165836587548256
  residual_std: 1.299389123916626
  prediction_std: 9.227533340454102
  target_std: 9.52761459350586
  normalized_mae: 0.02872487963020465
  normalized_rmse: 0.035493214133957486
  num_samples: 115
  target_range: 36.61000061035156
  prediction_range: 34.73887634277344

Test metrics (final):
  mae: 1.055832028388977
  mse: 1.860719919204712
  rmse: 1.364082079350327
  r2_score: 0.9792184829711914
  mape: 5.104335770010948
  residual_std: 1.3635083436965942
  prediction_std: 9.31137752532959
  target_std: 9.462409019470215
  normalized_mae: 0.029859502549892055
  normalized_rmse: 0.03857698121619927
  num_samples: 116
  target_range: 35.36000061035156
  prediction_range: 34.464412689208984
