Dataset: energy_efficiency (regression)
Penalty: 0.2
Seed: 19
Best fitness: -5.274901585518846
Final val loss: 6.71027700
Final penalty: 0.19621057
Model saved to: logs/regression/energy_efficiency/models/best_model_penalty_0.2_seed_19.pth

Final architecture & hyperparameters:
  num_layers: 4
  layer_sizes: [8, 10, 11, 12]
  activations: [3, 4, 1, 2]
  dropout_rates: [0.027, 0.466, 0.458, 0.073]
  batch_norms: [1, 1, 0, 1]
  learning_rate: 0.0158
  batch_size: 16
  patience: 18
  optimizer_type: 0
  init_type: 0
  l2_penalty: 0.0002

Validation metrics (final):
  mae: 1.962062954902649
  mse: 6.710277080535889
  rmse: 2.5904202517228527
  r2_score: 0.938438355922699
  mape: 11.19743064045906
  residual_std: 2.5841214656829834
  prediction_std: 9.54547119140625
  target_std: 10.440352439880371
  normalized_mae: 0.05463835064518425
  normalized_rmse: 0.07213646722107458
  num_samples: 115
  target_range: 35.909996032714844
  prediction_range: 22.996841430664062

Test metrics (final):
  mae: 2.345829963684082
  mse: 9.643741607666016
  rmse: 3.105437426139193
  r2_score: 0.9137845039367676
  mape: 10.809627920389175
  residual_std: 3.0275816917419434
  prediction_std: 9.620373725891113
  target_std: 10.576210021972656
  normalized_mae: 0.06386686612905018
  normalized_rmse: 0.08454771208391136
  num_samples: 116
  target_range: 36.72999954223633
  prediction_range: 22.987838745117188
