Dataset: energy_efficiency (regression)
Penalty: 0.0
Seed: 7
Best fitness: -3.9023155196853305
Final val loss: 6.20652656
Final penalty: 0.00000000
Model saved to: logs/regression/energy_efficiency/models/best_model_penalty_0.0_seed_7.pth

Final architecture & hyperparameters:
  num_layers: 4
  layer_sizes: [11, 6, 2, 10]
  activations: [3, 4, 2, 4]
  dropout_rates: [0.05, 0.134, 0.204, 0.138]
  batch_norms: [0, 1, 0, 1]
  learning_rate: 0.0287
  batch_size: 32
  patience: 23
  optimizer_type: 2
  init_type: 2
  l2_penalty: 0.0001

Validation metrics (final):
  mae: 1.9130518436431885
  mse: 6.206526279449463
  rmse: 2.4912900833603184
  r2_score: 0.9436553120613098
  mape: 10.921843349933624
  residual_std: 2.416748523712158
  prediction_std: 10.001713752746582
  target_std: 10.495369911193848
  normalized_mae: 0.05214095710945571
  normalized_rmse: 0.06790106071371611
  num_samples: 115
  target_range: 36.69000244140625
  prediction_range: 25.190074920654297

Test metrics (final):
  mae: 1.9699867963790894
  mse: 6.617609024047852
  rmse: 2.5724713844954334
  r2_score: 0.9296414256095886
  mape: 12.055440247058868
  residual_std: 2.3231630325317383
  prediction_std: 9.439736366271973
  target_std: 9.698221206665039
  normalized_mae: 0.054075947841790276
  normalized_rmse: 0.07061409176353892
  num_samples: 116
  target_range: 36.43000030517578
  prediction_range: 25.171710968017578
