Detail — Mlp

Metrics

• Accuracy: 96.71%
• Macro F1: 96.53%
• Weighted F1: 96.71%
• Best Val Acc: 97.90%

Config (JSON)

{
  "tag": "high_lr_fast",
  "layers": [
    512,
    256
  ],
  "dropouts": [
    0.1,
    0.2
  ],
  "input_dropout": 0.0,
  "activation": "relu",
  "optimizer": "adam",
  "lr": 0.002,
  "epochs": 70,
  "batch": 256,
  "use_smote": true,
  "test_size": 0.15,
  "val_size": 0.15,
  "early_stop_patience": 8,
  "reduce_lr": true,
  "rlr_factor": 0.5,
  "rlr_patience": 4
}

Source MD

md/mlp_high_lr_fast.md