{
  "experiment": "dinov3_gyuyig_tsugdri_binary",
  "run_subdir": null,
  "task": "gyuyig_tsugdri_binary_classification",
  "balanced_parquet_dir": null,
  "balanced_dataset_repo": "BDRC/gyuyig-tsugdri-binary-balanced-script-classification-dataset",
  "data_source": "hf",
  "benchmark_dataset_repo": null,
  "benchmark_per_parent": 60,
  "preprocess": {
    "train": "center_crop",
    "val": "center_crop",
    "test": "center_crop",
    "size": 224
  },
  "pooling": "cls_token",
  "training_config": {
    "lr_head_a": 0.0005,
    "lr_backbone_b": 1e-05,
    "lr_head_b": 0.0001,
    "lr_backbone_c": 1.5e-05,
    "lr_head_c": 5e-05,
    "epochs_a": 7,
    "epochs_b": 10,
    "epochs_c": 12,
    "unfreeze_blocks_b": 4,
    "unfreeze_blocks_c": 8,
    "skip_stage_c": false,
    "scheduler": "cosine_warmup",
    "warmup_epochs_a": 0,
    "warmup_epochs_b": 1,
    "warmup_epochs_c": 1,
    "warmup_start_factor": 0.01,
    "min_lr_ratio": 0.01,
    "weight_decay": 0.02,
    "grad_clip_norm": 1.0,
    "class_weight_mode": "custom",
    "class_weights": [
      1.0,
      1.0
    ],
    "label_smoothing": 0.05,
    "dropout": 0.1,
    "early_stop_patience": 6,
    "early_stop_min_delta": 0.001,
    "resolved_class_weights": [
      1.0,
      1.0
    ],
    "pooling": "cls_token",
    "lambda_consistency": 0.0,
    "flip_op": "rot180"
  },
  "resolved_class_weights": [
    1.0,
    1.0
  ],
  "warmstart": {
    "warmstart_repo": "BDRC/4-class-balanced-script-classifier",
    "warmstart_checkpoint": null,
    "warmstart_checkpoint_file": "final_model.pt",
    "checkpoint_test_metrics": {
      "loss": 0.42663901050885517,
      "accuracy": 0.9208333333333333,
      "macro_f1": 0.9208489161207983,
      "weighted_f1": 0.9208489161207983
    },
    "warmstart_pooling": "cls_token"
  },
  "checkpoint_selection": {
    "method": "best_val_macro_f1"
  },
  "best_checkpoint": {
    "path": "best_stage_c_last_blocks.pt",
    "epoch": 1,
    "val_macro_f1": 0.9164578111946533
  },
  "val_metrics": {
    "loss": 0.3914561231931051,
    "accuracy": 0.9166666666666666,
    "macro_f1": 0.9164578111946533,
    "weighted_f1": 0.9164578111946533,
    "auc_roc": 0.9311111111111111
  },
  "val_confusion_matrix": [
    [
      29,
      1
    ],
    [
      4,
      26
    ]
  ],
  "val_report": "              precision    recall  f1-score   support\n\n      Gyuyig       0.88      0.97      0.92        30\n     Tsugdri       0.96      0.87      0.91        30\n\n    accuracy                           0.92        60\n   macro avg       0.92      0.92      0.92        60\nweighted avg       0.92      0.92      0.92        60\n",
  "test_metrics": {
    "loss": 0.40474860469500223,
    "accuracy": 0.85,
    "macro_f1": 0.847930160518164,
    "weighted_f1": 0.847930160518164,
    "auc_roc": 0.9297222222222223
  },
  "test_confusion_matrix": [
    [
      58,
      2
    ],
    [
      16,
      44
    ]
  ],
  "test_report": "              precision    recall  f1-score   support\n\n      Gyuyig       0.78      0.97      0.87        60\n     Tsugdri       0.96      0.73      0.83        60\n\n    accuracy                           0.85       120\n   macro avg       0.87      0.85      0.85       120\nweighted avg       0.87      0.85      0.85       120\n",
  "history": {
    "stage_a": [
      {
        "epoch": 1,
        "train_loss": 0.8516533403169542,
        "train_acc": 0.4017857142857143,
        "lr_head": 5e-06,
        "val_loss": 0.8099705815315247,
        "val_accuracy": 0.36666666666666664,
        "val_macro_f1": 0.36666666666666664,
        "val_weighted_f1": 0.36666666666666664
      },
      {
        "epoch": 2,
        "train_loss": 0.6583722290538606,
        "train_acc": 0.6309523809523809,
        "lr_head": 0.0005,
        "val_loss": 0.6123360713322957,
        "val_accuracy": 0.65,
        "val_macro_f1": 0.6378269617706237,
        "val_weighted_f1": 0.6378269617706238
      },
      {
        "epoch": 3,
        "train_loss": 0.5344764121941158,
        "train_acc": 0.7291666666666666,
        "lr_head": 0.0004668412874366486,
        "val_loss": 0.5156712194283803,
        "val_accuracy": 0.7833333333333333,
        "val_macro_f1": 0.7827903091060986,
        "val_weighted_f1": 0.7827903091060985
      },
      {
        "epoch": 4,
        "train_loss": 0.4900844920249212,
        "train_acc": 0.7529761904761905,
        "lr_head": 0.00037624999999999996,
        "val_loss": 0.45860581994056704,
        "val_accuracy": 0.7666666666666667,
        "val_macro_f1": 0.7624434389140271,
        "val_weighted_f1": 0.7624434389140272
      },
      {
        "epoch": 5,
        "train_loss": 0.4637424250443776,
        "train_acc": 0.7857142857142857,
        "lr_head": 0.0002525,
        "val_loss": 0.43475709557533265,
        "val_accuracy": 0.7666666666666667,
        "val_macro_f1": 0.7624434389140271,
        "val_weighted_f1": 0.7624434389140272
      },
      {
        "epoch": 6,
        "train_loss": 0.4428629179795583,
        "train_acc": 0.7827380952380952,
        "lr_head": 0.00012875000000000007,
        "val_loss": 0.4212049206097921,
        "val_accuracy": 0.85,
        "val_macro_f1": 0.8499583217560434,
        "val_weighted_f1": 0.8499583217560432
      },
      {
        "epoch": 7,
        "train_loss": 0.4315942114307767,
        "train_acc": 0.8095238095238095,
        "lr_head": 3.815871256335142e-05,
        "val_loss": 0.4180481950441996,
        "val_accuracy": 0.85,
        "val_macro_f1": 0.8499583217560434,
        "val_weighted_f1": 0.8499583217560432
      }
    ],
    "stage_b": [
      {
        "epoch": 1,
        "train_loss": 0.42322553339458646,
        "train_acc": 0.8125,
        "lr_head": 1.0000000000000002e-06,
        "lr_backbone": 1.0000000000000001e-07,
        "val_loss": 0.42059502998987836,
        "val_accuracy": 0.8666666666666667,
        "val_macro_f1": 0.8666666666666667,
        "val_weighted_f1": 0.8666666666666667
      },
      {
        "epoch": 2,
        "train_loss": 0.43337753273191904,
        "train_acc": 0.8095238095238095,
        "lr_head": 0.0001,
        "lr_backbone": 1e-05,
        "val_loss": 0.4085062007109324,
        "val_accuracy": 0.8666666666666667,
        "val_macro_f1": 0.8666666666666667,
        "val_weighted_f1": 0.8666666666666667
      },
      {
        "epoch": 3,
        "train_loss": 0.4347181845278967,
        "train_acc": 0.7886904761904762,
        "lr_head": 9.701478472890248e-05,
        "lr_backbone": 9.701478472890248e-06,
        "val_loss": 0.3922509431838989,
        "val_accuracy": 0.9166666666666666,
        "val_macro_f1": 0.9164578111946533,
        "val_weighted_f1": 0.9164578111946533
      },
      {
        "epoch": 4,
        "train_loss": 0.4287990019434974,
        "train_acc": 0.7708333333333334,
        "lr_head": 8.84191999343894e-05,
        "lr_backbone": 8.841919993438941e-06,
        "val_loss": 0.3846385677655538,
        "val_accuracy": 0.85,
        "val_macro_f1": 0.849624060150376,
        "val_weighted_f1": 0.849624060150376
      },
      {
        "epoch": 5,
        "train_loss": 0.40112980774470736,
        "train_acc": 0.8214285714285714,
        "lr_head": 7.525e-05,
        "lr_backbone": 7.525e-06,
        "val_loss": 0.3696295181910197,
        "val_accuracy": 0.9,
        "val_macro_f1": 0.899888765294772,
        "val_weighted_f1": 0.899888765294772
      },
      {
        "epoch": 6,
        "train_loss": 0.3798623964900062,
        "train_acc": 0.8422619047619048,
        "lr_head": 5.909558479451306e-05,
        "lr_backbone": 5.909558479451306e-06,
        "val_loss": 0.365932967265447,
        "val_accuracy": 0.9,
        "val_macro_f1": 0.899888765294772,
        "val_weighted_f1": 0.899888765294772
      },
      {
        "epoch": 7,
        "train_loss": 0.3871297893070039,
        "train_acc": 0.8214285714285714,
        "lr_head": 4.190441520548695e-05,
        "lr_backbone": 4.190441520548696e-06,
        "val_loss": 0.3574748694896698,
        "val_accuracy": 0.8833333333333333,
        "val_macro_f1": 0.883300916921367,
        "val_weighted_f1": 0.883300916921367
      },
      {
        "epoch": 8,
        "train_loss": 0.39022268142019,
        "train_acc": 0.8005952380952381,
        "lr_head": 2.5750000000000013e-05,
        "lr_backbone": 2.575000000000001e-06,
        "val_loss": 0.3670339067776998,
        "val_accuracy": 0.85,
        "val_macro_f1": 0.849624060150376,
        "val_weighted_f1": 0.849624060150376
      },
      {
        "epoch": 9,
        "train_loss": 0.3902270041760944,
        "train_acc": 0.7976190476190477,
        "lr_head": 1.2580800065610596e-05,
        "lr_backbone": 1.2580800065610596e-06,
        "val_loss": 0.3644895474116007,
        "val_accuracy": 0.8666666666666667,
        "val_macro_f1": 0.8665183537263625,
        "val_weighted_f1": 0.8665183537263625
      }
    ],
    "stage_c": [
      {
        "epoch": 1,
        "train_loss": 0.4103491987500872,
        "train_acc": 0.8125,
        "lr_head": 5.000000000000001e-07,
        "lr_backbone": 1.5000000000000002e-07,
        "val_loss": 0.3914561231931051,
        "val_accuracy": 0.9166666666666666,
        "val_macro_f1": 0.9164578111946533,
        "val_weighted_f1": 0.9164578111946533
      },
      {
        "epoch": 2,
        "train_loss": 0.4436623099304381,
        "train_acc": 0.7678571428571429,
        "lr_head": 5e-05,
        "lr_backbone": 1.5e-05,
        "val_loss": 0.3777097463607788,
        "val_accuracy": 0.9166666666666666,
        "val_macro_f1": 0.9164578111946533,
        "val_weighted_f1": 0.9164578111946533
      },
      {
        "epoch": 3,
        "train_loss": 0.3808557249250866,
        "train_acc": 0.8363095238095238,
        "lr_head": 4.899745109695881e-05,
        "lr_backbone": 1.4699235329087644e-05,
        "val_loss": 0.3577085534731547,
        "val_accuracy": 0.8666666666666667,
        "val_macro_f1": 0.8665183537263625,
        "val_weighted_f1": 0.8665183537263625
      },
      {
        "epoch": 4,
        "train_loss": 0.3782187302907308,
        "train_acc": 0.8392857142857143,
        "lr_head": 4.6071024937571735e-05,
        "lr_backbone": 1.3821307481271522e-05,
        "val_loss": 0.3452535013357798,
        "val_accuracy": 0.8833333333333333,
        "val_macro_f1": 0.883300916921367,
        "val_weighted_f1": 0.883300916921367
      },
      {
        "epoch": 5,
        "train_loss": 0.38378763056936716,
        "train_acc": 0.8392857142857143,
        "lr_head": 4.145780316514581e-05,
        "lr_backbone": 1.2437340949543742e-05,
        "val_loss": 0.33214412728945414,
        "val_accuracy": 0.8833333333333333,
        "val_macro_f1": 0.883300916921367,
        "val_weighted_f1": 0.883300916921367
      },
      {
        "epoch": 6,
        "train_loss": 0.3539558429093588,
        "train_acc": 0.8571428571428571,
        "lr_head": 3.5531521571796694e-05,
        "lr_backbone": 1.0659456471539008e-05,
        "val_loss": 0.3277231236298879,
        "val_accuracy": 0.8833333333333333,
        "val_macro_f1": 0.883300916921367,
        "val_weighted_f1": 0.883300916921367
      },
      {
        "epoch": 7,
        "train_loss": 0.35197025750364574,
        "train_acc": 0.8511904761904762,
        "lr_head": 2.877229224726381e-05,
        "lr_backbone": 8.631687674179142e-06,
        "val_loss": 0.3296816150347392,
        "val_accuracy": 0.8666666666666667,
        "val_macro_f1": 0.8665183537263625,
        "val_weighted_f1": 0.8665183537263625
      }
    ]
  },
  "confusion_matrix": [
    [
      58,
      2
    ],
    [
      16,
      44
    ]
  ],
  "report": "              precision    recall  f1-score   support\n\n      Gyuyig       0.78      0.97      0.87        60\n     Tsugdri       0.96      0.73      0.83        60\n\n    accuracy                           0.85       120\n   macro avg       0.87      0.85      0.85       120\nweighted avg       0.87      0.85      0.85       120\n",
  "idx_to_label": {
    "0": "Gyuyig",
    "1": "Tsugdri"
  },
  "label_to_idx": {
    "Gyuyig": 0,
    "Tsugdri": 1
  },
  "model_id": "facebook/dinov3-vits16-pretrain-lvd1689m"
}