{
  "trained_at": "2026-07-23T11:57:41.358736+00:00",
  "base_model": "distilbert/distilbert-base-multilingual-cased",
  "dataset": "/content/drive/MyDrive/FishSTOP/data/processed/fishstop_train_complete.csv",
  "dataset_sha256": "c89e1f5d2bb92b5a5dd5beb8a235a798932ed19eb9d75c2f740a94b5fd450816",
  "rows": {
    "test": 3810,
    "train": 13581,
    "validation": 1906
  },
  "labels": {
    "0": "LEGITIMATE",
    "1": "MALICIOUS"
  },
  "epochs": 4,
  "best_validation_email_f1": 0.9872773536895675,
  "sources_by_split": {
    "test": [
      "enron",
      "github_phishing_pot",
      "nazario_phishing-2022",
      "nazario_phishing-2023",
      "nazario_phishing-2024",
      "nazario_phishing-2025",
      "spamassassin_easy_ham",
      "spamassassin_hard_ham",
      "spaphish",
      "zenodo_2024"
    ],
    "train": [
      "enron",
      "github_phishing_pot",
      "nazario_phishing-2022",
      "nazario_phishing-2023",
      "nazario_phishing-2024",
      "nazario_phishing-2025",
      "spamassassin_easy_ham",
      "spamassassin_hard_ham",
      "spaphish",
      "synthetic_modern_v2",
      "zenodo_2024"
    ],
    "validation": [
      "enron",
      "github_phishing_pot",
      "nazario_phishing-2022",
      "nazario_phishing-2023",
      "nazario_phishing-2024",
      "nazario_phishing-2025",
      "spamassassin_easy_ham",
      "spamassassin_hard_ham",
      "spaphish",
      "zenodo_2024"
    ]
  },
  "dataset_audit": {
    "class_counts_by_split": {
      "test": {
        "0": 2244,
        "1": 1566
      },
      "train": {
        "0": 7976,
        "1": 5605
      },
      "validation": {
        "0": 1122,
        "1": 784
      }
    },
    "sources_by_split_and_label": {
      "test": {
        "enron": {
          "0": 1555
        },
        "github_phishing_pot": {
          "1": 1180
        },
        "nazario_phishing-2022": {
          "1": 33
        },
        "nazario_phishing-2023": {
          "1": 66
        },
        "nazario_phishing-2024": {
          "1": 65
        },
        "nazario_phishing-2025": {
          "1": 83
        },
        "spamassassin_easy_ham": {
          "0": 494
        },
        "spamassassin_hard_ham": {
          "0": 48
        },
        "spaphish": {
          "0": 132,
          "1": 135
        },
        "zenodo_2024": {
          "0": 15,
          "1": 4
        }
      },
      "train": {
        "enron": {
          "0": 5442
        },
        "github_phishing_pot": {
          "1": 4132
        },
        "nazario_phishing-2022": {
          "1": 117
        },
        "nazario_phishing-2023": {
          "1": 233
        },
        "nazario_phishing-2024": {
          "1": 228
        },
        "nazario_phishing-2025": {
          "1": 291
        },
        "spamassassin_easy_ham": {
          "0": 1729
        },
        "spamassassin_hard_ham": {
          "0": 169
        },
        "spaphish": {
          "0": 464,
          "1": 471
        },
        "synthetic_modern_v2": {
          "0": 120,
          "1": 120
        },
        "zenodo_2024": {
          "0": 52,
          "1": 13
        }
      },
      "validation": {
        "enron": {
          "0": 778
        },
        "github_phishing_pot": {
          "1": 590
        },
        "nazario_phishing-2022": {
          "1": 17
        },
        "nazario_phishing-2023": {
          "1": 33
        },
        "nazario_phishing-2024": {
          "1": 33
        },
        "nazario_phishing-2025": {
          "1": 42
        },
        "spamassassin_easy_ham": {
          "0": 247
        },
        "spamassassin_hard_ham": {
          "0": 24
        },
        "spaphish": {
          "0": 66,
          "1": 67
        },
        "zenodo_2024": {
          "0": 7,
          "1": 2
        }
      }
    },
    "real_train_sources_per_label": {
      "0": [
        "enron",
        "spamassassin_easy_ham",
        "spamassassin_hard_ham",
        "spaphish",
        "zenodo_2024"
      ],
      "1": [
        "github_phishing_pot",
        "nazario_phishing-2022",
        "nazario_phishing-2023",
        "nazario_phishing-2024",
        "nazario_phishing-2025",
        "spaphish",
        "zenodo_2024"
      ]
    },
    "train_class_weights": {
      "0": 0.8513665997993982,
      "1": 1.211507582515611
    },
    "warnings": [
      "No language column is present, so multilingual coverage cannot be measured or stratified."
    ]
  },
  "validation_multi_chunk_emails": 421,
  "test_multi_chunk_emails": 812,
  "calibration": {
    "version": 1,
    "method": "temperature_scaling",
    "temperature": 1.7604503631591797,
    "threshold": 0.5985404849052429,
    "band": 0.0,
    "positive_label_id": 1,
    "validation_f1_at_threshold": 0.9872773536895675,
    "selective_accuracy": 0.9895068205666316,
    "selective_coverage": 1.0,
    "target_selective_accuracy": 0.95,
    "minimum_selective_coverage": 0.8,
    "model_type": "distilbert",
    "aggregation": "maximum_positive_logit_margin",
    "max_length": 512,
    "stride": 128,
    "max_chunks": 8,
    "dataset_sha256": "c89e1f5d2bb92b5a5dd5beb8a235a798932ed19eb9d75c2f740a94b5fd450816"
  },
  "test_metrics": {
    "accuracy": 0.9845144356955381,
    "precision": 0.9851899549259497,
    "recall": 0.9770114942528736,
    "f1": 0.9810836806668805,
    "confusion_matrix": [
      [
        2221,
        23
      ],
      [
        36,
        1530
      ]
    ],
    "selective_coverage": 1.0,
    "selective_accuracy": 0.9845144356955381
  },
  "test_metrics_by_source": {
    "enron": {
      "accuracy": 0.9948553054662379,
      "precision": 0.0,
      "recall": 0.0,
      "f1": 0.0,
      "confusion_matrix": [
        [
          1547,
          8
        ],
        [
          0,
          0
        ]
      ],
      "selective_coverage": 1.0,
      "selective_accuracy": 0.9948553054662379,
      "rows": 1555
    },
    "github_phishing_pot": {
      "accuracy": 0.9830508474576272,
      "precision": 1.0,
      "recall": 0.9830508474576272,
      "f1": 0.9914529914529915,
      "confusion_matrix": [
        [
          0,
          0
        ],
        [
          20,
          1160
        ]
      ],
      "selective_coverage": 1.0,
      "selective_accuracy": 0.9830508474576272,
      "rows": 1180
    },
    "nazario_phishing-2022": {
      "accuracy": 0.9696969696969697,
      "precision": 1.0,
      "recall": 0.9696969696969697,
      "f1": 0.9846153846153847,
      "confusion_matrix": [
        [
          0,
          0
        ],
        [
          1,
          32
        ]
      ],
      "selective_coverage": 1.0,
      "selective_accuracy": 0.9696969696969697,
      "rows": 33
    },
    "nazario_phishing-2023": {
      "accuracy": 1.0,
      "precision": 1.0,
      "recall": 1.0,
      "f1": 1.0,
      "confusion_matrix": [
        [
          0,
          0
        ],
        [
          0,
          66
        ]
      ],
      "selective_coverage": 1.0,
      "selective_accuracy": 1.0,
      "rows": 66
    },
    "nazario_phishing-2024": {
      "accuracy": 1.0,
      "precision": 1.0,
      "recall": 1.0,
      "f1": 1.0,
      "confusion_matrix": [
        [
          0,
          0
        ],
        [
          0,
          65
        ]
      ],
      "selective_coverage": 1.0,
      "selective_accuracy": 1.0,
      "rows": 65
    },
    "nazario_phishing-2025": {
      "accuracy": 0.9518072289156626,
      "precision": 1.0,
      "recall": 0.9518072289156626,
      "f1": 0.9753086419753086,
      "confusion_matrix": [
        [
          0,
          0
        ],
        [
          4,
          79
        ]
      ],
      "selective_coverage": 1.0,
      "selective_accuracy": 0.9518072289156626,
      "rows": 83
    },
    "spamassassin_easy_ham": {
      "accuracy": 0.9939271255060729,
      "precision": 0.0,
      "recall": 0.0,
      "f1": 0.0,
      "confusion_matrix": [
        [
          491,
          3
        ],
        [
          0,
          0
        ]
      ],
      "selective_coverage": 1.0,
      "selective_accuracy": 0.9939271255060729,
      "rows": 494
    },
    "spamassassin_hard_ham": {
      "accuracy": 0.9583333333333334,
      "precision": 0.0,
      "recall": 0.0,
      "f1": 0.0,
      "confusion_matrix": [
        [
          46,
          2
        ],
        [
          0,
          0
        ]
      ],
      "selective_coverage": 1.0,
      "selective_accuracy": 0.9583333333333334,
      "rows": 48
    },
    "spaphish": {
      "accuracy": 0.9213483146067416,
      "precision": 0.9253731343283582,
      "recall": 0.9185185185185185,
      "f1": 0.9219330855018587,
      "confusion_matrix": [
        [
          122,
          10
        ],
        [
          11,
          124
        ]
      ],
      "selective_coverage": 1.0,
      "selective_accuracy": 0.9213483146067416,
      "rows": 267
    },
    "zenodo_2024": {
      "accuracy": 1.0,
      "precision": 1.0,
      "recall": 1.0,
      "f1": 1.0,
      "confusion_matrix": [
        [
          15,
          0
        ],
        [
          0,
          4
        ]
      ],
      "selective_coverage": 1.0,
      "selective_accuracy": 1.0,
      "rows": 19
    }
  }
}