{
  "reviewed": "2026-09-23",
  "scope": "Extracted aggregates from stored project reports. No experiments rerun. Internal synthetic and curated evaluation, not independently validated real-world accuracy.",
  "sources": [
    {
      "file": "data/reports/neural_baseline_minilm_smoke_e5.json",
      "sha256": "0adb75f55a58b72891cf0a85b31f32ea273b447e492e1e87f2dec649ed3d44f7"
    },
    {
      "file": "data/reports/neural_baseline_minilm_smoke_e5_balanced.json",
      "sha256": "0ac343f3c476a815ffa8b2814b6776c5f05eefeea253e43e0fdfe5899942db4f"
    },
    {
      "file": "data/reports/minilm_large_v1_hybrid_balanced.json",
      "sha256": "99350223552743ac6caa500840903faa7f5a991baa77471230273e0cf884e783"
    },
    {
      "file": "data/reports/public_ocr_dataset_validation.json",
      "sha256": "c7ad681589f8c2984bb1f0c74a0c9384997d40a57bac7b2eb7333bc63bec55bb"
    },
    {
      "file": "data/reports/public_ocr_extraction_report.json",
      "sha256": "08901a2cb9b2c49e950f1aa65006851288b0b6210ac8f3ff877f63b4661dc6e6"
    }
  ],
  "smoke_negative_controls": {
    "positive_only": {
      "rows": 40,
      "predicted_entity_rows": 38
    },
    "with_training_negatives": {
      "rows": 40,
      "predicted_entity_rows": 0
    }
  },
  "larger_baseline": {
    "run": "minilm-large-v1",
    "training_rows": 12000,
    "epochs": 3,
    "metric": "Per-row entity-label count matching, not exact character-span matching.",
    "splits": {
      "dev": {
        "rows": 240,
        "exact_label_rows": 240,
        "no_entity_exact_rows": 0
      },
      "stress": {
        "rows": 120,
        "exact_label_rows": 120,
        "no_entity_exact_rows": 17
      },
      "test_clean": {
        "rows": 240,
        "exact_label_rows": 240,
        "no_entity_exact_rows": 0
      },
      "test_negative": {
        "rows": 160,
        "exact_label_rows": 160,
        "no_entity_exact_rows": 160
      },
      "test_noisy": {
        "rows": 240,
        "exact_label_rows": 240,
        "no_entity_exact_rows": 0
      }
    }
  },
  "document_preparation": {
    "extracted_rows": 1829,
    "validated_rows": 1829,
    "entity_labels": {},
    "caveat": "Schema validation and extraction do not establish detection accuracy or absence of personal information."
  }
}