{
  "measurement": "same_text_vendor_agreement_not_accuracy",
  "articles": 83,
  "identical_input_hashes_verified": true,
  "metrics": {
    "vendor_related_securities": {
      "articles": 83,
      "model_pairs": 56,
      "vendor_pairs": 0,
      "common_pairs": 0,
      "model_only_pairs": 56,
      "vendor_only_pairs": 0,
      "exact_sets": 53,
      "both_empty": 53,
      "nonempty_exact_sets": 0
    },
    "vendor_named_company_mentions": {
      "articles": 83,
      "model_pairs": 56,
      "vendor_pairs": 92,
      "common_pairs": 52,
      "model_only_pairs": 4,
      "vendor_only_pairs": 40,
      "exact_sets": 59,
      "both_empty": 42,
      "nonempty_exact_sets": 17
    }
  },
  "independent_gold_rows": 0,
  "quality_equivalence_established": false,
  "limitations": [
    "Known retrospective overlap subset, not a sufficient independent quality test.",
    "Vendor named companies, related securities and direct issuer identity have different scopes.",
    "Default-empty cached fields do not establish that the vendor supplied an explicit empty set.",
    "English quality is not measured by this Korean overlap subset."
  ],
  "inputs_sha256": {
    "results_same_text_20261107/predictions.json": "905546af7d6719f1fde8ec578fff8011f250c8ac255e80fc18836c47489b131a",
    "deepsearch_corpus_comparison_20261105/same_text_vendor_references.json": "56975bf968cea8abb2d362b7d140e366308ccbc0e340a338b236ce999793476e"
  },
  "generation_evidence": {
    "job": "20261107T071120Z-b095cc1342cb",
    "articles": 83,
    "generations": 298,
    "models": [
      {
        "family": "qwen3",
        "model_id": "mlx-community/Qwen3-14B-4bit",
        "revision": "a4d9b2df59d2c150bef02fcbe0d91046b7ca33a4"
      },
      {
        "family": "phi4",
        "model_id": "mlx-community/phi-4-4bit",
        "revision": "fc0f8f23d369dc29b55cad1d65cb5bf0dcbee910"
      }
    ],
    "valid_reviews": 273,
    "parse_failures": 25,
    "predictions_sha256": "905546af7d6719f1fde8ec578fff8011f250c8ac255e80fc18836c47489b131a",
    "raw_generations_sha256": "86c5432cfeee13a93abc6b233e8425cc6996c024398ac69df12d4d22ced109e0",
    "registry_sha256": "8980193d030f923080bcc9681c1fb80568f20741091bd21535ffc6275cb1c585",
    "models_sha256": "14eb4a052d7ce2f8249359cca5bacb1c14db5bfcd98ee01ee84f6b32a1306fb8",
    "verified": true
  }
}
