{
  "schema_version": 1,
  "created_at": "2026-07-14T07:23:50.679303+00:00",
  "status": "frozen_before_inference",
  "purpose": "Isolated English-gloss to Kuku Yalanji canonical-headword diagnostic.",
  "selection": {
    "entry_type": "Every @graph entry is considered; no semantic allowlist is used.",
    "form_rule": "Headword and English definition must each be one Unicode letter token, optionally internally hyphenated or apostrophized.",
    "case_rule": "Both forms must already be lowercase; names and uppercase grammatical abbreviations are excluded.",
    "grouping": "Casefolded NFC English gloss; every surviving headword in the group is an accepted reference."
  },
  "interpretation": {
    "primary_endpoint": "normalized accepted-reference exact match",
    "secondary_endpoints": [
      "raw accepted-reference exact match",
      "accepted-headword token containment",
      "minimum character error rate",
      "maximum sentence chrF++ across accepted references"
    ],
    "warning": "This is a dictionary-coverage probe, not an estimate of unrestricted translation quality. Lexicon material may have influenced training-corpus construction."
  },
  "counts": {
    "lexicon_graph_entries": 1640,
    "eligible_entries": 395,
    "unique_english_gloss_rows": 297,
    "rows_with_multiple_accepted_references": 68,
    "source_seen_as_isolated_training_input": 0,
    "target_seen_as_training_token": 192,
    "identity_translation": 14
  },
  "inputs": {
    "lexicon": {
      "path": "dictionaries/kuku_yalanji/lexicon.jsonld",
      "sha256": "a461612f8d2df805f4dd9f311e7a96702eee108cab04d752a6cbb7564b8b6993"
    },
    "training_file": {
      "path": "/mnt/donto-data/donto-resources/research/translation-training/kuku-yalanji-runpod-2026-06-30/prepared/v21.2-claude-balanced-replay/train.eng-gvn.jsonl",
      "sha256": "7cb97c1c1d1afd05d0b7ddd04f64180182bf9de261bf5473d07b28bdb43967f0",
      "rows": 22164
    },
    "elder_file": {
      "path": "/mnt/donto-data/donto-resources/research/translation-training/kuku-yalanji-runpod-2026-06-30/sources/elder-shared-sentence-pairs-2026-07-02/sentence_pairs.eng-gvn.jsonl",
      "sha256": "29a9bfbfecbe535b56eda3bf0ecf255a17190b438186b60b999c78e1ee4854d2",
      "rows": 43
    },
    "builder": {
      "path": "/mnt/donto-data/workspace/mobtranslate.com/training/translation/build_kuku_lexicon_probe.py",
      "sha256": "58e8d843bd51c5b286adb8d5df932298b0862742bf2bfec0fbe880e4bd9d1720"
    }
  },
  "probe": {
    "path": "/mnt/donto-data/donto-resources/research/translation-training/kuku-yalanji-runpod-2026-06-30/analysis/v22-lexicon-elder-version-probe-2026-07-14/single_gloss_probe.eng-gvn.jsonl",
    "sha256": "38ce15e36a60018d1e9c057f359a68f4c3bf7f8f09b4d027fa7ca82abd88cce7"
  },
  "combined_probe": {
    "path": "/mnt/donto-data/donto-resources/research/translation-training/kuku-yalanji-runpod-2026-06-30/analysis/v22-lexicon-elder-version-probe-2026-07-14/combined_lexicon_elder_probe.eng-gvn.jsonl",
    "sha256": "6e525f16eb7475c6bc5450b978d70db0fbf1a0d71f2af9f143fb80c4bf60c01f",
    "rows": 340
  }
}