{
  "purpose": "Mechanics-only smoke validation, not improved released weights or generalization",
  "adopted": false,
  "training": {
    "steps": 6,
    "seed": 42,
    "learningRate": 2e-05,
    "trainRecords": 28,
    "trainDomains": [
      "garden",
      "library",
      "museum",
      "workshop"
    ],
    "changedAdapterTensors": 240,
    "backboneAndReadoutFrozen": true,
    "seconds": 53.129103966
  },
  "developmentBefore": {
    "questions": 12,
    "answered": 12,
    "errors": 0,
    "correct": 12,
    "accuracyAllQuestions": 1.0,
    "brierAnswered": 0.0010257946398990957,
    "nllAnswered": 0.024988250780074998,
    "ece10Answered": 0.024658157503599232,
    "confidentWrongAt09": 0,
    "coverageByFixedThreshold": [
      {
        "threshold": 0.5,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.7,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.8,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.9,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.95,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.99,
        "accepted": 0,
        "coverage": 0.0,
        "errorRate": null
      }
    ],
    "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here."
  },
  "developmentAfter": {
    "questions": 12,
    "answered": 12,
    "errors": 0,
    "correct": 12,
    "accuracyAllQuestions": 1.0,
    "brierAnswered": 3.058330986685694e-05,
    "nllAnswered": 0.004121613352848992,
    "ece10Answered": 0.0041117529524682395,
    "confidentWrongAt09": 0,
    "coverageByFixedThreshold": [
      {
        "threshold": 0.5,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.7,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.8,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.9,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.95,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.99,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      }
    ],
    "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here."
  },
  "reload": {
    "complete": true,
    "adopted": false,
    "priorStrictGateFailed": true,
    "all240AdapterTensorsBitExact": true,
    "freshRepeat": {
      "maxProbabilityDelta": 0.0,
      "sameAllTopOptions": true,
      "questions": 12
    },
    "crossProcessReload": {
      "maxProbabilityDelta": 0.0,
      "sameAllTopOptions": true,
      "questions": 12
    },
    "originalStrictTolerancePassed": true,
    "calibrationAndTestPassed": true,
    "seconds": 196.0350929440001
  },
  "initialStrictReloadFailure": {
    "maxProbabilityDelta": 0.00048164095202640755,
    "tolerance": 1e-05,
    "failed": true,
    "interpretation": "Observed after compiled/HTTP work in the original process. Cause was not isolated. Preserved rather than erased by the successful fresh-process control."
  },
  "calibration": {
    "records": 4,
    "factor": 0.1353352832366127,
    "factorBounds": [
      0.1353352832366127,
      7.38905609893065
    ],
    "beforeNll": 0.003907485844979199,
    "afterNll": 0.0,
    "interpretation": "In-sample calibration fit; independent test needed before adoption",
    "lowerBoundSelected": true
  },
  "toyTest": {
    "questions": 12,
    "answered": 12,
    "errors": 0,
    "correct": 12,
    "accuracyAllQuestions": 1.0,
    "brierAnswered": 3.9934072114968966e-38,
    "nllAnswered": 0.0,
    "ece10Answered": 0.0,
    "confidentWrongAt09": 0,
    "coverageByFixedThreshold": [
      {
        "threshold": 0.5,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.7,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.8,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.9,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.95,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      },
      {
        "threshold": 0.99,
        "accepted": 12,
        "coverage": 1.0,
        "errorRate": 0.0
      }
    ],
    "coverageInterpretation": "Observed on this dataset; not a future error guarantee. Thresholds not tuned here."
  },
  "inputHashes": {
    "train.jsonl": "cef745ddee9bf2e1d8f3639660441505bfb5aacf2c5cf9de5261a17ac1ba0bb5",
    "development.jsonl": "c7da14026dbfd87b5dac9040c7b6ed33d3007b620aab018dc9c25ca5ce6d9059",
    "calibration.jsonl": "b951c82c4263d6b0ec058b704f1d3d616ae85f483dc3d6dfdad49bed61c7481c",
    "test.jsonl": "00baffcf3d9114d585294dd0955724ad9aff04475d3e9105584c9f1bbfbfbc8a"
  },
  "limitations": [
    "Authored toy templates recur across partitions; this is not an independent cross-domain generalization test",
    "Development was already 12/12 correct before training",
    "Calibration chose its lower search bound on a tiny all-correct sample; this is not evidence of deployment calibration",
    "No trained overlay or fitted temperature is adopted or included in the release"
  ]
}