{
  "schemaVersion": 1,
  "measuredAt": "2026-09-24T13:34:28Z",
  "machine": {
    "chip": "Apple M5 Max",
    "ramGb": 128,
    "os": "macOS 26.6.1",
    "arch": "arm64",
    "python": "3.12.13",
    "loadAverageAtStart": [
      35.6,
      47.3,
      46.7
    ],
    "loadAverageAtEnd": [
      43.9,
      47.3,
      46.8
    ]
  },
  "models": [
    {
      "id": "laya-typed-decisions",
      "source": "https://huggingface.co/convaiinnovations/laya/tree/55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851/typed-decisions",
      "sha": "55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851",
      "license": "Apache-2.0",
      "backend": "mlx 0.32.2 (Device(gpu, 0)), float32"
    },
    {
      "id": "jev",
      "source": "https://jev-ai.pro/api/v1/systemone (hosted, TypeSafe-compatible)",
      "sha": null,
      "license": "proprietary, hosted",
      "backend": "hosted API, requested model jev-1.13.0"
    }
  ],
  "corpus": {
    "rows": 48,
    "sha256": "6e7c1a4d1a12116ac2b45ccb30ae9fc46d4048dd664debd38bd6c86f0c68e029",
    "path": "experiments/laya-vs-jev/corpus.json",
    "byType": {
      "choice": 16,
      "score": 16,
      "noul": 16
    }
  },
  "perModel": {
    "laya-typed-decisions": {
      "accuracy": 0.8125,
      "calibratedAccuracy": 1.0,
      "p50Ms": 8.71,
      "p90Ms": 10.46,
      "maxMs": 71.77,
      "coldLoadS": 0.805,
      "costPer1kUsd": null,
      "detail": {
        "correct": 39,
        "answered": 48,
        "rows": 48,
        "highConfidenceAnswers": 2,
        "highConfidenceCorrect": 2,
        "byType": {
          "choice": {
            "correct": 15,
            "rows": 16
          },
          "score": {
            "correct": 11,
            "rows": 16
          },
          "noul": {
            "correct": 13,
            "rows": 16
          }
        },
        "firstCallMs": 38.4
      }
    },
    "jev": {
      "accuracy": 0.9583,
      "calibratedAccuracy": 1.0,
      "p50Ms": 378.43,
      "p90Ms": 828.31,
      "maxMs": 1176.36,
      "coldLoadS": null,
      "costPer1kUsd": 0.084463,
      "detail": {
        "correct": 46,
        "answered": 48,
        "rows": 48,
        "highConfidenceAnswers": 36,
        "highConfidenceCorrect": 36,
        "byType": {
          "choice": {
            "correct": 16,
            "rows": 16
          },
          "score": {
            "correct": 14,
            "rows": 16
          },
          "noul": {
            "correct": 16,
            "rows": 16
          }
        },
        "meanInputTokens": 349.0,
        "failures": {},
        "servedModels": [
          "jev-1.13.0"
        ]
      }
    }
  },
  "agreement": {
    "laya_vs_jev": {
      "percent": 77.08,
      "agree": 37,
      "comparedRows": 48
    }
  },
  "parity": {
    "rows": 48,
    "sameDecision": 48,
    "maxTopProbDelta": 0.002097,
    "maxAnyProbDelta": 0.002097,
    "threshold": 0.01,
    "mismatches": [],
    "passed": true,
    "note": "Reference probabilities are the upstream system_one output, which rounds to 4 decimals; rounding contributes at most 0.00005 to each delta.",
    "status": "passed",
    "reference": {
      "implementation": "convaiinnovations/laya@55cf4c4ebb4e rl_agent_api.RLAgent.system_one",
      "device": "cpu",
      "torch": "2.14.0",
      "transformers": "5.17.0",
      "threads": 6,
      "coldLoadS": 62.73,
      "p50Ms": 155.82,
      "p90Ms": 411.43,
      "maxMs": 538.0
    }
  },
  "weights": {
    "repo": "convaiinnovations/laya",
    "sha": "55cf4c4ebb4ebe31b2550e8bdf3bd21b99753851",
    "subdir": "typed-decisions",
    "fileSha256": {
      "typed-decisions/model.safetensors": "4fa56de72383a9d3efa9cfa78955733c81b9fc8067a587ca4beb82c78107a24e",
      "typed-decisions/rl_agent_config.json": "ebf0cd524d92342a6be5e48e9fca3d7c2babfb5a56ccd79d2171ef5d8c7f7be8",
      "typed-decisions/encoder/config.json": "5268d24ad3b77c8151de5dcb0762ba4391619aad9ab0bda33e36fb083cfeae6d",
      "typed-decisions/tokenizer/tokenizer.json": "6c8aaa9a542084f2457eab775d4eeb51f92a70c0fd9de28d5edb0ddec3c08d30",
      "typed-decisions/tokenizer/tokenizer_config.json": "08d4cf3ac4dca381759441b85b91a6d40e688471dcd33d15d6649eb0a9a854d1",
      "rl_agent_api.py": "be3b46819c9999c3ef88e0f2ecf6d3ab1cdfed1d9a8b466fc89811e34d44031b",
      "rl_common.py": "8d83611d480c971d640a7b7d3aa2f2219c5e8455e9cc2329fd073681bd8be23e"
    }
  },
  "runtime": {
    "mlx": "0.32.2",
    "numpy": "2.5.3",
    "python": "3.12.13"
  },
  "jevStatus": "run",
  "rows": [
    {
      "id": "c01",
      "type": "choice",
      "expected": "billing",
      "laya": {
        "decision": "billing",
        "topProb": 0.7591338924655915,
        "confidence": 0.41332557285797955,
        "confidenceSource": "reported",
        "ms": 9.52,
        "tokens": 65
      },
      "reference": {
        "decision": "billing",
        "topProb": 0.7595,
        "confidence": 0.4139,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "billing",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 1176.36,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 366
      }
    },
    {
      "id": "c02",
      "type": "choice",
      "expected": "technical",
      "laya": {
        "decision": "technical",
        "topProb": 0.7316081688493244,
        "confidence": 0.3973856469682404,
        "confidenceSource": "reported",
        "ms": 9.62,
        "tokens": 68
      },
      "reference": {
        "decision": "technical",
        "topProb": 0.7319,
        "confidence": 0.3978,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "technical",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 403.51,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 371
      }
    },
    {
      "id": "c03",
      "type": "choice",
      "expected": "sales",
      "laya": {
        "decision": "sales",
        "topProb": 0.4448032433273546,
        "confidence": 0.10552645711877995,
        "confidenceSource": "reported",
        "ms": 9.8,
        "tokens": 70
      },
      "reference": {
        "decision": "sales",
        "topProb": 0.4469,
        "confidence": 0.1066,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "sales",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 348.36,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 373
      }
    },
    {
      "id": "c04",
      "type": "choice",
      "expected": "account_access",
      "laya": {
        "decision": "account_access",
        "topProb": 0.6189163731718973,
        "confidence": 0.24490415883622574,
        "confidenceSource": "reported",
        "ms": 9.03,
        "tokens": 71
      },
      "reference": {
        "decision": "account_access",
        "topProb": 0.6182,
        "confidence": 0.2441,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "account_access",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 401.82,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 375
      }
    },
    {
      "id": "c05",
      "type": "choice",
      "expected": "cancellation",
      "laya": {
        "decision": "cancellation",
        "topProb": 0.8688251363720237,
        "confidence": 0.619034045356931,
        "confidenceSource": "reported",
        "ms": 8.73,
        "tokens": 67
      },
      "reference": {
        "decision": "cancellation",
        "topProb": 0.8688,
        "confidence": 0.619,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "cancellation",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 778.69,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 371
      }
    },
    {
      "id": "c06",
      "type": "choice",
      "expected": "data_export",
      "laya": {
        "decision": "data_export",
        "topProb": 0.6802907301992698,
        "confidence": 0.30439528249582115,
        "confidenceSource": "reported",
        "ms": 8.4,
        "tokens": 66
      },
      "reference": {
        "decision": "data_export",
        "topProb": 0.6801,
        "confidence": 0.3042,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "data_export",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 365.93,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 370
      }
    },
    {
      "id": "c07",
      "type": "choice",
      "expected": "docs",
      "laya": {
        "decision": "docs",
        "topProb": 0.37177649754785597,
        "confidence": 0.051005899135757704,
        "confidenceSource": "reported",
        "ms": 9.08,
        "tokens": 78
      },
      "reference": {
        "decision": "docs",
        "topProb": 0.372,
        "confidence": 0.0511,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "docs",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 835.1,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 375
      }
    },
    {
      "id": "c08",
      "type": "choice",
      "expected": "dependency",
      "laya": {
        "decision": "dependency",
        "topProb": 0.45309912690719706,
        "confidence": 0.07745798800674286,
        "confidenceSource": "reported",
        "ms": 8.68,
        "tokens": 75
      },
      "reference": {
        "decision": "dependency",
        "topProb": 0.4539,
        "confidence": 0.0781,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "dependency",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 925.46,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 380
      }
    },
    {
      "id": "c09",
      "type": "choice",
      "expected": "bugfix",
      "laya": {
        "decision": "bugfix",
        "topProb": 0.6884816539652869,
        "confidence": 0.31303901291185454,
        "confidenceSource": "reported",
        "ms": 8.82,
        "tokens": 73
      },
      "reference": {
        "decision": "bugfix",
        "topProb": 0.6894,
        "confidence": 0.3142,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "bugfix",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 833.95,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 371
      }
    },
    {
      "id": "c10",
      "type": "choice",
      "expected": "feature",
      "laya": {
        "decision": "feature",
        "topProb": 0.41584165240433424,
        "confidence": 0.07524677438732263,
        "confidenceSource": "reported",
        "ms": 9.24,
        "tokens": 62
      },
      "reference": {
        "decision": "feature",
        "topProb": 0.4158,
        "confidence": 0.0753,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "feature",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 790.16,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 363
      }
    },
    {
      "id": "c11",
      "type": "choice",
      "expected": "cooking",
      "laya": {
        "decision": "cooking",
        "topProb": 0.8161247988965661,
        "confidence": 0.5106832324739915,
        "confidenceSource": "reported",
        "ms": 9.05,
        "tokens": 57
      },
      "reference": {
        "decision": "cooking",
        "topProb": 0.8163,
        "confidence": 0.511,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "cooking",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 760.66,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 366
      }
    },
    {
      "id": "c12",
      "type": "choice",
      "expected": "sport",
      "laya": {
        "decision": "sport",
        "topProb": 0.8829538478425022,
        "confidence": 0.6469258632784647,
        "confidenceSource": "reported",
        "ms": 8.31,
        "tokens": 57
      },
      "reference": {
        "decision": "sport",
        "topProb": 0.8831,
        "confidence": 0.6473,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "sport",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 352.68,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 362
      }
    },
    {
      "id": "c13",
      "type": "choice",
      "expected": "finance",
      "laya": {
        "decision": "finance",
        "topProb": 0.8990096050848622,
        "confidence": 0.6839142994955683,
        "confidenceSource": "reported",
        "ms": 8.68,
        "tokens": 58
      },
      "reference": {
        "decision": "finance",
        "topProb": 0.8992,
        "confidence": 0.6843,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "finance",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 375.24,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 361
      }
    },
    {
      "id": "c14",
      "type": "choice",
      "expected": "travel",
      "laya": {
        "decision": "travel",
        "topProb": 0.8232304365429647,
        "confidence": 0.5262762799452873,
        "confidenceSource": "reported",
        "ms": 8.82,
        "tokens": 60
      },
      "reference": {
        "decision": "travel",
        "topProb": 0.8232,
        "confidence": 0.5261,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "travel",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 340.65,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 367
      }
    },
    {
      "id": "c15",
      "type": "choice",
      "expected": "test_failure",
      "laya": {
        "decision": "test_failure",
        "topProb": 0.5806502243077953,
        "confidence": 0.20657895624392209,
        "confidenceSource": "reported",
        "ms": 18.37,
        "tokens": 285
      },
      "reference": {
        "decision": "test_failure",
        "topProb": 0.581,
        "confidence": 0.2073,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "test_failure",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 374.62,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 592
      }
    },
    {
      "id": "c16",
      "type": "choice",
      "expected": "network",
      "laya": {
        "decision": "test_failure",
        "topProb": 0.38089800498930176,
        "confidence": 0.0572675053268753,
        "confidenceSource": "reported",
        "ms": 71.77,
        "tokens": 288
      },
      "reference": {
        "decision": "test_failure",
        "topProb": 0.3814,
        "confidence": 0.0574,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": "network",
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 378.48,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 588
      }
    },
    {
      "id": "s01",
      "type": "score",
      "expected": 2,
      "laya": {
        "decision": 2,
        "topProb": 0.7341656782673387,
        "confidence": 0.42577704347754297,
        "confidenceSource": "reported",
        "ms": 8.87,
        "tokens": 53
      },
      "reference": {
        "decision": 2,
        "topProb": 0.7343,
        "confidence": 0.426,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 2,
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 762.46,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 321
      }
    },
    {
      "id": "s02",
      "type": "score",
      "expected": 0,
      "laya": {
        "decision": 0,
        "topProb": 0.5098066559663468,
        "confidence": 0.07054873829079278,
        "confidenceSource": "reported",
        "ms": 9.31,
        "tokens": 58
      },
      "reference": {
        "decision": 0,
        "topProb": 0.5108,
        "confidence": 0.071,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 0,
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 340.59,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 326
      }
    },
    {
      "id": "s03",
      "type": "score",
      "expected": 1,
      "laya": {
        "decision": 1,
        "topProb": 0.46636021955145324,
        "confidence": 0.14482092005659752,
        "confidenceSource": "reported",
        "ms": 9.6,
        "tokens": 55
      },
      "reference": {
        "decision": 1,
        "topProb": 0.4664,
        "confidence": 0.1448,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 2,
        "topProb": 0.69,
        "confidence": 0.53,
        "confidenceSource": "reported",
        "ms": 382.96,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 323
      }
    },
    {
      "id": "s04",
      "type": "score",
      "expected": 0,
      "laya": {
        "decision": 0,
        "topProb": 0.9678830645196498,
        "confidence": 0.8723766670494076,
        "confidenceSource": "reported",
        "ms": 9.43,
        "tokens": 57
      },
      "reference": {
        "decision": 0,
        "topProb": 0.9679,
        "confidence": 0.8725,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 0,
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 315.05,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 326
      }
    },
    {
      "id": "s05",
      "type": "score",
      "expected": 3,
      "laya": {
        "decision": 3,
        "topProb": 0.6964760520043237,
        "confidence": 0.43190675411129986,
        "confidenceSource": "reported",
        "ms": 8.33,
        "tokens": 55
      },
      "reference": {
        "decision": 3,
        "topProb": 0.6969,
        "confidence": 0.4327,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 3,
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 363.08,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 325
      }
    },
    {
      "id": "s06",
      "type": "score",
      "expected": 2,
      "laya": {
        "decision": 2,
        "topProb": 0.4009655164149344,
        "confidence": 0.15852182214470179,
        "confidenceSource": "reported",
        "ms": 9.07,
        "tokens": 60
      },
      "reference": {
        "decision": 2,
        "topProb": 0.4014,
        "confidence": 0.1589,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 2,
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 787.41,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 329
      }
    },
    {
      "id": "s07",
      "type": "score",
      "expected": 1,
      "laya": {
        "decision": 1,
        "topProb": 0.4713206118893929,
        "confidence": 0.2586716134506597,
        "confidenceSource": "reported",
        "ms": 8.18,
        "tokens": 56
      },
      "reference": {
        "decision": 1,
        "topProb": 0.4716,
        "confidence": 0.2588,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 1,
        "topProb": 0.68,
        "confidence": 0.67,
        "confidenceSource": "reported",
        "ms": 387.95,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 326
      }
    },
    {
      "id": "s08",
      "type": "score",
      "expected": 0,
      "laya": {
        "decision": 2,
        "topProb": 0.3529342188100105,
        "confidence": 0.04788018761524626,
        "confidenceSource": "reported",
        "ms": 8.84,
        "tokens": 56
      },
      "reference": {
        "decision": 2,
        "topProb": 0.3529,
        "confidence": 0.048,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 0,
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 319.21,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 327
      }
    },
    {
      "id": "s09",
      "type": "score",
      "expected": 3,
      "laya": {
        "decision": 2,
        "topProb": 0.3705868008525366,
        "confidence": 0.06845622796903283,
        "confidenceSource": "reported",
        "ms": 8.49,
        "tokens": 66
      },
      "reference": {
        "decision": 2,
        "topProb": 0.3705,
        "confidence": 0.0685,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 3,
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 342.22,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 338
      }
    },
    {
      "id": "s10",
      "type": "score",
      "expected": 1,
      "laya": {
        "decision": 2,
        "topProb": 0.3899733663879514,
        "confidence": 0.116098727259209,
        "confidenceSource": "reported",
        "ms": 9.36,
        "tokens": 63
      },
      "reference": {
        "decision": 2,
        "topProb": 0.3901,
        "confidence": 0.1163,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 1,
        "topProb": 0.52,
        "confidence": 0.52,
        "confidenceSource": "reported",
        "ms": 326.86,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 331
      }
    },
    {
      "id": "s11",
      "type": "score",
      "expected": 2,
      "laya": {
        "decision": 2,
        "topProb": 0.4286091927162839,
        "confidence": 0.05436229638210399,
        "confidenceSource": "reported",
        "ms": 8.36,
        "tokens": 44
      },
      "reference": {
        "decision": 2,
        "topProb": 0.429,
        "confidence": 0.0545,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 2,
        "topProb": 1.0,
        "confidence": 0.99,
        "confidenceSource": "reported",
        "ms": 303.58,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 314
      }
    },
    {
      "id": "s12",
      "type": "score",
      "expected": 0,
      "laya": {
        "decision": 1,
        "topProb": 0.42805178037810276,
        "confidence": 0.02870472288338244,
        "confidenceSource": "reported",
        "ms": 8.59,
        "tokens": 46
      },
      "reference": {
        "decision": 1,
        "topProb": 0.4284,
        "confidence": 0.0287,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 0,
        "topProb": 0.99,
        "confidence": 0.98,
        "confidenceSource": "reported",
        "ms": 413.4,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 316
      }
    },
    {
      "id": "s13",
      "type": "score",
      "expected": 1,
      "laya": {
        "decision": 1,
        "topProb": 0.3901948454321524,
        "confidence": 0.04059721624527313,
        "confidenceSource": "reported",
        "ms": 9.26,
        "tokens": 61
      },
      "reference": {
        "decision": 1,
        "topProb": 0.391,
        "confidence": 0.041,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 2,
        "topProb": 0.65,
        "confidence": 0.65,
        "confidenceSource": "reported",
        "ms": 429.85,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 332
      }
    },
    {
      "id": "s14",
      "type": "score",
      "expected": 3,
      "laya": {
        "decision": 3,
        "topProb": 0.46816123523368186,
        "confidence": 0.20583982031499182,
        "confidenceSource": "reported",
        "ms": 8.41,
        "tokens": 58
      },
      "reference": {
        "decision": 3,
        "topProb": 0.4683,
        "confidence": 0.206,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 3,
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 358.11,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 328
      }
    },
    {
      "id": "s15",
      "type": "score",
      "expected": 2,
      "laya": {
        "decision": 2,
        "topProb": 0.7901166414013114,
        "confidence": 0.4514030559893971,
        "confidenceSource": "reported",
        "ms": 11.99,
        "tokens": 193
      },
      "reference": {
        "decision": 2,
        "topProb": 0.7904,
        "confidence": 0.4519,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 2,
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 863.53,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 471
      }
    },
    {
      "id": "s16",
      "type": "score",
      "expected": 0,
      "laya": {
        "decision": 1,
        "topProb": 0.5483554361104005,
        "confidence": 0.1909403130361933,
        "confidenceSource": "reported",
        "ms": 10.89,
        "tokens": 155
      },
      "reference": {
        "decision": 1,
        "topProb": 0.5487,
        "confidence": 0.191,
        "confidenceSource": "reported"
      },
      "jev": {
        "decision": 0,
        "topProb": 1.0,
        "confidence": 1.0,
        "confidenceSource": "reported",
        "ms": 348.26,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 426
      }
    },
    {
      "id": "n01",
      "type": "noul",
      "expected": true,
      "laya": {
        "decision": true,
        "topProb": 0.751625728364079,
        "confidence": 0.19130877687067427,
        "confidenceSource": "1-normalized-entropy",
        "ms": 8.31,
        "tokens": 54
      },
      "reference": {
        "decision": true,
        "topProb": 0.7519,
        "confidence": 0.19174721619680957,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": true,
        "topProb": 0.98,
        "confidence": 0.8585594574581792,
        "confidenceSource": "1-normalized-entropy",
        "ms": 325.44,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 294
      }
    },
    {
      "id": "n02",
      "type": "noul",
      "expected": false,
      "laya": {
        "decision": false,
        "topProb": 0.9994895203392229,
        "confidence": 0.9936811879913687,
        "confidenceSource": "1-normalized-entropy",
        "ms": 8.04,
        "tokens": 51
      },
      "reference": {
        "decision": false,
        "topProb": 0.9995,
        "confidence": 0.9937959407041683,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": false,
        "topProb": 0.99,
        "confidence": 0.9192068641040888,
        "confidenceSource": "1-normalized-entropy",
        "ms": 749.74,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 291
      }
    },
    {
      "id": "n03",
      "type": "noul",
      "expected": true,
      "laya": {
        "decision": false,
        "topProb": 0.8351702274631619,
        "confidence": 0.35426029073015575,
        "confidenceSource": "1-normalized-entropy",
        "ms": 8.02,
        "tokens": 53
      },
      "reference": {
        "decision": false,
        "topProb": 0.8356,
        "confidence": 0.355267396925095,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": true,
        "topProb": 0.96,
        "confidence": 0.7577078109175851,
        "confidenceSource": "1-normalized-entropy",
        "ms": 404.96,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 293
      }
    },
    {
      "id": "n04",
      "type": "noul",
      "expected": false,
      "laya": {
        "decision": false,
        "topProb": 0.8306296832842431,
        "confidence": 0.3437376990278531,
        "confidenceSource": "1-normalized-entropy",
        "ms": 8.13,
        "tokens": 51
      },
      "reference": {
        "decision": false,
        "topProb": 0.8311,
        "confidence": 0.344817752050641,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": false,
        "topProb": 0.94,
        "confidence": 0.6725550808455236,
        "confidenceSource": "1-normalized-entropy",
        "ms": 369.9,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 291
      }
    },
    {
      "id": "n05",
      "type": "noul",
      "expected": true,
      "laya": {
        "decision": true,
        "topProb": 0.6396757898094683,
        "confidence": 0.05704802101841677,
        "confidenceSource": "1-normalized-entropy",
        "ms": 8.15,
        "tokens": 58
      },
      "reference": {
        "decision": true,
        "topProb": 0.6404,
        "confidence": 0.05764934176095671,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": true,
        "topProb": 0.99,
        "confidence": 0.9192068641040888,
        "confidenceSource": "1-normalized-entropy",
        "ms": 384.31,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 325
      }
    },
    {
      "id": "n06",
      "type": "noul",
      "expected": false,
      "laya": {
        "decision": false,
        "topProb": 0.9660439266921953,
        "confidence": 0.7861410706589302,
        "confidenceSource": "1-normalized-entropy",
        "ms": 8.06,
        "tokens": 56
      },
      "reference": {
        "decision": false,
        "topProb": 0.9662,
        "confidence": 0.786895495383761,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": false,
        "topProb": 0.94,
        "confidence": 0.6725550808455236,
        "confidenceSource": "1-normalized-entropy",
        "ms": 825.89,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 328
      }
    },
    {
      "id": "n07",
      "type": "noul",
      "expected": true,
      "laya": {
        "decision": true,
        "topProb": 0.6880787805682438,
        "confidence": 0.10462125751421714,
        "confidenceSource": "1-normalized-entropy",
        "ms": 8.14,
        "tokens": 48
      },
      "reference": {
        "decision": true,
        "topProb": 0.6882,
        "confidence": 0.10475966580011042,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": true,
        "topProb": 0.99,
        "confidence": 0.9192068641040888,
        "confidenceSource": "1-normalized-entropy",
        "ms": 381.11,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 288
      }
    },
    {
      "id": "n08",
      "type": "noul",
      "expected": false,
      "laya": {
        "decision": false,
        "topProb": 0.9999026871120789,
        "confidence": 0.9985627242199436,
        "confidenceSource": "1-normalized-entropy",
        "ms": 7.9,
        "tokens": 47
      },
      "reference": {
        "decision": false,
        "topProb": 0.9999,
        "confidence": 0.9985269664716718,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": false,
        "topProb": 0.99,
        "confidence": 0.9192068641040888,
        "confidenceSource": "1-normalized-entropy",
        "ms": 378.37,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 287
      }
    },
    {
      "id": "n09",
      "type": "noul",
      "expected": true,
      "laya": {
        "decision": false,
        "topProb": 0.6904978077426528,
        "confidence": 0.10740201107040759,
        "confidenceSource": "1-normalized-entropy",
        "ms": 8.28,
        "tokens": 58
      },
      "reference": {
        "decision": false,
        "topProb": 0.6909000000000001,
        "confidence": 0.10786817002315141,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": true,
        "topProb": 0.9,
        "confidence": 0.5310044064107189,
        "confidenceSource": "1-normalized-entropy",
        "ms": 401.65,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 298
      }
    },
    {
      "id": "n10",
      "type": "noul",
      "expected": false,
      "laya": {
        "decision": false,
        "topProb": 0.7587729824044718,
        "confidence": 0.2029251805325053,
        "confidenceSource": "1-normalized-entropy",
        "ms": 8.22,
        "tokens": 54
      },
      "reference": {
        "decision": false,
        "topProb": 0.759,
        "confidence": 0.2033007066096726,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": false,
        "topProb": 0.92,
        "confidence": 0.5978208097977271,
        "confidenceSource": "1-normalized-entropy",
        "ms": 386.58,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 294
      }
    },
    {
      "id": "n11",
      "type": "noul",
      "expected": true,
      "laya": {
        "decision": false,
        "topProb": 0.5674211339670663,
        "confidence": 0.013155894299230475,
        "confidenceSource": "1-normalized-entropy",
        "ms": 8.04,
        "tokens": 46
      },
      "reference": {
        "decision": false,
        "topProb": 0.5674,
        "confidence": 0.013147622578773888,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": true,
        "topProb": 0.93,
        "confidence": 0.634076349099777,
        "confidenceSource": "1-normalized-entropy",
        "ms": 339.2,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 286
      }
    },
    {
      "id": "n12",
      "type": "noul",
      "expected": false,
      "laya": {
        "decision": false,
        "topProb": 0.8499407254010768,
        "confidence": 0.3900113804554042,
        "confidenceSource": "1-normalized-entropy",
        "ms": 8.27,
        "tokens": 41
      },
      "reference": {
        "decision": false,
        "topProb": 0.8503000000000001,
        "confidence": 0.39091095485208727,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": false,
        "topProb": 0.99,
        "confidence": 0.9192068641040888,
        "confidenceSource": "1-normalized-entropy",
        "ms": 357.1,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 281
      }
    },
    {
      "id": "n13",
      "type": "noul",
      "expected": true,
      "laya": {
        "decision": true,
        "topProb": 0.7171702893943899,
        "confidence": 0.140722102709136,
        "confidenceSource": "1-normalized-entropy",
        "ms": 8.18,
        "tokens": 57
      },
      "reference": {
        "decision": true,
        "topProb": 0.7175,
        "confidence": 0.14116508699737174,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": true,
        "topProb": 0.99,
        "confidence": 0.9192068641040888,
        "confidenceSource": "1-normalized-entropy",
        "ms": 359.72,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 297
      }
    },
    {
      "id": "n14",
      "type": "noul",
      "expected": false,
      "laya": {
        "decision": false,
        "topProb": 0.9656931739168734,
        "confidence": 0.7844495094191678,
        "confidenceSource": "1-normalized-entropy",
        "ms": 8.07,
        "tokens": 51
      },
      "reference": {
        "decision": false,
        "topProb": 0.9657,
        "confidence": 0.7844823780056923,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": false,
        "topProb": 0.99,
        "confidence": 0.9192068641040888,
        "confidenceSource": "1-normalized-entropy",
        "ms": 349.61,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 291
      }
    },
    {
      "id": "n15",
      "type": "noul",
      "expected": true,
      "laya": {
        "decision": true,
        "topProb": 0.7375593534513174,
        "confidence": 0.16959294659296786,
        "confidenceSource": "1-normalized-entropy",
        "ms": 11.42,
        "tokens": 172
      },
      "reference": {
        "decision": true,
        "topProb": 0.7378,
        "confidence": 0.16995191061473192,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": true,
        "topProb": 0.99,
        "confidence": 0.9192068641040888,
        "confidenceSource": "1-normalized-entropy",
        "ms": 325.86,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 406
      }
    },
    {
      "id": "n16",
      "type": "noul",
      "expected": false,
      "laya": {
        "decision": false,
        "topProb": 0.6663463213981842,
        "confidence": 0.08138415373832364,
        "confidenceSource": "1-normalized-entropy",
        "ms": 10.28,
        "tokens": 156
      },
      "reference": {
        "decision": false,
        "topProb": 0.6674,
        "confidence": 0.08243924558097626,
        "confidenceSource": "1-normalized-entropy"
      },
      "jev": {
        "decision": false,
        "topProb": 0.98,
        "confidence": 0.8585594574581793,
        "confidenceSource": "1-normalized-entropy",
        "ms": 312.3,
        "status": "ok",
        "servedModel": "jev-1.13.0",
        "inputTokens": 393
      }
    }
  ],
  "notes": [
    "Backend: every model component (ModernBERT-large encoder, decision head, scorer, act head) runs in MLX on the Apple GPU in float32; no component falls back to torch-MPS. Tokenization uses the Hugging Face `tokenizers` library (Rust, CPU), which is not a model component.",
    "coldLoadS = reading and upcasting the pinned safetensors plus the tokenizer, after Python imports and after the one-off download. The first call (38 ms, Metal kernel compilation) is excluded from p50/p90/max, which cover one call per corpus row, batch size 1, sequential.",
    "costPer1kUsd is null for Laya: it runs locally and no per-call price applies; electricity and hardware were not measured.",
    "Jev: 48 sequential calls, 15 s timeout, requested jev-1.13.0; served-model labels seen: jev-1.13.0. Latency is wall time from this Mac over the public internet, including failed calls. Accuracy counts an unanswered row as wrong.",
    "Jev costPer1kUsd = measured mean input tokens per call x $0.242 per 1M input tokens, a vendor price CLAIM (jev-ai.pro/pricing, read 2026-09-24, highest listed annual-plan rate; listed range $0.124-$0.242; output tokens free). It is not an invoice.",
    "calibratedAccuracy = accuracy of the answers whose confidence is >= 0.9 (confidence = the model's reported value; for Laya noul answers, which carry none, 1 - normalized entropy of [1-p, p]).",
    "Decisions: choice = top key; score = most probable level; noul = probability >= 0.5.",
    "Corpus: 48 hand-written public rows authored before any model ran; the expected answers are the author's judgment. A small authored corpus is not a benchmark; it supports mechanism and parity claims, not general accuracy claims.",
    "Vendor figures (Laya eval/results.md, jev-ai.pro/compare) are CLAIMS and are not reproduced here.",
    "The Mac was shared with other work during the run (load average [35.6, 47.3, 46.7] at start, [43.9, 47.3, 46.8] at end); latencies are under that load."
  ]
}
