{
  "model": "Qwen/Qwen3.6-35B-A3B-FP8",
  "api_base": "http://localhost:8000/v1",
  "timestamp": "2026-05-05T12:54:26.166412",
  "total_evaluated": 1200,
  "summary": {
    "total_episodes": 1200,
    "detection_accuracy": 0.6066666666666667,
    "miss_rate": 0.5490909090909091,
    "false_alarm_rate": 0.26153846153846155,
    "precision": 0.5933014354066986,
    "recall": 0.4509090909090909,
    "f1_security": 0.5123966942148761,
    "threat_type_accuracy": 0.4637096774193548,
    "parse_failure_rate": 0.0,
    "breakdown": {
      "TP": {
        "total": 550,
        "correct": 248,
        "accuracy": 0.4509090909090909
      },
      "FP": {
        "total": 400,
        "correct": 282,
        "accuracy": 0.705
      },
      "TN": {
        "total": 250,
        "correct": 198,
        "accuracy": 0.792
      }
    },
    "per_sq": {
      "SQ1": {
        "total": 190,
        "accuracy": 0.6789473684210526,
        "miss_count": 53,
        "false_alarm_count": 8
      },
      "SQ2": {
        "total": 240,
        "accuracy": 0.5958333333333333,
        "miss_count": 70,
        "false_alarm_count": 27
      },
      "SQ3": {
        "total": 290,
        "accuracy": 0.503448275862069,
        "miss_count": 67,
        "false_alarm_count": 77
      },
      "SQ4": {
        "total": 290,
        "accuracy": 0.596551724137931,
        "miss_count": 69,
        "false_alarm_count": 48
      },
      "SQ5": {
        "total": 190,
        "accuracy": 0.7210526315789474,
        "miss_count": 43,
        "false_alarm_count": 10
      }
    }
  },
  "errors": {
    "total_errors": 605,
    "error_distribution": {
      "FALSE_ALARM": 170,
      "MISS": 302,
      "WRONG_TYPE": 133
    },
    "error_by_sq": {
      "SQ1": {
        "FALSE_ALARM": 8,
        "MISS": 53
      },
      "SQ2": {
        "FALSE_ALARM": 27,
        "MISS": 70,
        "WRONG_TYPE": 23
      },
      "SQ3": {
        "FALSE_ALARM": 77,
        "MISS": 67,
        "WRONG_TYPE": 50
      },
      "SQ4": {
        "FALSE_ALARM": 48,
        "MISS": 69,
        "WRONG_TYPE": 31
      },
      "SQ5": {
        "FALSE_ALARM": 10,
        "MISS": 43,
        "WRONG_TYPE": 29
      }
    },
    "error_by_category": {
      "device_fault": 53,
      "fire_gas": 63,
      "water_damage": 33,
      "intrusion": 137,
      "behavioral_anomaly": 85,
      "child_specific": 29,
      "elderly_specific": 35
    }
  },
  "pipeline": "EGPv2.1"
}