{
  "model": "Qwen/Qwen3.5-9B",
  "api_base": "http://localhost:8000/v1",
  "timestamp": "2026-05-04T14:48:06.999403",
  "total_evaluated": 1200,
  "summary": {
    "total_episodes": 1200,
    "detection_accuracy": 0.5875,
    "miss_rate": 0.38727272727272727,
    "false_alarm_rate": 0.42,
    "precision": 0.5524590163934426,
    "recall": 0.6127272727272727,
    "f1_security": 0.5810344827586207,
    "threat_type_accuracy": 0.5459940652818991,
    "parse_failure_rate": 0.013333333333333334,
    "breakdown": {
      "TP": {
        "total": 550,
        "correct": 337,
        "accuracy": 0.6127272727272727
      },
      "FP": {
        "total": 400,
        "correct": 226,
        "accuracy": 0.565
      },
      "TN": {
        "total": 250,
        "correct": 142,
        "accuracy": 0.568
      }
    },
    "per_sq": {
      "SQ1": {
        "total": 190,
        "accuracy": 0.5736842105263158,
        "miss_count": 41,
        "false_alarm_count": 39
      },
      "SQ2": {
        "total": 240,
        "accuracy": 0.6125,
        "miss_count": 43,
        "false_alarm_count": 49
      },
      "SQ3": {
        "total": 290,
        "accuracy": 0.4896551724137931,
        "miss_count": 80,
        "false_alarm_count": 67
      },
      "SQ4": {
        "total": 290,
        "accuracy": 0.5448275862068965,
        "miss_count": 39,
        "false_alarm_count": 89
      },
      "SQ5": {
        "total": 190,
        "accuracy": 0.7842105263157895,
        "miss_count": 10,
        "false_alarm_count": 29
      }
    }
  },
  "errors": {
    "total_errors": 648,
    "error_distribution": {
      "FALSE_ALARM": 273,
      "PARSE_FAIL": 14,
      "MISS": 208,
      "WRONG_TYPE": 153
    },
    "error_by_sq": {
      "SQ1": {
        "FALSE_ALARM": 39,
        "PARSE_FAIL": 2,
        "MISS": 40
      },
      "SQ2": {
        "FALSE_ALARM": 49,
        "PARSE_FAIL": 2,
        "MISS": 42,
        "WRONG_TYPE": 28
      },
      "SQ3": {
        "FALSE_ALARM": 67,
        "PARSE_FAIL": 3,
        "MISS": 78,
        "WRONG_TYPE": 39
      },
      "SQ4": {
        "FALSE_ALARM": 89,
        "PARSE_FAIL": 5,
        "MISS": 38,
        "WRONG_TYPE": 50
      },
      "SQ5": {
        "FALSE_ALARM": 29,
        "PARSE_FAIL": 2,
        "WRONG_TYPE": 36,
        "MISS": 10
      }
    },
    "error_by_category": {
      "device_fault": 40,
      "fire_gas": 53,
      "water_damage": 37,
      "intrusion": 81,
      "behavioral_anomaly": 79,
      "child_specific": 28,
      "elderly_specific": 43
    }
  },
  "pipeline": "EGP"
}