{
  "model": "Qwen/Qwen3.5-9B",
  "api_base": "http://localhost:8000/v1",
  "timestamp": "2026-05-03T06:31:18.653202",
  "total_evaluated": 1200,
  "avg_latency_seconds": 74.59,
  "api_errors": 1,
  "summary": {
    "total_episodes": 1200,
    "detection_accuracy": 0.5433333333333333,
    "miss_rate": 0.3109090909090909,
    "false_alarm_rate": 0.58,
    "precision": 0.5013227513227513,
    "recall": 0.6890909090909091,
    "f1_security": 0.5803981623277181,
    "threat_type_accuracy": 0.46965699208443273,
    "parse_failure_rate": 0.025,
    "breakdown": {
      "TP": {
        "total": 550,
        "correct": 379,
        "accuracy": 0.6890909090909091
      },
      "FP": {
        "total": 400,
        "correct": 157,
        "accuracy": 0.3925
      },
      "TN": {
        "total": 250,
        "correct": 116,
        "accuracy": 0.464
      }
    },
    "per_sq": {
      "SQ1": {
        "total": 190,
        "accuracy": 0.531578947368421,
        "miss_count": 27,
        "false_alarm_count": 62
      },
      "SQ2": {
        "total": 240,
        "accuracy": 0.5708333333333333,
        "miss_count": 29,
        "false_alarm_count": 74
      },
      "SQ3": {
        "total": 290,
        "accuracy": 0.5103448275862069,
        "miss_count": 81,
        "false_alarm_count": 61
      },
      "SQ4": {
        "total": 290,
        "accuracy": 0.45517241379310347,
        "miss_count": 17,
        "false_alarm_count": 141
      },
      "SQ5": {
        "total": 190,
        "accuracy": 0.7052631578947368,
        "miss_count": 17,
        "false_alarm_count": 39
      }
    }
  },
  "errors": {
    "total_errors": 749,
    "error_distribution": {
      "FALSE_ALARM": 374,
      "MISS": 155,
      "PARSE_FAIL": 19,
      "WRONG_TYPE": 201
    },
    "error_by_sq": {
      "SQ1": {
        "FALSE_ALARM": 62,
        "MISS": 25,
        "PARSE_FAIL": 2,
        "WRONG_TYPE": 1
      },
      "SQ2": {
        "FALSE_ALARM": 73,
        "PARSE_FAIL": 4,
        "WRONG_TYPE": 37,
        "MISS": 26
      },
      "SQ3": {
        "FALSE_ALARM": 61,
        "MISS": 77,
        "WRONG_TYPE": 47,
        "PARSE_FAIL": 4
      },
      "SQ4": {
        "FALSE_ALARM": 139,
        "PARSE_FAIL": 7,
        "WRONG_TYPE": 68,
        "MISS": 12
      },
      "SQ5": {
        "FALSE_ALARM": 39,
        "WRONG_TYPE": 48,
        "MISS": 15,
        "PARSE_FAIL": 2
      }
    },
    "error_by_category": {
      "device_fault": 26,
      "water_damage": 14,
      "fire_gas": 20,
      "intrusion": 145,
      "behavioral_anomaly": 85,
      "child_specific": 28,
      "elderly_specific": 38
    }
  }
}