{
  "model": "deepseek-v4-flash",
  "api_base": "https://uni-api.cstcloud.cn/v1",
  "timestamp": "2026-05-04T02:17:43.163035",
  "total_evaluated": 1200,
  "avg_latency_seconds": 62.46,
  "api_errors": 0,
  "summary": {
    "total_episodes": 1200,
    "detection_accuracy": 0.49833333333333335,
    "miss_rate": 0.09454545454545454,
    "false_alarm_rate": 0.8446153846153847,
    "precision": 0.47564469914040114,
    "recall": 0.9054545454545454,
    "f1_security": 0.6236693800876644,
    "threat_type_accuracy": 0.3755020080321285,
    "parse_failure_rate": 0.004166666666666667,
    "breakdown": {
      "TP": {
        "total": 550,
        "correct": 498,
        "accuracy": 0.9054545454545454
      },
      "FP": {
        "total": 400,
        "correct": 59,
        "accuracy": 0.1475
      },
      "TN": {
        "total": 250,
        "correct": 41,
        "accuracy": 0.164
      }
    },
    "per_sq": {
      "SQ1": {
        "total": 190,
        "accuracy": 0.47368421052631576,
        "miss_count": 8,
        "false_alarm_count": 92
      },
      "SQ2": {
        "total": 240,
        "accuracy": 0.5,
        "miss_count": 7,
        "false_alarm_count": 113
      },
      "SQ3": {
        "total": 290,
        "accuracy": 0.4517241379310345,
        "miss_count": 20,
        "false_alarm_count": 138
      },
      "SQ4": {
        "total": 290,
        "accuracy": 0.4827586206896552,
        "miss_count": 10,
        "false_alarm_count": 140
      },
      "SQ5": {
        "total": 190,
        "accuracy": 0.6157894736842106,
        "miss_count": 7,
        "false_alarm_count": 66
      }
    }
  },
  "errors": {
    "total_errors": 913,
    "error_distribution": {
      "FALSE_ALARM": 549,
      "WRONG_TYPE": 311,
      "MISS": 50,
      "PARSE_FAIL": 3
    },
    "error_by_sq": {
      "SQ1": {
        "FALSE_ALARM": 92,
        "WRONG_TYPE": 10,
        "MISS": 7,
        "PARSE_FAIL": 1
      },
      "SQ2": {
        "FALSE_ALARM": 113,
        "WRONG_TYPE": 59,
        "MISS": 7
      },
      "SQ3": {
        "FALSE_ALARM": 138,
        "PARSE_FAIL": 2,
        "WRONG_TYPE": 96,
        "MISS": 19
      },
      "SQ4": {
        "FALSE_ALARM": 140,
        "WRONG_TYPE": 79,
        "MISS": 10
      },
      "SQ5": {
        "FALSE_ALARM": 66,
        "WRONG_TYPE": 67,
        "MISS": 7
      }
    },
    "error_by_category": {
      "device_fault": 17,
      "fire_gas": 36,
      "water_damage": 25,
      "intrusion": 140,
      "behavioral_anomaly": 87,
      "child_specific": 29,
      "elderly_specific": 27
    }
  }
}