{
  "baseline": {
    "created_at": "2026-11-12T07:54:46Z",
    "manifest": {
      "cost": {
        "currency": "USD",
        "is_estimate": true,
        "notice": "按字符启发式 token 估算与原配置单价计算，不代表供应商账单",
        "pricing": {
          "currency": "USD",
          "input_per_1k_tokens": 0.00015,
          "is_estimate": true,
          "label": "示例单价（非供应商报价）",
          "output_per_1k_tokens": 0.0006
        },
        "total": 7.32e-05
      },
      "created_at": "2026-11-12T07:54:46Z",
      "dataset": {
        "items": 14,
        "path": "feedback.jsonl",
        "sha256": "5837588c00a0a0d848b6b815768b6dc769a797d2fa3e9e970fe1d9aba61a2869"
      },
      "execution": {
        "concurrency": 1,
        "max_retries": 2,
        "retry_backoff_seconds": 0.05,
        "serial_reproducible": true,
        "timeout_seconds": 2.0
      },
      "model": {
        "is_mock": true,
        "name": "offline-feedback-rules-v1",
        "notice": "离线确定性规则模拟，未调用真实模型",
        "parameters": {
          "max_tokens": 128,
          "seed": 42,
          "temperature": 0
        },
        "provider": "mock"
      },
      "prompt": {
        "description": "基础分类指令与统一输出格式",
        "name": "feedback-classifier",
        "template": "你是用户反馈分类助手。请将用户反馈分类为 咨询、投诉、建议、其他，并输出简短理由。\n输出格式：类别｜理由\n只输出一行，不要解释分类过程。\n\n用户反馈：\n${input}\n",
        "version_id": "p_6764c526720c1ab0089ef693c6472983aea73d9f98e4de230381ea6d9de55ac2"
      }
    },
    "name": "反馈分类-v1-基础指令",
    "reproducibility_hash": "2bd20cb9dc2fdf0084c76512a4e83bb6c2d06f550cb83410f45d4f38f216290e",
    "run_id": "run_453d906d772cf609da93",
    "summary": {
      "cost_currency": "USD",
      "cost_is_estimate": true,
      "estimated_total_cost": 7.32e-05,
      "estimated_total_tokens": 362,
      "failed": 0,
      "metrics": {
        "contains": {
          "correct": 1,
          "evaluated": 14,
          "mean": 0.07142857142857142
        },
        "exact_match": {
          "correct": 1,
          "evaluated": 14,
          "mean": 0.07142857142857142
        },
        "regex": {
          "correct": 11,
          "evaluated": 14,
          "mean": 0.7857142857142857
        }
      },
      "pricing": {
        "currency": "USD",
        "input_per_1k_tokens": 0.00015,
        "is_estimate": true,
        "label": "示例单价（非供应商报价）",
        "output_per_1k_tokens": 0.0006
      },
      "succeeded": 14,
      "token_count_is_estimate": true,
      "total": 14,
      "total_latency_ms": 0.0
    }
  },
  "by_tag": {
    "一般评价": {
      "baseline": {
        "contains": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "candidate": {
        "contains": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "count": 1,
      "delta": {
        "contains": 0.0,
        "exact_match": 0.0,
        "regex": 0.0
      }
    },
    "产品功能": {
      "baseline": {
        "contains": {
          "correct": 0,
          "evaluated": 2,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 2,
          "mean": 0.0
        },
        "regex": {
          "correct": 2,
          "evaluated": 2,
          "mean": 1.0
        }
      },
      "candidate": {
        "contains": {
          "correct": 2,
          "evaluated": 2,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 2,
          "evaluated": 2,
          "mean": 1.0
        },
        "regex": {
          "correct": 2,
          "evaluated": 2,
          "mean": 1.0
        }
      },
      "count": 2,
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      }
    },
    "价格": {
      "baseline": {
        "contains": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "candidate": {
        "contains": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "count": 1,
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      }
    },
    "关键词歧义": {
      "baseline": {
        "contains": {
          "correct": 0,
          "evaluated": 2,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 2,
          "mean": 0.0
        },
        "regex": {
          "correct": 0,
          "evaluated": 2,
          "mean": 0.0
        }
      },
      "candidate": {
        "contains": {
          "correct": 2,
          "evaluated": 2,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 2,
          "evaluated": 2,
          "mean": 1.0
        },
        "regex": {
          "correct": 2,
          "evaluated": 2,
          "mean": 1.0
        }
      },
      "count": 2,
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 1.0
      }
    },
    "其他": {
      "baseline": {
        "contains": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "candidate": {
        "contains": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "count": 1,
      "delta": {
        "contains": 0.0,
        "exact_match": 0.0,
        "regex": 0.0
      }
    },
    "咨询": {
      "baseline": {
        "contains": {
          "correct": 0,
          "evaluated": 4,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 4,
          "mean": 0.0
        },
        "regex": {
          "correct": 3,
          "evaluated": 4,
          "mean": 0.75
        }
      },
      "candidate": {
        "contains": {
          "correct": 3,
          "evaluated": 4,
          "mean": 0.75
        },
        "exact_match": {
          "correct": 3,
          "evaluated": 4,
          "mean": 0.75
        },
        "regex": {
          "correct": 4,
          "evaluated": 4,
          "mean": 1.0
        }
      },
      "count": 4,
      "delta": {
        "contains": 0.75,
        "exact_match": 0.75,
        "regex": 0.25
      }
    },
    "客服": {
      "baseline": {
        "contains": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "candidate": {
        "contains": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "count": 1,
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      }
    },
    "常规": {
      "baseline": {
        "contains": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "candidate": {
        "contains": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "count": 1,
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      }
    },
    "建议": {
      "baseline": {
        "contains": {
          "correct": 0,
          "evaluated": 4,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 4,
          "mean": 0.0
        },
        "regex": {
          "correct": 2,
          "evaluated": 4,
          "mean": 0.5
        }
      },
      "candidate": {
        "contains": {
          "correct": 4,
          "evaluated": 4,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 4,
          "evaluated": 4,
          "mean": 1.0
        },
        "regex": {
          "correct": 4,
          "evaluated": 4,
          "mean": 1.0
        }
      },
      "count": 4,
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.5
      }
    },
    "投诉": {
      "baseline": {
        "contains": {
          "correct": 0,
          "evaluated": 5,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 5,
          "mean": 0.0
        },
        "regex": {
          "correct": 5,
          "evaluated": 5,
          "mean": 1.0
        }
      },
      "candidate": {
        "contains": {
          "correct": 5,
          "evaluated": 5,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 5,
          "evaluated": 5,
          "mean": 1.0
        },
        "regex": {
          "correct": 5,
          "evaluated": 5,
          "mean": 1.0
        }
      },
      "count": 5,
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      }
    },
    "物流": {
      "baseline": {
        "contains": {
          "correct": 0,
          "evaluated": 2,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 2,
          "mean": 0.0
        },
        "regex": {
          "correct": 2,
          "evaluated": 2,
          "mean": 1.0
        }
      },
      "candidate": {
        "contains": {
          "correct": 1,
          "evaluated": 2,
          "mean": 0.5
        },
        "exact_match": {
          "correct": 1,
          "evaluated": 2,
          "mean": 0.5
        },
        "regex": {
          "correct": 2,
          "evaluated": 2,
          "mean": 1.0
        }
      },
      "count": 2,
      "delta": {
        "contains": 0.5,
        "exact_match": 0.5,
        "regex": 0.0
      }
    },
    "理由泛化": {
      "baseline": {
        "contains": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "candidate": {
        "contains": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "count": 1,
      "delta": {
        "contains": 0.0,
        "exact_match": 0.0,
        "regex": 0.0
      }
    },
    "疑问词干扰": {
      "baseline": {
        "contains": {
          "correct": 0,
          "evaluated": 2,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 2,
          "mean": 0.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 2,
          "mean": 0.5
        }
      },
      "candidate": {
        "contains": {
          "correct": 2,
          "evaluated": 2,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 2,
          "evaluated": 2,
          "mean": 1.0
        },
        "regex": {
          "correct": 2,
          "evaluated": 2,
          "mean": 1.0
        }
      },
      "count": 2,
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.5
      }
    },
    "稳定性": {
      "baseline": {
        "contains": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "candidate": {
        "contains": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "count": 1,
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      }
    },
    "订单": {
      "baseline": {
        "contains": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "exact_match": {
          "correct": 0,
          "evaluated": 1,
          "mean": 0.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "candidate": {
        "contains": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "exact_match": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        },
        "regex": {
          "correct": 1,
          "evaluated": 1,
          "mean": 1.0
        }
      },
      "count": 1,
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      }
    }
  },
  "candidate": {
    "created_at": "2026-11-12T07:54:46Z",
    "manifest": {
      "cost": {
        "currency": "USD",
        "is_estimate": true,
        "notice": "按字符启发式 token 估算与原配置单价计算，不代表供应商账单",
        "pricing": {
          "currency": "USD",
          "input_per_1k_tokens": 0.00015,
          "is_estimate": true,
          "label": "示例单价（非供应商报价）",
          "output_per_1k_tokens": 0.0006
        },
        "total": 0.00013575
      },
      "created_at": "2026-11-12T07:54:46Z",
      "dataset": {
        "items": 14,
        "path": "feedback.jsonl",
        "sha256": "5837588c00a0a0d848b6b815768b6dc769a797d2fa3e9e970fe1d9aba61a2869"
      },
      "execution": {
        "concurrency": 1,
        "max_retries": 2,
        "retry_backoff_seconds": 0.05,
        "serial_reproducible": true,
        "timeout_seconds": 2.0
      },
      "model": {
        "is_mock": true,
        "name": "offline-feedback-rules-v1",
        "notice": "离线确定性规则模拟，未调用真实模型",
        "parameters": {
          "max_tokens": 128,
          "seed": 42,
          "temperature": 0
        },
        "provider": "mock"
      },
      "prompt": {
        "description": "增加意图优先级和具体理由约束",
        "name": "feedback-classifier",
        "template": "你是用户反馈分类助手。请将用户反馈分类为 咨询、投诉、建议、其他，并输出简短理由。\n先识别用户意图：明确的建议意图优先于普通业务关键词；故障、损坏、长期未解决等强烈不满优先于疑问；其余询问归为咨询，避免仅凭“退款”等单个词误判。\n理由要具体简洁，直接概括反馈事项，避免“用户提出疑问”或“用户表达不满”等泛泛表述。\n输出格式严格为：类别｜理由\n只输出一行，不要输出其他内容。\n\n用户反馈：\n${input}\n",
        "version_id": "p_0ff357957dab904f8df0c248fab8fb32c079806d4e3945a1780bce66a0ee14bb"
      }
    },
    "name": "反馈分类-v2-意图与具体理由",
    "reproducibility_hash": "a9b47db125cb78a038063cc210af3d0d7e128b99049c789ee4474b7d8864708c",
    "run_id": "run_4b312ce41e8290fe5186",
    "summary": {
      "cost_currency": "USD",
      "cost_is_estimate": true,
      "estimated_total_cost": 0.00013575,
      "estimated_total_tokens": 779,
      "failed": 0,
      "metrics": {
        "contains": {
          "correct": 13,
          "evaluated": 14,
          "mean": 0.9285714285714286
        },
        "exact_match": {
          "correct": 13,
          "evaluated": 14,
          "mean": 0.9285714285714286
        },
        "regex": {
          "correct": 14,
          "evaluated": 14,
          "mean": 1.0
        }
      },
      "pricing": {
        "currency": "USD",
        "input_per_1k_tokens": 0.00015,
        "is_estimate": true,
        "label": "示例单价（非供应商报价）",
        "output_per_1k_tokens": 0.0006
      },
      "succeeded": 14,
      "token_count_is_estimate": true,
      "total": 14,
      "total_latency_ms": 0.0
    }
  },
  "comparison_id": "cmp_26ab1f23a7b50cbd6468",
  "cost": {
    "baseline": 7.32e-05,
    "basis": "按字符启发式 token 估算与各实验配置单价计算，不代表供应商账单",
    "candidate": 0.00013575,
    "currency": "USD",
    "delta": 6.255e-05,
    "is_estimate": true
  },
  "created_at": "2026-11-12T07:54:46Z",
  "dataset_sha256": "5837588c00a0a0d848b6b815768b6dc769a797d2fa3e9e970fe1d9aba61a2869",
  "records": [
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 22,
          "total": 25
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 1.0
        },
        "output": "咨询｜用户提出疑问"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 52,
          "total": 55
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "咨询｜询问发货时间"
      },
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      },
      "expected": "咨询｜询问发货时间",
      "id": "ship-time",
      "input": "请问订单一般多久发货？",
      "pattern": "^咨询｜.+$",
      "tags": [
        "咨询",
        "常规"
      ]
    },
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 23,
          "total": 26
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 1.0
        },
        "output": "投诉｜用户表达不满"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 53,
          "total": 56
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "投诉｜商品到货损坏"
      },
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      },
      "expected": "投诉｜商品到货损坏",
      "id": "broken-screen",
      "input": "收到的手机屏幕碎了，怎么处理？",
      "pattern": "^投诉｜.+$",
      "tags": [
        "投诉",
        "疑问词干扰"
      ]
    },
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 23,
          "total": 26
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 1.0
        },
        "output": "投诉｜用户表达不满"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 53,
          "total": 56
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "投诉｜物流迟迟未送达"
      },
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      },
      "expected": "投诉｜物流迟迟未送达",
      "id": "slow-delivery",
      "input": "包裹十天了还没到，太慢了。",
      "pattern": "^投诉｜.+$",
      "tags": [
        "投诉",
        "物流"
      ]
    },
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 23,
          "total": 26
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 1.0
        },
        "output": "建议｜用户提出建议"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 52,
          "total": 55
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "建议｜增加深色模式"
      },
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      },
      "expected": "建议｜增加深色模式",
      "id": "dark-mode",
      "input": "建议App增加深色模式。",
      "pattern": "^建议｜.+$",
      "tags": [
        "产品功能",
        "建议"
      ]
    },
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 24,
          "total": 27
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 1.0
        },
        "output": "建议｜用户提出建议"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 54,
          "total": 57
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "建议｜支持导出Excel"
      },
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      },
      "expected": "建议｜支持导出Excel",
      "id": "excel-export",
      "input": "希望能支持导出Excel，方便做账。",
      "pattern": "^建议｜.+$",
      "tags": [
        "产品功能",
        "建议"
      ]
    },
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 22,
          "total": 25
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 1.0
        },
        "output": "咨询｜用户提出疑问"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 52,
          "total": 55
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "咨询｜询问地址修改方法"
      },
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      },
      "expected": "咨询｜询问地址修改方法",
      "id": "change-address",
      "input": "如何修改收货地址？",
      "pattern": "^咨询｜.+$",
      "tags": [
        "咨询",
        "订单"
      ]
    },
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 24,
          "total": 27
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 1.0
        },
        "output": "投诉｜用户表达不满"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 54,
          "total": 57
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "投诉｜客服未及时解决问题"
      },
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      },
      "expected": "投诉｜客服未及时解决问题",
      "id": "support-loop",
      "input": "客服两次都说稍后处理，到现在没人解决。",
      "pattern": "^投诉｜.+$",
      "tags": [
        "客服",
        "投诉"
      ]
    },
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 23,
          "total": 26
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 0.0
        },
        "output": "咨询｜用户提出疑问"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 52,
          "total": 55
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "建议｜提供发票抬头模板"
      },
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 1.0
      },
      "expected": "建议｜提供发票抬头模板",
      "id": "invoice-template",
      "input": "能不能增加发票抬头模板？",
      "pattern": "^建议｜.+$",
      "tags": [
        "建议",
        "疑问词干扰"
      ]
    },
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 22,
          "total": 25
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "其他｜无法归入以上类别"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 52,
          "total": 55
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "其他｜无法归入以上类别"
      },
      "delta": {
        "contains": 0.0,
        "exact_match": 0.0,
        "regex": 0.0
      },
      "expected": "其他｜无法归入以上类别",
      "id": "color-praise",
      "input": "这个配色挺好看的。",
      "pattern": "^其他｜.+$",
      "tags": [
        "一般评价",
        "其他"
      ]
    },
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 22,
          "total": 25
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 1.0
        },
        "output": "投诉｜用户表达不满"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 52,
          "total": 55
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "投诉｜会员价格偏高"
      },
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      },
      "expected": "投诉｜会员价格偏高",
      "id": "membership-price",
      "input": "你们的会员太贵了。",
      "pattern": "^投诉｜.+$",
      "tags": [
        "价格",
        "投诉"
      ]
    },
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 23,
          "total": 26
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 1.0
        },
        "output": "投诉｜用户表达不满"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 53,
          "total": 56
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "投诉｜应用频繁闪退"
      },
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 0.0
      },
      "expected": "投诉｜应用频繁闪退",
      "id": "app-crash",
      "input": "界面经常闪退，重装也没用。",
      "pattern": "^投诉｜.+$",
      "tags": [
        "投诉",
        "稳定性"
      ]
    },
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 23,
          "total": 26
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 0.0
        },
        "output": "投诉｜用户表达不满"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 52,
          "total": 55
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "建议｜简化退款流程"
      },
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 1.0
      },
      "expected": "建议｜简化退款流程",
      "id": "refund-suggestion",
      "input": "建议退款流程更简单一些。",
      "pattern": "^建议｜.+$",
      "tags": [
        "关键词歧义",
        "建议"
      ]
    },
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 24,
          "total": 27
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 0.0
        },
        "output": "投诉｜用户表达不满"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 54,
          "total": 57
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 1.0,
          "exact_match": 1.0,
          "regex": 1.0
        },
        "output": "咨询｜询问退款到账时间"
      },
      "delta": {
        "contains": 1.0,
        "exact_match": 1.0,
        "regex": 1.0
      },
      "expected": "咨询｜询问退款到账时间",
      "id": "refund-arrival",
      "input": "已经申请退款了，想问问什么时候到账？",
      "pattern": "^咨询｜.+$",
      "tags": [
        "关键词歧义",
        "咨询"
      ]
    },
    {
      "baseline": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 22,
          "total": 25
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 1.0
        },
        "output": "咨询｜用户提出疑问"
      },
      "candidate": {
        "attempts": 1,
        "error": null,
        "estimated_tokens": {
          "completion": 3,
          "is_estimate": true,
          "method": "ceil(Unicode字符数/4)，仅为估算",
          "prompt": 52,
          "total": 55
        },
        "latency_ms": 0.0,
        "metrics": {
          "contains": 0.0,
          "exact_match": 0.0,
          "regex": 1.0
        },
        "output": "咨询｜用户提出疑问"
      },
      "delta": {
        "contains": 0.0,
        "exact_match": 0.0,
        "regex": 0.0
      },
      "expected": "咨询｜询问物流单号查询方法",
      "id": "tracking-number",
      "input": "物流单号在哪里查看？",
      "pattern": "^咨询｜.+$",
      "tags": [
        "咨询",
        "物流",
        "理由泛化"
      ]
    }
  ],
  "schema_version": 1,
  "summary": {
    "baseline": {
      "contains": {
        "correct": 1,
        "evaluated": 14,
        "mean": 0.07142857142857142
      },
      "exact_match": {
        "correct": 1,
        "evaluated": 14,
        "mean": 0.07142857142857142
      },
      "regex": {
        "correct": 11,
        "evaluated": 14,
        "mean": 0.7857142857142857
      }
    },
    "candidate": {
      "contains": {
        "correct": 13,
        "evaluated": 14,
        "mean": 0.9285714285714286
      },
      "exact_match": {
        "correct": 13,
        "evaluated": 14,
        "mean": 0.9285714285714286
      },
      "regex": {
        "correct": 14,
        "evaluated": 14,
        "mean": 1.0
      }
    },
    "delta": {
      "contains": 0.857143,
      "exact_match": 0.857143,
      "regex": 0.214286
    }
  },
  "type": "promptlens_comparison"
}