{
  "slug": "reasoning-verification",
  "category": "推理与验证",
  "title": "多想几步就会更对吗？推理时计算、投票与过程验证",
  "dek": "从思维链、自一致性和过程监督出发，分析额外计算如何改善候选选择，以及错误相关性为何会破坏投票直觉。",
  "english": "Extra inference computation can expand search and selection. We distinguish generated rationales, answer aggregation, and process verification, using an explicitly hypothetical independent-vote calculation.",
  "refs": [
    "cot",
    "consistency",
    "verify"
  ],
  "sections": [
    [
      "摘要",
      "推理时增加计算，可以用于生成更长候选、采样更多路径或验证中间步骤。这些策略消耗的资源不同，也依赖不同的可靠性条件。本文比较思维链提示、自一致性与过程监督，并用一个独立投票模型说明：更多候选只有在错误结构合适时，才可能转化为更高正确率。",
      [
        "cot",
        "consistency",
        "verify"
      ]
    ],
    [
      "1. 中间文字不是正确性的证明",
      "思维链研究展示，在被研究模型与任务中，带中间步骤的示例能改善算术、常识与符号推理表现。这个结果说明提示组织可能影响任务求解，并不保证生成的每一步都有效。读者应把最终答案正确性、步骤有效性和解释是否忠实于实际计算分开看，不能以叙述连贯代替检验。",
      [
        "cot"
      ]
    ],
    [
      "2. 自一致性把计算花在多路径上",
      "自一致性方法采样不同推理路径，再对答案进行汇总选择。它利用“不同路径可能汇聚到正确答案”的结构，区别于只沿单一贪心路径生成。但如果模型反复复用同一个错误前提，样本数量增加也可能只是重复错误。有效候选多样性应与抽样次数分开报告。",
      [
        "consistency"
      ]
    ],
    [
      "3. 过程监督关注中间判断",
      "Let's Verify Step by Step 比较最终结果监督与过程监督，在特定数学任务和候选选择设置中报告了过程监督的优势。其摘要中的 78% 属于 MATH 测试集的代表性子集及论文的评估过程，不能改写为“解决全部数学问题的准确率”。本文不以该数字绘制通用能力排行榜。",
      [
        "verify"
      ]
    ],
    [
      "4. 一个透明的投票推导",
      "假定每次候选独立、正确概率恒为 p，取奇数 n 次并以多数票决定，成功概率为从 k=(n+1)/2 到 n 的二项概率之和。图 2 用 p=0.6、0.7、0.8 与 n=1、3、5、7、9 计算。这里的独立、同分布与可归并为同一正确答案都是假设，并不是对语言模型样本的实测判断。",
      []
    ],
    [
      "5. 为什么独立假设经常不够",
      "同一个模型可能共享知识缺口、提示偏差和题目误读，即使采样得到不同文字，错误仍然相关。形式上“有九个答案”不代表有九份独立证据。外部检索、程序执行或可证明的约束检查，有机会引入不同于语言重述的信息；是否真的有效，仍需针对任务测量。",
      [
        "consistency",
        "verify"
      ]
    ],
    [
      "6. 比较时必须给出预算",
      "一个方法若使用更多词元、更多候选或更强验证器，质量变化就与计算投入共同发生。本文建议同时报告每题总词元、调用次数、超时率、最终正确率与完整延迟，并保留按难度分组的结果。只报告最容易题目的平均耗时，会掩盖长尾；只展示最好一次答案，会掩盖选择成本。",
      []
    ],
    [
      "7. 建立可检查的验证边界",
      "对于算术，执行计算并核对单位；对于代码，使用隔离测试与预期行为；对于文献事实，核对来源与版本。验证器也应有自己的误判评估。若验证器本质上仍是同类模型的主观评分，应公开这一依赖，不应把它写成外部实验已经确认。这里列出的是设计原则，本站未据此声称完成新的模型实验。",
      []
    ],
    [
      "验证器的能力也有预算条件",
      "过程检查增加了另一类计算投入。如果验证器能访问标准答案、特殊工具或更多上下文，应与候选模型的可用信息区分开。若在一百个候选中挑出最好答案，最终正确率应与生成一百个候选的成本同时展示。把这种结果与单次生成直接比较，可以作为系统层比较，却不能省略资源差异后宣称基础模型推理能力同幅度提高。",
      []
    ],
    [
      "后续实验可以如何设计",
      "选择一组可执行判定答案的题目，在固定总词元或固定延迟两个预算下分别比较单路径、多采样投票和带验证器选择。保存每条候选与最后选择，检查正确答案是否已经出现却未被选中。再按错误原因区分候选生成不足、答案归并错误与验证器误判。对答案不可唯一归并的任务，应先定义等价关系，不能直接套多数票。",
      []
    ],
    [
      "局限与结论",
      "本稿的投票曲线是数学示意，不是任何模型的测试成绩。额外计算能扩大候选空间，真正决定价值的是候选生成、错误相关性与验证质量。一个可复核的推理系统，应让读者看到它如何选择答案、使用了多少资源，以及哪些环节仍可能同时出错。",
      []
    ]
  ],
  "diagram": [
    "问题与预算",
    "先固定时间、词元或调用上限",
    "生成候选路径",
    "中间步骤是可检查对象",
    "汇总相同答案",
    "文字不同不一定信息独立",
    "核对中间步骤",
    "过程监督或外部可执行检查",
    "选择并报告结果",
    "包含失败、超时与资源消耗",
    "分析相关错误",
    "共同前提错误不能靠重复消除"
  ],
  "figureNote": "图 1 是独立同分布候选的二项多数票解析结果，p=0.6/0.7/0.8；不是模型实验，自相关错误不满足该模型。",
  "table": {
    "headers": [
      "投入方式",
      "主要观察",
      "需要另行验证"
    ],
    "rows": [
      [
        "更长路径",
        "单条求解过程",
        "步骤是否有效"
      ],
      [
        "更多样本",
        "候选覆盖与一致答案",
        "错误是否相关"
      ],
      [
        "过程检查",
        "中间判断与最终选择",
        "验证器是否会误判"
      ]
    ]
  },
  "version": 4,
  "kind": "ai_assisted_research_guide",
  "author": "Codex（AI）",
  "published_date": "2026-10-16",
  "disclosure": "站点用户公开委托，Codex（AI）依据一手资料整理；封面由 ImageGen 生成，数据图由代码绘制；未经独立同行评审，未开展新的模型实验。",
  "limitations": "文献导读与方法分析；不是原创实验论文、正式期刊录用或专业资质认证。解析示例不代表本站或当前产品的实测表现。",
  "sources": [
    {
      "id": "cot",
      "title": "Chain-of-Thought Prompting Elicits Reasoning in Large Language Models",
      "authors": "Wei et al.",
      "date": "2022-01-28",
      "url": "https://arxiv.org/html/2201.11903v6",
      "locator": "v6; method and limitations",
      "note": "Intermediate-step exemplars improved studied arithmetic, symbolic and commonsense reasoning tasks; generated explanation is not guaranteed correct.",
      "accessed_at": "2026-10-15",
      "rights": "Original source remains under its own license; only attributed factual synthesis, no source figures or long excerpts."
    },
    {
      "id": "consistency",
      "title": "Self-Consistency Improves Chain of Thought Reasoning in Language Models",
      "authors": "Wang et al.",
      "date": "2022-03-21",
      "url": "https://arxiv.org/abs/2203.11171v4",
      "locator": "v4 abstract",
      "note": "Samples multiple reasoning paths and marginalizes to choose a consistent answer. Correlated model errors can invalidate independent-vote intuitions.",
      "accessed_at": "2026-10-15",
      "rights": "Original source remains under its own license; only attributed factual synthesis, no source figures or long excerpts."
    },
    {
      "id": "verify",
      "title": "Let's Verify Step by Step",
      "authors": "Lightman et al.",
      "date": "2023-05-31",
      "url": "https://arxiv.org/abs/2305.20050v1",
      "locator": "v1 abstract; process vs outcome supervision",
      "note": "Process supervision is compared to outcome supervision on MATH; a reported 78 percent is for a representative subset with the paper's selection procedure, not all mathematics.",
      "accessed_at": "2026-10-15",
      "rights": "Original source remains under its own license; only attributed factual synthesis, no source figures or long excerpts."
    }
  ],
  "figures": [
    {
      "slug": "reasoning-verification",
      "role": "conceptual-cover",
      "generator": "OpenAI built-in image_gen",
      "ai_generated": true,
      "alt": "金色路径穿过深蓝石质迷宫的推理与验证概念插画。",
      "original_sha256": "acb4ce12bafa54bc1eacb86dd82862b41d1c48f5ade7d65d33d1d645b8895a7c",
      "prompt": "Premium scientific journal cover art, landscape 3:2, no text, numbers, logos, watermark or borders. A breathtaking editorial still-life with gallery-grade sculptural material, warm golden details, cinematic but natural illumination, believable shadows and highly tactile surfaces. Intelligent contemporary magazine art direction, refined color palette, memorable composition that reads at thumbnail size. Openly disclosed conceptual art, never an actual experiment or graph. Avoid flat diagrams, sci-fi brains, robot heads, computer chips, neon networks, cheap gradients and generic stock imagery. Topic: reasoning with verification. A large intricate spiral labyrinth carved into a dark blue slate circular disk, viewed obliquely from above like a monumental landscape model. A single continuous fine polished gold path winds from the outer edge to an open central chamber and rises into a graceful thin arc. Alternative paths stay shadowed. Deep midnight blue and foggy blue-gray setting, pale ivory side light, moody gold highlights. Extraordinary precision, clear navigable abstract geometry, dramatic scale, beautifully composed without clutter. No lettering anywhere.",
      "files": [
        {
          "file": "assets/covers/reasoning-verification.png",
          "width": 1280,
          "height": 853,
          "bytes": 1871247,
          "sha256": "39b95f04d384da6909164a1bcc42ee6f8ba1fd5f8c27bea161f3c1c851dcbc45",
          "url": "/guides/assets/v2-reasoning-verification-39b95f04d384.png"
        },
        {
          "file": "assets/covers/reasoning-verification-480.png",
          "width": 480,
          "height": 320,
          "bytes": 279458,
          "sha256": "e6437eb53cd1884cd3d528b832e46d06afcdf65333d1c509b7e003bbbc5f8571",
          "url": "/guides/assets/v2-reasoning-verification-480-e6437eb53cd1.png"
        }
      ],
      "url": "/guides/assets/v2-reasoning-verification-39b95f04d384.png",
      "thumbnail_url": "/guides/assets/v2-reasoning-verification-480-e6437eb53cd1.png"
    },
    {
      "slug": "reasoning-verification",
      "role": "data-chart",
      "file": "assets/charts/reasoning-verification.png",
      "width": 1800,
      "height": 1125,
      "sha256": "dd8911425b6bd1361ee690a4a166161a6b49e32f72f4af488533dcff1818800a",
      "bytes": 130405,
      "alt": "图表：是独立同分布候选的二项多数票解析结果，p=0.6/0.7/0.8；不是模型实验，自相关错误不满足该模型。",
      "data": {
        "kind": "analytical",
        "n": [
          1,
          3,
          5,
          7,
          9
        ],
        "probability_by_p": {
          "0.6": [
            0.6,
            0.648,
            0.68256,
            0.7102080000000001,
            0.73343232
          ],
          "0.7": [
            0.7,
            0.7839999999999999,
            0.8369199999999999,
            0.873964,
            0.90119134
          ],
          "0.8": [
            0.8,
            0.8960000000000001,
            0.94208,
            0.966656,
            0.98041856
          ]
        },
        "assumptions": "independent identically distributed votes; unique grouped correct answer"
      },
      "data_source_sha256": "963323d4a7a0cad4aebd986f09a55251a7c21f68c757d21b76245a128675e13e",
      "ai_generated": true,
      "disclosure": "AI-assisted deterministic rendering of unchanged v1 data; not a new experiment.",
      "url": "/guides/assets/v2-reasoning-verification-dd8911425b6b.png"
    }
  ],
  "peer_reviewed": false,
  "previous_version": "/guides/reasoning-verification/v3/",
  "revision": "Mathematical typography only; scientific prose, numeric data, sources and figures are unchanged.",
  "source_language": "zh-Hans",
  "original_version": 2,
  "original_article_sha256": "8bd370ec6f3f1255cc0398ce7a7a48dd3d55499c150330a7a99e31d3250f9d04",
  "interface_localizations": {
    "en": {
      "title": "Does more thinking mean better answers? Reasoning and verification",
      "dek": "Read chain-of-thought, self-consistency and process supervision through their assumptions and correlated errors.",
      "translation_method": "AI-assisted translation",
      "review_status": "not independently language-reviewed"
    },
    "zh": {
      "title": "多想几步就会更对吗？推理时计算、投票与过程验证",
      "dek": "从思维链、自一致性和过程监督出发，分析额外计算如何改善候选选择，以及错误相关性为何会破坏投票直觉。",
      "translation_method": "original",
      "review_status": "original"
    },
    "es": {
      "title": "¿Pensar más mejora las respuestas? Razonamiento y verificación",
      "dek": "Examinar la cadena de pensamiento, la autoconsistencia y la supervisión del proceso con sus supuestos y errores correlacionados.",
      "translation_method": "AI-assisted translation",
      "review_status": "not independently language-reviewed"
    },
    "ja": {
      "title": "長く考えれば正しくなる？推論時の計算と検証",
      "dek": "思考連鎖、自己整合性、過程の監督を、仮定と誤りの相関から読む。",
      "translation_method": "AI-assisted translation",
      "review_status": "not independently language-reviewed"
    },
    "ar": {
      "title": "هل التفكير الأطول ينتج إجابات أفضل؟ الاستدلال والتحقق",
      "dek": "قراءة سلسلة التفكير والاتساق الذاتي والإشراف على العملية عبر افتراضاتها وأخطائها المترابطة.",
      "translation_method": "AI-assisted translation",
      "review_status": "not independently language-reviewed"
    }
  },
  "presentation_revision": {
    "kind": "scientific_math_typography",
    "source_head": "57294bb73b50b2da740f2e8ac63a3d4d7f9a6843",
    "source_v3_sha256": "c99e5715673351ab1ca6bfc2b2c994f4fa4ab4ae2ad4a2f38a94421a3aa5b9a5",
    "annotations_sha256": "2d57369e64d81dc061ddf3bb16d964ebe272b0a04a6a351b888d3470592554f6"
  }
}