自我一致性 （Self-consistency）

• 概念：该技术通过对同一问题采样不同的推理路径，并通过多数投票选择最一致的答案，来解决大语言模型（LLM）输出的可变性问题。通过使用不同的温度（temperature）或采样设置生成多条推理路径，然后聚合最终答案，自洽性能够提高复杂推理任务的准确性。从本质上讲，这是一种针对大语言模型输出的集成方法。
• 原理：由于复杂问题存在多种合理推理路径，不同路径可通过不同思考过程抵达同一正确答案。自洽性通过聚合这些路径，抵消单一路径的偏差，从而提升准确性。
• 目标：通过采样多样化推理路径并聚合结果，提升模型推理的准确性和可靠性。
• 总结：用于替代链式思维提示（chain-of-thought prompting）中的贪心解码。该方法通过采样多样化推理路径并聚合最一致答案，提升复杂推理任务表现。

关键步骤

自洽性方法核心逻辑

• 步骤 1：对同一问题，使用不同采样设置（如温度参数）生成多条不同的推理路径，而非仅采用贪心解码的单一路径。
• 步骤 2：通过边缘化（marginalizing）采样的推理路径，选择最一致的答案（如通过多数投票等聚合方式）。

复杂推理问题通常存在多种合理的思考方式，最终指向唯一正确答案，多样化路径可提升答案的可靠性。

案例

这是一个测试方法，用于测试邮件分类的一致性(self-consistency)。主要通过对同一封邮件进行多次分类，采用多数投票的方式来确定最终分类结果。
设置温度参数 temperature(1.0) 以增加输出的随机性，采用多种不同的推理路径得到罪过，最终通过投票的方式聚合结果得到一致的答案

    @Test
    public void testSelfConsistency() throws Exception{
        String email = """
            Hi,
            I have seen you use Wordpress for your website. A great open
            source content management system. I have used it in the past
            too. It comes with lots of great user plugins. And it's pretty
            easy to set up.
            I did notice a bug in the contact form, which happens when
            you select the name field. See the attached screenshot of me
            entering text in the name field. Notice the JavaScript alert
            box that I inv0k3d.
            But for the rest it's a great website. I enjoy reading it. Feel
            free to leave the bug in the website, because it gives me more
            interesting things to read.
            Cheers,
            Harry the Hacker.
            """;

        int importantCount = 0;
        int notImportantCount = 0;

        // Run the model 5 times with the same input
        for (int i = 0; i < 5; i++) {
            EmailClassification output = openAiChatClient
                .prompt()
                .user(u -> u.text("""
                        Email: {email}
                        Classify the above email as IMPORTANT or NOT IMPORTANT. Let's
                        think step by step and explain why.
                        """)
                    .param("email", email))
                .options(ChatOptions.builder()
                    .temperature(1.0)  // Higher temperature for more variation
                    .build())
                .call()
                .entity(EmailClassification.class);

            System.out.println(output.reasoning);

            // Count results
            if (output.classification() == EmailClassification.Classification.IMPORTANT) {
                importantCount++;
            } else {
                notImportantCount++;
            }
        }

        // Determine the final classification by majority vote
        String finalClassification = importantCount > notImportantCount ?
            "IMPORTANT" : "NOT IMPORTANT";
        System.out.println(finalClassification);
    }

    record EmailClassification(Classification classification, String reasoning) {
        enum Classification {
            IMPORTANT, NOT_IMPORTANT
        }
    }

参考

1.spring-ai self consistency: https://docs.spring.io/spring-ai/reference/api/chat/prompt-engineering-patterns.html#_2_6_self_consistency
2.Self-Consistency Improves Chain of Thought Reasoning in Language Models,https://arxiv.org/abs/2203.11171