用两个独立种子比对消除大模型采样偏差，提升生成随机性的准确性。
Mutual Debiasing via Dual-Seed Comparison for Probabilistic Sampling in Large Language Models
- 通过对比两个大模型生成的种子字符序值，构造去偏伪均匀变量。
- 在96%测试场景中优于现有方法，显著改善采样分布精度。
- 适用于选择题生成与可控文生图，无需额外工具或训练。
尽管大语言模型在推理和决策方面表现卓越，但高保真概率采样仍是持续挑战。生成随机变量时，模型普遍存在系统性偏差，扭曲目标概率分布。现有方法多依赖单一自生成种子，继承模型固有偏差。为此，本文提出双种子比对（DSC）机制，一种透明、无需工具的协议：利用两个独立生成的种子，比较其字符级序值构建比特序列，经转换与归一化得到伪均匀变量，并通过逆累积分布函数映射至目标分布。实证结果表明，DSC在96%评估场景中显著优于现有方法。此外，基于DSC比对算子的任务适配变体，在多项选择题生成和属性约束文本到图像提示中均提升了分布控制能力。
原文摘要 · Abstract (English)
Although Large Language Models (LLMs) demonstrate remarkable capabilities in reasoning and decision-making, high-fidelity probabilistic sampling remains a persistent challenge. When generating random variables, LLMs consistently exhibit systematic biases that warp the target probability distributions. Current approaches often rely on a single, self-generated seed, which inherits model-specific biases. To overcome this vulnerability, we introduce Dual-Seed Comparison (DSC), a transparent, tool-free protocol that utilizes two independent LLM-generated seeds to neutralize bias. DSC compares the character-level ordinal values of the two seeds to construct a bit sequence, converts and normalizes this sequence into a pseudo-uniform variate, and then maps the variate to the target distribution through the inverse cumulative distribution function (CDF). Empirical results show that DSC substantially outperforms existing methods across 96\% of evaluated settings. Beyond direct sampling, task-adapted variants based on the DSC comparison operator improve distributional control in MCQ generation and attribute-constrained text-to-image prompting.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。