大模型采样不随机，导致智能体决策不可靠。
The Illusion of Stochasticity in LLMs
- 用实验证明大模型无法准确从指定分布中采样
- 即使顶级模型也难以将内部概率映射到输出
- 对依赖随机性的智能体系统有重大影响
本文揭示，可靠随机采样是大型语言模型（LLMs）作为智能体运行的基本要求，却长期未被满足。智能体常需从观测数据推断出的概率分布中采样，而这一过程需由LLM模拟实现。然而，标准强化学习智能体依赖外部采样机制，而LLM无法将其内部概率估计有效转化为随机输出，形成显著缺陷。通过在多个模型家族、模型规模、提示风格及分布上进行严格实证分析，我们揭示了该问题的普遍性。关键发现是：尽管前沿模型能将给定随机种子映射到目标分布，但其直接从特定分布采样的能力存在根本性缺陷。
原文摘要 · Abstract (English)
In this work, we demonstrate that reliable stochastic sampling is a fundamental yet unfulfilled requirement for Large Language Models (LLMs) operating as agents. Agentic systems are frequently required to sample from distributions, often inferred from observed data, a process which needs to be emulated by the LLM. This leads to a distinct failure point: while standard RL agents rely on external sampling mechanisms, LLMs fail to map their internal probability estimates to their stochastic outputs. Through rigorous empirical analysis across multiple model families, model sizes, prompting styles, and distributions, we demonstrate the extent of this failure. Crucially, we show that while powerful frontier models can convert provided random seeds to target distributions, their ability to sample directly from specific distributions is fundamentally flawed.
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。