大型语言模型中推理时引发的概率变换的经验表征
Empirical Characterization of Inference-Time Elicited Probability Transformations in Large Language Models
AI总结:
本研究通过经验观察发现,在多种推理时流程(如思维链、自我细化、检索增强和验证器引导修订)下,候选答案的概率变换遵循近似的对数比率关系,并分析了其系数变化和鲁棒性。
摘要:
大型语言模型越来越依赖推理时程序,如思维链推理、自我细化、检索增强和验证器引导修订,但这些程序下引发的概率变换结构仍不清楚。我们研究外部引发的候选答案概率分配,并观察到重复出现的近似对数比率关系:\[ \log \tilde q_t(i) = α_t \left( \log q_t(i) + \log b_t(i) \right) + c_t, \] 其中 $q_t$ 和 $\tilde q_t$ 分别是引发前和引发后的概率,$b_t$ 是外部构建的证据信号,$α_t$ 是提示配置的经验描述符。在来自 GPQA Diamond、TheoremQA、MMLU-Pro 和 ARC-Challenge 的 4,975 个推理问题上,对多个指令微调模型系列进行评估,我们在约 $1.3 \times 10^5$ 个候选级观测上观察到近似对数比率关系,平均 $R^2 \approx 0.76$。系数在不同引发设置下变化,但定性相似的关系在评估条件下持续存在。使用替代统计表示、提示配置、保留评估和 token 级对数概率的鲁棒性分析表明,观察到的结构不依赖于特定的提示程序或概率估计方法。主要贡献不是代数形式本身(它与广义贝叶斯更新和概率变换框架相关),而是经验观察:在受控条件下,多样化的推理时提示流程反复表现出可复现的对数比率结构。该框架为分析推理时 LLM 流程中的校准、证据放大、不确定性传播和交互敏感性提供了协议敏感的视角。
英文摘要:
Large language models increasingly rely on inference-time procedures such as chain-of-thought reasoning, self-refinement, retrieval augmentation, and verifier-guided revision, yet the structure of elicited probability transformations under these procedures remains poorly understood. We study externally elicited probability assignments over candidate answers and observe recurring approximate log-ratio relationships: \[ \log \tilde q_t(i) = α_t \left( \log q_t(i) + \log b_t(i) \right) + c_t, \] where $q_t$ and $\tilde q_t$ are pre- and post-elicitation probabilities, $b_t$ is an externally constructed evidence signal, and $α_t$ is an empirical descriptor of the prompting configuration. Across 4,975 reasoning problems from GPQA Diamond, TheoremQA, MMLU-Pro, and ARC-Challenge, evaluated on multiple instruction-tuned model families, we observe approximate log-ratio relationships with mean $R^2 \approx 0.76$ over about $1.3 \times 10^5$ candidate-level observations. Coefficients vary across elicitation settings, but qualitatively similar relationships persist across evaluated conditions. Robustness analyses using alternative statistical representations, prompting configurations, held-out evaluation, and token-level log-probabilities suggest that the observed structure is not tied to one prompting procedure or probability estimation method. The main contribution is not the algebraic form itself, which is related to generalized Bayesian updating and probability-transformation frameworks, but the empirical observation that diverse inference-time prompting pipelines repeatedly exhibit reproducible log-ratio structure under controlled conditions. The framework provides a protocol-sensitive perspective for analyzing calibration, evidence amplification, uncertainty propagation, and interaction sensitivity in inference-time LLM pipelines.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。