学习可泛化的预测器的超额描述长度
Excess Description Length of Learning Generalizable Predictors
AI总结:
本文提出EDL框架,用于量化微调提取的预测结构并评估模型泛化能力。
摘要:
理解微调是否激发潜在能力还是教授新能力是语言模型评估和安全性的基本问题。我们开发了一个正式的信息论框架,用于量化微调从训练数据集中提取的预测结构,并将其写入模型参数的程度。我们的核心量,超额描述长度(EDL),通过预quential编码定义,衡量在使用进化模型(在线训练)依次编码训练标签所需的位数与最终训练模型下的残余编码成本之间的差距。我们证明EDL在期望下非负,随着数据量趋于无限时收敛于超额描述长度,并提供预期泛化增益的界限。通过一系列玩具模型,我们澄清了学习中的常见困惑:为什么随机标签导致EDL接近零,单个示例如何消除关于数据分布所描述的规则的许多位不确定性,为什么在稀有输入上学习的结构对预期泛化贡献比例很小,以及格式学习如何创造与能力获取不同的早期瞬态。该框架为经验观察到能力激发和教学表现出质的不同扩展特征提供了严谨的基础。
英文摘要:
Understanding whether fine-tuning elicits latent capabilities or teaches new ones is a fundamental question for language model evaluation and safety. We develop a formal information-theoretic framework for quantifying how much predictive structure fine-tuning extracts from the train dataset and writes into a model's parameters. Our central quantity, Excess Description Length (EDL), is defined via prequential coding and measures the gap between the bits required to encode training labels sequentially using an evolving model (trained online) and the residual encoding cost under the final trained model. We establish that EDL is non-negative in expectation, converges to surplus description length in the infinite-data limit, and provides bounds on expected generalization gain. Through a series of toy models, we clarify common confusions about information in learning: why random labels yield EDL near zero, how a single example can eliminate many bits of uncertainty about the underlying rule(s) that describe the data distribution, why structure learned on rare inputs contributes proportionally little to expected generalization, and how format learning creates early transients distinct from capability acquisition. This framework provides rigorous foundations for the empirical observation that capability elicitation and teaching exhibit qualitatively distinct scaling signatures.
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。