量化语言模型记忆容量，发现每参数约3.6比特。
How much do language models memorize?
- 区分无意记忆与泛化能力，精准测量模型记忆量。
- 模型容量达每参数3.6比特，数据量超限时开始泛化。
- 适合关注模型隐私与容量极限的研究者参考。
我们提出一种新方法，用于估计模型对特定数据点的掌握程度，并据此测量现代语言模型的容量。以往研究难以区分记忆与泛化。我们正式将记忆分为两部分：无意记忆（模型对特定数据集的存储信息）和泛化能力（模型对真实数据生成过程的理解）。当完全消除泛化后，可计算总记忆量，从而估算模型容量：我们的测量表明，GPT类模型的容量约为每参数3.6比特。我们在不同规模的数据集上训练语言模型，观察到模型在容量耗尽前持续记忆，之后“顿悟”现象出现，无意记忆减少，泛化开始。我们训练了数百个参数从50万到15亿的Transformer语言模型，建立了模型容量、数据规模与成员推断之间的缩放规律。
原文摘要 · Abstract (English)
We propose a new method for estimating how much a model knows about a datapoint and use it to measure the capacity of modern language models. Prior studies of language model memorization have struggled to disentangle memorization from generalization. We formally separate memorization into two components: unintended memorization, the information a model contains about a specific dataset, and generalization, the information a model contains about the true data-generation process. When we completely eliminate generalization, we can compute the total memorization, which provides an estimate of model capacity: our measurements estimate that GPT-style models have a capacity of approximately 3.6 bits per parameter. We train language models on datasets of increasing size and observe that models memorize until their capacity fills, at which point "grokking" begins, and unintended memorization decreases as models begin to generalize. We train hundreds of transformer language models ranging from $500K$ to $1.5B$ parameters and produce a series of scaling laws relating model capacity and data size to membership inference.
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。