Figureure 1 · Foundation Models and IDES T: Intra-Dataset CorrelationFigure 1. Foundation Models and IDES T: Intra-Dataset Correlation. Linear probe accuracy of pretrained SSL models on ImageNet (left), iNat-18 (middle left), iNat-21 (middle right), SUN397 (right) versus IDES T on each respective dataset. Each point corresponds to a model checkpoint; point size reflects the number of parameters. We report Kendall’s $\tau \stackrel { \bar { \in } } { \in } [ - 1 , 1 ]$ and Spearman’s $\rho \in [ \bar { - } 1 , 1 ]$ . Correlations across all four benchmarks demonstrate IDES T’s ability to provide insights into models’ representation quality.这张图/表用于判断 IdEst 的实验收益来自哪里。重点看替换、消融或跨模型设置下趋势是否一致，而不是只看单个最高分。Figureure 7 · Overview of Joint-Embedding ArchitecturesFigure 7. Overview of Joint-Embedding Architectures. Two separate data augmentation operators are sampled from the same distribution $( t , t ^ { \prime } \sim \tau )$ and applied to each image in a batch I to obtain two views, X and X′. An encoder network $f$ and a projection head g are trained to maximize agreement between the resulting embeddings. After training, the encoder $f$ and its representations Y are retained for downstream tasks.这张图概括 IdEst 的整体方法流程。阅读时先看模块之间传递的训练信号，再看作者如何把目标拆成可优化的子问题。
核心问题
它和通用视觉自监督的关系在于：用 MST intrinsic dimension 作为 SSL 表征质量代理，目标是减少 linear probe 成本和超参敏感性。
方法拆解
用 MST intrinsic dimension 作为 SSL 表征质量代理，目标是减少 linear probe 成本和超参敏感性
