Data for "Causal perturbation reveals a subspace-level dissociation between semantic geometry and classification performance in artificial neural networks"
收藏官方服务：
资源简介：
Artificial neural networks output-layer activations extracted from: the ImageNet-1K validation set, the 1,000 NSD shared images, and the 1,854 THINGS-odd-one-out images, Human neural responses in visually related cortical regions to the 1,000 shared images from the public NSD
提供机构：
Zenodo创建时间：
2026-07-14
