pytorch: ground truth similarity matrix

按照真实标签排序pair-wise相似度矩阵的Pytorch代码

本文仅作留档，用于输出可视化

• Inputs: Ground-truths $\mathbf{Y}\in {\mathbb{R}}^{n\times 1}$, Similarity matrix $\mathbf{A}\in {\mathbb{R}}^{n\times n}$
• Outputs: Block diagnal matrix of $\mathbf{A}$ sorted by $\mathbf{Y}$ for visualization.

示例：大规模预训练模型CLIP特征在CIFAR10数据集上的 ground truth similarity matrix

import torch
import matplotlib.pyplot as pltY = labels.argsort()
A = logits.mm(logits.t())
# ground truth similarity matrix
A_sorted = A[Y][:, Y]# visualize
# A should be normalized to [0,1]
plt.imshow(A_sorted , cmap='Blues', vmin=0, vmax=1)
plt.colorbar()
plt.set_title('Entitle here')
plt.savefig('save_name.png')
plt.close()