[LAB 08 · 图像重建：遮挡损失与表示评价](/course/notebooks/b5ba27f2c5ba4ec0922208a7d5dc80a4.html)

# 运行准备与结果身份 · 实现代码
[独立阅读](/course/notebook-blocks/4bb149c8de264021b5927384daedb5f7.html)
按问题继续

索引恢复、去噪/遮挡ConvAE训练、PCA基线、独立冻结评价
展开完整实现代码
```
"""Actual denoising and masked convolutional AE on held-out digit images; not original ViT MAE."""
from pathlib import Path
import copy, hashlib, json, platform, time
import numpy as np
import torch
from torch import nn
from torch.nn import functional as F
from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split
from sklearn.neighbors import KNeighborsClassifier
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt

torch.set_num_threads(2)
SEED=20260922

SOURCE_SHA256='9f7f2ed888f5ea118ad1fc622ff64e1aa1e53d5ae162698fcbfb005f45201825'
print("source",SOURCE_SHA256,"torch",torch.__version__)
```