Chapter 2 uses this code to get the model labels:
with open('../data/p1ch2/imagenet_classes.txt') as f:
labels = [line.strip() for line in f.readlines()]
However, it is never made clear where the file imagenet_classes.txt comes from.
It turns out that they are already in torchvision.models:
from torchvision import models
labels = models.ResNet101_Weights.DEFAULT.meta['categories']
It would be useful to update the book to make this clear, or at least put it in the errata.