labels in IMBD
AlexanderTrg opened this issue Β· comments
AlexanderTrg commented
π Bug
all labels in train ds from IMBD are 1
from torchtext.datasets import IMDB
train_iter = IMDB(split='train')
labels = [label for label, text in train_iter]
print(sum(labels)/len(labels))
1.0
pytorch 2.0.0+cu118
torchtext 0.15.1+cpu
Nayef Ahmed commented
Closing because this is a duplicate of #2041. Let's track the issue there