C.W.K.
Stream
Quiz · 8 questions

🧠 Neural Networks — torch.nn

Modules, layers, parameters. Subclassing nn.Module is half the framework.

Level 0Tensor Curious
0 XP0/62 lessons0/13 achievements
0/120 XP to next level120 XP to go0% complete

Quiz

01Why call model(x) instead of model.forward(x)?
02What does model.eval() change?
03Which normalization is the standard in Transformer blocks?
04What goes wrong if you store submodules in a regular Python list?
05What's the shape of nn.Conv2d(3, 16, 3).weight?
06What does F.cross_entropy(logits, labels) expect as its first argument?
Hint
If the loss does softmax internally, what should you pass it?
07Which activation is the modern default for Transformer FFN blocks?
08What is a 'buffer' in nn.Module?
Spotted a bug or have feedback on this page?Report an Issue

Comments 0

🔔 Reply notifications (sign in)
Sign inPlease sign in to comment.

No comments yet — be the first.