Quiz · 7 questions

🔷 Tensors — The Foundation

Shape, dtype, device, stride. The four numbers you'll check ten thousand times.

Level 0Tensor Curious

0 XP0/62 lessons0/13 achievements

0/120 XP to next level120 XP to go0% complete

01What does torch.zeros_like(x) guarantee compared to torch.zeros(x.shape)?

Hint

Think about what makes 'expected cuda:0 got cpu' errors disappear.

02Which operator performs matrix multiplication?

03After tt = t.transpose(0, 1) on a contiguous 2D tensor, which is true?

Hint

Did the bytes move, or did the recipe for walking them change?

04Why do you need .detach() in output.detach().cpu().numpy()?

05On Apple Silicon, what's the recommended check for the GPU?

06Which dtype combo is the modern default for mixed-precision training on Ampere+ GPUs?

Hint

What dtype eliminates the need for a GradScaler?

07What's the trailing underscore convention in PyTorch (add_, zero_)?

Spotted a bug or have feedback on this page?Report an Issue

Comments 0

No comments yet — be the first.