Gradient Checkpointing Quiz

Answers and explanations appear as you take the quiz.

1. What does gradient checkpointing primarily trade in order to save memory?
   • Extra compute (recomputation) for lower memory use
   • Model accuracy for speed
   • Disk space for network bandwidth
   • Precision for batch size
2. Why are activations normally stored during the forward pass?
   • To display them to the user
   • Because backpropagation needs them to compute gradients
   • To speed up data loading
   • They are never actually needed
3. In a simplified chain of N similarly sized layers, what can checkpoint boundaries spaced about sqrt(N) layers apart achieve in the textbook schedule?
   • It stores no activations and performs no recomputation
   • It guarantees constant memory for every model graph
   • Activation storage can scale on the order of sqrt(N) checkpoints
   • It makes optimizer-state memory scale as sqrt(N)
4. What most determines gradient-checkpointing runtime overhead in a practical model?
   • Only the model’s file extension
   • A fixed 20–30% slowdown for every model
   • The number of labels in the dataset
   • Which operations are recomputed, checkpoint placement, graph structure, and hardware
5. In PyTorch, which utility is commonly used to apply gradient checkpointing to a module?
   • torch.utils.checkpoint
   • torch.optim.SGD
   • torch.nn.Dropout
   • torch.cuda.empty_cache
6. What happens to operations inside a checkpointed region during backward?
   • Their gradients are skipped entirely
   • Their forward computations are rerun to regenerate needed activations
   • Their weights are permanently discarded
   • They are executed only during inference
7. Which training techniques can be combined with activation checkpointing to reduce different memory costs?
   • Only increasing the number of labels
   • Turning off backpropagation
   • FSDP for sharded model state and mixed precision for lower-precision tensors
   • Replacing the model with a tokenizer
8. What is selective activation checkpointing intended to choose?
   • Which training labels are removed
   • Which model outputs are published
   • Which GPU becomes the tokenizer
   • Which operations to save and which to recompute based on their memory and runtime costs