Learn PyTorch

Get started with PyTorch exercises. Learn to read the requirements, interpret test failures, and recognize checks for behavior beyond numerical output.

Understand tensor shapes, strides and shared storage. Practise view and reshape, and learn when a shape change can reuse memory or needs a copy.

Select tensor elements with slices, index arrays and masks. Distinguish views from copies and understand how indexing affects shape and shared storage.

Predict which tensor shapes can broadcast and what shape the result will have. Add dimensions deliberately and catch unintended pairwise operations.

Reduce tensors along selected dimensions and use keepdim to preserve axes. Handle dtypes, numerical accuracy and ties when choosing reduction operations.

Express matrix products, batch dimensions and tensor contractions with matmul and einsum. Check dimension labels and broadcasting before computing a result.

Compare clone, detach, copy_, output buffers and in-place updates. Track shared storage and learn when mutation conflicts with the required behavior.

Choose tensor dtypes and reason about promotion, division, rounding and accumulated error. Compare floating-point results with suitable tolerances.

Inspect automatic differentiation graphs and gradients. Practise leaf tensors, gradient accumulation, no_grad and detach, and understand what each changes.

Write code that respects tensor devices and avoids unnecessary data movement. Examine transfer and copy behavior using the checks provided by each exercise.

Define modules with parameters, buffers and child modules. Save and restore state, switch training modes, and reason about shared weights.

Assemble batches with stack, cat, padding and masks. Write collation logic and make shuffling reproducible while preserving the intended shapes.

Measure tensor-code performance and investigate unnecessary loops, allocations and synchronization. Compare changes using the benchmarks available for each exercise.