Deep Learning with PyTorch

This chapter covers PyTorch from tensors to production training loops. Each section includes runnable code examples.

• Tensors: Creating tensors, dtypes, device placement, views vs copies, broadcasting
• Autograd: Computational graphs, backward(), gradient accumulation, custom autograd functions
• nn.Module: Module anatomy, parameter registration, saving/loading state
• Data Loading: Dataset, DataLoader, IterableDataset, collate_fn, streaming
• Training Loop: Complete training loop with mixed precision, checkpointing, and logging
• Debugging: Anomaly detection, shape debugging, NaN detection, memory profiling