Mixed precision training;

Gradient accumulation strategies;

Memory checkpointing for large models;

PyTorch GPU acceleration;

Tensor operations with PyTorch;

PyTorch optimization techniques;

Quiz #2;

Introduction to Dask

Distributed GPU processing;

Dask-CUDA integration;

Scaling workloads beyond one GPU;

RAPIDS ecosystem introduction;

cuDF, cuML, cuGraph fundamentals;

Single-GPU workflow examples;

Computational bottlenecks;

Warp divergence & occupancy;

Bandwidth-bound vs compute-bound tasks;