This tutorial is best followed in the order below, as later topics build on concepts introduced earlier.
Loading premade datasets, building custom and robust datasets, augmenting data, and constructing DataLoaders.
data.md
Training baseline models, evaluating performance, building modular networks, preventing overfitting, and applying transfer learning.
accuracy.md
Improving training with optimizers, learning-rate schedulers, regularization, and hyperparameter tuning.
optimization.md
Understanding model predictions using saliency maps, Grad-CAM, and feature visualizations.
interpretability.md
Understanding performance bottlenecks, data loading efficiency, and gradient accumulation.
efficiency.md
Exporting using ONNX and preparing models for real-world inference through pruning and quantization.
deployment.md