Utilities

General utilities for models, training, and visualization.

Models

ConvNet([num_classes, dropout_rate, ...])	Simple CNN for MNIST-sized inputs (28x28).
ResNet18([num_classes, input_channels])	ResNet-18 for CIFAR-10 sized inputs (32x32).
MLP(input_dim[, hidden_dims, num_classes, ...])	Multi-layer perceptron for tabular data.

Training

train_epoch(model, train_loader, criterion, ...)	Train model for one epoch.
seed_everything([seed])	Set random seeds for reproducibility.

Visualization