Deep Learning

1What deep learning is & when to use itHow deep learning differs from classic ML, where it shines, and when a simpler model is the smarter choice.8 min 2Tensors & autograd in PyTorchCreate and reshape tensors, move them to a GPU, and let autograd compute gradients for you.8 min 3Neural networks explainedNeurons, layers, activations and the intuition behind backpropagation and gradient descent.9 min 4Your first PyTorch modelBuild, train and evaluate a complete image classifier on Fashion-MNIST using nn.Module.8 min 5Training loops, loss functions & optimizersChoose the right loss, pick and tune optimizers and learning-rate schedules, and write a reusable training loop.9 min 6Overfitting & regularisationSpot overfitting and fix it with dropout, weight decay, early stopping, batch norm and more data.9 min 7Data pipelines: Dataset, DataLoader & augmentationLoad your own images or CSVs with custom Datasets, batch them efficiently and augment them correctly.8 min 8CNNs & transfer learningHow convolutions, pooling and pretrained backbones work, and how to fine-tune one on your own images.9 min 9Sequence models: RNN, LSTM & GRUModel ordered data such as daily sales or sentences with recurrent networks, and know their limits.9 min 10Attention & transformersUnderstand self-attention, multi-head attention and the transformer block that powers modern LLMs and vision models.10 min 11Deploying deep learning modelsExport with TorchScript or ONNX, shrink with quantisation, and serve predictions from FastAPI or Django.9 min 12Capstone: sweet image classifier end to endPlan, build, evaluate and deploy a production-style image classifier for a sweet shop, step by step.9 min