Volume 7 — Deep Learning¶
Neural networks, understood
Neural networks from the perceptron to the transformer, with backpropagation derived rather than asserted, and the practical training concerns that dominate real work.
Outlined, not yet written
This volume is fully outlined but the lessons are not yet authored. The chapter list below is the plan of record, not a promise that content exists.
What you will be able to do¶
- Explain backpropagation without hand-waving
- Diagnose a training run that is not converging
- Read a model architecture diagram and implement it
- Fine-tune a pretrained model responsibly
Chapters¶
- From linear models to neural networks
- Backpropagation, derived
- Activation functions and initialisation
- Optimisers and learning rates
- Regularisation and overfitting
- Convolutional networks
- Sequence models and attention
- The transformer architecture
- Training practicalities: batching, mixed precision, checkpoints
- Transfer learning and fine-tuning