Keras

1. The Functional API

   Building models as graphs of layers with multiple inputs and outputs.

2. The Sequential model

   Stacking layers linearly and incrementally building a model.

3. Making new layers and models via subclassing

   The Layer and Model classes, build, call, and state.

4. Training and evaluation with built-in methods

   compile, fit, evaluate, predict, losses, metrics, and callbacks.

5. Customizing what happens in fit

   Overriding train_step and writing custom training logic.

6. Writing your own callbacks

   Hooks for logging, early stopping, and learning rate schedules.

7. Saving and serialization

   The .keras format, weights saving, and custom object serialization.

8. Transfer learning and fine-tuning

   Freezing layers and fine-tuning pretrained models.

9. Quantization

   Reducing model size and latency with quantization.

10. Distributed training

    Data and model parallel training with the Keras distribution API.

11. Multi-backend Keras

    Running Keras on JAX, TensorFlow, and PyTorch backends.

12. KerasHub and pretrained models

    Pretrained models and tasks from KerasHub.

Each section links to the official page. Titles and summaries by Extra Credit; the docs belong to their authors.