Exercises

The following exercises are programming challenges, combining the expressive power of the TensorFlow Python API and the advantages brought by other programming languages:

1. What is a checkpoint file?
2. What is a SavedModel file?
3. What are the differences between a checkpoint and a SavedModel?
4. What is a SignatureDef?
5. Can a checkpoint have a SignatureDef?
6. Can a SavedModel have more than one SignatureDef?
7. Export a computational graph as a SavedModel that computes the batch matrix multiplication; the returned dictionary must have a meaningful key value.
8. Convert the SavedModel defined in the previous exercise into its TensorFlow.js representation.
9. Use the model.json file we created in the previous exercise to develop a simple web page that computes the multiplication of matrices chosen by the user.
10. Restore the semantic segmentation model defined in Chapter 8, Semantic Segmentation...