http://d2l.ai/chapter_deep-learning-computation/read-write.html
Exercises and my silly answers
- Even if there is no need to deploy trained models to a different device, what are the practical
benefits of storing model parameters?
- we can create checkpoints within the code and go back if required.
- Assume that we want to reuse only parts of a network to be incorporated into a network
of a different architecture. How would you go about using, say the first two layers from a
previous network in a new network?
- this works
net[0].weight = nn.Parameter(clone.state_dict()['linear1.weight'])
net[1] = another_net[1]
- How would you go about saving the network architecture and parameters? What restrictions
would you impose on the architecture
- this works
torch.save(net, 'net')and then load it into a new network.