get_model_summary

[cyrusmvahid]

Description

I would like to print my model summary (keras style) or access the network and take a look at layers, weights, collect_params, and number of parameters.
It would be nice to add a get_network_summary or an immutable network object returned by the estimator.

[geoalgo]

Hi @cyrusmvahid

The example examples/evaluate_model.py can be adapted as follow:

train_transformation, train_model = estimator.train_model(dataset.train)
print(train_model.collect_params())

This will give you access to network weights if you want to investigate weights.

For printing model summary, we could do something but the print will be dependent on whether the network is hybridized or not.
E.g., we could add in the trainer code something like this that may do the trick:

...
loss.backward()
trainer.step(batch_size)

first_batch = batch_no == 1 and epoch_no == 0

if not self.hybridize and first_batch:
    net.summary(*inputs)