Load and optimize package

In this example, we show how to optimize an existing .nav package. After the optimization is completed, a new package is saved.

We recommend running this example in NVIDIA NGC PyTorch or TensorFlow2 container.

For PyTorch nav package run

./optimize.py --input-path torch_linear.nav --output-path optimized_torch_linear.nav --no-defaults

or for TensorFlow2 package:

./optimize.py --input-path tensorflow_linear.nav --output-path optimized_tensorflow_linear.nav --no-defaults

The --no-defaults sets the optimize parameters to the package parameters. Default behaviour is to use default parameters.