python -m experimental_experiment lighten and unlighten

lighten

    usage: lighten [-h] -i INPUT -o OUTPUT [-v VERBOSE]
    
    Removes the weights from a heavy model, stores statistics to restore random
    weights.
    
    options:
      -h, --help            show this help message and exit
      -i INPUT, --input INPUT
                            onnx model to lighten
      -o OUTPUT, --output OUTPUT
                            onnx model to output
      -v VERBOSE, --verbose VERBOSE
                            verbosity
    
    This is mostly used to write unit tests without adding a big onnx file to the
    repository.

unlighten

    usage: unlighten [-h] -i INPUT -o OUTPUT [-v VERBOSE]
    
    Restores random weights for a model reduces with command lighten, the command
    expects to find a file nearby with extension '.stats'.
    
    options:
      -h, --help            show this help message and exit
      -i INPUT, --input INPUT
                            onnx model to unlighten
      -o OUTPUT, --output OUTPUT
                            onnx model to output
      -v VERBOSE, --verbose VERBOSE
                            verbosity
    
    This is mostly used to write unit tests without adding a big onnx file to the
    repository.