How to prune LPRnet model?

I have trained a model with lprnet and want to get advantage of the pruning. I have tried to prune the model with

!lprnet prune
usage: lprnet [-h] [--gpus GPUS] [--gpu_index GPU_INDEX [GPU_INDEX ...]]
              [--use_amp] [--log_file LOG_FILE]
              {evaluate,export,inference,train} ...
lprnet: error: invalid choice: 'prune' (choose from 'evaluate', 'export', 'inference', 'train')

I see there is us_lprnet_pruned model in NGC. Would like to know if I can prune my LPRnet model?

LPRnet does not support pruning yet.

Usually the inference speed is quick for LPRnet.
Do you have any idea why you need pruning?