I am trying to understand the Peoplenet(https://ngc.nvidia.com/catalog/models/nvidia:tlt_peoplenet) model given in the TLT framework.
In the NGC container, Nvidia hosts 2 flavours of the model unpruned and pruned. Starting from the unpruned weights, it is possible to prune and retrain the model to get a pruned version. https://developer.nvidia.com/blog/training-custom-pretrained-models-using-tlt/
My query is what is the pruning threshold and other parameters used in the hosted pruned model? Because depending on the parameters like pruning threshold the architecture is getting changed, which might affect the accuracy.
Does anyone has any ideas please share the pruning parameters used or Is this protected by Nvidia?