Is re-training required on a model pruned using TFMOT?

Question

I am trying to prune a pre-trained model using TFMOT (Tensorflow model Optimization ToolKit). Is it necessary to re-train the pruned model to get a reduced gzip size ? Witho