Pruned LeNet300-100 using Iterative pruning based on https://arxiv.org/abs/1506.02626
Steps to follow:
-
create conda enviornment with tensorflow>2.0 and matplotlib.
-
Run main.py and update quality_parameter value to change pruning theshold. (Iterative training (prune+retrain) in single iteration and run until we get pruning accuracy >= original model accuracy. Threshold is calculated based on standard deviation of weights and sensitivity/quality factor).
-
Results are uploaded in results folder. original model, pruned model, and retarin model accuracy & compression rate during iteraive pruning.
![Pruning Results]
