Issue with using sparse matrix representation for a pruned model

Soumya_Kundu · September 12, 2024, 7:26am

Similar thread discussing the same.

gist.github.com

https://gist.github.com/aymuos15/10c8348148d3de5954a6b4cc5cfebf2c

Pruning_nuances_in_pytorch.py

import torch
import torch.nn as nn
import torch.nn.utils.prune as prune
import numpy as np
import matplotlib.pyplot as plt

class SimpleCNN(nn.Module):
    def __init__(self, conv1_out=32, conv2_out=64):
        super(SimpleCNN, self).__init__()
        self.conv1 = nn.Conv2d(1, conv1_out, 3, 1)

This file has been truncated. show original

This link shows how to actually remove the weights via structured pruning.