import torch
import torchvision
import torchvision.transforms as transforms
import torch.nn as nn
from torch.autograd import Variable
from torch.utils.data.sampler import SubsetRandomSampler
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline

from torchvision.models import vgg16_bn # bn = batch normalization
import copy

vgg_base = vgg16_bn(pretrained=True)
# freeze old layers
for param in vgg_base.parameters():
    param.requires_grad = False

#print(vgg_base)
for m in vgg_base.modules():
    if isinstance(m, nn.Conv2d):
        first_weights = m.weight.data
        #print(first_weights)
        break

fig = plt.figure()
plt.figure(figsize=(10,10))
for index, filter in enumerate(first_weights):
    print(type(filter))
    #print(type(np.abs(filter)))
    plt.subplot(8, 8, index+1)
    plt.imshow(filter[:,:,:].numpy())
    plt.axis('off')

fig.show()