CODEWITHSUNDEEP
pythonmachine-learningneural-networkdeep-learningpytorch
Infintyyy
Infintyyy

Reputation: 959

How to iterate over layers in Pytorch

Let's say I have a network model object called m. Now I have no prior information about the number of layers this network has. How can create a for loop to iterate over its layer? I am looking for something like:

Weight=[]
for layer in m._modules:
    Weight.append(layer.weight)

Upvotes: 26

Views: 52862

Answers (5)

iacob
iacob

Reputation: 24251

You can use the children method:

for module in model.children():
    # ...

Or, if you want to flatten Sequential layers:

for module in model.modules():
    if not isinstance(module, nn.Sequential):
        # ...

Upvotes: 5

Charlie Parker
Charlie Parker

Reputation: 5109

you can do this too:

for name, m in mdl.named_children():
    print(name)
    print(m.parameters())

Reference:

# https://discuss.pytorch.org/t/how-to-get-the-module-names-of-nn-sequential/39682
# looping through modules but get the one with a specific name

import torch
import torch.nn as nn

from collections import OrderedDict

params = OrderedDict([
    ('fc0', nn.Linear(in_features=4,out_features=4)),
    ('ReLU0', nn.ReLU()),
    ('fc1L:final', nn.Linear(in_features=4,out_features=1))
])
mdl = nn.Sequential(params)

# throws error
# mdl['fc0']

for m in mdl.children():
    print(m)

print()

for m in mdl.modules():
    print(m)

print()

for name, m in mdl.named_modules():
    print(name)
    print(m)

print()

for name, m in mdl.named_children():
    print(name)
    print(m)

Upvotes: 3

oopsi
oopsi

Reputation: 2039

Assuming m is your module, then you can do:

for layer in m.children():
    weights = list(layer.parameters())

Upvotes: 8

kmario23
kmario23

Reputation: 61375

You can simply get it using model.named_parameters(), which would return a generator which you can iterate on and get the tensors, its name and so on.

Here is the code for resnet pretrained model:

In [106]: resnet = torchvision.models.resnet101(pretrained=True)

In [107]: for name, param in resnet.named_parameters(): 
     ...:     print(name, param.shape)

which would output

conv1.weight torch.Size([64, 3, 7, 7])
bn1.weight torch.Size([64])
bn1.bias torch.Size([64])
layer1.0.conv1.weight torch.Size([64, 64, 1, 1])
layer1.0.bn1.weight torch.Size([64])
layer1.0.bn1.bias torch.Size([64])
........
........ and so on

You can find some discussion on this topic in how-to-manipulate-layer-parameters-by-its-names/

Upvotes: 3

Wasi Ahmad
Wasi Ahmad

Reputation: 37721

Let's say you have the following neural network.

import torch
import torch.nn as nn
import torch.nn.functional as F

class Net(nn.Module):

    def __init__(self):
        super(Net, self).__init__()
        # 1 input image channel, 6 output channels, 5x5 square convolution
        # kernel
        self.conv1 = nn.Conv2d(1, 6, 5)
        self.conv2 = nn.Conv2d(6, 16, 5)
        # an affine operation: y = Wx + b
        self.fc1 = nn.Linear(16 * 5 * 5, 120)
        self.fc2 = nn.Linear(120, 84)
        self.fc3 = nn.Linear(84, 10)

    def forward(self, x):
        # define the forward function 
        return x

Now, let's print the size of the weight parameters associated with each NN layer.

model = Net()
for name, param in model.named_parameters():
    print(name, param.size())

Output:

conv1.weight torch.Size([6, 1, 5, 5])
conv1.bias torch.Size([6])
conv2.weight torch.Size([16, 6, 5, 5])
conv2.bias torch.Size([16])
fc1.weight torch.Size([120, 400])
fc1.bias torch.Size([120])
fc2.weight torch.Size([84, 120])
fc2.bias torch.Size([84])
fc3.weight torch.Size([10, 84])
fc3.bias torch.Size([10])

I hope you can extend the example to fulfill your needs.

Upvotes: 19

Related Questions