How do you visualize neural network architectures?

Question

When writing a paper / making a presentation about a topic which is about neural networks, one usually visualizes the networks architecture.

What are good / simple ways to visualize common architectures automatically?

See also: How can a neural network architecture be visualized with Keras? — Martin Thoma, Nov 17 '16 at 09:52
Just found https://www.reddit.com/r/MachineLearning/comments/4sgsn9/drawing_cnn_architectures/ — Martin Thoma, Mar 09 '17 at 10:10
I wrote Simple diagrams of convoluted neural networks with a survey of deep learning visualization approaches (both manual and automatic). I got a lot of inspiration, and links, from this thread - thx! — Piotr Migdal, Sep 17 '18 at 20:00

score 92 · Answer 1 · edited Aug 29 '22 at 01:27

92

I recently created a tool for drawing NN architectures and exporting SVG, called NN-SVG

edited Aug 29 '22 at 01:27

Christoph Rackwitz

103
4

answered May 10 '18 at 14:41

Alex Lenail

1,021
7
2

1

Download SVG doesn't work – image357 Jan 23 '19 at 16:13
works for me 1/23/19. If you're still having an issue, please feel free to open an issue. – Alex Lenail Jan 23 '19 at 23:28
2

this is the only right answer – ArtificiallyIntelligent Feb 23 '19 at 15:52
awesome tool. However, I noticed that in AlexNet style, the dimensions of the Tensors were mistakenly represented (width and height dimensions) – FlySoFast Apr 29 '19 at 16:01
5

Awesome, how can i visualize LSTM and attention? – keramat Jul 14 '20 at 18:53
@AlexLenail is this project open source? – Sabito stands with Ukraine Oct 05 '20 at 15:21
2

https://github.com/alexlenail/NN-SVG – Alex Lenail Oct 07 '20 at 02:07
Really awesome tool! – Javier TG Nov 21 '20 at 19:56
Does it support adding Batchnorm and Dropout layers? – Mayank Mishra Mar 06 '21 at 10:44

Martin Thoma · Answer 2 · 2019-09-23T08:55:47.800

52

Tensorflow, Keras, MXNet, PyTorch

If the neural network is given as a Tensorflow graph, then you can visualize this graph with TensorBoard.

Here is how the MNIST CNN looks like:

You can add names / scopes (like "dropout", "softmax", "fc1", "conv1", "conv2") yourself.

Interpretation

The following is only about the left graph. I ignore the 4 small graphs on the right half.

Each box is a layer with parameters that can be learned. For inference, information flows from bottom to the top. Ellipses are layers which do not contain learned parameters.

The color of the boxes does not have a meaning.

I'm not sure of the value of the dashed small boxes ("gradients", "Adam", "save").

edited Sep 23 '19 at 08:55

answered Jul 18 '16 at 19:59

Martin Thoma

18,880
35
95
169

it is good, I am trying to avoid the name like conv1, conv2 etc, I want to make all the name of conv later as CONV, How I will do?? – Sudip Das Mar 26 '18 at 13:03
+1. It's not only for TF though: MXNet and Pytorch have some support too – Jakub Bartczuk Jul 03 '18 at 16:08
@SudipDas You can add names in the code to the layers, which will show up as you plot it. – Ben Nov 27 '18 at 16:16
How I will show the name of each layer as "CONV", if I write it as "CONV" of each layer then I will get error, cause each layer should have a unique name as tf rules, BUT I want to know, is there any other way to overcome this problem?? @Ben – Sudip Das Nov 27 '18 at 16:22
@SudipDas Oh well, that does not work. I think you can only overcome this problem by storing it and editing it yourself. Maybe you can cheat by adding invisible characters to the name, but I would recommend giving them unique names anyway. – Ben Nov 27 '18 at 16:32
Yes, absolutely, Thank you so much for your reply @Ben – Sudip Das Nov 28 '18 at 13:04
The link is dead – onof Sep 23 '19 at 07:10
1

@onof I fixed the link – Martin Thoma Sep 23 '19 at 08:55
The link is dead again. – Psychotechnopath Jun 04 '21 at 14:24

score 33 · Answer 3 · edited Jun 16 '20 at 11:08

There is an open source project called Netron

Netron is a viewer for neural network, deep learning and machine learning models.

Netron supports ONNX (.onnx, .pb), Keras (.h5, .keras), CoreML (.mlmodel) and TensorFlow Lite (.tflite). Netron has experimental support for Caffe (.caffemodel), Caffe2 (predict_net.pb), MXNet (-symbol.json), TensorFlow.js (model.json, .pb) and TensorFlow (.pb, .meta).

Franck Dernoncourt · Answer 4 · 2020-12-29T23:41:57.220

21

In Caffe you can use caffe/draw.py to draw the NetParameter protobuffer:

In Matlab, you can use view(net)

Keras.js:

Also, see Can anyone recommend a Network Architecture visualization tool? (Reddit/self.MachineLearning).

edited Dec 29 '20 at 23:41

answered Jul 19 '16 at 00:43

Franck Dernoncourt

5,690
10
40
76

score 21 · Answer 5 · answered Mar 27 '18 at 16:04

I would add ASCII visualizations using keras-sequential-ascii (disclaimer: I am the author).

A small network for CIFAR-10 (from this tutorial) would be:

       OPERATION           DATA DIMENSIONS   WEIGHTS(N)   WEIGHTS(%)

           Input   #####     32   32    3
          Conv2D    \|/  -------------------       896     2.1%
            relu   #####     30   30   32
    MaxPooling2D   Y max -------------------         0     0.0%
                   #####     15   15   32
          Conv2D    \|/  -------------------     18496    43.6%
            relu   #####     13   13   64
    MaxPooling2D   Y max -------------------         0     0.0%
                   #####      6    6   64
         Flatten   ||||| -------------------         0     0.0%
                   #####        2304
           Dense   XXXXX -------------------     23050    54.3%
         softmax   #####          10

For VGG16 it would be:

       OPERATION           DATA DIMENSIONS   WEIGHTS(N)   WEIGHTS(%)

          Input   #####      3  224  224
     InputLayer     |   -------------------         0     0.0%
                  #####      3  224  224
  Convolution2D    \|/  -------------------      1792     0.0%
           relu   #####     64  224  224
  Convolution2D    \|/  -------------------     36928     0.0%
           relu   #####     64  224  224
   MaxPooling2D   Y max -------------------         0     0.0%
                  #####     64  112  112
  Convolution2D    \|/  -------------------     73856     0.1%
           relu   #####    128  112  112
  Convolution2D    \|/  -------------------    147584     0.1%
           relu   #####    128  112  112
   MaxPooling2D   Y max -------------------         0     0.0%
                  #####    128   56   56
  Convolution2D    \|/  -------------------    295168     0.2%
           relu   #####    256   56   56
  Convolution2D    \|/  -------------------    590080     0.4%
           relu   #####    256   56   56
  Convolution2D    \|/  -------------------    590080     0.4%
           relu   #####    256   56   56
   MaxPooling2D   Y max -------------------         0     0.0%
                  #####    256   28   28
  Convolution2D    \|/  -------------------   1180160     0.9%
           relu   #####    512   28   28
  Convolution2D    \|/  -------------------   2359808     1.7%
           relu   #####    512   28   28
  Convolution2D    \|/  -------------------   2359808     1.7%
           relu   #####    512   28   28
   MaxPooling2D   Y max -------------------         0     0.0%
                  #####    512   14   14
  Convolution2D    \|/  -------------------   2359808     1.7%
           relu   #####    512   14   14
  Convolution2D    \|/  -------------------   2359808     1.7%
           relu   #####    512   14   14
  Convolution2D    \|/  -------------------   2359808     1.7%
           relu   #####    512   14   14
   MaxPooling2D   Y max -------------------         0     0.0%
                  #####    512    7    7
        Flatten   ||||| -------------------         0     0.0%
                  #####       25088
          Dense   XXXXX ------------------- 102764544    74.3%
           relu   #####        4096
          Dense   XXXXX -------------------  16781312    12.1%
           relu   #####        4096
          Dense   XXXXX -------------------   4097000     3.0%
        softmax   #####        1000

score 18 · Answer 6 · answered Jan 22 '18 at 10:48

18

Keras

The keras.utils.vis_utils module provides utility functions to plot a Keras model (using graphviz)

The following shows a network model that the first hidden layer has 50 neurons and expects 104 input variables.

plot_model(model, to_file='model.png', show_shapes=True, show_layer_names=True)

answered Jan 22 '18 at 10:48

mingxue

281
2
3

1

Can I use it in LaTex compatible format? – hola Jul 11 '19 at 00:01
they use it so obviously you can.. probably just embed the image like any other figure – sivi Feb 13 '20 at 16:38

score 13 · Answer 7 · answered Dec 11 '17 at 12:33

13

Here is yet another way - dotnets, using Graphviz, heavily inspired by this post by Thiago G. Martins.

dotnets example

answered Dec 11 '17 at 12:33

bytesinflight

231
2
4

Jesse · Answer 8 · 2020-01-29T15:18:45.327

13

I've been working on a drag-and-drop neural network visualizer (and more). Here's an example of a visualization for a LeNet-like architecture. Models with fan-out and fan-in are also quite easily modeled. You can visit the website at https://math.mit.edu/ennui/

The open-source implementation is available at https://github.com/martinjm97/ENNUI.

edited Jan 29 '20 at 15:18

answered Apr 09 '19 at 19:52

Jesse

231
2
4

my browser keeps crashing when press Train – Dan D. Sep 25 '19 at 04:25
1

Thanks for checking it out. Yes, this bug just popped up recently and seems to be a result of some recent changes to WebGL on Chrome. Everything should work on Firefox. I'll update you when I know more. – Jesse Sep 25 '19 at 20:19
tks, your visualiser is amazing, looks greater than tf playground :) – Dan D. Sep 26 '19 at 01:50
1

Thank you! Let me know if you have issues or ideas. We have fun things like code generation too! – Jesse Sep 26 '19 at 16:01
1

Bug fixes are in and the implementation has been open-sourced! – Jesse Jan 29 '20 at 15:19

score 13 · Answer 9 · answered Oct 12 '20 at 09:27

13

PlotNeuralNet LaTex tool

This solution is not automatically generated (you need to construct the graph by yourself) but the PlotNeuralNet github repo allows you to build images directly from LaTex, and the result is great ! See for example the image below from the README :

or my example :

answered Oct 12 '20 at 09:27

Rémi Boutin

131
1
2

2

This is a really good visualization! Does it require the user to have Latex downloaded on the hard drive, or is it possible to this through Overleaf / ShareLatex? I am having some problems getting it to work on Overleaf. – SimpleProgrammer Apr 21 '21 at 16:38
Hello, They do have examples running on Overleaf in the Readme file of the package. It should not be a problem. – Rémi Boutin Apr 23 '21 at 11:52

score 12 · Answer 10 · answered Mar 05 '18 at 12:34

The Python package conx can visualize networks with activations with the function net.picture() to produce SVG, PNG, or PIL Images like this:

Conx is built on Keras, and can read in Keras' models. The colormap at each bank can be changed, and it can show all bank types.

More information can be found at: http://conx.readthedocs.io/en/latest/

score 10 · Answer 11 · answered Jul 21 '16 at 17:32

In R, nnet does not come with a plot function, but code for that is provided here.

Alternatively, you can use the more recent and IMHO better package called neuralnet which features a plot.neuralnet function, so you can just do:

data(infert, package="datasets")
plot(neuralnet(case~parity+induced+spontaneous, infert))

neuralnet is not used as much as nnet because nnet is much older and is shipped with r-cran. But neuralnet has more training algorithms, including resilient backpropagation which is lacking even in packages like Tensorflow, and is much more robust to hyperparameter choices, and has more features overall.

You should add the updated link for the code of NNet in R https://beckmw.wordpress.com/2013/11/14/visualizing-neural-networks-in-r-update/ — wacax, May 10 '18 at 16:47

score 7 · Answer 12 · edited Jul 03 '18 at 15:10

7

There are some novel alternative efforts on neural network visualization.

Please see these articles:

Stunning 'AI brain scans' reveal what machines see as they learn new skills

Inside an AI 'brain' - What does machine learning look like?

These approaches are more oriented towards visualizing neural network operation, however, NN architecture is also somewhat visible on the resulting diagrams.

Examples:

edited Jul 03 '18 at 15:10

EliaCereda

103
2

answered May 17 '17 at 19:02

VividD

656
7
18

40

Please explain what we see here. It looks beautiful, but I don't understand how the fancy images support understanding the operation of the network. – Martin Thoma Mar 27 '18 at 17:15
I don't like your derogatory usage of "fancy images" term. @Martin – VividD Mar 27 '18 at 17:25
One could call your diagram "lego boxes" as well, and ask the same question. This is not an appropriate conversation style for this site. – VividD Mar 28 '18 at 07:06
21

I didn't mean to attack you, but your overly defensive answer without actually answering my question speaks for itself. - I added an "interpretation" part to the "lego boxes" diagram. – Martin Thoma Mar 28 '18 at 07:29
2

By the way: The second link is dead. – Martin Thoma Mar 28 '18 at 07:37
10

@MartinThoma It's clearly data art, not data viz (vide https://lisacharlotterost.github.io/2015/12/19/Meaning-and-Beauty-in-Data-Vis/). – Piotr Migdal Apr 02 '18 at 14:04
3

Not sure how is this useful, in fact those labels could be anything. – phoxis Jan 10 '19 at 15:04
To like is one thing, but isn't it accurate to call the images fancy? I think it's an appropriate conversation style if it conveys the intended message, and in this case I guess that was that the images look more interesting than what they are useful. When it comes to the lego boxes (which I think would be a fair description), they communicate certain aspects of the architecture very effectively (number of layers, shape of layers, connections, etc.), which is not something I immediately see in these images. If you become good at interpreting them, it might be another thing, though. – HelloGoodbye Jan 24 '23 at 13:00

score 5 · Answer 13 · answered Jan 22 '18 at 12:00

5

You can read the popular paper Understanding Neural Networks Through Deep Visualization which discusses visualization of convolutional nets. Its implementation not only displays each layer but also depicts the activations, weights, deconvolutions and many other things that are deeply discussed in the paper. It's code is in caffe'. The interesting part is that you can replace the pre-trained model with your own.

answered Jan 22 '18 at 12:00

Green Falcon

14,058
9
57
98

that's about showing the weights/activations, not the structure... at least from what I can see in the paper and the repo's readme – Christoph Rackwitz Aug 28 '22 at 09:54
@ChristophRackwitz Yes actually. Most of the other answers are about the structure. Mine is about what you've mentioned. By the way, it was about visualisation of neural nets. – Green Falcon Aug 28 '22 at 12:45

Ali Mirzaei · Answer 14 · 2019-01-26T14:43:12.020

5

Tensorspace-JS is a fantastic tool for 3d visualization of network architecture:

https://tensorspace.org/

and here is a nice post about how to write a program:

https://medium.freecodecamp.org/tensorspace-js-a-way-to-3d-visualize-neural-networks-in-browsers-2c0afd7648a8

edited Jan 26 '19 at 14:43

answered Jan 25 '19 at 07:56

Ali Mirzaei

151
1
4

Could you provide a link to this tool? – Piotr Migdal Jan 25 '19 at 14:34
1

@PiotrMigdal I updated the answer. – Ali Mirzaei Jan 26 '19 at 14:43

score 5 · Answer 15 · answered Oct 05 '20 at 09:56

Tensorflow / Keras / Python

I wrote a small python package called visualkeras that allows you to directly generate the architecture from your keras model.

Install via pip install visualkeras

And then it's as simple as:

import visualkeras
visualkeras.layered_view(<model>)

There are lots of options to tweak it and I am working on more visualizations. Also, always open for PRs or feature requests.

Here's what VGG16 looks like:

Thomas Wagenaar · Answer 16 · 2017-05-15T18:41:41.730

4

Not per se nifty for papers, but very useful for showing people who don't know a lot of about neural networks what their topology may look like. This Javascript library (Neataptic) lets you visualise your network:

edited May 15 '17 at 18:41

answered Apr 10 '17 at 08:22

Thomas Wagenaar

1,128
8
7

score 4 · Answer 17 · answered Jan 25 '19 at 13:31

4

Netscope is my everyday tool for Caffe models.

answered Jan 25 '19 at 13:31

Dmytro Prylipko

836
5
10

score 2 · Answer 18 · answered May 29 '22 at 07:15

2

I have found one amazing website. You just need to upload your h5 model, Then you will get a beautiful visualization within a few seconds. Check it out!

answered May 29 '22 at 07:15

Aravind R

136
3

1

Thanks, already covered in https://datascience.stackexchange.com/a/30642/843 – Franck Dernoncourt May 29 '22 at 07:20

Eduardo Lozano · Answer 19 · 2020-06-12T00:05:35.960

1

You can use eiffel2, which you can install using pip:

python -m pip install eiffel2

Just import builder from eiffel and provide a list of neurons per layer in your network as an input.

Example:

from eiffel2 import builder

builder([1, 10, 10, 5, 5, 2, 1])
# or the following if you want to have a dark theme
builder([1, 10, 10, 5, 5, 2, 1], bmode="night")

Output:

To see more about eiffel2 visit the Github repository:

https://github.com/Ale9806/Eiffel2/blob/master/README.md

edited Jun 12 '20 at 00:05

answered Jun 09 '20 at 10:16

Eduardo Lozano

9
2

I just figured out Eiffel does not have support anymore, use eiffel2 instead – Eduardo Lozano Jun 10 '20 at 02:28
1

Session crashed. – keramat Jul 14 '20 at 18:57

score 0 · Answer 20 · answered Mar 19 '23 at 10:30

For a solution for PyTorch I'd add TorchView.

It is as easy as:

from torchview import draw_graph
model = MLP()
batch_size = 2
device='meta' -> no memory is consumed for visualization
model_graph = draw_graph(model, input_size=(batch_size, 128), device='meta')
model_graph.visual_graph

Which yields:

It has many customization options as well.

psychicmachinist · Answer 21 · 2023-12-15T08:52:26.853

I'll add a plug for my recent project, TorchExplorer (live demo here). It's sort of a combination of Netron and wandb.watch. It can:

Interactively traverse model architectures, showing input/output tensor sizes and module parameters
Visualize module input/output tensors, parameters, and associated gradients as histograms over the course of training (modeled off of wandb.watch)
Directly integrate with weights and biases or serve standalone with a simple torchexplorer.watch call

How do you visualize neural network architectures?

21 Answers21

Tensorflow, Keras, MXNet, PyTorch

Interpretation

PlotNeuralNet LaTex tool

Tensorflow / Keras / Python

device='meta' -> no memory is consumed for visualization

Linked

Related