I want help finding the visualization tool can draw similar architecture as given in the image below

enter image description here

Keras visualization produces something similar to this graph. But I'm working in Pytorch. I tried using Netron and Graphviz, both produce graphs that do not show the branching and merging properly.

enter image description here

This is the plot I rendered in Graphviz for another branching and merging architecture.

Here's the python code I used to render the plot

graph {
    graph [rankdir=RL]
    input [label=Input]
    conv1 [label="Conv 1"]
    conv2 [label="Conv 2"]
    conv3 [label="Conv 3"]
    conv4 [label="Conv 4"]
    conv5 [label="Conv 5"]
    conv6 [label="Conv 6"]
    conv7 [label="Conv 7"]
    conv8 [label="Conv 8"]
    conv9 [label="Conv 9"]
    batch1 [label="Batch Norm1"]
    batch2 [label="Batch Norm2"]
    batch3 [label="Batch Norm3"]
    caps1_a [label="Caps 1(a)"]
    caps1_b [label="Caps 1(b)"]
    caps2_a [label="Caps 2(a)"]
    caps2_b [label="Caps 2(b)"]
    caps3_a [label="Caps 3(a)"]
    caps3_b [label="Caps 3(b)"]
    sum1 [label=Sum]
    sum2 [label=Sum]
    sum3 [label=Sum]
    stack [label=Stack]
    sum4 [label=Sum]
    softmax [label=Softmax]
    input -- conv1 [constraint=true]
    conv1 -- conv2 [constraint=false]
    conv2 -- conv3 [constraint=false]
    conv3 -- conv4 [constraint=false]
    conv4 -- conv5 [constraint=false]
    conv5 -- conv6 [constraint=false]
    conv6 -- conv7 [constraint=false]
    conv7 -- conv8 [constraint=false]
    conv8 -- conv9 [constraint=false]
    conv3 -- batch1 [constraint=false]
    conv6 -- batch2 [constraint=false]
    conv9 -- batch3 [constraint=false]
    batch1 -- caps1_a [constraint=false]
    batch2 -- caps2_a [constraint=false]
    batch3 -- caps3_a [constraint=false]
    caps1_a -- caps1_b [constraint=false]
    caps2_a -- caps2_b [constraint=false]
    caps3_a -- caps3_b [constraint=false]
    caps1_b -- sum1 [constraint=false]
    caps2_b -- sum2 [constraint=false]
    caps3_b -- sum3 [constraint=false]
    sum1 -- stack [label=w1 constraint=false]
    sum2 -- stack [label=w2 constraint=false]
    sum3 -- stack [label=w3 constraint=false]
    stack -- sum4 [constraint=false]
    sum4 -- softmax [constraint=false]
    input -- conv2 [label=k1 constraint=false]
    conv2 -- conv4 [label=k2 constraint=false]
    conv4 -- conv6 [label=k3 constraint=false]
    conv6 -- conv8 [label=k4 constraint=false]
  • $\begingroup$ Can you show the, dot / graphviz, code you used to create the image. Maybe you need some hidden nodes and edges and / or some things with rank = same / rankdir \ LR. $\endgroup$
    – albert
    May 28, 2020 at 9:06
  • $\begingroup$ I have added the code in the post $\endgroup$ May 28, 2020 at 13:14
  • $\begingroup$ This is python code and not graphviz / dot. $\endgroup$
    – albert
    May 28, 2020 at 13:40
  • $\begingroup$ I've edited it to the dot source now. Sorry $\endgroup$ May 28, 2020 at 15:59

2 Answers 2


You could use Tensorboard, via the tensorboardX interface.

That allows you to load models from PyTorch (and Chainer, MXNet etc.) into Tensorboard. This will then show you the full graph interactively.

From the homepage:

enter image description here


draw.io is a great visualization tool, it helped me draw diagrams as shown in the question.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.