Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Tensor Visualization #122

Closed
cybertk opened this issue Jul 9, 2018 · 5 comments
Closed

Tensor Visualization #122

cybertk opened this issue Jul 9, 2018 · 5 comments

Comments

@cybertk
Copy link

cybertk commented Jul 9, 2018

Sometimes, we need to inspect the layer's weights/bias, or the statistics info such as mean, standard error and etc.

Is it possible to visualize these info in Netron?

@lutzroeder
Copy link
Owner

Weights will show if they are stored in the model file. What model format are you using and can you share a sample file? #65 is tracking renderings other than plain numbers.

Stats are usually not stored in model files.

@cybertk
Copy link
Author

cybertk commented Jul 11, 2018

I can see the raw weights/bias data in the panel, but it's hard to read and understand especially for a large size kernel in the fully connected layer.

Actually what I asked is the plot/picture you mentioned in #65, in addition to some statistics info. It will help a lot to understand the tensor data.

Another proposal is exporting/save the tensor to a file, then use the existing tools to visualize them.

@lutzroeder
Copy link
Owner

lutzroeder commented Jul 12, 2018

If you have ideas what this should look like feel free to add them to #65 and #125. For exporting a tensor, what format do you have in mind and which tools are you using for visualization?

@nrupatunga
Copy link

In mxnet, models weights are saved in .params and architecture in .symbol,
And the weights are shown empty, How do we visualize the weights?

@lutzroeder
Copy link
Owner

@nrupatunga created #285 to track this separately.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants