search

f-dangel / cockpit

Cockpit: A Practical Debugging Tool for Training Deep Neural Networks
MIT License
474 stars 24 forks source link

Monai 3D Resnet models #21

Closed Mushtaqml closed 2 years ago

Mushtaqml commented 2 years ago

Hi!

Thank you for this project, indeed it is quite interesting work.

I am curious to know that if we can use it to interpret/observe the training of a 3D resnet model?

Regards

f-dangel commented 2 years ago

Hi,

thanks for your question. As many of Cockpit's quantities rely on BackPACK, your model architecture must be supported to get the full view. You can take a look at the supported layers here.

If your model is not supported by BackPACK, you can still configure a Cockpit with quantities that rely purely on PyTorch's autodiff. To do that, you can tweak the advanced exampe from the docs.

Best, Felix

Mushtaqml commented 2 years ago

@f-dangel Thank you for the response.

As it is a resnet model, I don't think BackPACK supports it. I am looking into the advanced example, and will update you in a while, if it works.

f-dangel commented 2 years ago

Sounds good.

By the way, BackPACK has basic support for ResNets from version 1.4.0 on. Check out this example to see how it works.

fsschneider commented 2 years ago

I am closing this issue as it is stale. Please feel free to open it again if you have further questions.