jetcard icon indicating copy to clipboard operation
jetcard copied to clipboard

Published 20 hours ago • verified publisher icon NVIDIA-AI-IOT

Slow training in jupyter notebook on Jetson nano

Open gprashmi opened this issue 6 years ago • 2 comments

I’m running a PyTorch model in jupyter notebook on Jetson nano and it has been very slow. The actual training doesn’t even seem to start and everything just freezes when I run the training cell. Can someone please suggest a solution for this?

gprashmi avatar Jul 15 '19 21:07 gprashmi

Hi gprashmi,

Thanks for reaching out! Could you provide more information?

  1. What model are you attempting train?

  2. What batch size

Anything else that may be helpful in addressing this.

Thanks, John

jaybdub avatar Jul 23 '19 21:07 jaybdub

I am working on Dog breed classifier using CNN model with 5 convolution layers and max pool layers,3 Linear layers. The number of dog breed classes is 133 and the model batch size is 10, number of epochs is 5. I am facing the same issue when I run the train block for epoch=1

gprashmi avatar Jul 23 '19 21:07 gprashmi