Hi there,
every time i use the “python train.py” command from the “Hello AI World” to retrain a neural-network with my own dataset, i get a “Segmentation fault (core dumped)” after the last epoch and after the model is saved.
Does anyone know why this occurs and if it has any effect on the retrained models perofmance, or is it something i can ignore?
Ty,
Michael.
Hi @michaelmueller1994, you can safely ignore it, as the error only occurs when PyTorch is done running and Python is unloading the modules. It doesn’t impact the models at all, and the issue has been fixed in the PyTorch wheel that will be posted in the upcoming JetPack 4.4 production release. Apologies for the inconvienience.
Thank you so much for the quick reply :)