I am having an issue with my Jupyter Notebook restarting when trying to execute a data loader command (specifically, SegmentationDataLoaders.from_label_fun). I was able to run run all the examples in lesson 1, so I know I have a functioning installation of fastai v2 on my Windows computer. However, for some reason with my own data I am having an issue.
I have read online that it could be a CPU memory issue, but just visually monitoring the memory, it’s not close to maxing out. I have reduced batch size and have the num_workers=0 code in as well.
Thoughts on how to debug? I am not getting any info on the error causing the restart.
This is on a Windows 10 machine. Error I get is:
‘The kernel appears to have died. It will restart automatically.’
What is your installation config? and do you have a GPU?
Just to be safe try a conda install, it worked best for me. Also most of fastai runs on gpu, idk why you would have a cpu memory error.
I installed it via pip. I have a GeoForce GTX Titan X on Windows 10. I am able to run the Lesson 1 2020 class. I am not sure why my it crashes when using my own data…
Could you maybe share your code and the error stacktrace?
Also how did you install CUDA?
Sorry for the late response, but I moved over to Linux to get it to work.
@al24 Hi, I’m facing the some problem, I’m working on windows 11 with just CPU: Intel(R) Core™ i5-10210U CPU @ 1.60GHz 2.11 GHz 16 GB.
what was the main cause for this ? and if it works for you in linux, could you please explain why? It’s related to what factors?