[FIXED] issue terminate called after throwing an instance of 'c10::CUDAError'

Issue

I am having issues with running yolov5 on colab. I was able to run the code fine when I had I had more classes, and a slightly smaller dataset, but now I have decreased the amount of classes and 70 instances when the overall one has 3400 instances. Now I am getting this error.

terminate called after throwing an instance of 'c10::CUDAError'

Other times I will get

cuda assertion index >= -sizes[i] && index < sizes[i] && index out of bounds

any idea what could be causing this and how to fix it?

Solution

The issue was that I was not outputting any labels that I said existed. So for example if I claimed that there would be labels "0","1","2" in the training dataset. There was no instances of label "2".

Answered By - Nikita Belooussov