Fastai on Kaggle Kernels

Hello all! I was trying to implement the fastai lessons on kaggle kernels by uploading the notebook and running it on the kernels. I would get various different errors like bus errors and other errors that seem to be memory-related. This was true even when I decreased the batch size. I found this kernel though which seems to be the same exact notebook like the one on GitHub. Yet it is so far running fine, as I write this post. What is the difference between these two kernels, and what will I have to do to avoid these errors in the future?

In addition, I have noticed that the training speed on Kaggle Kernels is quite slow compared to those demonstrated by Jeremy in the video. Why is that the case?

Did you enable GPU support for the kernel? Also I’m not sure, but you may need to enable internet to be able to download the pretrained models.

yep I had both GPU and internet enabled… it had downloaded the data and the pretrained models perfectly fine

Can you share that kernel?