[Errno 20] Not a directory: 'data/MFo/train/.DS_Store'

I got a directory error. Is there any specific directory structure that ConvLearner function expects.
I created dataset with train and valid splits. Some things which I didn’t do while creating directory 1. creating a test set 2. sample set. Here is screenshot ,

you need to delete hiden OS`s .DS_Store files in you train/valid folders

Thanks but in previous version of this course we used to use glob which use listdir and fnmatch under the hood and also checks for a leading ‘.’, not by using fnmatch. This helped me a lot during migrating data to server side using Jupyter. I was curious if we are using same in the Convlearner function.

I did not check glob. Deleting those files just helped in my case.

1 Like