Hi everyone,
thanks for the really cool library.
The RandomCrop augmentation forces my data to be on the CPU. After looking into the code (and some of the similar issues here), it seems that the samples_cropped variable needs to have its device specified. I will submit a pull request soon with the change, let me know if I should write a test for it (not very familiar with testing but I can try).
Hi everyone, thanks for the really cool library. The
RandomCrop
augmentation forces my data to be on the CPU. After looking into the code (and some of the similar issues here), it seems that thesamples_cropped
variable needs to have its device specified. I will submit a pull request soon with the change, let me know if I should write a test for it (not very familiar with testing but I can try).