According to the main page, the e4e folder is part of the unknown methods, which should have been made with the CelebA dataset and have the real data included in the in the folder, but the folder structure is 3 folders with images:
e4e/cdf/Celeb-real/inversions/XXXXX.jpg
e4e/cdf/Youtube-real/inversions/XXXXX.jpg
e4e/ff/XXX/XXXXX.jpg
There's the latents.pt files too but that should've been just the weight.
As the folders inside it were /cdf/ and /ff/, it seems it was made from the Celeb-DF and FF++ datasets instead, so it made sense that there were no real data inside the folder.
So, there seems to be a small mistake in the category and it should be corrected to accordingly.
According to the main page, the e4e folder is part of the unknown methods, which should have been made with the CelebA dataset and have the real data included in the in the folder, but the folder structure is 3 folders with images:
e4e/cdf/Celeb-real/inversions/XXXXX.jpg
e4e/cdf/Youtube-real/inversions/XXXXX.jpg
e4e/ff/XXX/XXXXX.jpg
There's the latents.pt files too but that should've been just the weight.
As the folders inside it were /cdf/ and /ff/, it seems it was made from the Celeb-DF and FF++ datasets instead, so it made sense that there were no real data inside the folder.
So, there seems to be a small mistake in the category and it should be corrected to accordingly.