Daniele Mascali

Very nice code! I found a small bug that results in missing one cloth category (cape) and in confusing some cateegories a bit. In data/aligned_dataset.py at line 82 you add...

Based on the paper, the attention appears to be shifted by 2 in both the frequency and time dimensions. However, the code specifies shift_size=(2,0). Shouldn't it be shift_size=(2,2) instead?

Thank you for this great repo! I noticed in the decoder block that the third parameter is calculated by dividing the decoder embedding dimension by the number of heads in...