Runtime Error: quantize.py #5
Description
Nice work! When I want to train this model on my data (Obama video), I receive this runtime error:
The size of tensor a (196) must match the size of tensor b (256) at non-singleton dimension 1
This comes from the chamfer_compare function in VPGC_Pytorch_2/Grid_Codebook_Modeling/taming/modules/vqvae/quantize.py from this line : dist = torch.sum((z - c) ** 2, -1)
Can someone please help to solve this?
I have not modified this code and the folders are organized and pre-processed as the authors suggested.