Issue of generating coarse aligned results on UDIS testing data

[tlliao]

Hi there! Thanks for the great work!

Trying to run on UDIS testing data to generate coarse aligned results, but failing with the following error, what I'm doing wrong?

Namespace(device='0', exist_ok=False, half=False, img_size=512, name='exp', project='runs/infer', rmse=False, source='data/udis.yaml', task='test', visualize=False, weights=['weights/align/udis_align_variant.pt'])
CUDA:0 (NVIDIA GeForce RTX 3090, 24259.6875MB)

Fusing layers... 
No module named 'thop'
Model Summary: 100 layers, 10181056 parameters, 0 gradients
0it [00:00, ?it/s]
Traceback (most recent call last):
  File "inference_align.py", line 176, in <module>
    infer()
  File "/home/x/anaconda3/envs/udis2/lib/python3.8/site-packages/torch/autograd/grad_mode.py", line 26, in decorate_context
    return func(*args, **kwargs)
  File "inference_align.py", line 100, in infer
    vertices_offsets, warped_imgs, warped_ones = model(imgs)
  File "/home/x/anaconda3/envs/udis2/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/mnt/cv_liao_DL/UnsupDIS-pytorch/models/yolo.py", line 329, in forward
    return self.model[-1](feature1, feature2, x2, m2)
  File "/home/x/anaconda3/envs/udis2/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/mnt/cv_liao_DL/UnsupDIS-pytorch/models/yolo.py", line 141, in forward
    off = self.m[i](x).unsqueeze(-1)  # [bs, 8, 1], for matrix multiplication
  File "/home/x/anaconda3/envs/udis2/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/home/x/anaconda3/envs/udis2/lib/python3.8/site-packages/torch/nn/modules/container.py", line 117, in forward
    input = module(input)
  File "/home/x/anaconda3/envs/udis2/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/home/x/anaconda3/envs/udis2/lib/python3.8/site-packages/torch/nn/modules/linear.py", line 93, in forward
    return F.linear(input, self.weight, self.bias)
  File "/home/x/anaconda3/envs/udis2/lib/python3.8/site-packages/torch/nn/functional.py", line 1690, in linear
    ret = torch.addmm(bias, input, weight.t())
**RuntimeError: mat1 dim 1 must match mat2 dim 0**```