MATLAB to Pytorch-Geometric

I have a .mat file, with a struct. One of the columns in the struct is 43249 cells containing a different adjacency matrix of different sizes. The other struct columns are 43249 cells containing single digit int. My goal is to right a DL program that will train on the the ints and graphs, and predict graphs.

I am new to pytorch, so can someone point me in the right direction? For instance, I am able to add all the matrices to a dictionary or a list, have pytorch convert them all to tensors, and then i try to use loader on them, but i get the following:

raceback (most recent call last):
  File "<input>", line 1, in <module>
  File "/home/anthony/anaconda3/envs/GeoDL/lib/python3.8/site-packages/torch/utils/data/", line 435, in __next__
    data = self._next_data()
  File "/home/anthony/anaconda3/envs/GeoDL/lib/python3.8/site-packages/torch/utils/data/", line 475, in _next_data
    data = self._dataset_fetcher.fetch(index)  # may raise StopIteration
  File "/home/anthony/anaconda3/envs/GeoDL/lib/python3.8/site-packages/torch/utils/data/_utils/", line 47, in fetch
    return self.collate_fn(data)
  File "/home/anthony/anaconda3/envs/GeoDL/lib/python3.8/site-packages/torch_geometric/data/", line 34, in __call__
    return self.collate(batch)
  File "/home/anthony/anaconda3/envs/GeoDL/lib/python3.8/site-packages/torch_geometric/data/", line 17, in collate
    return default_collate(batch)
  File "/home/anthony/anaconda3/envs/GeoDL/lib/python3.8/site-packages/torch/utils/data/_utils/", line 55, in default_collate
    return torch.stack(batch, 0, out=out)
RuntimeError: stack expects each tensor to be equal size, but got [14, 14] at entry 0 and [16, 16] at entry 3

My other question is: is what i am asking possible? Thanks for the feedback.