Input type (MPSFloatType) and weight type (torch.FloatTensor) should be the same

Hello, I am on M1 Mac, with MPS, and the solution is here in case somebody encounters the same issue (just a quick, dirty workaround) is go to site-packages/codeformer/app.py

replace line 71:
```
# REPLACE THIS LINE
# device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
device = torch.device('mps')
```

Start with site-packages/codeformer/app.py at line 71 and inspect how the device is selected. Reproduce the reported input/weight type mismatch on an M1 Mac with MPS, then verify that the device selection avoids the error. Done means the application runs without the reported type mismatch.

Written by the indexing model from the issue text.