dengdong

The workaround below works fine on my m1max macbook. In file `lib_layerdiffusion/models.py`, find ``` median = torch.median(result, dim=0).values ``` and modify like this ``` if self.load_device == torch.device("mps"): ''' In...

to reproduce the crash in a simple scenario, try the code below on your m series macbook: ``` import torch a = torch.randn(8, 1, 4, 512, 512) mps_device = torch.device("mps")...