Fix torch multi-GPU --device error (#1701)

utils/torch_utils.py

@@ -75,13 +75,14 @@ def time_synchronized():
     return time.time()
 
 
-def profile(x, ops, n=100, device=torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')):
+def profile(x, ops, n=100, device=None):
     # profile a pytorch module or list of modules. Example usage:
     #     x = torch.randn(16, 3, 640, 640)  # input
     #     m1 = lambda x: x * torch.sigmoid(x)
     #     m2 = nn.SiLU()
     #     profile(x, [m1, m2], n=100)  # profile speed over 100 iterations
     
+    device = device or torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')    
     x = x.to(device)
     x.requires_grad = True
     print(torch.__version__, device.type, torch.cuda.get_device_properties(0) if device.type == 'cuda' else '')