and the other ones..

really need to unify this file better.

codes/models/audio/music/transformer_diffusion12.py

@@ -511,6 +511,7 @@ class TransformerDiffusionWithMultiPretrainedVqvae(nn.Module):
         # higher gradients. Ideally we would use parameter groups, but ZeroRedundancyOptimizer makes this trickier than
         # directly fiddling with the gradients.
         for p in scaled_grad_parameters:
+            if hasattr(p, 'grad') and p.grad is not None:
                 p.grad *= .2
 
 
@@ -566,6 +567,7 @@ class TransformerDiffusionWithCheaterLatent(nn.Module):
         # higher gradients. Ideally we would use parameter groups, but ZeroRedundancyOptimizer makes this trickier than
         # directly fiddling with the gradients.
         for p in scaled_grad_parameters:
+            if hasattr(p, 'grad') and p.grad is not None:
                 p.grad *= .2