Problem with torch.pow()

vision
2

I do not know if this helps, but i found a way to circumvent the issue though I still dont understand why the original issue came up.

alpha=torch.clamp(alpha,min=0.1,max=2)
with

m = torch.nn.Threshold(0.1,0.1)
alpha=m(alpha)

Now allows me to use x**alpha[0][0]. Think there is a bug with clamp.

1 Like