Difference of 0*a + b and b tensor

Hellow everyone.

Is there any difference between 0*a + b tensor and b tensor in autograd?

In detailed, if tensor y=0*x+b and y=b is same at the back-propagation stage?


The former one requires more computation as it will compute grad_x *0 but the result itself is the same

1 Like

Thank you for reply!