Normalize a vector to [0,1]

8

Broadcasting wasn’t available in version 0.1.12.
You could try:

tensor_vec = tensor_vec / tensor_vec.sum(0).expand_as(tensor_vec)
5 Likes