How can I do clamp using a tensor as threshold?

Hi, i’m going to apply some kind of constraint to the output. However, the thresholds for each elements are different. I noticed that both torch.clamp and nn.HardTanh take scalars as argument. Are there any workaround for this?


torch.min may help

import torch as t
a = t.randn(4,3) # size : batch_size x element_num
b = t.randn(1,3) # constrain for each element
results = t.min(a,b)
1 Like