is there any way torch.mode can be applied over multiple dimensions
for example
import numpy as npimport torchx = np.random.randint(10, size=(3, 5))y = torch.tensor(x)lets say y has
[[6 3 7 3 0][2 5 7 9 7][6 1 4 6 3]]torch.mode should return a size 3 tensor [3,7,6]
without using a loop
Best Answer
Use the dimension attribute in torch to select which dimension should be reduced using mode operator.
torch.mode(y, dim = 1)[0]Will give you the desired answer.