Take
take(input, index) -> Tensor
Returns a new tensor with the elements of input
at the given indices.
The input tensor is treated as if it were viewed as a 1-D tensor. The result
takes the same shape as the indices.
Examples
if (torch_is_installed()) {
src = torch_tensor(matrix(c(4,3,5,6,7,8), ncol = 3, byrow = TRUE))
torch_take(src, torch_tensor(c(1, 2, 5), dtype = torch_int64()))
}
#> torch_tensor
#> 4
#> 3
#> 7
#> [ CPUFloatType{3} ]