torch.median

torch.median(input) → Tensor

Returns the median of the values in input.

Note

The median is not unique for input tensors with an even number of elements. In this case the lower of the two medians is returned. To compute the mean of both medians, use torch.quantile() with q=0.5 instead.

Warning

This function spreads the (sub)gradient evenly over every element equal to the median, while median(dim=0) sends it to the element at the returned index.

Parameters:

input (Tensor) – the input tensor.

Example:

>>> a = torch.randn(1, 3)
>>> a
tensor([[ 1.5219, -1.5212,  0.2202]])
>>> torch.median(a)
tensor(0.2202)

torch.median(input, dim=-1, keepdim=False, *, out=None)

Returns a namedtuple (values, indices) where values contains the median of each row of input in the dimension dim, and indices contains the index of the median values found in the dimension dim.

By default, dim is the last dimension of the input tensor.

If keepdim is True, the output tensors are of the same size as input except in the dimension dim where they are of size 1. Otherwise, dim is squeezed (see torch.squeeze()), resulting in the outputs tensor having 1 fewer dimension than input.

Note

The median is not unique for input tensors with an even number of elements in the dimension dim. In this case the lower of the two medians is returned. To compute the mean of both medians in input, use torch.quantile() with q=0.5 instead.

Warning

indices does not necessarily contain the first occurrence of each median value found, unless it is unique. The exact implementation details are device-specific: CUDA returns the first occurrence, while other devices may return another one. Do not expect the same indices, or the same gradients, on CPU and GPU.

Parameters:

• input (Tensor) – the input tensor.

• dim (int, optional) – the dimension to reduce. If None, all dimensions are reduced.

• keepdim (bool, optional) – whether the output tensor has dim retained or not. Default: False.

Keyword Arguments:

out ((Tensor, Tensor), optional) – The first tensor will be populated with the median values and the second tensor, which must have dtype long, with their indices in the dimension dim of input.

Example:

>>> a = torch.randn(4, 5)
>>> a
tensor([[ 0.2505, -0.3982, -0.9948,  0.3518, -1.3131],
        [ 0.3180, -0.6993,  1.0436,  0.0438,  0.2270],
        [-0.2751,  0.7303,  0.2192,  0.3321,  0.2488],
        [ 1.0778, -1.9510,  0.7048,  0.4742, -0.7125]])
>>> torch.median(a, 1)
torch.return_types.median(values=tensor([-0.3982,  0.2270,  0.2488,  0.4742]), indices=tensor([1, 4, 4, 3]))