Rate this Page
★★★★★

torch.quantile#

torch.quantile(input,q,dim=None,keepdim=False,*,interpolation='linear',out=None)→Tensor#

Computes the q-th quantiles of each row of theinput tensor along the dimensiondim.

To compute the quantile, we map q in [0, 1] to the range of indices [0, n] to find the locationof the quantile in the sorted input. If the quantile lies between two data pointsa<b withindicesi andj in the sorted order, result is computed according to the giveninterpolation method as follows:

  • linear:a+(b-a)*fraction, wherefraction is the fractional part of the computed quantile index.

  • lower:a.

  • higher:b.

  • nearest:a orb, whichever’s index is closer to the computed quantile index (rounding down for .5 fractions).

  • midpoint:(a+b)/2.

Ifq is a 1D tensor, the first dimension of the output represents the quantiles and has sizeequal to the size ofq, the remaining dimensions are what remains from the reduction.

Note

By defaultdim isNone resulting in theinput tensor being flattened before computation.

Parameters

  • input (Tensor) – the input tensor.

  • q (float orTensor) – a scalar or 1D tensor of values in the range [0, 1].

  • dim (int,optional) – the dimension to reduce.

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

Keyword Arguments

  • interpolation (str,optional) – interpolation method to use when the desired quantile lies between two data points.Can belinear,lower,higher,midpoint andnearest.Default islinear.

  • out (Tensor,optional) – the output tensor.

Example:

>>>a=torch.randn(2,3)>>>atensor([[ 0.0795, -1.2117,  0.9765],        [ 1.1707,  0.6706,  0.4884]])>>>q=torch.tensor([0.25,0.5,0.75])>>>torch.quantile(a,q,dim=1,keepdim=True)tensor([[[-0.5661],        [ 0.5795]],        [[ 0.0795],        [ 0.6706]],        [[ 0.5280],        [ 0.9206]]])>>>torch.quantile(a,q,dim=1,keepdim=True).shapetorch.Size([3, 2, 1])>>>a=torch.arange(4.)>>>atensor([0., 1., 2., 3.])>>>torch.quantile(a,0.6,interpolation='linear')tensor(1.8000)>>>torch.quantile(a,0.6,interpolation='lower')tensor(1.)>>>torch.quantile(a,0.6,interpolation='higher')tensor(2.)>>>torch.quantile(a,0.6,interpolation='midpoint')tensor(1.5000)>>>torch.quantile(a,0.6,interpolation='nearest')tensor(2.)>>>torch.quantile(a,0.4,interpolation='nearest')tensor(1.)