torch.logsumexp#

torch.logsumexp(input,dim,keepdim=False,*,out=None)#

Returns the log of summed exponentials of each row of theinputtensor in the given dimensiondim. The computation is numericallystabilized.

For summation index$j$j given bydim and other indices$i$i, the result is

$$\text{logsumexp}(x{)}_i=\log \sum _j\exp (x_{ij})$$logsumexp(x)i​=logj∑​exp(xij​)

Ifkeepdim isTrue, the output tensor is of the same sizeasinput except in the dimension(s)dim where it is of size 1.Otherwise,dim is squeezed (seetorch.squeeze()), resulting in theoutput tensor having 1 (orlen(dim)) fewer dimension(s).

Parameters

• input (Tensor) – the input tensor.

• dim (int ortuple ofints) – the dimension or dimensions to reduce.

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

Keyword Arguments

out (Tensor,optional) – the output tensor.

Example:

>>>a=torch.randn(3,3)>>>torch.logsumexp(a,1)tensor([1.4907, 1.0593, 1.5696])>>>torch.dist(torch.logsumexp(a,1),torch.log(torch.sum(torch.exp(a),1)))tensor(1.6859e-07)