torch.Size#

Created On: Apr 19, 2024 | Last Updated On: Jun 18, 2025

torch.Size is the result type of a call totorch.Tensor.size(). It describes the size of all dimensionsof the original tensor. As a subclass oftuple, it supports common sequence operations like indexing andlength.

Example:

>>>x=torch.ones(10,20,30)>>>s=x.size()>>>storch.Size([10,20,30])>>>s[1]20>>>len(s)3

classtorch.Size(iterable=(),/)#

count(value,/)#

Return number of occurrences of value.

index(value,start=0,stop=9223372036854775807,/)#

Return first index of value.

Raises ValueError if the value is not present.

numel()→int#

Returns the number of elements atorch.Tensor with the given size would contain.

More formally, for a tensorx=tensor.ones(10,10) with sizes=torch.Size([10,10]),x.numel()==x.size().numel()==s.numel()==100 holds true.

Example:

>>>x=torch.ones(10,10)>>>s=x.size()>>>storch.Size([10, 10])>>>s.numel()100>>>x.numel()==s.numel()True

Warning

This function does not return the number of dimensions described bytorch.Size, but instead the numberof elements atorch.Tensor with that size would contain.