Reference guides
latest (dev)
Copy
Latest development documentation · Updated 2026-10-08
Size
tensorplay.Size is the lightweight object that holds a tensor’s shape. It behaves like a
tuple of integers.
import tensorplay as tp
x = tp.zeros(2, 3, 4)
print(x.shape) # tensorplay.Size([2, 3, 4])
print(x.size()) # same
print(len(x.shape)) # 3
size() on a tensor accepts an optional dimension to return a single axis length:
print(x.size(0)) # 2
print(x.size(-1)) # 4
A Size is created automatically by the shape of a tensor, and you can build one directly:
s = tp.Size([3, 4])
print(s) # tensorplay.Size([3, 4])
print(tuple(s)) # (3, 4)
Because it is a tuple-like sequence, a Size can be unpacked or sliced, and is what reshape
expects as its target shape.
Sequence of tensor dimension sizes (tuple subclass). |
Where to go next
Tensor attributes — the rest of a tensor’s metadata.
Help improve this page
Found an error, an unclear step, or a missing example?
Was this page helpful?

