giagrad.Tensor.empty#

classmethod Tensor.empty(*shape, dtype=<class 'numpy.float32'>, **kwargs) → Tensor[source]#

Creates a tensor filled with uninitialized data.

Datatype is np.float32 by default.

Parameters:

  • shape (int, ...) – A variable number of integers defining the shape of the output tensor.

  • **kwargs – Parameters passed to the Tensor class initializer.

Examples

>>> Tensor.empty(2, 3, requires_grad=True, dtype=np.float64)
tensor: [[4.67662529e-310 0.00000000e+000 4.67596337e-310]
         [6.94592882e-310 6.94611561e-310 6.94609055e-310]]