Tensor
Detatch

Tensor.detatch

Returns the same Tensor detached from the computation graph and with the required_grad flag set to False.

Usage

from tinygrad.tensor import Tensor
 
# create a tensor
tensor = Tensor([[1, 2, 3, 4, 5]])
 
# detatch tensor
print(tensor.detach())

Return value

<Tensor <LB (1, 5) dtypes.float op=LoadOps.FROM st=ShapeTracker(shape=(1, 5), views=[View(shape=(1, 5), strides=(0, 1), offset=0, mask=None, contiguous=True, shape_strides=((5, 1),))], var_vals={})> on GPU with grad None>
DeepwalkDevice
;