L2 Loss.
tf.nn.l2_loss(
t: Annotated[Any, TV_L2Loss_T], name=None
) -> Annotated[Any, TV_L2Loss_T]
Computes half the L2 norm of a tensor without the sqrt:
output = sum(t ** 2) / 2
Args | |
|---|---|
t
|
A Tensor. Must be one of the following types: half, bfloat16, float32, float64.
Typically 2-D, but may have any dimensions.
|
name
|
A name for the operation (optional). |
Returns | |
|---|---|
A Tensor. Has the same type as t.
|