renorm¶

paddle. renorm ( x, p, axis, max_norm ) [source]

renorm

This operator is used to calculate the p-norm along the axis, suppose the input-shape on axis dimension has the value of T, then the tensor is split into T parts, the p-norm should be calculated for each part, if the p-norm for part i is larger than max-norm, then each element in part i should be re-normalized at the same scale so that part-i’ p-norm equals max-norm exactly, otherwise part-i stays unchanged.

Parameters

• x (Tensor) – The input Tensor

• p (float) – The power of the norm operation.

• axis (int) – the dimension to slice the tensor.

• max-norm (float) – the maximal norm limit.

Returns

the renorm Tensor.

Return type

Tensor

Examples

import paddle
input = [[[2.0,2,-2],[3,0.3,3]],[[2,-8,2],[3.1,3.7,3]]]
x = paddle.to_tensor(input,dtype='float32')
y = paddle.renorm(x, 1.0, 2, 2.05)
print(y)

# [[[ 0.40594056, 0.29285714, -0.41000000], # [ 0.60891086, 0.04392857, 0.61500001]], # [[ 0.40594056, -1.17142856, 0.41000000], # [ 0.62920785, 0.54178572, 0.61500001]]])