l1_loss

paddle.nn.functional. l1_loss ( input, label, reduction='mean', name=None ) [source]

This operator computes the L1 Loss of Tensor input and label as follows.

If reduction set to 'none', the loss is:

\[Out = \lvert input - label \rvert\]

If reduction set to 'mean', the loss is:

\[Out = MEAN(\lvert input - label \rvert)\]

If reduction set to 'sum', the loss is:

\[Out = SUM(\lvert input - label \rvert)\]
Parameters

  • input (Tensor) – The input tensor. The shapes is [N, *], where N is batch size and * means any number of additional dimensions. It’s data type should be float32, float64, int32, int64.

  • label (Tensor) – label. The shapes is [N, *], same shape as input . It’s data type should be float32, float64, int32, int64.

  • reduction (str, optional) – Indicate the reduction to apply to the loss, the candicates are 'none' | 'mean' | 'sum'. If reduction is 'none', the unreduced loss is returned; If reduction is 'mean', the reduced mean loss is returned. If reduction is 'sum', the reduced sum loss is returned. Default is 'mean'.

  • name (str, optional) – Name for the operation (optional, default is None). For more information, please refer to Name.

Returns

Tensor, the L1 Loss of Tensor input and label.

If reduction is 'none', the shape of output loss is [N, *], the same as input . If reduction is 'mean' or 'sum', the shape of output loss is [1].

System Message: WARNING/2 (/usr/local/lib/python3.8/site-packages/paddle/nn/functional/loss.py:docstring of paddle.nn.functional.loss.l1_loss, line 36); backlink

Inline emphasis start-string without end-string.

Examples

import paddle

input = paddle.to_tensor([[1.5, 0.8], [0.2, 1.3]])
label = paddle.to_tensor([[1.7, 1], [0.4, 0.5]])

l1_loss = paddle.nn.functional.l1_loss(input, label)
print(l1_loss.numpy())
# [0.35]

l1_loss = paddle.nn.functional.l1_loss(input, label, reduction='none')
print(l1_loss.numpy())
# [[0.20000005 0.19999999]
# [0.2        0.79999995]]

l1_loss = paddle.nn.functional.l1_loss(input, label, reduction='sum')
print(l1_loss.numpy())
# [1.4]