celoss.py 2.2 KB
Newer Older
1
# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve.
B
Bin Lu 已提交
2 3 4 5 6 7 8 9 10 11 12 13 14
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

G
gaotingquan 已提交
15 16
import warnings

B
Bin Lu 已提交
17
import paddle
18
import paddle.nn as nn
B
Bin Lu 已提交
19 20
import paddle.nn.functional as F

G
gaotingquan 已提交
21 22
from ppcls.utils import logger

B
Bin Lu 已提交
23

24
class CELoss(nn.Layer):
C
cuicheng01 已提交
25 26 27 28
    """
    Cross entropy loss
    """

29 30 31 32 33
    def __init__(self, epsilon=None):
        super().__init__()
        if epsilon is not None and (epsilon <= 0 or epsilon >= 1):
            epsilon = None
        self.epsilon = epsilon
B
Bin Lu 已提交
34

35
    def _labelsmoothing(self, target, class_num):
36
        if len(target.shape) == 1 or target.shape[-1] != class_num:
37
            one_hot_target = F.one_hot(target, class_num)
B
Bin Lu 已提交
38 39
        else:
            one_hot_target = target
40 41
        soft_target = F.label_smooth(one_hot_target, epsilon=self.epsilon)
        soft_target = paddle.reshape(soft_target, shape=[-1, class_num])
B
Bin Lu 已提交
42 43
        return soft_target

44 45 46 47 48 49 50 51
    def forward(self, x, label):
        if isinstance(x, dict):
            x = x["logits"]
        if self.epsilon is not None:
            class_num = x.shape[-1]
            label = self._labelsmoothing(label, class_num)
            x = -F.log_softmax(x, axis=-1)
            loss = paddle.sum(x * label, axis=-1)
D
dongshuilong 已提交
52
        else:
53 54 55 56 57 58
            if label.shape[-1] == x.shape[-1]:
                label = F.softmax(label, axis=-1)
                soft_label = True
            else:
                soft_label = False
            loss = F.cross_entropy(x, label=label, soft_label=soft_label)
59
        loss = loss.mean()
60
        return {"CELoss": loss}
C
cuicheng01 已提交
61 62


G
gaotingquan 已提交
63 64 65 66 67
class MixCELoss(object):
    def __init__(self, *args, **kwargs):
        msg = "\"MixCELos\" is deprecated, please use \"CELoss\" instead."
        logger.error(DeprecationWarning(msg))
        raise DeprecationWarning(msg)