PaddleClas/ppcls/arch/backbone/model_zoo/distillation_models.py

66 lines
2.2 KiB
Python
Raw Normal View History

2020-08-28 17:43:27 +08:00
# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
2020-04-17 13:01:19 +08:00
#
2020-08-28 17:43:27 +08:00
# 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
2020-04-17 13:01:19 +08:00
#
# http://www.apache.org/licenses/LICENSE-2.0
#
2020-08-28 17:43:27 +08:00
# 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.
2020-04-17 13:01:19 +08:00
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import math
import paddle
2020-09-13 16:50:39 +08:00
import paddle.nn as nn
2020-04-17 13:01:19 +08:00
from .resnet_vd import ResNet50_vd
from .mobilenet_v3 import MobileNetV3_large_x1_0
from .resnext101_wsl import ResNeXt101_32x16d_wsl
__all__ = [
2020-04-20 11:54:09 +08:00
'ResNet50_vd_distill_MobileNetV3_large_x1_0',
2020-04-17 13:01:19 +08:00
'ResNeXt101_32x16d_wsl_distill_ResNet50_vd'
]
2020-09-13 16:50:39 +08:00
class ResNet50_vd_distill_MobileNetV3_large_x1_0(nn.Layer):
2021-01-19 13:54:27 +08:00
def __init__(self, class_dim=1000, freeze_teacher=True, **args):
2020-08-28 17:43:27 +08:00
super(ResNet50_vd_distill_MobileNetV3_large_x1_0, self).__init__()
2020-04-17 13:01:19 +08:00
2020-08-28 17:43:27 +08:00
self.teacher = ResNet50_vd(class_dim=class_dim, **args)
self.student = MobileNetV3_large_x1_0(class_dim=class_dim, **args)
2021-01-19 13:54:27 +08:00
if freeze_teacher:
for param in self.teacher.parameters():
param.trainable = False
2020-08-28 17:43:27 +08:00
2021-01-19 13:54:27 +08:00
def forward(self, x):
teacher_label = self.teacher(x)
student_label = self.student(x)
2020-08-28 17:43:27 +08:00
return teacher_label, student_label
2020-04-17 13:01:19 +08:00
2020-09-13 16:50:39 +08:00
class ResNeXt101_32x16d_wsl_distill_ResNet50_vd(nn.Layer):
2021-01-19 13:54:27 +08:00
def __init__(self, class_dim=1000, freeze_teacher=True, **args):
super(ResNeXt101_32x16d_wsl_distill_ResNet50_vd, self).__init__()
2020-08-29 11:49:27 +08:00
self.teacher = ResNeXt101_32x16d_wsl(class_dim=class_dim, **args)
self.student = ResNet50_vd(class_dim=class_dim, **args)
2021-01-19 13:54:27 +08:00
if freeze_teacher:
for param in self.teacher.parameters():
param.trainable = False
2020-04-17 13:01:19 +08:00
2021-01-19 13:54:27 +08:00
def forward(self, x):
teacher_label = self.teacher(x)
student_label = self.student(x)
return teacher_label, student_label