• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1# Copyright 2021 Huawei Technologies Co., Ltd
2#
3# Licensed under the Apache License, Version 2.0 (the "License");
4# you may not use this file except in compliance with the License.
5# You may obtain a copy of the License at
6#
7# http://www.apache.org/licenses/LICENSE-2.0
8#
9# Unless required by applicable law or agreed to in writing, software
10# distributed under the License is distributed on an "AS IS" BASIS,
11# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12# See the License for the specific language governing permissions and
13# limitations under the License.
14# ============================================================================
15
16import numpy as np
17import pytest
18
19import mindspore.context as context
20import mindspore.nn as nn
21from mindspore import Tensor
22from mindspore.nn import Dense
23from mindspore.nn import TrainOneStepCell, WithLossCell
24from mindspore.nn.optim import SGD
25from mindspore.ops import operations as P
26
27context.set_context(mode=context.GRAPH_MODE, device_target="CPU")
28
29class NetSGD(nn.Cell):
30    def __init__(self):
31        super(NetSGD, self).__init__()
32        self.batch_size = 1
33        self.reshape = P.Reshape()
34        weight = Tensor(np.ones([10, 16]).astype(np.float32) * 0.01)
35        self.fc1 = Dense(16, 10, weight_init=weight)
36
37    def construct(self, input_x):
38        output = self.reshape(input_x, (self.batch_size, -1))
39        output = self.fc1(output)
40        return output
41
42
43@pytest.mark.level0
44@pytest.mark.platform_x86_cpu
45@pytest.mark.env_onecard
46def test_SGD():
47    epoch = 3
48    net = NetSGD()
49    learning_rate = 0.1
50    momentum = 0.9
51    dampening = 0.0
52    weight_decay = 0.0
53    nesterov = True
54    loss_scale = 1.0
55
56    optimizer = SGD(filter(lambda x: x.requires_grad, net.get_parameters()), learning_rate, momentum, dampening,
57                    weight_decay, nesterov, loss_scale)
58    criterion = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction='mean')
59    net_with_criterion = WithLossCell(net, criterion)
60    train_network = TrainOneStepCell(net_with_criterion, optimizer)  # optimizer
61    train_network.set_train()
62    losses = []
63    for _ in range(epoch):
64        data = Tensor(np.arange(0, 16).reshape(1, 1, 4, 4).astype(np.float32) * 0.01)
65        label = Tensor(np.array([0]).astype(np.int32))
66        loss = train_network(data, label)
67        losses.append(loss.asnumpy())
68
69    last_loss = 100.0
70    for loss in losses:
71        assert last_loss > loss
72        last_loss = loss
73