1# Copyright 2020 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# ============================================================================ 15import numpy as np 16import pytest 17 18import mindspore.context as context 19import mindspore.nn as nn 20from mindspore import Tensor 21from mindspore.nn import TrainOneStepCell, WithLossCell 22from mindspore.nn.optim import Momentum 23from mindspore.ops import operations as P 24 25context.set_context(mode=context.GRAPH_MODE, device_target="Ascend") 26 27 28class LeNet(nn.Cell): 29 def __init__(self): 30 super(LeNet, self).__init__() 31 self.relu = P.ReLU() 32 self.batch_size = 32 33 34 self.conv1 = nn.Conv2d(1, 6, kernel_size=5, stride=1, padding=0, has_bias=False, pad_mode='valid') 35 self.conv2 = nn.Conv2d(6, 16, kernel_size=5, stride=1, padding=0, has_bias=False, pad_mode='valid') 36 self.pool = nn.MaxPool2d(kernel_size=2, stride=2) 37 self.reshape = P.Reshape() 38 self.fc1 = nn.Dense(400, 120) 39 self.fc1.matmul.add_prim_attr("primitive_target", "CPU") 40 self.fc1.bias_add.add_prim_attr("primitive_target", "CPU") 41 self.fc2 = nn.Dense(120, 84) 42 self.fc2.matmul.add_prim_attr("primitive_target", "CPU") 43 self.fc2.bias_add.add_prim_attr("primitive_target", "CPU") 44 self.fc3 = nn.Dense(84, 10) 45 self.fc3.matmul.add_prim_attr("primitive_target", "CPU") 46 self.fc3.bias_add.add_prim_attr("primitive_target", "CPU") 47 48 def construct(self, input_x): 49 output = self.conv1(input_x) 50 output = self.relu(output) 51 output = self.pool(output) 52 output = self.conv2(output) 53 output = self.relu(output) 54 output = self.pool(output) 55 output = self.reshape(output, (self.batch_size, -1)) 56 output = self.fc1(output) 57 output = self.relu(output) 58 output = self.fc2(output) 59 output = self.relu(output) 60 output = self.fc3(output) 61 return output 62 63 64def train(net, data, label): 65 learning_rate = 0.01 66 momentum = 0.9 67 68 optimizer = Momentum(filter(lambda x: x.requires_grad, net.get_parameters()), learning_rate, momentum) 69 criterion = nn.SoftmaxCrossEntropyWithLogits(sparse=True) 70 net_with_criterion = WithLossCell(net, criterion) 71 train_network = TrainOneStepCell(net_with_criterion, optimizer) # optimizer 72 train_network.set_train() 73 res = train_network(data, label) 74 print("+++++++++Loss+++++++++++++") 75 print(res) 76 print("+++++++++++++++++++++++++++") 77 diff = res.asnumpy()[0] - 2.3025851 78 assert np.all(diff < 1.e-6) 79 80 81@pytest.mark.level1 82@pytest.mark.platform_arm_ascend_training 83@pytest.mark.platform_x86_ascend_training 84@pytest.mark.env_onecard 85def test_lenet(): 86 data = Tensor(np.ones([32, 1, 32, 32]).astype(np.float32) * 0.01) 87 label = Tensor(np.ones([32]).astype(np.int32)) 88 net = LeNet() 89 train(net, data, label) 90